Evidence for the Presence of an F-Type ATP Synthase Involved in Sulfate Respiration in Desulfovibrio vulgaris

Abstract

Using a library of genomic DNA from Desulfovibrio vulgaris Miyazaki F, a strict anaerobe, and two synthetic deoxyoligonucleotide probes designed for F-type ATPases, the genes for open reading frames (ORFs) 1 to 5 were cloned and sequenced. The predicted protein sequences of the gene products indicate that they are composed of 172, 488, 294, 471, and 134 amino acids, respectively, and that they share considerable identity at the amino acid level with δ, α, γ, β, and ɛ subunits found in other F-type ATPases, respectively. Furthermore, a component carrying ATPase activity was partially purified from the cytoplasmic membrane fraction of the D. vulgaris Miyazaki F cells. The N-terminal amino acid sequences of three major polypeptides separated by sodium dodecyl sulfate–12% polyacrylamide gel electrophoresis were identical to those of the products predicted by the sequences of ORF-2, ORF-3, and ORF-4, suggesting that an F-type ATPase is functioning in the D. vulgaris Miyazaki F cytoplasmic membrane. The amount of the F-type ATPase produced in the D. vulgaris Miyazaki F cells is similar to that in the Escherichia coli cells cultured aerobically. It indicates that the enzyme works as an ATP synthase in the D. vulgaris Miyazaki F cells in connection with sulfate respiration.

The ATPases have been classified into three major groups, the F, V, and P types (25). F-type ATPases are the most common H⁺-translocating ATP synthases. They consist of two parts, namely F_o, a membrane-embedded portion, and F₁, a soluble domain. The F_o portion generally serves as a proton channel, but in some cases it may also serve for the transport of Na⁺ (10). The F₁ portion is composed of five subunits, i.e., α, β, γ, δ, and ɛ, with stoichiometry of 3:3:1:1:1. The genes encoding these proteins form a cluster, not only in aerobic and anaerobic eubacteria but also in archaebacteria (30). F-type ATPases are located in the membranes of bacteria, chloroplasts, and mitochondria and catalyze the hydrolysis or synthesis of ATP coupling with H⁺ (or Na⁺) transport across a membrane. It is accepted that aerobic organisms use the enzyme mainly for synthesizing ATP; e.g., the enzyme functions as an ATP synthase. In contrast, this is not yet well established for strict anaerobes, which are unable to utilize molecular oxygen as a terminal electron acceptor. It has been anticipated that F-type ATPases usually hydrolyze ATP to pump out H⁺ in strict anaerobes. However, Propionigenium modestum (11) and Acetobacterium woodii (27), strictly anaerobic bacteria, have Na⁺-translocating ATP synthases, which synthesize ATP. The generation of an Na⁺ gradient in P. modestum is coupled to a decarboxylation reaction by Na⁺-translocating membrane-bound methylmalonyl-coenzyme A (CoA) decarboxylase (11). Moorella thermoacetica (the former name was Clostridium thermoaceticum [5]), which is also an obligatory anaerobic bacterium, was indicated to have a H⁺-translocating ATP synthase (13). This is thought to synthesize ATP by using the H⁺ gradient generated by an anaerobic electron transport that involves electron carriers, such as cytochromes b₅₅₄ and b₅₅₉, menaquinone, rubredoxin, ferredoxin, and a flavoprotein, although the ultimate electron acceptor of the electron transport chain has not been identified yet.

