A Na+-translocating Pyrophosphatase in the Acetogenic Bacterium Acetobacterium woodii*

The anaerobic acetogenic bacterium Acetobacterium woodii employs a novel type of Na+-motive anaerobic respiration, caffeate respiration. However, this respiration is at the thermodynamic limit of energy conservation, and even worse, in the first step, caffeate is activated by caffeyl-CoA synthetase, which hydrolyzes ATP to AMP and pyrophosphate. Here, we have addressed whether or not the energy stored in the anhydride bond of pyrophosphate is conserved by A. woodii. Inverted membrane vesicles of A. woodii have a membrane-bound pyrophosphatase that catalyzes pyrophosphate hydrolysis at a rate of 70–120 milliunits/mg of protein. Pyrophosphatase activity was dependent on the divalent cation Mg2+. In addition, activity was strictly dependent on Na+ with a Km of 1.1 mm. Hydrolysis of pyrophosphate was accompanied by 22Na+ transport into the lumen of the inverted membrane vesicles. Inhibitor studies revealed that 22Na+ transport was primary and electrogenic. Next to the Na+-motive ferredoxin:NAD+ oxidoreductase (Fno or Rnf), the Na+-pyrophosphatase is the second primary Na+-translocating enzyme in A. woodii.

The anaerobic acetogenic bacterium Acetobacterium woodii is the prime example for an organism that has based its bioenergetics on Na ϩ instead of H ϩ (1-4). The energy-conserving pathways such as carbonate respiration (Wood-Ljungdahl pathway) (5,6) are Na ϩ -motive; the F 1 F 0 -ATP synthase uses Na ϩ as coupling ion (7,8); and the flagellar motor is powered by an electrochemical sodium ion gradient (9). In addition, A. woodii is able to grow in the apparent absence of an electrochemical proton gradient, indicating that the growth-essential secondary transport systems are also driven by ⌬ Na ϩ. 2 A. woodii not only grows by CO 2 reduction, but it can also reduce the double bond of certain phenylacrylates, such as caffeate, in a process called caffeate respiration (3, 10 -13), in which caffeate is reduced to hydrocaffeate, and a Na ϩ gradient is established across the cytoplasmic membrane that drives ATP synthesis (11). In this anaerobic respiration, oxidation of sugars or hydrogen yields reduced ferredoxin that is reoxidized by a membrane-bound Na ϩ -translocating ferredoxin:NAD ϩ oxidoreductase (Fno/Rnf) (14,15). This is the only coupling site known in caffeate respiration, and thermo-dynamic calculations show that this reaction allows for the transport of 1-2 mol of Na ϩ . Assuming that 3-4 mol of Na ϩ / mol of ATP are translocated, 3-4 mol of caffeate have to be reduced to synthesize 1 mol of ATP. Given this unfavorable energy balance, it was surprising to see that the electron acceptor caffeate is activated to caffeyl-CoA prior to its reduction by NADH via a hypothetical caffeyl-CoA reductase-Etf complex. Caffeate activation is catalyzed by an AMP-dependent caffeyl-CoA synthetase that hydrolyzes ATP to AMP and PP i (35). Thus, two energy-rich phosphate bonds have to be invested. Here, we have addressed whether or not some energy may be saved by conserving the energy of the PP i bond in the form of an electrochemical ion (Na ϩ ) gradient across the membrane. To search for such an activity, we have used an inverted membrane vesicle system of A. woodii, 22 Na ϩ , as a tracer and pyrophosphate as an energy source.

