V-Type H+-ATPase/synthase from a thermophilic eubacterium, Thermus thermophilus. Subunit structure and operon.

V-type ATPase (V(o)V(1)) capable of ATP-driven H(+) pumping and of H(+) gradient driven ATP synthesis was isolated from a thermophilic eubacterium, Thermus thermophilus. When the enzyme was analyzed by gel electrophoresis in the presence of sodium dodecyl sulfate, it showed eight polypeptide bands of which four were subunits of V(1). We also isolated the V(o)V(1) operon, containing nine genes in the order of atpG-I-L-E-X-F-A-B-D, which encoded proteins with molecular sizes of 13, 43, 10, 20, 35, 11, 64, 53, and 25 kDa, respectively. The last four genes were identified as those for V(1) subunits; atpA, B, D, and F encoded the A, B, gamma, and delta subunits, respectively. The first five genes, atpG-atpX, were identified as genes for the V(o) subunits. The product of atpL, the proteolipid subunit, lacked a 19-amino acid presequence and, unlike V-type ATPases, contained two membrane-spanning domains rather than four. The hydrophobic 43-kDa product of atpI is the smallest member so far found of the eukaryotic 100-kDa subunit family. Its electrophoretic band overlapped with the band of the A subunit. Therefore, all the gene products were found in our purified V(o)V(1). We isolated the A(3)B(3) subcomplex reconstituted from the isolated subunits and the A(3)B(3)gamma subcomplex from subunit-expressing Escherichia coli. Electron microscopic observation of these subcomplexes revealed that the gamma subunit of V(1) filled the central cavity of A(3)B(3) and might be central subunit, similar to the gamma subunit of F(1)-ATPase.

F-type H ϩ -ATPase (F o F 1 ) and vacuolar type H ϩ -ATPase (V o V 1 ) are two classes of a superfamily of H ϩ -translocating ATPases (1,2). F o F 1 is responsible for ATP synthesis and is present in inner membranes of mitochondria, thylakoid membranes of chloroplasts, and plasma membranes of bacteria (2,3). These enzymes are composed of a hydrophobic membrane part (F o ) and a hydrophilic catalytic part (F 1 ). F o forms a proton channel across the membranes and is composed of at least three kinds of transmembrane subunits, one of which is the most evolutionarily conserved 8-kDa proteolipid subunit (2). F 1 , which can easily be separated from the membranes and shows ATPase activity, is composed of two kinds of nucleotide binding subunits, ␣ and ␤, and three minor subunits, ␥, ␦, and ⑀, with a stoichiometry ␣ 3 ␤ 3 ␥␦⑀ (2, 3). The ␥ subunit penetrates the central cavity of the hexagonal ␣ 3 ␤ 3 like a shaft (4) and rotates when F 1 hydrolyzes ATP (5).
On the other hand, V o V 1 is present in the membranes of yeast vacuoles (6), clathrin-coated vesicles (7), chromaffine granules (8), lysosomes (9), and plant vacuoles (10) in eukaryotic cells, and it pumps H ϩ into vesicles. Like F o F 1 , the V o V 1 is composed of a water-soluble set of subunits (V 1 ) and a membrane-integral set of subunits (V o ). The two major A and B subunits and other minor subunits comprise the V 1 domain (1). The A and B subunits are homologous to the ␤ and ␣ subunits of F o F 1 . By analogy to F 1 , V 1 has also been proposed to have a central shaft subunit that rotates in the A 3 B 3 hexagon. However, an equivalent subunit with indisputable sequence homology to the F 1 ␥ subunit has not been identified in V o V 1 . The hydrophobic V o is thought to be composed of at least five different subunits (1,11). The 16-kDa proteolipid subunit containing four transmembrane domains is the major subunit of V o (1). Unlike F o F 1 , V 1 dissociates into individual subunits with concomitant loss of ATPase activity, and V o does not show activity as a H ϩ channel by itself once V 1 is detached from V o (12).
