The Rnf Complex Is an Energy-Coupled Transhydrogenase Essential To Reversibly Link Cellular NADH and Ferredoxin Pools in the Acetogen Acetobacterium woodii

Ferredoxin and NAD⁺ are key electron carriers in anaerobic bacteria, but energetically, they are not equivalent, since the redox potential of ferredoxin is lower than that of the NADH/NAD⁺ couple. We describe by mutant studies in Acetobacterium woodii that the main function of Rnf is to energetically link cellular pools of ferredoxin and NAD⁺. When ferredoxin is greater than NADH, exergonic electron flow from ferredoxin to NAD⁺ generates a chemiosmotic potential. This is essential for energy conservation during autotrophic growth. When NADH is greater than ferredoxin, Rnf works in reverse. This reaction is essential for growth on low-energy substrates to provide reduced ferredoxin, indispensable for biosynthesis and CO₂ reduction. Our studies put a new perspective on the cellular function of the membrane-bound ion-translocating Rnf complex widespread in bacteria.

ABSTRACT

The Rnf complex is a respiratory enzyme that catalyzes the oxidation of reduced ferredoxin to the reduction of NAD⁺, and the negative free energy change of this reaction is used to generate a transmembrane ion gradient. In one class of anaerobic acetogenic bacteria, the Rnf complex is believed to be essential for energy conservation and autotrophic growth. We describe here a methodology for markerless mutagenesis in the model bacterium of this class, Acetobacterium woodii, which enabled us to delete the rnf genes and to test their in vivo role. The rnf mutant did not grow on H₂ plus CO₂, nor did it produce acetate or ATP from H₂ plus CO₂, and ferredoxin:NAD⁺ oxidoreductase activity and Na⁺ translocation were also completely lost, supporting the hypothesis that the Rnf complex is the only respiratory enzyme in this metabolism. Unexpectedly, the mutant also did not grow on low-energy substrates, such as ethanol or lactate. Oxidation of these substrates is not coupled to the reduction of ferredoxin but only of NAD⁺, and we speculated that the growth phenotype is caused by a loss of reduced ferredoxin, indispensable for biosynthesis and CO₂ reduction. The electron-bifurcating hydrogenase of A. woodii reduces ferredoxin, and indeed, the addition of H₂ to the cultures restored growth on ethanol and lactate. This is consistent with the hypothesis that endergonic reduction of ferredoxin with NADH is driven by reverse electron transport catalyzed by the Rnf complex, which renders the Rnf complex essential also for growth on low-energy substrates.

IMPORTANCE Ferredoxin and NAD⁺ are key electron carriers in anaerobic bacteria, but energetically, they are not equivalent, since the redox potential of ferredoxin is lower than that of the NADH/NAD⁺ couple. We describe by mutant studies in Acetobacterium woodii that the main function of Rnf is to energetically link cellular pools of ferredoxin and NAD⁺. When ferredoxin is greater than NADH, exergonic electron flow from ferredoxin to NAD⁺ generates a chemiosmotic potential. This is essential for energy conservation during autotrophic growth. When NADH is greater than ferredoxin, Rnf works in reverse. This reaction is essential for growth on low-energy substrates to provide reduced ferredoxin, indispensable for biosynthesis and CO₂ reduction. Our studies put a new perspective on the cellular function of the membrane-bound ion-translocating Rnf complex widespread in bacteria.

INTRODUCTION

Rhodobacter nitrogen fixation (rnf) genes were first described in the 1990s in Rhodobacter capsulatus. Deletion of the rnf genes led to a mutant that was no longer able to fix nitrogen (1). The similarity of the subunits/enzyme predicted from the genes to the ion-translocating NADH:quinone oxidoreductase (Nqr), a respiratory enzyme found in many bacteria (2–5), led to the hypothesis that the Rnf complex drives endergonic reverse electron transfer from NADH to a low-potential electron acceptor required for nitrogen fixation, at the expense of the electrochemical proton gradient across the membrane. However, experimental evidence supporting this hypothesis is still lacking. Since their discovery in the 1990s, Rnf complexes and rnf genes entered a long “sleep” and came back to life with the discovery of rnf genes in the anaerobe Clostridium kluyveri in 2008. It was postulated that the Rnf complex in C. kluyveri catalyzes the reverse reaction, the oxidation of reduced ferredoxin with the reduction of NAD⁺ (6). It was speculated that this exergonic electron transfer reaction provides the energy for the “uphill” transport of ions across the membrane, thus generating a transmembrane electrochemical ion gradient that is hypothesized to synthesize ATP by a membrane-bound ATP synthase (7). Again, experimental evidence for this assumption is still lacking, but nevertheless, the presence of rnf genes in many anaerobes has since then be taken as an indication that this assumed novel type of respiratory chain is widespread in anaerobes (8).