Gram-negative eubacterium Desulfovibrio vulgaris Miyazaki F is a strict anaerobe that uses sulfate as a terminal electron acceptor. While electron transport proteins involved in the energy transduction in the sulfate-reducing bacteria have been extensively investigated (24), little is known about the ATPase (4, 18). Although D. vulgaris produces 2 mol of ATP through the oxidation of 2 mol of lactate to acetic acid, they are used out for the formation of adenosine 5′-phosphosulfate during the reduction of a sulfate ion (20). Thus, for living activity, the bacterium should have other ATP-generating systems. One of the possible sources is an energy conversion system coupled to the electron transport. Odom and Peck proposed a chemiosmotic hydrogen cyclic model as a general mechanism for energy coupling in Desulfovibrio species (21). It predicts generation of the proton gradient through the oxidation of hydrogen molecules coupled to the sulfate reduction. Actually, an increase of the proton concentration in the bulk phase was observed as a function of hydrogen consumption (4, 17). However, there has been no report on the identification of a H⁺-translocating ATP synthase in Desulfovibrio. This enzyme has a twofold importance in Desulfovibrio, namely as a key enzyme in the energy conversion system and as a milestone in the evolution of the biological energy conversion systems. In this study, we report the cloning and sequencing of the genes of the α, β, γ, δ, and ɛ subunits of F₁-ATPase and a partial purification of the enzyme, which confirm the presence of F-type ATPase in D. vulgaris Miyazaki F for the first time.

MATERIALS AND METHODS

Materials and bacterial strains.

The pUC118 vector, restriction enzymes, modifying enzymes, and Ex Taq polymerase were purchased from Takara Shuzo Co., Ltd. (Kusatsu, Japan). Radiochemicals [α-³⁵S]dCTP (specific radioactivity, 400 Ci mmol⁻¹, 10 mCi ml⁻¹) and [γ-³²P]dATP (3,000 Ci mmol⁻¹, 10 mCi ml⁻¹) were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden) and were used for dideoxynucleotide sequencing and 5′-end labeling, respectively. Synthetic oligonucleotides, Hybond-N⁺ filters, Thermo Sequenase cycle sequencing kits, a MonoQ column (5-mm inner diameter by 5 cm), and phenyl Sepharose resin were also obtained from Amersham Pharmacia Biotech. Deoxyribonuclease I (DNase I) and phenylmethylsulfonyl fluoride (PMSF) were from Sigma Chemical Co. (St. Louis, Mo.). The vector λEMBL3, Gigapack II gold packaging extracts, Escherichia coli JM109, and XL1-Blue MRA(P2) were obtained from Toyobo Co., Ltd. (Osaka, Japan). All other reagent-grade chemicals and antibiotics were obtained from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). D. vulgaris Miyazaki F cells were grown in the Postgate C medium (26) at 37°C.

Construction of a genomic library of D. vulgaris Miyazaki F.

D. vulgaris Miyazaki F cells were lysed with 0.5% (wt/vol) sodium dodecyl sulfate (SDS) and treated with 100 μg of proteinase K per ml. After cell wall debris, polysaccharides, and remaining proteins were removed by selective precipitation with CTAB (cethyltrimethylammonium bromide) (19), the genomic DNA was recovered from the resulting supernatant by isopropanol precipitation. Two deoxyoligonucleotide primers for PCR were designed according to the known amino acid sequences of the β subunits of F₁-ATPases (F₁β). Primer 1 was a 23-mer with the nucleotide sequence 5′-GGCGGCGCGGGCGTGGGCAAGAC-3′ (corresponding to the amino acid sequence of G-G-A-G-V-G-K-T-V in the so-called P-loop region), while primer 2 was a 27-mer with the nucleotide sequence 5′-GTCGGTAAGGTCGTCCGCGGGCACGTA-3′. This is complementary to the base sequence corresponding to the amino acid sequence of Y-V-P-A-D-D-L-T-D in the central domain. The known codon preference caused by the high GC content (65%) of D. vulgaris DNA (26) was taken into account in the design of these deoxyoligonucleotides. About 500-bp DNA fragments were amplified from the D. vulgaris Miyazaki F genomic DNA by PCR. The amplified 500-bp DNA fragments were phosphorylated with T4 polynucleotide kinase and ligated with pUC19 vector previously cut with SmaI. The resultant plasmid was sequenced completely by using two universal primers, 5′-CAGGAAACAGCTATGAC-3′ and 5′-GTTTTCCCAGTCACGAC-3′. On the basis of the DNA sequence obtained, four kinds of new sequencing primers (primers 3, 4, 5, and 6) for the F₁β gene (atpD) were synthesized. Their sequences are 5′-CCTTGGTGGCCTGCAGGAACGCATC-3′ (25-mer), 5′-GATGCGTTCCTGCAGGCCACCAAGG-3′ (25-mer), 5′-GGTGTTGGCGAGCGTACCCG-3′ (20-mer), and 5′-CGGGTACGCTCGCCAACACC-3′ (20-mer), respectively. These oligonucleotides were also used for the hybridization experiments described below.