We have previously identified V o V 1 in prokaryotic cells at first in archea and then in a thermophilic eubacterium, Thermus thermophilus (13)(14)(15)(16). Because T. thermophilus, a strict aerobe, does not have F o F 1 but carries out oxidative phosphorylation, V o V 1 is responsible for ATP synthesis in this eubacterium, and indeed, the V o V 1 purified from this bacterium can catalyze ATP synthesis driven by H ϩ flow (17). Based on SDSpolyacrylamide gel electrophoresis (PAGE), 1 the isolated T. thermophilus V o V 1 has been thought to be composed of eight kinds of polypeptides with the following apparent molecular sizes: 100, 66, 56, 38, 30, 24, 13, and 12 kDa (18). Four of them, A (66 kDa), B (56 kDa), ␥ (30 kDa), and ␦ (12 kDa), are found in the purified V 1 , whereas the 100-, 38-, 24-, and 13-kDa polypeptides are thought to be V o subunits. In this study, to determine the subunit composition, we compared the nine genes in the T. thermophilus V o V 1 operon and partial peptide sequences of polypeptides in the isolated V o V 1 . Furthermore, we reported the electron microscopic observation of the subcomplexes.

EXPERIMENTAL PROCEDURES
Purification of V o V 1 from T. thermophilus Plasma Membrane-Culturing and disruption of T. thermophilus cells were carried out as described (16). Membranes were precipitated by centrifugation at 100,000 ϫ g for 15 min and washed with 50 mM Tris-SO 4 (pH 8.0) containing 5 mM MgSO 4 five times to avoid contamination with soluble proteins. The washed membranes (30 g) were suspended in 100 ml of 50 mM Tris-SO 4 (pH 8.0), 5 mM MgSO 4 , and 3% Triton X-100 (w/v), and the suspension was sonicated for 2 min. Debris and insoluble materials were removed by centrifugation at 100,000 ϫ g for 15 min, and the supernatant was applied to a DEAE-Sephacel column (3 ϫ 10 cm) equilibrated with 50 mM Tris-SO 4 (pH 8.0), 5 mM MgSO 4 , and 0.2% Triton X-100. The column was washed with 200 ml of the same buffer. Proteins were eluted with a linear Na 2 SO 4 gradient (0 -0.2 M). Fractions containing V o V 1 were identified by polyacrylamide gel electrophoresis analysis in the presence of an alkyl ether sulfate (Softy 12, LION Corp.; Ref. 18). They were combined and concentrated to 1 ml by ultrafiltration using a Mini-Module, model NM-3 (Funakoshi). The concentrated solution was applied to a Sephacryl S-300 column (1.8 ϫ 90 cm) equilibrated with 50 mM Tris-SO 4 (pH 8.0), 5 mM MgSO 4 , 50 mM Na 2 SO 4 , and 0.1% Triton X-100. Proteins were eluted with the same buffer, and the fractions were analyzed by polyacrylamide gel electrophoresis in the presence of alkyl ether sulfate and SDS. The fractions containing V o V 1 were combined and dialyzed against 50 mM Tris-SO 4 (pH 8.0), 5 mM MgSO 4 for 5 h. The dialyzed solution was applied to a Poros HQ/M (4.6 ϫ 10 cm, PerSeptive Biosystems Corp.) equilibrated with 50 mM Tris-SO 4 (pH 8.0), 5 mM MgSO 4 , and 0.1% Triton X-100. Proteins were eluted with a linear NaCl gradient (0 -0.5 M). The purity of each fraction was analyzed by SDS-PAGE. The fractions containing V o V 1 were combined, and the V o V 1 was stored at 4°C until use.