A long-standing search for the answer to the question of how strictly anaerobic acetogenic bacteria conserve energy led us to the rnf genes and the Rnf complex in Acetobacterium woodii (9). Acetogenic bacteria are an ecologically important group of strict anaerobes that can grow on a number of different organic substrates (10–14). The oxidation of these substrates is, in most cases, coupled to the reduction of carbon dioxide to acetate via the Wood-Ljungdahl pathway (WLP). Acetogenic bacteria can also grow autotrophically by converting hydrogen plus carbon dioxide to acetate in the WLP (15). The WLP is the only pathway known for carbon dioxide fixation that is coupled to the net synthesis of ATP (16). How energy is conserved for growth has been an enigma for decades, but recently, we have discovered that ATP is synthesized by electron transport phosphorylation (17). Membranes of A. woodii catalyzed electron transfer from reduced ferredoxin to NAD⁺, and this activity was coupled to primary and electrogenic Na⁺ transport across the cytoplasmic membrane (17, 18), thus establishing a Na⁺ motive force. Vice versa, a transmembrane Na⁺ gradient established by ATP hydrolysis was able to drive the reverse reaction, endergonic ferredoxin reduction with NADH as a reductant (17). The activity was solubilized from membranes of A. woodii and partly purified, and peptide mass fingerprinting of polypeptides present in the preparation revealed that they are encoded by the rnf genes (9). Since then, we and others have tried to purify Rnf complexes from acetogens and other bacteria to prove their function as respiratory enzymes, but so far, this has been without success due to the inherent problems with stability of the detergent-solubilized complex.

To circumvent the problems inherent to the biochemical approach, a genetic approach was used in addition. Inside-out membrane vesicles of the rnf-containing archaeon Methanosarcina acetivorans catalyzed Na⁺ transport coupled to ferredoxin:heterodisulfide oxidoreductase activity. Ionophore studies revealed that Na⁺ transport was primary and electrogenic. Deletion of the rnf genes led to a mutant unable to grow on acetate and that lost the ferredoxin:heterodisulfide oxidoreductase activity and Na⁺ transport (19). These data are consistent with the hypothesis that the Rnf complex of M. acetivorans is Na⁺ translocating and is the entry point of electrons derived from reduced ferredoxin into the electron transport chain leading to the acceptor, the heterodisulfide. The anaerobic bacterium Bacteroides fragilis also has rnf genes and has an Na⁺-dependent ferredoxin:NAD⁺ oxidoreductase activity at the cytoplasmic membrane (8). Deletion of the rnf genes led to a loss of membrane-bound ferredoxin:NAD⁺ oxidoreductase activity. The acetogen Clostridium ljungdahlii also has rnf genes, and it was suggested that its mode of energy conservation is different from that of A. woodii by using protons instead of Na⁺. Interruption of the rnfA and rnfB genes was achieved by a single homologous crossover yielding an apparently instable integrant. However, the integrant lost its ability to grow autotrophically (20). Moreover, the deletion of the rnf genes in Clostridium thermocellum led to a decrease in ethanol production, showing that this complex is needed for the supply of NADH for ethanol production (21).

The currently best understood acetogen is A. woodii, but the physiological role of the Rnf complex remained to be established. To this end, we have deleted the rnf genes from the chromosome of A. woodii and describe here the phenotype of the mutant. These analyses clearly demonstrate that the Rnf complex is essential for autotrophic growth. We also demonstrate a novel function of the Rnf complex. An rnf deletion mutant cannot grow heterotrophically on low-energy substrates, but growth can be restored by the addition of molecular hydrogen. This is consistent with the hypothesis that the Rnf complex provides reduced ferredoxin at the expense of NADH, generated from low-energy substrates by reverse electron transport. In sum, the Rnf complex is a membrane-bound enzyme complex tightly coupled to the membrane potential and essential for the connection of cellular ferredoxin and pyridine nucleotide pools.

RESULTS

Deletion and verification of the rnf operon of A. woodii.

