A Na+ A1AO ATP synthase with a V‐type c subunit in a mesophilic bacterium

A1AO ATP synthases with a V‐type c subunit have only been found in hyperthermophilic archaea which makes bioenergetic analyses impossible due to the instability of liposomes at high temperatures. A search for a potential archaeal A1AO ATP synthase with a V‐type c subunit in a mesophilic organism revealed an A1AO ATP synthase cluster in the anaerobic, acetogenic bacterium Eubacterium limosum KIST612. The enzyme was purified to apparent homogeneity from cells grown on methanol to a specific activity of 1.2 U·mg−1 with a yield of 12%. The enzyme contained subunits A, B, C, D, E, F, H, a, and c. Subunit c is predicted to be a typical V‐type c subunit with only one ion (Na+)‐binding site. Indeed, ATP hydrolysis was strictly Na+‐dependent. N,N′‐dicyclohexylcarbodiimide (DCCD) inhibited ATP hydrolysis, but inhibition was relieved by addition of Na+. Na+ was shown directly to abolish binding of the fluorescence DCCD derivative, NCD‐4, to subunit c, demonstrating a competition of Na+ and DCCD/NCD‐4 for a common binding site. After incorporation of the A1AO ATP synthase into liposomes, ATP‐dependent primary transport of 22Na+ as well as ΔµNa+‐driven ATP synthesis could be demonstrated. The Na+ A1AO ATP synthase from E. limosum is the first ATP synthase with a V‐type c subunit from a mesophilic organism. This will enable future bioenergetic analysis of these unique ATP synthases.


Introduction
Despite all the differences in the reactions generating the electrochemical ion potential across the membrane of living cells, the ATP synthase is the universal enzyme present in every living cell to harvest electrochemical energy and transform it into biological useful fuel, ATP [1,2]. All known ATP synthases use the same principle mechanisms of energy conservation and they evolved from a common ancestor [3]. ATP synthases are rotary machines that are composed of two motor domains connected by one central stalk and at least one peripheral stalk. The membrane-embedded motor, A O , consists of the c ring, composed of multiple copies of the c subunit, and the stator, subunit a.
The membrane-embedded domain translocates ions (H + or Na + ) across the membrane leading to rotation of the c ring against the stator. The resulting rotational energy is transmitted via the central stalk to the A 1 motor which then drives ATP synthesis. This operation is fully reversible, resulting in ATP hydrolysis coupled to ion transport across the membrane [2,4].
Evolutionary, archaeal A 1 A O ATP synthases are more closely related to V 1 V O ATPases, mainly present in vacuoles of eukarya, than to F 1 F O ATP synthases found in bacteria [2]. V 1 V O ATPases are designed by nature as ATP-driven ion pumps and are not able to synthesize ATP in vivo [2,5,6]. In most archaeal A 1 A O ATP synthases, the subunit c consists of two transmembrane helices with one ion-binding site as seen in F 1 F O ATP synthases. In the course of evolution, the c subunit in eukaryotic V 1 V O ATPases [7,8] underwent gene duplication, giving rise to a proteolipid with four transmembrane helices [9]. Additionally, one ion-binding site was lost resulting in a rotor missing half the number of ion-binding sites. This binding site composition is generally seen as the reason for the inability of V 1 V O ATPases to synthesize ATP in vivo [7,10]. Notably, some A 1 A O ATP synthases have a c subunit with four transmembrane helices but only one ionbinding site [6]. This structural feature would predict that the enzyme has lost its ATP synthesis capability. However, since no other ATP synthase genes are found in these archaea, one has to postulate that such an enzyme is able to synthesize ATP. This is indeed substantiated by measuring ATP synthesis in whole cells and membrane vesicles [11]. At this point in time, intact A 1 A O ATP synthases could only be purified from hyperthermophilic archaea [12][13][14][15][16]. However, these hyperthermophilic ATP synthases are inactive at temperatures below 80°C, with temperature optima between 80°C and 100°C. Under these conditions, all commonly used liposome systems are inactive and suitable lipids for these conditions could not be obtained so far, despite all efforts.
Inspection of genome sequences from mesophilic bacteria and archaea revealed one gene cluster in Eubacterium limosum (T opt = 37°C) potentially encoding an A 1 A O ATP synthase with an unusual V-type c subunit [17]. We have purified the enzyme from membranes of E. limosum and will demonstrate that it is a Na + -dependent A 1 A O ATP synthase containing a Vtype like c subunit capable of ATP synthesis.