ATP driven H ϩ Translocation and Light-induced ATP Synthesis by the V o V 1 -Proteoliposomes containing V o V 1 and bacteriorhodopsin (bR) were reconstituted at 25°C in the reaction buffer (25 mM potassium phosphate buffer (pH 7.3), 50 mM K 2 SO 4 , and 50 mM Na 2 SO 4 ) according to the procedure described by Richard and co-workers (19 -21) with minor modification. Unilamellar liposomes were prepared by reverse phase evaporation using phosphatidylcholine and resuspended at a lipid concentration of 4 mg/ml. For reconstitution of proteoliposomes, 850 l of liposome solutions were mixed with 40 l of 20% (w/v) Triton X-100, 10 l of V o V 1 or authentic F o F 1 solution (3 mg protein/ml), and 50 l of bR solution (4 mg protein/ml). Then, 40 l of 14.8% of n-octyl-␤-D-glucopyranoside was added. After a 2-min incubation, 10 l of 20 mM pyranine solution in ethanol and wet Bio-Beads (80 mg) were added and incubated for 1 h at room temperature while stirring. Addition of Bio-Beads and the 1-h incubation was repeated three times. The mixtures were washed twice with the reaction buffer by centrifugation (30 min, 150,000 ϫ g), and the pellet was suspended in 1 ml of the reaction buffer. A 200-l aliquot of the proteoliposomes was mixed with 600 l of the reaction buffer, and 2 mM ADP (final concentration) was added. The reaction mixture was preilluminated by the 300 W slide projector for 15 min on a magnet stirred at 40°C. The reaction was started by addition of 2 mM (final concentration) MgSO 4 . At 10-min intervals, 50-l aliquots were taken, and the reaction was stopped by addition of the same volume of 4% trichloroacetic acid. The amount of ATP was determined by the luciferin-luciferase assay method using a luminescense reader BLR201 (Aloka). ATP driven proton translocation into bR-V o V 1 liposomes was detected by decrease of pyranine fluorescence. An aliquot of 150 l of bR-V o V 1 liposomes was preincubated at 25°C in 2 ml of the reaction buffer supplemented with 7 mM MgSO 4 . The reaction was started by adding 2 mM ATP-Mg. Excitation and emission wavelengths were 460 and 510 nm, respectively. Excess fluorecence from pyranine molecules not entrapped into bR-V o V 1 liposomes was quenched by addition of 20 mM p-xylene-bispyridinium bromide.
Cloning and Sequencing of Entire atp Operon-To obtain the 3Ј region of the operon, the genomic DNA of T. thermophilus was digested completely with StuI. DNA fragments of 2-3 kilobase pairs were ligated into the HincII site of pUC118. The recombinant plasmids obtained were then introduced into Escherichia coli strain JM109. Based on the nucleotide sequence of the cloned 6.2-kilobase pair BamHI fragment (22), which contains the 5Ј region of the ␥ subunit gene, an oligonucleotide was synthesized and used to screen the transformed colonies (23). Cloning of the 5Ј region of the atp operon was carried out in a similar manner. The nucleotide sequences were determined by the dideoxy chain termination method after subcloning the restriction fragments into pUC118 vector. All DNA sequences were confirmed by sequencing both strands. Computer searches to find proteins with sequence similarity were carried out using the protein sequence data bank of the National Biomedical Research Foundation and the SWISS-PROT protein sequence data bank. Comparisons of two sequences, hydropathy analysis, and multiple sequence alignments were carried out with GENETYX (Software Development Co., Ltd). Sequence data used for comparisons were taken from the references cited.