The rnf genes Awo_c22060 (rnfC), Awo_c22050 (rnfD), Awo_c22040 (rnfG), Awo_c22030 (rnfE), Awo_c22020 (rnfA), and Awo_c22010 (rnfB) were deleted from the chromosome of A. woodii by double homologous recombination using a suicide plasmid (pMTL-JPB23). This plasmid contained homology arms identical to the upstream and downstream regions of the rnf operon leaving 6 bp behind the start codon of rnfC and 3 bp in front of the stop codon of rnfB. After homologous recombination, as described in Materials and Methods, the strain was tested for plasmid loss through thiamphenicol sensitivity, and the 5,119-bp deletion was confirmed by a flanking PCR using primers AW_rnf_FlankF and AW_rnf_FlankR (Fig. 1; see also Table S2 in the supplemental material), followed by Sanger sequencing of the entire region.

FIG 1.

Deletion of the rnf operon in A. woodii. Agarose gel showing PCR amplifications of the rnf operon deletion with primers AW_rnf_FlankF and AW_rnf_FlankR annealing to regions outside the recombination regions used for deletion of the operon. Lane M shows the reference marker (GeneRuler 1-kb DNA ladder; Thermo Fisher Scientific, Waltham, MA). Lane ΔrnfCDGEAB shows the resulting PCR fragment of the rnf deletion mutant, and lane WT shows the fragment amplified with genomic DNA of the wild type.

It was expected that the Rnf complex is not essential for growth on fructose, and indeed, the rnf mutant grew on fructose. To test whether deletion of the rnf genes led to a loss of ferredoxin:NAD⁺ oxidoreductase (Fno) activity, cells were grown on fructose and membranes were prepared; indeed, we found that Fno activity could not be determined (Fig. 2B). In addition, inverted membrane vesicles lost the Fno-coupled Na⁺ translocation activity (Fig. 2C). These data demonstrate that the rnf genes encode proteins catalyzing the Fno activity and the coupled Na⁺ translocation. To address the physiological role of the Rnf complex, the growth of A. woodii on different substrates was examined.

FIG 2.

Fd_red-dependent NAD⁺ reduction and Na⁺ transport activity. (A and B) Fno activities of the wild type (A) and rnf deletion mutant (B) were measured in anoxic cuvettes filled with 950 μl of 20 mM Tris-HCl buffer containing, 20 mM NaCl, 3 mM DTE, and 4 μM resazurin (pH 7.7) at a pressure of 0.5 × 10⁵ Pa CO. Ten microliters of ferredoxin (3 mM), 5 μl carbon monoxide dehydrogenase (CODH; 30 mg/ml), and washed membranes (200 μg) were added. The reaction was started with addition of 30 μl NAD⁺ (100 mM), and reduction of NAD⁺ at 340 nm (red line) and oxidation of Fd_red at 430 nm (black line) were documented. (C) For analysis of Na⁺ translocation, inverted membrane vesicles (IMV; protein concentration, 3 mg/ml) were incubated in 20 mM Tris-HCl buffer containing 2 mM DTE and 2 μM resazurin (pH 7.7) at a pressure of 0.5 × 10⁵ Pa CO, and 10 μl ferredoxin (3 mM) and 0.5 μCi/ml ²²Na⁺ were added. After incubation for 30 min, Fd was reduced by the addition of 5 μl CODH (30 mg/ml), and the reaction was started by the addition of NAD⁺. Transport rates were calculated from the initial slopes and plotted against the Na⁺ concentration. ■, A. woodii wild type with addition of NAD⁺; ▲, A. woodii wild type without addition of NAD⁺; ▼, A. woodii Δrnf mutant after addition of NAD⁺; ◆, A. woodii Δrnf mutant without addition of NAD⁺. This figure represents one of two obtained data sets showing comparable results.

The Rnf complex is essential for autotrophic growth and ATP synthesis in A. woodii.

In contrast to the wild type, the rnf mutant did not grow on H₂ plus CO₂. Next, we tested whether nongrowing resting cells of the rnf mutant are still able to convert H₂ plus CO₂ to acetate. Therefore, cells were grown on fructose, harvested by centrifugation, and resuspended in buffer. As can be seen in Fig. 3A, acetogenesis from H₂ plus CO₂ was largely reduced but still possible with only 20% activity. This gave us the opportunity to determine ATP synthesis coupled to acetogenesis from H₂ plus CO₂ in the rnf mutant. Upon addition of H₂ plus CO₂, resting cells of the wild type not only started to form acetate, but ATP was synthesized shortly after the substrate pulse. In contrast, the rnf mutant did not synthesize ATP (Fig. 3B). This provides clear evidence that the Rnf complex is the only respiratory enzyme active during autotrophic acetogenesis.