For the preparation of genomic library, the genomic DNA was partially digested with Sau3AI. The digested fragments were then subjected to size fractionation via isopycnic centrifugation at 140,000 × g for 24 h by using a 5 mM EDTA–20 mM Tris-HCl buffer (pH 8.0) containing 40% (wt/vol) sucrose. The fractions containing 15 to 20 kbp of DNA fragments were diluted threefold with a 1 mM EDTA–10 mM Tris-HCl buffer (pH 8.0) and collected by ethanol precipitation. These 15- to 20-kbp DNA fragments were ligated into the vector λEMBL3, which had been cut with BamHI previously. The ligated DNA was then reconstituted into λ phage by using Gigapack II gold packaging extracts and were transfected to E. coli XL1-Blue MRA(P2). This strain allows the transfection of only recombinant phages by Spi/P2 selection (12). The obtained λ phages with the D. vulgaris Miyazaki F genomic library were amplified to approximately 10⁸ PFU as stocks.

Cloning and dideoxy sequencing.

The plaques carrying 15- to 20-kbp DNA fragments of the D. vulgaris Miyazaki F genomic library were blotted onto Hybond-N⁺. It was found by autoradiography that 34 per 1,000 plaques hybridize with primer-3 and primer-5. A phage stock was prepared from a positive plaque, and the hybridization with the primers was reconfirmed. Then, the phage stock was amplified to approximately 10⁸ PFU.

The λ phage DNA was isolated from the phage lysate by the Qiagen lambda starter kit (Funakoshi, Tokyo, Japan). The purified λ DNA (λMK27, hereafter) contained a 17-kbp insert. It was digested with either SalI, PstI, or both PstI and SphI, and then separated electrophoretically in a 0.7% agarose gel. The separated fragments were subcloned into the same site of pUC118 vectors. These recombinant plasmids were designated pMK1, -2, -3, and -4, respectively. The pMK1 and pMK2 plasmids contained SalI inserts of 4,700 and 750 bp, respectively, and pMK3 and pMK4 contained a PstI insert of 3,000 bp and a PstI-SphI insert of 1,200 bp, respectively. The pMK1 vector was then digested with both SphI and SalI, and the resultant SphI-SalI fragment (900 bp) was resubcloned into pUC118. This pMK1 derivative vector was designated pMK11. Furthermore, the SalI insert of 4,700 bp in pMK1 was digested with Sau3AI and the resultant 660-bp fragment was resubcloned into a BamHI site of pUC118. This pMK1 derivative vector was designated pMK12. Deletion mutants of the pMK1 insert in different lengths were also obtained by using exonuclease III and Mung bean nuclease. The pMK2 vector was digested with PstI, or with both PstI and SalI, and the resultant three fragments were individually subcloned into pUC118, yielding pMK21 (100-bp insert), pMK22 (250-bp insert), and pMK23 (400-bp insert). All plasmids mentioned above were subjected to restriction mapping. The coding regions of these plasmid inserts were sequenced on both strands, and each base was completely determined by the dideoxy chain termination method (28). The plasmids used in this work are summarized in Table 1.

TABLE 1.

The genomic DNAs of phages and plasmids used in this work

Vector	Relevant characteristic	Reference
λ-phage DNA
λEMBL3		12
λMK27	17-kbp insert containing atp genes	This work
Plasmids
pUC118	Ampr	33
pMK1	4,700-bp SalI insert in pUC118	This work
pMK11	900-bp SalI-SphI insert in pUC118	This work
pMK12	660-bp Sau3AI insert in pUC118	This work
pMK2	750-bp SalI insert in pUC118	This work
pMK21	100-bp PstI insert in pUC118	This work
pMK22	250-bp SalI-PstI insert in pUC118	This work
pMK23	400-bp PstI insert in pUC118	This work
pMK3	3,000-bp PstI insert in pUC118	This work
pMK4	1,200-bp PstI-SphI insert in pUC118	This work

Purification of F₁-ATPase from D. vulgaris Miyazaki F.