Preparation of A 3 B 3 ␥ and A 3 B 3 -
The DNA fragment coding A, B, and D was polymerase chain reaction amplified using exTaq polymerase (TaKaRa). The sequences of the amplified DNA fragments were confirmed using an ABI373A sequencer. The DNA fragment obtained was ligated into the EcoRI/SalI site of pTrc 99A (Amersham Pharmacia Biotech). After transformation of E. coli. strain JM103, the expression of A 3 B 3 ␥ was induced by addition of 0.4 mM (final concentration) isopropyl-1-thio-␤-D-galactopyranoside. Purification of the recombinant A 3 B 3 ␥ complex expressed in E. coli was carried out as follows. The E. coli harboring the expression vector were grown in 2 liters of 2ϫ YT containing 0.2 mM ampicillin at 37°C. About 20 g of cells were obtained from 1 liter of overnight culture medium. Cells were suspended in 50 mM Tris-Cl (pH 8.0), 50 mM NaCl, and 0.1 mM EDTA and disrupted by sonication. The debris was removed by brief centrifugation. The cell lysate was incubated at 65°C for 30 min, and insoluble denatured proteins were removed by centrifugation for 30 min at 60,000 ϫ g. The supernatant was applied to a DEAE-Sephacel column (2 ϫ 20 cm) equilibrated with 50 mM Tris-Cl (pH 8.0), 0.1 mM EDTA (buffer A). After the column was washed with buffer A, proteins were eluted with a 0 -0.5 M NaCl linear gradient in buffer A. The fractions containing A 3 B 3 ␥ were combined. Supplemented with solid ammonium sulfate to 1.2 M and applied to a butyl-toyopearl column (2 ϫ 10 cm) equilibrated with buffer A plus 1.2 M ammonium sulfate. Proteins were eluted with a 1.2-0 M ammonium sulfate reverse gradient in buffer A. The fractions containing A 3 B 3 ␥ were collected, and proteins were precipitated by addition of solid ammonium sulfate to 1.5 M. The precipitated proteins were collected by centrifugation, dissolved in a minimum volume of buffer A, and applied to the Sephacryl S-300 column (1.8 ϫ 90 cm) equilibrated with buffer A plus 0.1 M Na 2 SO 4 . The proteins were eluted with the same buffer, and the fractions containing A 3 B 3 ␥ were collected and stored at 4°C until use. A 3 B 3 complex was reconstituted from isolated A and B subunits prepared as described previously (17). Each subunit was precipitated by adding solid ammonium sulfate and then dissolved in 50 mM Tris-SO 4 (pH 8.0) and 5 mM MgCl 2 . The final protein concentration of each subunit was 50 mg/ml. The subunits were mixed at an equal molar ratio. Then, 49 l of the mixture was supplemented with 1 l of 0.1 M ATP-Mg and incubated at room temperature for 3 h. The incubated mixture was subjected to gel permeation high performance liquid chromatography (G-3000SWXL, Tosoh) and eluted with 50 mM Tris-Cl (pH 7.5) and 0.1 M Na 2 SO 4 . The fractions containing the A 3 B 3 complex were collected and analyzed immediately.
Electron Microscopy-The freshly prepared samples were diluted with 50 mM Tris-HCl (pH 8.0) to final protein concentrations of 10 -100 g/ml, and the specimens on carbon support films were embedded in negative stain (2% uranyl acetate) using the drop method (24). Specimens were examined with a Hitachi HF-2000 electron microscopy equipped with a minimal-dose system and an anti-contamination device. Grids were scanned at low magnification under dark illumination to reduce radiation damage. After adjusting the astigmatism and focus at adjacent areas, the photographs of the target specimen area were recorded on Kodak SO163 film using an electron dose of about 30 e/A 2 . The accelerating voltage was 200 kV, and the direct magnification was 50,000. Micrographs were digitized with a scanner (SCAI, Zeiss) at a pixel size of 0.14 nm for single particle analysis. About 40 particles were selected. The images were rotated to maximize correlation with each other and averaged after rejecting the particles that correlated poorly. All images were processed using the software we developed: the extensible object-oriented system (25).
Others-ATPase activity was measured at 25°C with an enzymecoupled ATP regenerating system (17). One unit is defined as the activity that hydrolyses 1 mol of ATP/min. The peptide sequence of subunits was determined as follows. Purified V o V 1 was boiled in the SDS sample buffer for 5 min and subjected to 18 or 12% SDS-PAGE. The separated subunits were transferred to polyvinylidene difluoride membranes (Immobilon, Bio-Rad) and visualized by CBB-staining. Each subunit was analyzed using a peptide sequencer. bR and F o F 1 prepared from thermophilic Bacillus PS3 were gifts from Dr. Dirk Bald.