FIG 3.

Acetogenesis and ATP synthesis from H₂ plus CO₂ by resting cells of A. woodii. Resting cells (protein concentration, 0.8 mg/ml) in 40 mM imidazole buffer containing 20 mM KCl, 20 mM NaCl, 20 mM MgSO₄, 2 mM DTE, and 4 μM resazurin (pH 7.0) were incubated at 30°C for 10 min. Production of acetate (A) and ATP (B) was monitored over time after exchanging the headspace to H₂ plus CO₂ at a pressure of 1.0 × 10⁵ Pa. ▲, A. woodii wild type; ■, A. woodii Δrnf mutant; ▼, A. woodii wild type without addition of H₂ plus CO₂; ◆, H₂ plus CO₂ without addition of resting cells.

The Rnf complex is not essential for heterotrophic growth on fructose in A. woodii.

Fructose is converted by A. woodii via glycolysis, pyruvate ferredoxin oxidoreductase (PFOR), and phosphotransacetylase and acetate kinase to acetate, yielding 2 mol acetate, 2 mol CO₂, 2 mol NADH, 2 mol reduced ferredoxin, and 2 mol ATP per mol sugar. The 4 mol reducing equivalents is then shuttled to the electron acceptor CO₂, which is reduced in the WLP to acetate (22). As the WLP is Rnf dependent, one would expect only 2 mol acetate per fructose in the case of the rnf mutant. Wild-type cells growing on fructose in complex medium containing yeast extract produced acetate as the only product; formate and molecular hydrogen were not detected. Concomitant with acetate production, the pH fell from 7.3 to 6.34. The growth rate was 0.214/h, and the final optical density (OD) was 2.2 (Fig. 4A). The fructose-to-acetate ratio was 1:2.5. The rnf mutant showed biphasic growth kinetics. In the first very short phase, the OD increased to 0.25 to 0.3, with a doubling time of 9.1 h. In the second phase (25 h), the doubling time was 7 h, and the final OD was 2.2. In the first phase, fructose was apparently not consumed and acetate was not produced. Overall, 1 mol fructose was converted to 2.05 mol acetate, demonstrating that the rnf deletion resulted in a reduction of acetate production, as expected. A pyrE mutant did not have the biphasic growth phenotype, indicating that this phenotype was caused by the rnf deletion. The biphasic growth phenotype was not observed in minimal medium, but again, the rnf mutant produced only 80% of the acetate of the wild type (wild type, 1:2.55; rnf mutant, 1:2.06) (Fig. 5). These data are consistent with the notion that the WLP is not operative in the rnf mutant. However, this raised the question about the fate of the reducing equivalents (2 NADH and 2 reduced ferredoxin) in the rnf mutant. Formate, alanine, ethanol, lactate, and mannitol were not found, neither in complex nor in minimal medium. No unaccounted peaks in the gas chromatography or high-performance liquid chromatography (HPLC) were detected either. Hydrogen was also not detected.

FIG 4.

Growth of A. woodii wild type and the rnf mutant in complex medium. The wild type (A) and rnf mutant (B) were grown on fructose (40 mM) in 120-ml anoxic serum bottles. Fructose concentration (▲) was measured enzymatically, acetate concentration (▼) was measured by gas chromatography, the pH (♢) was measured by a pH electrode, and the OD (■) was measured photometrically at 600 nm.

FIG 5.

Growth of A. woodii wild type and the rnf deletion mutant in minimal medium. The wild type (A) and rnf mutant (B) were grown on fructose (40 mM) in 120-ml anoxic serum bottles. Fructose (▲) and acetate concentrations (▼) were measured enzymatically, and the OD at 600 nm (■) was measured photometrically.

Growth on low-energy substrates is impaired in the rnf mutant but restored by the addition of molecular hydrogen.

A. woodii grows on certain low-energy substrates whose oxidation can only be coupled for thermodynamic reasons to the reduction of NAD⁺ but not to ferredoxin. However, both electron donors are required for operation of the WLP (14). Ethanol is oxidized by A. woodii to acetate, yielding 2 mol NADH (23). According to the current metabolic model for ethanol conversion in A. woodii, the Rnf complex is essential for the reduction of ferredoxin with NADH as a reductant, since there is no other way to reduce ferredoxin at the expense of NADH. Indeed, the rnf mutant did not grow on ethanol. If our hypothesis is correct, the addition of molecular hydrogen should restore growth, since the electron-bifurcating hydrogenase can reduce ferredoxin and thus should relieve growth inhibition. This was indeed observed, as after the addition of hydrogen to the culture, growth was possible and ethanol was converted to acetate (Fig. 6A). However, growth was biphasic again, with ethanol being consumed in both phases, but acetate started to appear only at the end of the first phase.