Preparation of cytoplasmic membrane was previously described (23). A 3-ml suspension of cytoplasmic membrane (15 mg of protein/ml) was mixed with an equal volume of chloroform. This mixture was vigorously shaken for 2 min and centrifuged in six Eppendorf tubes at 15,000 rpm at 4°C for 15 min by using a Tomy MRX-150 microcentrifuge. The clear supernatant was diluted 10 times in ultrapure water and applied on a MonoQ column (5-mm inner diameter by 5 cm) equilibrated with a 1 mM EDTA–20 mM Tris-HCl (pH 8.0) solution (buffer A). Fractions containing ATPase activity were obtained by elution of buffer A with a linear NaCl gradient (0 to 0.5 M). Solid ammonium sulfate was added to the collected fraction. The final ammonium sulfate concentration was 30% (wt/wt). After centrifugation for 30 min at 15,000 rpm, the supernatant was subjected to phenyl Sepharose column chromatography (5-mm inner diameter by 5 cm). The column was preequilibrated with buffer B (3 mM EDTA, 10 mM Tris-HCl, pH 7.4) containing 1 M (NH₄)₂SO₄. The fraction containing the ATPase activity was obtained by elution of buffer B with a reverse linear (NH₄)₂SO₄ gradient (1 to 0 M).

Other methods.

N-terminal amino acid sequence analysis was performed by using a Shimadzu PPSQ-10 protein sequencer equipped with a Shimadzu C-R7A analyzer (Kyoto, Japan) after transferring a protein from a sodium dodecyl sulfate–12% polyacrylamide gel electrophoresis (SDS–12% PAGE) gel onto a nylon filter. Protein concentration was determined by the bicinchronic acid method (29), using bovine serum albumin as a standard (Pierce Chemical Company). The ATPase activity was measured at 25°C in the presence of an ATP-regenerating system (32). The putative genes were originally identified as open reading frames (ORFs) by a software, Genetyx-Mac version 9.0 (Software Development Co., Ltd.). Then, a function was inferred from DNA homology matching. Homologies were determined by utilizing the BLAST network service (2) from National Center for Biotechnology Information.

Nucleotide sequence accession number.

The nucleotide sequence data reported in this paper will appear in the DDBJ, EMBL, and GenBank nucleotide sequence databases with the accession no. AB022018.

RESULTS

Cloning and DNA sequencing of the genes of ORFs 1 to 5.

A region in the 17-kbp insert of λMK27 genomic DNA was subcloned into a pUC118 vector, and the nucleotide sequence was determined. Five ORFs that were preceded by possible Shine-Dalgarno sequences were found in it. However, no possible promoter and terminator regions could be found. Therefore, these five ORFs seemed to be transcribed by the same promoter, located upstream of this DNA fragment.

ORF-4 comprises 1,416 nucleotides, which begins with an ATG initiation codon at nucleotide 3,078 from the 5′ end of the region shown in Fig. 1. The sequences similar to the primers initially designed for PCR were also in it. The derived amino acid sequence (included in Fig. 1 as well) composed of 471 amino acid residues is highly homologous with those of F₁-ATPase β subunits (Table 2). Its size is also reasonable in comparison with those from other bacterial species. The crystal structure of the F₁-ATPase from the bovine mitochondrion has shown that each of β and α subunits consists of three domains; N-terminal, central, and C-terminal domains (1). The corresponding regions of the ORF-4 product from D. vulgaris Miyazaki F are the region of 1 to 82 (N-terminal domain), 83 to 357 (central domain), and 358 to 471 (C-terminal domain) in terms of amino acid sequence. In the central domain, the P loop, which interacts with the phosphate region of a bound nucleotide, and the region around Glu-188 are highly homologous with those of other β subunits. The side chain of Glu-188 is assumed to catalyze ATP hydrolysis (1). Now, we can conclude that the ORF-4 product is the F₁-ATPase β subunit.