Properties of Isolated
thermophilus membranes using two column chromatography steps, i.e. DEAE cellulose and Sepharose 6B. However, a large amount of V o V 1 was lost because of dissociation into V 1 and V o during the step of DEAE cellulose column chromatography, and the final preparation still contained some contamination (18). We found that the dissociation of V o V 1 was fairly suppressed by using DEAE-Sephacel instead of DEAE cellulose. To further improve the purity of V o V 1 , we also used ion exchange fast protein liquid chromatography for purification. The resultant isolated V o V 1 migrated as an apparently single band on nondenaturing gel electrophoresis with no apparent contamination (Fig. 1a). The specific ATPase activity of the purified V o V 1 was 4.4 (Ϯ 0.7) units/mg at 25°C, comparable with the specific activity of V 1 -ATPase (5.2 units/mg; Ref. 17). Finally, more than 10 mg of V o V 1 was routinely obtained from 100 g of wet cells using this improved purification procedure.
SDS-PAGE of the purified enzyme showed eight polypeptides with apparent molecular sizes of 66, 56, 38, 30, 24, 13, 12, and 9 kDa (Fig. 1b), four of which (66, 56, 30, and 12 kDa) were subunits of V 1 . After a short lag because of an initially inhibited form of V o V 1 (17), an apparent H ϩ pumping activity was observed (Fig. 1c). The lower rate of quenching by V o V 1 than by F o F 1 may be a reflection of ADP inhibition of V o V 1 (17). V o V 1 proteoliposomes reconstituted by dialysis or the freeze-thaw method (26) failed to show H ϩ pumping activity (date not shown). The V o V 1 preparation was also capable of ATP synthesis; vesicles containing bR and V o V 1 synthesized ATP by illumination (Fig. 1d). Neither bafilomycine A1 nor azide inhibitors of eukaryotic V o V 1 (1,27) and F 1 -ATPase (28), respectively, affected the illumination-driven ATP synthesis of T. thermophilus V o V 1 (Fig. 1d).
atp Operon Contains Nine Genes-We cloned the entire operon of the V o V 1 , which contained nine genes in the order of atpG-I-L-E-X-F-A-B-D (Fig. 2). The sizes of the G-D genes were 360, 1188, 297, 561, 969, 324, 1749, 1434, and 681 base pairs, and the approximate molecular sizes of the encoded proteins are 13, 43, 10, 20, 35, 12, 64, 53, and 25 kDa, respectively. All of the nine genes have Shine-Dalgaerno sequences in their upstream regions (29). At 17-50 bases downstream from the last codon of atpD, there is transcriptional termination signal (5Ј-GCCCGGGGGGTTCAGGCCCCCCGGGCTTTTTCTTT-3Ј), suggesting that atpD is the last gene in the operon. At 71-32 bases upstream from the initiation codon of atpG, there is a consensus sequence for a promoter of the T. thermophilus atp operon (5Ј-TTGACCTGCATCCTCCGCCGCCTAAGTATACTT-AGGCGGGG-3Ј, consensus sequences and a probable transcriptional start point being underlined). Therefore, the operon contains no gene(s) other than the nine genes.
atpI Encodes a Hydrophobic 43-kDa Subunit-The atpI gene is predicted to encode a 43-kDa polypeptide, but we could not find a protein band at the corresponding position in 18% SDS-PAGE (Fig. 1b). Therefore, we prepared an anti-atpI peptide antibody raised against the deduced peptide sequence 359 GHM-LQPIRLLWVEFFTKF 376 as an antigen and analyzed V o V 1 with Western blotting. A 60-kDa polypeptide was stained with the antibody (data not shown). The position of the stained band overlapped with that of the A subunit, but the atpI product is not a component of V 1 because there was no positive band in the V 1 preparation. In fact, a new band appeared just below the band of the A subunit when V o V 1 was analyzed with SDS-PAGE at a lower (12%) gel concentration (Fig. 3a, arrow). The N-terminal amino acid sequence of this polypeptide was VIAP-MEKLVLAGPKG, which agreed with the predicted N-terminal sequence of the atpI product. Thus, the atpI product was contained in the purified V o V 1 preparation, and we call it the 43-kDa subunit hereafter. The C-terminal ϳ60-amino acid stretch of the 43-kDa subunit has sequence similarity to those of rat VPP1 (96 kDa; Ref. 30 (Fig. 3b). The hydropathy plot of the 43-kDa subunit indicated that this stretch contained a large hydrophobic region that is commonly observed for the homologues mentioned above (Fig. 3c).