FIG 6.

Growth of the A. woodii rnf deletion mutant on ethanol plus H₂ (A) and lactate plus H₂ (B) in complex medium. Cells were grown on 40 mM ethanol (A) or 80 mM lactate (B). H₂ was used in each culture at a pressure of 1.0 × 10⁵ Pa. Cell growth at 600 nm (■), pH (◆), acetate (▲), and ethanol (●) were followed over time.

Lactate is a low-energy substrate whose oxidation yields reduced ferredoxin (in the pyruvate ferredoxin oxidoreductase reaction) but also NADH (in the lactate dehydrogenase reaction). Furthermore, lactate oxidation with NAD⁺ as an electron acceptor is endergonic and driven by electron confurcation with reduced ferredoxin as a coreductant (24). Therefore, one would expect the Rnf complex to be essential for this metabolic scenario, as it is the only enzyme available to reduce ferredoxin at the expense of NADH. Indeed, growth of the rnf mutant on lactate was completely impaired. Growth on lactate could be restored by the addition of H₂. Also in contrast to ethanol, the growth rate was only 54% of that of the wild type under the same conditions, and the final OD was only 0.45, compared to 1.64 for the wild type (Fig. 6B).

DISCUSSION

Analyses of the amino acid sequences derived from DNA sequences predict that the Rnf complex is membrane integral and contains, in its simplest form, six subunits. It has FeS clusters and flavins as electron carriers (8, 25). The Rnf complex catalyzes exergonic electron transfer from reduced ferredoxin (E₀′ = −500 to −450 mV) to NAD⁺ (E₀′ = −320 mV), which is coupled to electrogenic translocation of ions across the membrane (18). Some Rnf complexes use Na⁺ as a coupling ion, and others use protons, and this correlates with the ion specificity of the ATP synthase available (25). The potential difference from reduced ferredoxin to NAD⁺ allows for the translocation of around one ion per electron transferred (26). The topology and function of Rnf subunits are far from being fully understood, but RnfC is believed to bind NADH and RnfB ferredoxin (25, 27). The path of the electron and the ion is completely obscure.

The physiological role of the Rnf complex has been addressed in Rhodobacter capsulatus (impaired in nitrogen fixation) (1) and in one methanogen, M. acetivorans, which showed an impaired growth phenotype on acetate, together with a loss of ferredoxin:heterodisulfide oxidoreductase and Na⁺ translocation activity after deletion of the rnf genes (19). So far, the physiological role has only been studied in one acetogen, Clostridium ljungdahlii (20). Interruption of the rnf genes in C. ljungdahlii led to a loss of nitrogenase activity, as well as growth on H₂ plus CO₂ (20). Loss of growth on H₂ plus CO₂ was also observed in this study for the model acetogen A. woodii.

Acetogenesis from H₂ plus CO₂ according to

$$4H_2+2C{O}_2\to acetate+2H_{2}O\mathrm{\Delta }{{\mathit{G}}_0}^{\prime }=-95.0kJ/mol$$ (1)

as carried out by acetogenic bacteria is clearly at the thermodynamic limit of life. Considering the hydrogen partial pressures that have been determined in anaerobic ecosystems (28), the free energy change of this reaction is around −20 kJ/mol, just enough to synthesize about one-third of an ATP when considering a physiological phosphorylation potential of around 60 kJ/mol (26). The CO₂ reduction pathway in A. woodii (WLP) requires 4 reducing equivalents, with 2 in the form of NADH, 1 in the form of reduced ferredoxin, and 1 in the form of hydrogen. NADH is the reductant for the methylene-tetrahydrofolate dehydrogenase and methylene-tetrahydrofolate reductase, reduced ferredoxin for the CO dehydrogenase, and H₂ for the hydrogen-dependent CO₂ reductase (14). How is this stoichiometry achieved? The electron-bifurcating hydrogenase oxidizes 3 mol H₂, yielding 1.5 mol NADH and 1.5 mol reduced ferredoxin (29). One mole of reduced ferredoxin is used for the CO dehydrogenase reaction, which leaves 0.5 mol reduced ferredoxin and 1.5 mol NADH. The electrons are balanced out in the Rnf complex that oxidizes 0.5 mol reduced ferredoxin, coupled to the reduction of 0.5 mol NAD⁺. Electron transfer is coupled to the membrane potential across the cytoplasmic membrane and drives the export of one Na⁺ per electron (14). The ATP synthase has a stoichiometry of 3.3 Na⁺/ATP (30, 31), and thus, 0.3 ATP can be made according to the equation

$$4H_2+2C{O}_2+0.3ADP+0.3P_i\to acetate+2H_{2}O+0.3ATP$$ (2)

As demonstrated in this study, deletion of the rnf genes inhibited autotrophic growth and led to a loss of Fno activity and Na⁺ transport, demonstrating that the Rnf complex is essential for energy conservation during autotrophic acetogenesis.