EXPERIMENTAL PROCEDURES
Growth of Cells and Preparation of Vesicles-A. woodii (DSM 1030) was grown under anaerobic conditions using 20 mM fructose as substrate as described (5,7). The preparation of vesicles was done as described but slightly modified. For preparation of vesicles, the growth medium was supplemented with 420 mM sucrose and 8.1 mM MgSO 4 . 5 liters of medium were inoculated (4%), and the absorbance was followed at 600 nm. At A 600 ϭ 0.7-0.9, 70 g of penicillin G/ml were added to the medium to induce protoplast formation. During further incubation, A. woodii formed protoplasts as monitored by microscopic observations. After 20 h, the culture consisted almost entirely of spherical forms that were highly sensitive to low osmolarity. These protoplasts were harvested aerobically by centrifugation (6250 ϫ g, 20 min, 4°C) and washed in 25 mM Pipes (sodium-free)/KOH buffer (pH 6.8) containing 25 mM MgSO 4 and 420 mM sucrose. After washing the protoplasts, they were resuspended in a total volume of 300 ml and incubated with lysozyme (1 mg/ml) for 30 min at room temperature. The protoplasts were centrifuged (6250 ϫ g, 20 min, 4°C) and resuspended in 10 -20 ml of Pipes/KOH buffer (pH 6.8) containing 25 mM MgSO 4 and 420 mM sucrose. These protoplasts were passed through a French pressure cell at 41 megapascals and centrifuged three times (4500 ϫ g, 35 min, 4°C). The resulting supernatant (i.e. crude vesicles) was centrifuged further by ultracentrifugation (120,000 ϫ g, 40 min, 4°C). The pellet was washed in Pipes/ KOH buffer (pH 6.8) containing 25 mM MgSO 4 and 420 mM sucrose and centrifuged again. The resulting pellet was resus-* This work was supported by a grant from the Deutsche Forschungsgemeinschaft (SFB 807). 1 To whom correspondence should be addressed. pended in the same buffer in a volume of 3 ml. Protein concentrations were determined by the method of Bradford (16).
PP i Hydrolysis-Pyrophosphatase (PPase) 3 activity was determined by measuring the released phosphate using a method described by Heinonen and Lathi (17).
Measurement of Na ϩ Translocation-The experiments were performed under aerobic conditions in Pipes/KOH buffer (pH 6.8) containing 25 mM MgSO 4 and 420 mM sucrose as described (7). The sodium and protein concentrations used are indicated. In a 1.5-ml Eppendorf cup, the vesicles, buffer, supplements (17 M valinomycin, 1 mM N,NЈ-dicyclohexylcarbodiimide, NaCl, and 150 mM KCl), and 22 NaCl (final activity of 0.5 Ci/ml) were combined and incubated at 30°C for 120 min to assure equilibration of 22 Na ϩ before the reaction was started with potassium pyrophosphate (final concentration of 1 mM or as indicated). 100 M N,N,N,NЈ-tetracyclohexyl-1,2-phenylenedioxydiacetamide (ETH 2120) and 3,5-di-tert-butylhydroxybenzylidenemalonitrile (SF 6847) were added 6 min before the addition of PP i . 100-l samples were withdrawn from the cup and passed over a column (0.5 ϫ 3.2 cm) of Dowex 50-WX8 (100 -200-mesh) (7). The vesicles were collected by washing the column with 1 ml of 420 mM sucrose. The radioactivity in the eluate was determined by liquid scintillation counting. For elucidating the effect of Mg 2ϩ on the transport activity, the vesicles were pre-pared in the same buffer except that MgSO 4 was omitted. MgSO 4 and EDTA were added from stock solutions before the addition of 22 NaCl.
Determination of Na ϩ Concentration-The Na ϩ concentration in the buffer was determined with an Orion 84-11 ROSS sodium electrode (Thermo Electron Corp., Witchford, UK) as described (18).