atpL Encodes the 8-kDa Proteolipid Subunit-atpL encodes a highly hydrophobic 10-kDa polypeptide with three apparent transmembrane domains (Fig. 4b), and its sequence shows significant sequence similarity to the sequences of the proteolipid subunits of F o F 1 and V o V 1 (Fig. 4a). The glutamic acid residue that plays an essential role in H ϩ conduction (2) is conserved at position 57 in the predicted transmembrane domain. Previously, no protein band of the corresponding molecular size was found in SDS-PAGE of V o V 1 (Fig. 4c, lane 2) (18). However, a 5-min boiling prior to electrophoresis in 2% SDS resulted in appearance of a new 9-kDa polypeptide band with concomitant disappearance of the 100-kDa band, which was previously interpreted as a candidate for a subunit of V o (18) (Fig. 4c, shown by an asterisk). Analysis of the N-terminal sequence of the 9-kDa band (and also the 100-kDa band) revealed that the protein in the band was a product of atpL but is also shown for reference. d, effect of various inhibitors on ATP synthesis. bR-V o V 1 liposomes were preincubated at 25°C for 10 min in the presence of inhibitors before preillumination. The ATP synthesis of bR-F o F 1 liposomes is also shown for reference. Activity of ATP synthesis was expressed as amounts of synthesized ATP/mg of ATP synthase. Other experimental details are described under "Experimental Procedures". lacked the N-terminal 19 amino acid residues that constituted the putative first transmembrane domain (Fig. 4b). Consequently, the proteolipid subunit of T. thermophilus V o V 1 contains two rather than three transmembrane helices. The real molecular size of the subunit is approximately 8 kDa, and we therefore call it the 8-kDa subunit hereafter.
The Operon Contains All Genes for Nine Subunits of V o V 1 -It was already known that atpA and atpB encode the A and B subunits of V 1 part (22). To identify the remaining five gene products, we compared the partial peptide sequences of the putative subunit bands in SDS-PAGE and deduced amino acid sequences of the genes (data not shown). The results showed that the products of atpF and atpD are ␦ and ␥ subunits of V 1 part, respectively, and that the products of atpG, atpE, and atpX are 13-, 24-, and 38-kDa subunits of V o part, respectively. The names 24-and 38-kDa subunits are changed to 20-and 35-kDa subunit hereafter. These V o subunits are hydrophilic proteins rather than transmembrane protein; hence these subunits may be associated directly or indirectly with the hydrophobic subunits of V o . Thus, the products of all of nine genes in the operon are found in the V o V 1 preparation, and the V o V 1 is composed of nine kinds of subunits, four in V 1 2. Correspondence between gene products of atp operon and subunits of V o V 1. The letters G, I, L, F, X, F, A, B, and D represent open reading frames in the atp operon. Correspondence between gene products of atp operon and subunits of V o V 1 is indicated by arrows. was reconstituted from isolated A and B subunits (each Ͼ30 mg/ml) in the presence of 2 mM ATP-Mg and separated from free A and B subunits by gel permeation HPLC (Fig. 5a). The reconstitution was dependent on the presence of ATP-Mg (Ͼ0.1 mM) and very high concentrations of isolated A and B subunits. Although the isolated A 3 B 3 complex at a practical concentration (ϳ1 mg/ml) dissociated to the individual subunits when exposed to ATP-Mg or ADP-Mg, it was stable for at least 1 day at room temperature in the absence of nucleotide. Another subcomplex, A 3 B 3 ␥, was purified from subunit-expressing E. coli. The purified A 3 B 3 ␥ migrated as a single protein band in a nondenaturing PAGE with little contamination (data not shown) and was confirmed by SDS-PAGE to contain three subunits (Fig. 5c). This complex was as stable as V 1 and showed ATPase activity of 1.5 units/mg at 25°C in the presence of an ATP regenerating system. The reason for the lower activity of recombinant A 3 B 3 ␥ than V 1 (A 3 B 3 ␥␦) is not clear at present. Further studies will be necessary to clarify the characteristics of the ␦ subunit.