Acetogenic bacteria are metabolically very versatile and can also grow by oxidation of sugars, alcohols, or carbonic acids (10–13). The electrons gained from oxidative reactions are channeled into the WLP, which can be regarded as an electron sink. However, as outlined above, the WLP in A. woodii requires hydrogen, NADH, and reduced ferredoxin in a defined stoichiometry of 1:2:1. Growth on the substrate ethanol involves in the first step an alcohol dehydrogenase and an aldehyde dehydrogenase, which catalyze the oxidation of ethanol to acetyl coenzyme A (acetyl-CoA) via acetaldehyde (23). Both dehydrogenases are coupled to the reduction of NAD⁺, and thus, the question arises of how the stoichiometry of 1:2:1 of hydrogen, NADH, and reduced ferredoxin is established based on 4 mol NADH produced from 2 mol ethanol. Two moles NADH goes directly into the WLP, and the other 2 mol NADH is converted to 1 mol reduced ferredoxin and 1 mol H₂ by the concerted action of three enzymes.

The Rnf complex catalyzes

$$1.5NADH+3{{Na}^{+}}_{out}+1.5{Fd}_{ox}\to 1.5NA{D}^{+}+3N{a^{+}}_{in}+1.5{Fd}_{red}$$ (3)

where Na⁺_out is Na⁺ out of the cell, Fd_ox is oxidized ferredoxin, and Na⁺_in is Na⁺ into the cell.

The sodium ion potential is established by ATP hydrolysis according to

$$0.9ATP+3{{Na}^{+}}_{in}\to 0.9ADP+3N{a^{+}}_{out}$$ (4)

The electron-bifurcating hydrogenase catalyzes

$$0.5NADH+0.5{Fd}_{red}\to 1H_2$$ (5)

Altogether, this allows for acetogenesis according to

$$2ethanol+2C{O}_2+1.1ADP\to 3acetate+1.1ATP$$ (6)

The fact that the rnf mutant did not grow on ethanol is in accordance with the model suggesting that a lack of reduced ferredoxin blocks the WLP. This blockage should be overcome by the addition of molecular hydrogen to the culture since the electron-bifurcating hydrogenase reduces ferredoxin with H₂ as a reductant. This was indeed observed. Taken together, these results give strong evidence that a second cellular function of the Rnf complex allows for growth on low-energy substrates which are only coupled to the reduction of NAD⁺. The same is true for lactate. The oxidation of lactate only yields reduced ferredoxin in the PFOR reaction, which is reoxidized by the lactate dehydrogenase. Additional reduced ferredoxin is required for the WLP. Our data are consistent with the hypothesis that additional ferredoxin is reduced by the Rnf complex at the expense of NADH, which has been previously suggested (24).

Fructose is oxidized by A. woodii to 2 mol acetyl-CoA and 2 mol CO₂, reducing 2 mol NAD⁺ and ferredoxin (Fig. 7). The electrons are then directed into the WLP, in which CO₂ serves as an electron sink. Two moles NADH and 1 mol reduced ferredoxin are used directly as reductant in the WLP, whereas 1 mol reduced ferredoxin has to be converted to molecular hydrogen, the reductant for the hydrogen-dependent CO₂ reductase. This is achieved by the concerted action of the two redox balancing systems, the soluble electron-bifurcating hydrogenase and the energy-coupled membrane-bound Rnf complex, as follows:

$$0.5{Fd}_{red}+1N{a^{+}}_{out}+0.5NA{D}^{+}\to 0.5{Fd}_{ox}+1N{a^{+}}_{in}+0.5NADH$$ (7)

$$0.5NADH+0.5{Fd}_{red}\to H_2+0.5NA{D}^{+}+0.5{Fd}_{ox}$$ (8)

FIG 7.