RESULTS
Inverted membrane vesicles (IMVs) of A. woodii prepared in Pipes/KOH buffer (pH 6.8) that also contained 25 mM MgSO 4 , 420 mM sucrose, and 1.9 mM NaCl were incubated at 30°C. Upon the addition of PP i , it was hydrolyzed with a rate of 70 -120 milliunits/mg (1 unit corresponding to 1 mol of PP i /min at 1.9 mM NaCl). PP i hydrolysis was dependent on the external Na ϩ concentration. When no sodium was added to the assay, an activity of 68 -74 milliunits/mg was measured (resulting from residual Na ϩ in the buffer and vesicles), but increasing Na ϩ concentrations led to increasing activities in a Michaelis-Menten-type fashion (Fig. 1A). The double-reciprocal plot is shown in Fig. 1B, and a K m of 1.1 Ϯ 0.4 mM and a V max of 170 Ϯ 12 milliunits/mg of protein were obtained. KCl and LiCl had only a marginal stimulatory effect. Pyrophosphate hydrolysis was dependent on the pyrophosphate concentration. Highest activities were obtained at 1 mM PP i ( Fig.  2A). Pyrophosphate hydrolysis was accompanied by 22 Na ϩ transport into the vesicles (Fig. 2B). As expected, 22 Na ϩ transport increased with the pyrophosphate concentration. Also, at  Na ؉ Pyrophosphatase from A. woodii FEBRUARY 25, 2011 • VOLUME 286 • NUMBER 8 lower PP i concentrations, the Na ϩ gradient established collapsed after some time, which may be due to PP i depletion. There was a negligible activity at the lowest Na ϩ concentration achieved in the assay (900 M, taking into account the buffer, vesicles, and potassium pyrophosphate), but transport activity increased with increasing Na ϩ concentration in a Michaelis-Menten fashion (Fig. 3A). The corresponding Lineweaver-Burk plot is shown in Fig. 3B. The K m for Na ϩ transport was determined to be 7.2 Ϯ 1.4 mM, and the V max was 3.6 Ϯ 0.25 nmol/min/mg of protein. This gives a Na ϩ /PP i ratio of 0.02.
The intravesicular Na ϩ concentration in the IMVs (and thus the magnitude of the established Na ϩ gradient) was also dependent on the concentration of Na ϩ . At 0.9 mM NaCl, a 10-fold accumulation was observed. Increasing Na ϩ concentrations led to decreasing Na ϩ gradients. At 1.9 mM NaCl, the accumulation factor was 9-fold, and at 40.9 mM NaCl, it was only 2.2-fold. Because the accumulation factor was higher at lower sodium concentrations, all subsequent experiments were performed in the presence of 1.9 mM NaCl.
It has been described previously that the PPase activity in Rhodospirillum rubrum and the activity of PPase isolated from mung beans are Mg 2ϩ -dependent (19,20). To test the effect of Mg 2ϩ on the PP i -dependent Na ϩ transport at IMVs of A. woodii, IMVs had to be prepared in Mg 2ϩ -free buffer. Unfortunately, when IMVs were prepared in the absence of MgSO 4 , the yield was very low, and the activity was very low, indicating that MgSO 4 stabilizes the IMVs. When these IMVs were added to Mg 2ϩ -free assay buffer, Na ϩ transport (at 1.9 mM NaCl) was very low, with 0.004 nmol/min/mg of protein.
The addition of Mg 2ϩ to a final concentration of 25 mM stimulated Na ϩ transport by 2000% to 0.08 nmol/min/mg of protein. When the Mg 2ϩ -containing assay was supplemented with EDTA, Na ϩ transport dropped to 0.01 nmol/min/mg of protein. These data show that PPase-driven Na ϩ transport requires Mg 2ϩ .
If transport of 22 Na ϩ is electrogenic, a membrane potential should be established that in turn should slow down 22 Na ϩ transport. To test this, we used valinomycin in combination with KCl to dissipate the potential (⌬⌿). KCl itself (0, 50, 100, and 150 mM) had no effect on Na ϩ transport. However, in combination with valinomycin, 22 Na ϩ transport was stimu-lated slightly (Fig. 4). This experiment indicates that PP idriven 22 Na ϩ transport is electrogenic.
PP i -dependent Na ϩ transport should lead to the generation of a sodium ion potential that then could be used by the Na ϩ F 1 F 0 -ATP synthase of A. woodii to synthesize ATP. If this is true, inhibition of the Na ϩ F 1 F 0 -ATP synthase should lead to higher levels of 22 Na ϩ in the lumen of the vesicles. This was indeed observed. In the presence of the F 0 -directed inhibitor N,NЈ-dicyclohexylcarbodiimide (1 mM, no extra sodium added to the assay), PP i -driven accumulation of 22 Na ϩ was increased by 25%. However, the Na ϩ transport rate was not affected.