Results
Genome organization, genes, and subunits of the The genes Eli_2184-2192 encode an enzyme with highest similarity to an archaeal A 1 A O ATP synthase (Fig. 1A). Upstream of the A 1 A O ATP synthase operon on the reverse strand Eli_2183 (117 bp) encodes for a hypothetical protein with 41% identity to dynein protein 3 of Saccharomyces cerevisiae (Table 1). Eli_2184 is 309 bp long and encodes for subunit H with a molecular mass of 11.4 kDa ( Table 1). The encoded protein is 53% similar to subunit H of the A 1 A O ATP synthase from Methanothermobacter marburgensis. The next gene of the operon (Eli_2185) has a 3 bp overlap with Eli_2184. It is 1875 bp long and encodes for subunit a of the hydrophobic A O domain with a molecular mass of 69.8 kDa. The deduced gene product is 31% and 26% identical to the corresponding subunit from Methanocaldococcus jannaschii [14] and Methanosarcina mazei G€ o1 [18], respectively. Most interestingly, 14 bp downstream of Eli_2185 is the 474bp-long gene encoding the rotor subunit c (Eli_2186) with a predicted mass of a 16 kDa and four transmembrane helices (Fig. 1B). The similarity/identity to subunit c of Pyrococcus furiosus is 51/35%. Recently, we have shown that the c subunit from P. furiosus has indeed four transmembrane helices but only one ion (Na + )-binding site (Q. . .ET) [19] which is also conserved in Thermococcus onnurineus and in subunit c of E. limosum. As in P. furiosus, there is no second proton or sodium ion-binding site conserved. Eli_2187 (606 bp) is located 36 bp downstream of Eli_2186, and the deduced protein (22.6 kDa) is annotated as subunit E of V 1 V O ATPases. No significant homology to the corresponding protein sequence of subunit E in A 1 A O ATP synthases was found. Four base pair downstream of Eli_2187 is the 972-bp-long gene Eli_2188 which encodes for a subunit C with a molecular mass of 36.4 kDa. The identity/similarity of the deduced protein is 26/47% to subunit C of the A 1 A O ATP synthase from P. furiosus [12]. Ten base pair downstream of Eli_2188 is 318-bp-long Eli_2189 which encodes a polypeptide with a molecular mass of 11.3 kDa. 51/ 27% of the residues are similar/identical to subunit F of the A 1 A O ATPase from M. barkeri [20]. Eli_2190 (1806 bp) is located 26 bp downstream of Eli_2189 and encodes for subunit A with a molecular mass of 66.8 kDa. The deduced gene product is 59% identical to the corresponding subunit from M. jannaschii and T. onnurineus. Eli_2191 (1410 bp) encodes for subunit B with a molecular mass of 52.3 kDa. 71, 67, and 30% of the residues are identical to subunit B of the A 1 A O ATPase from P. furiosus, T. onnurineus, and M. barkeri, respectively. Four base pair downstream of Eli_2191 is Eli_2192 with a length of 636 bp. The protein has a molecular mass of 29.4 kDa. Sequence alignments showed 42% and 39% identity to subunit D of M. jannaschii and T. onnurineus, respectively. The next gene downstream of the putative ATP synthase operon Eli_2193 (138 bp) is located in 3 0 ? 5 0 direction. For this gene, no homologues were found.

Purification of the A 1 A O ATP synthase from E. limosum
To determine the effect of the growth substrate on ATPase activity, E. limosum KIST612 was grown on complex medium with 50 mM glucose, 50 mM  , 50 mM formate, 101 kPa H 2 + CO 2 , or 101 kPa CO as sole carbon and energy source. Cells were harvested during exponential growth, washed membranes were prepared, and ATP hydrolysis was determined. Washed membranes of H 2 + CO 2 -grown cells had the highest ATPase activity of 68.3 mUÁmg À1 . The activity of washed membranes from methanol-, formate-, and CO-grown cells was similar and slightly below the latter one (54, 41.3, and 50.7 mUÁmg À1 ). Membranes from glucose-grown cells showed the lowest activity with 24 mUÁmg À1 . Due to the low yield of cell mass from H 2 + CO 2 -grown cultures (final OD 600 = 0.4), methanol was selected as growth substrate (final OD 600 = 1.2) for large-scale cultivation. For purification of the A 1 A O ATP synthase, washed membranes of around 15 g cells (wet mass) were prepared and solubilized with 1 mg n-dodecyl-b-maltoside (DDM) per mg membrane protein.
The ATP synthase was further purified by a sucrose density gradient (20-65%), and the subsequent purification by an anion exchange chromatography step (DEAE) as well as a size-exclusion chromatography (Superose 6 column) resulted in 23.6-fold enrichment  of the activity with a specific activity of 1277 mUÁmg À1 and a yield of 12.2% (Table 2).