Redox balancing and energy conservation of the fructose metabolism in A. woodii. By metabolizing fructose to 2 mol CO₂ and 2 mol acetate, 4 mol ATP is generated. The Rnf complex creates an Na⁺ gradient by oxidizing 0.5 mol Fd_red used by the ATP synthase for the synthesis of 0.3 mol ATP. Additional reducing equivalents are disposed of in the WLP. THF, tetrahydrofolate; Co-FeS-P, corrinoid iron-sulfur protein.

Sum:

$$1{Fd}_{red}+1{N{a}^{+}}_{out}\to H_2+1{{Na}^{+}}_{in}$$ (9)

Thus, in the rnf mutant, the WLP is inactive due to a lack of the first reductant of the pathway. However, cells still did grow and produced only two-thirds of the acetate of the wild type, which is plausible since the WLP accounts for one-third of the total acetate (22). However, the question is the fate of the 2 mol NADH and reduced ferredoxin. Although A. woodii has a lactate dehydrogenase (24), lactate was not found as a product in the mutant. Another obvious product, hydrogen, could also not be detected. Neither ethanol, alanine, formate, nor mannitol was found. Although we can only speculate about the additional electron pathway in the rnf mutant, it should be mentioned that cells of A. woodii grown in the absence of Na⁺ also did not employ the WLP, and the additional reduced end products could not be accounted for either (32). Since the rnf mutant also grew on mineral medium on fructose, reduction of the components of yeast extract is not the explanation.

To conclude, our data confirmed that the Rnf complex is the sole respiratory enzyme during autotrophic acetogenesis in A woodii. In addition, we also demonstrate that the Rnf complex is essential for acetogenesis from low-energy heterotrophic substrates, such as ethanol or lactate. These findings put a new perspective on the Rnf complex. It is not only the essential respiratory enzyme for autotrophic acetogenesis, but it is also the only enzyme able to generate a high-energy intermediate from low-potential electron donors, reduced ferredoxin, which is required for the WLP and biosynthesis of cell material by carboxylation of acetyl-CoA or by fixation of nitrogen (Fig. 8). The inherent reversibility of this respiratory enzyme allows the cells to cope with various low-energy scenarios and may have made life on Earth possible by allowing cells to grow on H₂ plus CO₂ and on reduced C₂ substrates. The same principle also applies to C₁ substrates, such as methanol or methylsulfides.

FIG 8.

Physiological role of the Rnf complex in A. woodii. The Rnf complex is connected to the membrane potential and functions as a ferredoxin:NAD⁺ transhydrogenase. During growth on H₂ or CO, Fd_red is oxidized by Rnf and a transmembrane electronical Na⁺ potential is established and used for autotrophic growth. Heterotrophic growth on low-energy substrates yields net ATP by substrate-level phosphorylation but not enough or no Fd_red. Here, the Rnf is energized by ATP hydrolysis (reverse chemiosmosis) (red arrows) and drives the “uphill” transport of electrons from NADH to ferredoxin. Please note that exact stoichiometries are not considered in this figure.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains used in this study are listed in Table S1 in the supplemental material. A. woodii DSM1030 was grown at 30°C under anaerobic conditions, as previously described (33, 34). The complex medium used for genetic manipulations and growth studies was used as described previously (32), and unless otherwise stated, 20 mM fructose was used as carbon source. Gaseous substrates were used at a pressure of 1.0 × 10⁵ Pa. Minimal medium for growth studies was used as described previously (32), in which yeast extract was omitted and greater amounts of 0.2 g/liter KH₂PO₄, 1.35 g/liter NH₄Cl, and 1.5 ml/liter selenite-tungsten solution were used, and 10 μg/ml of d/l-pantothenic acid was added. If necessary, 1 μg/ml uracil or 1 mg/ml 5-fluoroorotic acid (5-FOA) was added. Escherichia coli was cultivated in lysogeny broth (LB) medium (35). Antibiotics were added as appropriate.

Construction of the rnf operon deletion mutant.