Electron micrographs of the negatively stained A 3 B 3 , A 3 B 3 ␥, and A 3 B 3 ␥␦ (V 1 ) are shown in Fig. 6 (a-c), respectively. A small central cavity filled with stain can be seen in many particles of the A 3 B 3 complex. In contrast, no such cavity was found in the centers of the particles of the A 3 B 3 ␥ and A 3 B 3 ␥␦ complexes. Averaged top views of the particles are shown in Fig. 6 (d-f). The central cavity of A 3 B 3 can be identified more easily after averaging the particles.
To conform the assignment of ␥-subunit, we examined the statistical significance of the difference among the averaged A 3 B 3 , A 3 B 3 ␥, and V 1 by analysis of variance and Student's t test. Fig. 6g shows the results obtained by analysis of variance, which indicates that the central masses significantly differ between A 3 B 3 , A 3 B 3 ␥, and V 1 . Fig. 6h shows that the difference is due to the central pore of A 3 B 3 when A 3 B 3 is compared with A 3 B 3 ␥ by Student's t test. Both methods show that the central mass is lost in an A 3 B 3 complex with the confidence level of 99.7% when it is compared with V 1 and/or A 3 B 3 ␥. The findings indicate that the central part of A 3 B 3 ␥ and V 1 correspond to the ␥ subunit.  the reconstitution of V o V 1 into liposome was carried out by reverse phase method (19 -21). As shown in Fig. 1c, the ATPdriven H ϩ pumping by V o V 1 was clearly observed. In a previous study, we demonstrated that the ATPase activity of V o V 1 was rapidly decreased during turnover because of ADP inhibition (17). The lower rate of H ϩ pumping by V o V 1 than by F o F 1 may be a reflection of ADP inhibition of V o V 1 . On the other hand, the bR-V o V 1 -co-reconstituted liposomes synthesized ATP driven by proton motive force (Fig. 1d), and the rate of ATP synthesis was higher than that of F o F 1 . Thus, it is concluded that the purified V o V 1 is an ATP synthase/H ϩ pump. This is the first report showing that V o V 1 is capable of both ATP synthesis driven by proton motive force and proton translocation across membranes driven by ATP hydrolysis. Bafilomycin A1, a specific inhibitor of eukaryotic V o V 1 that does not affect the activity of prokaryotic V o V 1 (1,(13)(14)(15)(16), did not show any inhibition of the ATP synthesis reaction catalyzed by V o V 1 . Azide, which inhibits the ATPase activity of F o F 1 because of the entrapment of the inhibitory ADP in the catalytic site, does not inhibit the ATP synthesis reaction of either F o F 1 (28) or V o V 1 (Fig. 1d). In this respect, the V o V 1 ATP synthase/H ϩ pump of T. thermophilus is more similar to an F o F 1 ATP synthase than to eukaryotic H ϩ pump V o V 1 .
atp Operon Encodes All Subunits of the V o V 1 -Comparison of peptide sequence and deduced amino acid sequence of the genes in the atp operon enabled us to conclude that T. thermophilus V o V 1 is composed of nine kinds of subunits. The putative 100-kDa subunit turned out to be an aggregated form of the 8-kDa proteolipid subunit, and a previously undetected hydrophobic 43-kDa subunit was identified in the protein band migrating with the A subunit in SDS-PAGE. The V o V 1 operon of T. thermophilus shares the commonly observed feature of V o V 1 and F o F 1 operons in various prokaryotes that genes of the membrane part (V o or F o ) precede those of the soluble part (V 1 or F 1 ).