Subunit composition of the A 1 A O ATP synthase from E. limosum
The preparation contained 13 proteins ( Fig. 2) with apparent molecular masses of 67 kDa (subunit A), 58 kDa (subunit B), 39 kDa (subunit C), 26 kDa (subunit D), 25 kDa (subunit E), 11 kDa (subunit F), 12 kDa (subunit H), and 14 kDa (subunit c). As seen before with A 1 A O ATP synthases from hyperthermophilic archaea, there is a protein above subunit A which could be a SDS-resistant ac subcomplex [12,15]. Single subunit a was not detected. The identity of the subunits was proven by MALDI-TOF-MS (Fig. 3) and immunological with antibodies generated and specific against the purified subunits. The 14 kDa protein was identified as subunit c by peptide mass fingerprinting. Notably, the loops connecting TMH1 and 2 as well as TMH3 and 4 were detected, demonstrating that subunit c indeed contains four transmembrane helices (Fig. 4).

Basic biochemical properties of ATP hydrolysis
ATPase activity was detected over a pH range between 5.0 and 9.0, with optimal activity between pH 7.0 and 7.5 ( Fig. 5A). pH values lower than 6.0 and higher than 8.0 decreased the activity very strongly (9.5% activity at pH 5.0, 10% activity at pH 9.0). ATP hydrolysis activity was optimal between 30°C and 40°C ( Fig. 5B). After 45°C, the activity decreased sharply. Methanol stimulated ATP hydrolysis activity as described before [21] (Fig. 5C). Optimal activity was determined with 20% methanol.
Na + dependence of ATP hydrolysis ATP hydrolysis was strictly dependent on the Na + concentration of the assay buffer ( Fig. 6). At the contaminating amount of 71 µM Na + , ATP hydrolysis was only 326 mUÁmg À1 but activity was stimulated fourfold by the addition of Na + . The Na + dependence followed a classical Michaelis-Menten kinetics; full activity was observed at 0.6 mM and half maximal was at 0.3 mM. The addition of Li + also stimulated ATP hydrolysis; full activity was observed at 2 mM and half maximal activity was at 1 mM. K + did not stimulate ATP hydrolysis.
DCCD and Na + compete for binding to subunit c DCCD is a common inhibitor for ATP synthases that covalently binds to the highly conserved protonated carboxylate (E138 in helix 4) in subunit c, thereby inhibiting proton or Na + binding and transport [22,23]. In all known Na + ATP synthases, the access of DCCD is blocked in the presence of Na + , thus reducing the inhibitory effect of DCCD on the enzyme [12,19,[24][25][26]. DCCD inhibited the A 1 A O ATP synthase from E. limosum with half maximal inhibition at 80 µM DCCD (Fig 7). When the enzyme was preincubated with Na + , DCCD inhibition was relieved. NCD-4 is a fluorescent analogue of DCCD and bound to subunit c (Fig. 8). Preincubation with Na + prevented NCD-4 binding. These data are consistent with the proposal that Na + is the coupling ion of the A 1 A O ATP synthase from E. limosum.