All plasmids were constructed in E. coli HB101 (Promega, Madison, WI) and transformed into A. woodii, as previously described (36). Plasmid pMTL-JPB20 was made to delete 300 bp from the C terminus of the A. woodii pyrE gene (Awo_c16210) by allelic coupled exchange (ACE) and is equivalent to the previously described pyrE ACE knockout (KO) vector pMTL-JH12 (37). Both vectors carry the Clostridium perfringens catP marker (specifying resistance to thiamphenicol), but the pIM13 plasmid replicon and the Clostridium acetobutylicum-derived pyrE locus homology arms (HAs) of pMTL-JH12 have been replaced in pMTL-JPB20 by the replication region of plasmid pCD6 and the A. woodii-derived pyrE locus HAs. The pyrE locus homology arms comprised a 208-bp left homology arm (LHA) (NCBI RefSeq accession number NC_016894.1, positions 1909130 to 1909337) and a larger 491-bp right homology arm (RHA) (NCBI RefSeq accession number NC_016894.1, positions 1909638 to 1910128). Single-crossover integrants and double-crossover mutants were obtained sequentially, as previously described (38), using 5-FOA at 1 mg/ml and A. woodii-specific growth medium/conditions, as described in the above section.

Plasmid pMTL-JPB23 is the allelic exchange KO vector containing the deletion cassette needed to create the in-frame deletion of rnfCDGEAB in the A. woodii pyrE mutant. It is a suicide vector which lacks a Gram-positive replicon and contains a heterologous pyrE gene (cac_0027) from C. acetobutylicum ATCC 824 that is used as a counterselectable marker, equivalent to the heterologous pyrE of pMTL-AMH101 (39). Homology arms were designed to flank the A. woodii rnf operon of 596 bp (LHA) (NCBI RefSeq accession number NC_016894.1, positions 2597290 to 2597885) and 601 bp (RHA) (NCBI RefSeq accession number NC_016894.1, positions 2591570 to 2592170). They were assembled as a single DNA fragment using splicing by overlap extension PCR (SOE-PCR) and the oligonucleotides listed in Table S2, and they were cloned, following cleavage with the restriction enzymes NotI and NheI, into the equivalent sites of pMTL-JPB23.

This rnf deletion cassette retained the three starting codons of rnfC and two end codons of rnfB without affecting the 5′ untranslated region (5′-UTR) and 3′-UTR. Plasmid pMTL-JPB23 was transformed by electroporation into the A. woodii pyrE mutant strain and plated onto minimal agar medium supplemented with 1 μg/ml uracil and 35 μg/ml thiamphenicol. Thiamphenicol-resistant colonies, representing single-crossover integrants, were restreaked onto agar minimal medium containing 1 μg/ml uracil and 1 mg/ml 5-FOA–agar plates to select for 5-FOA-resistant cells in which the plasmid had excised and been lost from the population. The resulting colonies were screened with the flanking oligonucleotides listed in Table S2, followed by Sanger sequencing of the entire cloned region, to distinguish wild-type revertants from the desired mutants containing the 5,119-bp rnf deletion.

Measurement of ferredoxin:NAD⁺ oxidoreductase activity.

The measurement of Fno activity was conducted as described previously (40). The buffer used was 20 mM Tris-HCl (pH 7.7) also containing 20 mM NaCl, 2 mM dithioerythritol (DTE), and 4 μM resazurin.

Measurement of ²²Na⁺ translocation.

²²Na⁺ translocation experiments using inverted membrane vesicles were performed under anaerobic conditions in 20 mM Tris-HCl buffer (pH 7.7), 2 mM DTE, and 2 μM resazurin (to monitor redox state) at 30°C in a shaking water bath, as described previously (41).

Experiments with cell suspensions.

For the experiments with resting cells, A. woodii (DSM 1030) was grown as described above at 30°C under anaerobic conditions using 20 mM fructose to an optical density at 600 nm (OD₆₀₀) of ∼1.5. Cells were harvested by centrifugation (11,300 × g, 10 min, 4°C) and washed twice with imidazole buffer (50 mM imidazole, supplemented with 20 mM MgSO₄, 2 mM DTE [pH 6.7 to 6.8]). The cells were suspended in suspension buffer (40 mM imidazole, supplemented with 20 mM KCl, 20 mM MgSO₄, 20 mM NaCl, 2 mM DTE, 4 μM resazurin [pH 7]) to a concentration of 1 mg/ml in 115-ml glass bottles. The bottles contained a final volume of 12 ml of buffer together with cells and substrate and were incubated at 30°C in a shaking water bath. All steps were carried out under strictly anaerobic conditions in an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI).

Analytical methods.

Protein concentrations were determined by the method of Schmidt et al. (42), Lowry et al. (43), or Bradford (44), according to the respective assay. Fructose and acetate concentrations were measured enzymatically (R-Biopharm AG, Darmstadt, Germany). Acetate and H₂ concentrations were also determined by using gas chromatography, as previously described (45, 46), and other unknown products were analyzed by high-performance liquid chromatography, as previously described (47). ATP yields were determined as described previously (48).