V o Contains Two Transmembrane Subunits-T. thermophilus V o V 1 contains two kinds of transmembrane subunits, a 43-kDa subunit and an 8-kDa subunit. This is in contrast to F o F 1 , which contains three kinds of transmembrane subunits. The 43-kDa subunit is so far the smallest member of the 100-kDa subunit family of V o V 1 , and its C-terminal hydrophobic stretch probably represents an indispensable part for all V o V 1 . The 8-kDa proteolipid subunit of T. thermophilus V o V 1 is also unique. Although the proteolipid subunits of F o F 1 have molecular sizes of 6 -9 kDa and two transmembrane helices (2,3), the corresponding subunits of V o V 1 are double sized, ϳ17 kDa, and have four transmembrane helices (1). Sequence similarity between the N-and C-terminal halves suggests that they have arisen from gene duplication of the two-helix proteolipid (41). An exception is the V o V 1 from S. acidocaldarius, whose proteolipid subunit has two transmembrane helices after the loss of a hydrophobic presequence (39,42). Similarly, although T. thermophilus atpL encodes a 10-kDa protein with three predicted transmembrane helices, the mature subunit did not have the presequence corresponding to the N-terminal transmembrane helix. Therefore, the proteolipid subunit of T. thermophilus belongs to the two-helix class.
The ␥ Subunit May Be the Central Subunit-Because A 3 B 3 in T. thermophilus V 1 should form a stator barrel and the subunit composition of V 1 is A 3 B 3 ␥␦, the candidates for the central rotor shaft subunit should be either the ␥ or ␦ subunit. Our electron microscopic study indicated that the central mass in V 1 is mainly composed of the ␥ subunit. The predicted secondary structure of ␥ subunit reveals the existence of several long ␣ helices, which are prerequisite for the shaft structure (data not shown). It is also worth mentioning that the features of A 3 B 3 and A 3 B 3 ␥ are very similar to those of the ␣ 3 ␤ 3 and ␣ 3 ␤ 3 ␥ complexes of F 1 -ATPase from a thermophilic Bacillus PS3. The ␣ 3 ␤ 3 complex is unstable and tends to dissociate when exposed to nucleotide. The ␣ 3 ␤ 3 ␥ complex is as stable as F 1 -ATPase and FIG. 5. Elution profiles of A 3 B 3 , A 3 B 3 ␥, and A 3 B 3 ␥␦ (V 1 ) complexes from gel permeation HPLC. The isolated A and B subunits were mixed with 5 mM ATP, and the mixture was fractionated by gel permeation HPLC (a), and the purity was then checked by rechromatography (b). c, analysis of A 3 B 3 , A 3 B 3 ␥, and A 3 B 3 ␥␦ (V 1 ) by 15% SDS-PAGE. Each subcomplex was purified by gel permeation HPLC before electrophoresis. Other experimental details are under "Experimental Procedures." FIG. 6. Electron micrographs of the A 3 B 3 , A 3 B 3 ␥, and A 3 B 3 ␥␦ (V 1 ). a-c, overviews of negatively stained A 3 B 3 , A 3 B 3 ␥, and A 3 B 3 ␥␦, respectively, on the carbon support films. Proteins are shown in white. Many particles are observed in an end-on orientation indicated by arrowheads. The scale bar represents 50 nm. d-f, the averaged top views of A 3 B 3 (n ϭ 24), A 3 B 3 ␥ (n ϭ 23), and A 3 B 3 ␥␦ (n ϭ 22) are shown in d, e, and f, respectively. Proteins are shown in white. The scale bar for d-f represents 5 nm. g, difference of density among A 3 B 3 , A 3 B 3 ␥, and V 1 by analysis of variance (25). Significant difference of density among them is shown in white. h, difference of density between A 3 B 3 and A 3 B 3 ␥ by Student's t test. Significantly lower density regions are shown in white.
can be expressed in E. coli. Electron microscopic images of the ␣ 3 ␤ 3 have a central cavity, but those of the ␣ 3 ␤ 3 ␥ do not (24). Taken together, the findings appear to indicate that the ␥ subunit of V o V 1 is a counterpart of the ␥ subunit of F 1 -ATPase. It may be possible to extend the conclusion to the ␥ subunit homologues of other organisms.