Na + transport and ATP synthesis in proteoliposomes
In order to unequivocally proof Na + transport by the A 1 A O ATP synthase, the enzyme was reconstituted into liposomes consisting of phosphatidylcholine type II S from soybeans. Liposomes were generated by sonification [27]. Upon addition of ATP, 22 Na + was transported into the lumen of the vesicles with a rate of 0.6 nmol Na + Ámin À1 Ámg À1 protein, leading to a final accumulation of 9.2 nmol Na + Ámg À1 protein (Fig. 9). 22 Na + transport was not observed in the absence of ATP or in the presence of the sodium ionophore N,N, N,N 0 -tetra-cyclo-hexyl-1,2-phenylenedioxydiacetamide (ETH 2120). The protonophore 3,3 0 ,4 0 ,5-tetrachlorosalicylanilide (TCS) slightly stimulated 22 Na + transport. These data are consistent with a primary and electrogenic Na + transport catalyzed by the A 1 A O ATP synthase. Next, we investigated whether the enzyme is capable of ATP synthesis despite its V-type like c subunit. Therefore, the enzyme was again reconstituted into liposomes and an electrochemical Na + potential (DµNa + ) of around 270 mV was applied. DµNa + was generated by a sodium ion potential (DpNa) of 75 mV combined with a K + /valinomycin diffusion potential of around 195 mV. Upon addition of 2 µM valinomycin, ATP was synthesized with a rate of about 54 nmolÁmin À1 Ámg À1 protein (Fig. 10). The data demonstrate that the A 1 A O ATP synthase with its Vtype like c subunit from E. limosum is capable of ATP synthesis.

Discussion
Many archaea like methanogens, Thermococcus, or Pyrococcus live under extreme conditions: at high temperatures of > 80°C, sometimes paired with another extreme, a metabolism that allows for the production of only a small fraction of an ATP per mol of substrate oxidized [28][29][30]. This life at the thermodynamic limit requires adaptions in the ATP-generating mechanisms. ATP is synthesized exclusively or in addition by a chemiosmotic mechanism that involves unusual respiratory enzyme such as membrane-bound, iontranslocating hydrogenases (progenitors of complex I) or a formate dehydrogenase [30,31]. Often, the free energy change is so small that it allows for the translocation of only one ion per mol of substrate oxidized, sometimes, as in the case of T. onnurineus, even less than that [31]. Special adaptions in the ATP synthase also contribute to this lifestyle. One adaption is the use of Na + instead of H + as coupling ion, the other a c ring with a reduced number of ion translocation sites [28,29,32]. This would allow the enzyme to synthesize ATP at lower costs; however, this would only work if the phosphorylation potential in these cells is also lower [33].
In F 1 F O ATP synthases, the minimal ion/ATP ratio is 2.7, restricted by the minimal number of 8 c subunits with two transmembrane helices to form a ring structure. However, a doubling of the size of the individual c subunits forming the ring increases the stability of the ring structure due to its longer size and possibly allows a further reduction of the ion/ATP ratio. Indeed, the c ring of the A 1 A O ATP synthase of P. furiosus is proposed to have 10 [13] or 9 [34] subunits.
Measurements of energetic parameter such as phosphorylation potential, membrane potential, and ion transport are scarce in these hyperthermophilic archaea, due to their difficult cultivation and inherent problems of measuring these values at temperatures at 80-100°C. A solution to this dilemma comes with acetogenic bacteria such as Acetobacterium woodii or, here, E. limosum. Both are mesophiles but like most other acetogenic bacteria live at the thermodynamic edge of life. A. woodii grows heterotrophically but also autotrophically on H 2 + CO 2 [35]. The phosphorylation potential in these cells was measured to be much below the 'textbook-knowledge' value of 60 kJÁmol À1 : 37.9 AE 1.3 kJÁmol À1 for fructose, 32.1 AE 0.3 for H 2 + CO 2 , and 30.2 AE 0.9 for CO [36]. This would allow ATP synthesis with only a fraction of ions. Indeed, the c ring of the Na + F 1 F O ATP synthase from A. woodii is unique since it contains one V-type c subunit along with nine F-type c subunits, the first and so far, only F/V-hybrid motor found in nature [37]. The ATP synthase of E. limosum has a V-type like c subunit, but, as shown here, is able to synthesize ATP. With this mesophilic enzyme in hand, the road is paved to analyze the dependence of ATP synthesis on energetic parameters in these unusual organisms whose metabolism is considered to be maybe the first metabolic scenario on early Earth.

Cultivation of cells and preparation of washed membranes
Eubacterium limosum KIST612 was cultivated in anoxic phosphate-buffered basal medium (PBBM) or carbonate-buffered      For experiments under Na + -free conditions, the protein was eluted with buffer D without addition of NaCl. The concentration of Na + was determined by using a Na + -selective electrode (ORION TM 8411BN ROSS Ò ; Thermo Fisher Scientific  GmbH, Dreieich, Germany). All preparations were routinely analyzed by SDS/PAGE, using the buffer system of Sch€ agger and von Jagow [39]. Proteins were visualized by staining with Coomassie brilliant blue G250 (Sigma-Aldrich Chemie GmbH, Steinheim, Germany) [40].