Pyrophosphatase activity could be directly or indirectly (via a primary H ϩ gradient in combination with a Na ϩ /H ϩ antiporter) coupled to Na ϩ transport. To discriminate between these possibilities, we used protonophores or sodium ionophores and analyzed their effect on 22 Na ϩ transport. The protonophore SF 6847 (100 M) did not inhibit 22 Na ϩ accumulation (Fig. 5). The effectiveness of the uncoupler was ensured by its ability to dissipate an artificial ⌬pH created by a NH 4 ϩ diffusion potential (21). The sodium ionophore ETH 2120 (100 M) completely inhibited 22 Na ϩ accumulation; moreover, a previously established Na ϩ gradient was dissipated immediately upon the addition of ETH 2120 (Fig. 5). These   . Na ؉ dependence of PP i -dependent transport activity. Membrane vesicles (protein concentration of 3.6 mg/ml) were incubated in 25 mM Pipes/KOH buffer (pH 6.8) containing 25 mM MgSO 4 , 420 mM sucrose, and different NaCl concentrations. Transport rates were calculated from the initial slopes and plotted against the Na ϩ concentration in a Michaelis-Menten (A) or Lineweaver-Burk (B) plot. Data from two independent experiments are shown. data demonstrate that 22 Na ϩ transport is directly linked to pyrophosphatase activity.

DISCUSSION
Pyrophosphatases are commonly found enzymes. Next to soluble pyrophosphatases, membrane-bound enzymes have been found in bacteria, archaea, and plants (22)(23)(24)(25). It has been shown that membrane-bound PPases couple PP i hydrolysis with proton or sodium ion transport (23, 26 -29). Proton-pumping PPases acidify vacuoles in plants (22). In bacteria, they have been proposed to support energy conservation under conditions of energy deprivation (30,31), and a light-induced maintenance of PP i levels by PPase has been suggested based on experiments with Rhodobacter capsulatus (32). PP i is a by-product of many biosynthetic processes for macromolecules, such as RNA and DNA synthesis, polysaccharide synthesis, formation of fatty acyl-CoA, and amino acid activation, and is formed in several reactions for substrate activation by anaerobic bacteria, e.g. activation of sulfate by sulfate-reducing bacteria or in activation of benzoate derivatives during anaerobic degradation of these compounds (23,33).
Here, we have demonstrated a Na ϩ -pumping pyrophosphatase at IMVs of A. woodii. The coupling efficiency was low, indicating leakiness of the vesicle system. Additionally, a second non-Na ϩ -motive pyrophosphatase may contribute to PP i hydrolysis activity. This enzyme may be membrane-attached and thus also explain the different K m values for 22 Na ϩ transport and PP i hydrolysis.
The overall ATP gain of caffeate respiration, as it is understood now, is very low. Electron flow from ferredoxin (Ϫ500 to Ϫ420 mV) to NAD ϩ (Ϫ320 mV) is accompanied with a free energy change of only Ϫ35 to Ϫ19 kJ/mol, and therefore, the Fno/Rnf complex allows for the synthesis of about onethird to one-half of an ATP. Thus, 2-3 mol of hydrogen have to be oxidized to get 1 ATP. Therefore, its highly economic for the cells to save the energy stored in PP i . The standard free energy stored in the PP i bond is equivalent to Ϫ22 kJ/mol. Assuming a membrane potential of Ϫ200 mV, this would allow for translocation of 1 Na ϩ . Consequently, 3 mol of PP i have to be hydrolyzed for the production of 1 mol of ATP. This is only a small amount of energy saved, but rationalizing that caffeate respiration is at the thermodynamic limit, this adds another piece to the overall energy balance. Additionally, energy may be conserved by electron bifurcation from NADH to ferredoxin in the process of caffeyl-CoA reduction and the subsequent Fno/Rnf complex-catalyzed ferredoxin oxidation (34,15). Furthermore, in the steady state of caffeate respiration, the activation of caffeate is likely to be catalyzed by an energy-saving CoA loop from hydrocaffeyl-CoA to caffeate (35).
In summary, caffeate respiration involves an initial activation of the caffeate prior to its reduction by a AMP and PP i , forming caffeyl-CoA synthetase, yielding caffeyl-CoA. The produced PP i , as shown here, can be used by a membranebound pyrophosphatase to translocate Na ϩ across the cytoplasmic membrane in A. woodii. Therefore, the energy stored in the PP i bond is not lost but reinvested and conserved in a chemiosmotic Na ϩ potential (Fig. 6).