Antibody generation and western blot
For expression of genes encoding for the single subunits A, B, C, D, and E, the amplified fragment was cloned into pTrc99a and each corresponding plasmid was transformed in Escherichia coli DK8. E. coli cultures were grown in 2xYT medium at 37°C, and gene expression was induced at an OD 600 of 1.0 by addition of 1 mM isopropyl-b-D-1-thiogalactopyranoside (IPTG). After 4 h of growth, cells were harvested, washed, and disrupted by a French press. After removal of cell debris, each single subunit was purified by Ni 2+ -NTA chromatography. All subunits were separated and analyzed by SDS/PAGE, using the buffer system of Sch€ agger and von Jagow [39] and visualized by staining with Coomassie brilliant blue G250 (Sigma-Aldrich Chemie GmbH) [40]. The subunits cut from a Coomassie-stained SDS gel were used to immunize rabbits. Western blotting with SDS/PAGE gels was performed as described previously [41].

Determination of ATPase activity
The ATPase activity of washed membranes isolated from cells grown on different substrates was measured in a buffer containing 100 mM Tris, 100 mM maleic acid, 10 mM MgCl 2 , 20% methanol (v/v), pH 7.3. After incubation for 3 min at 37°C, the reaction was started by the addition of 2.5 mM Na 2 -ATP. The ATPase activity was determined by a discontinuous assay following the ATP-dependent formation of inorganic phosphate, according to the method of Heinonen and Lahti [42], as previously described [43]. To determine the effect of pH and temperature, the enzyme was preincubated for 3 min at the pH (5-9) or temperature (20-50°C) indicated, respectively. The reaction temperature for the determination of the pH optimum was set to 37°C. The buffer used for the pH optima determination was 50 mM MES, 50 mM Tris, 50 mM HEPES, 50 mM CHES, 10 mM MgCl 2 , 20% methanol. For inhibitor studies with N 0 ,N 0 -dicyclohexylcarbodiimide (DCCD), the reaction mixture was preincubated at room temperature for 30 min before the reaction was started by addition of Tris-ATP. For determination of ATPase activity in the absence of sodium ions, the activity was determined in 100 mM Tris/ HCl, 5 mM MgCl 2 , and 20% methanol as described above.
Labeling with NCD-4 Purified ATP synthase was labeled with NCD-4 (solved in ethanol) by incubation with 60 or 200 nmol of NCD-4 overnight at room temperature in the presence or absence of 200 nM NaCl. After incubation, the protein was separated by SDS/PAGE (12.5%) and fluorescence was detected after excitation with UV light.

Measurement of 22 Na + translocation
Na + translocation experiments in proteoliposomes were performed according to Ref. [27] in a buffer composed of 50 mM Tris/HCl, 10 mM MgCl 2 , pH 7.3. The sodium ionophore ETH 2120 and the protonophore TCS were added from ethanolic stock solutions and controls received the 3) resulting in an outside concentration of 10 mM NaCl, thus creating a sodium ion potential (DpNa). Upon addition of 2 µM valinomycin and 5 mM ADP, a membrane potential (DΨ) was induced and the reaction was started. ATP synthesis was measured at 37°C, and ATP was determined by luciferin-luciferase assay (blue, circles). In the red plot (triangles up), no ADP was added. Each value represents the mean and standard deviation of three independent experiments (n = 3).

3020
The . The proteoliposomes, the buffer, supplements, and 22 NaCl (carrier-free, final activity 0.5 lCiÁmL À1 ) were combined in a reaction tube to a final Na + concentration of 4.5 mM and incubated at 37°C for 30 min to assure equilibration of 22 Na + , before the reaction was started by addition of 2.5 mM Tris-ATP (final concentration). Eighty microlitre samples were withdrawn and passed over a column (0.5 9 3.2 cm) of Dowex 50-WX8 (100-200-mesh) [43]. The proteoliposomes were collected by washing the column with 1 mL of 420 mM sucrose. The radioactivity in the eluate was determined by liquid scintillation counting.