FIG. 1.

Nucleotide sequences of the atpH, atpA, atpG, atpD, and atpC genes from D. vulgaris Miyazaki F and deduced amino acid sequences of the five atp gene products. Nucleotides are numbered from the 5′ end of the 4,950-bp region presented. The putative ribosome binding sites from the five atp genes are doubly underlined. The P-loop amino acid sequence of the β-subunit is shaded.

TABLE 2.

The amino acid sequence homologies between those encoded by D. vulgaris Miyazaki F atp genes and those of other bacteria

Gene	Protein characteristics		% Identity (% similarity) with gene product ofa:
	Subunit	Size (aa)	A. aeolicus	M. thermoacetica	E. coli	PS3	B. stearothermophilus	P. modestum
atpA	α	488	65 (81)	64 (80)	54 (70)	64 (82)	64 (82)	63 (82)
atpD	β	471b	64 (77)	72 (83)	68 (79)	70 (83)	68 (80)	68 (78)
atpG	γ	294b	38 (63)	42 (61)	40 (65)	44 (65)	45 (66)	41 (65)
atpH	δ	172	24 (44)	28 (54)	31 (61)	22 (48)	24 (50)	28 (54)
atpC	ɛ	134	34 (51)	46 (68)	38 (61)	46 (63)	46 (63)	37 (59)

^aReferences of organisms are as follows: Aquifex aeolicus (9), M. thermoacetica (7), E. coli (35), thermophilic bacterium PS3 (22), Bacillus stearothermophilus (accession no. D38060 in the GenBank database), and P. modestum (accession no. X58461 in the EMBL database). 

^bThe initial methionine was removed in the gene products. 

ORF-2 comprises 1,467 nucleotides, which begins with an ATG initiation codon at nucleotide 654 from the 5′ end. The derived amino acid sequence (included in Fig. 1) composed of 488 amino acid residues is highly homologous with those of F₁-ATPase α subunits (Table 2) and is a reasonable size compared with those from other bacterial species. A sequence corresponding to the P loop (G-D-R-Q-T-G-K-T starting from nucleotide 1,158) is also present. Consequently, the ORF-2 product must be the F₁-ATPase α subunit.

ORF-3 comprises 885 nucleotides, which begins with an ATG initiation codon at nucleotide 2,176 from the 5′ end. The derived amino acid sequence (included in Fig. 1) composed of 294 amino acid residues shows certain homology with those of F₁-ATPase γ subunits (Table 2). The relatively low homology is a general feature among the F₁-ATPase γ subunits from various bacterial species. ORF-3 has little homology with other subunits of the F₁-ATPases of other bacterial species. Furthermore, the size of the ORF-3 product is reasonable in comparison with γ subunits from other bacterial species. Therefore, the ORF-3 product should be the F₁-ATPase γ subunit.

ORFs 1 and 5 comprise 519 and 405 nucleotides, respectively. The ORF-1 begins with a GTG initiation codon at 131, and the derived amino acid sequence (included in Fig. 1) composed of 172 amino acid residues is a little homologous with those of F₁-ATPase δ subunits of various species (Table 2). ORF-5 begins with an ATG initiation codon at 4,504, and the derived amino acid sequence (included in Fig. 1) composed of 134 amino acid residues shows certain homology with those of F₁-ATPase ɛ subunits of various species. Furthermore, ORF-1 and ORF-5 have little homology with other subunits of other bacteria. The sizes of these products also are reasonable as the δ and ɛ subunits, respectively. Now, it can be concluded that the ORF-1 and ORF-5 products are the F₁-ATPase δ and ɛ subunits, respectively.

The organization of the genes mentioned above is presented in Fig. 2. The F_oF₁-ATPase genes are known to form a cluster in general (22). The gene arrangement in the order of the δ, α, γ, β, and ɛ subunit genes in D. vulgaris Miyazaki F is in accordance with those found for other species (34). This also supports the conclusion that D. vulgaris Miyazaki F has an F-type ATPase (or ATP synthase). Since the operon of F-type ATPase genes usually has F_o genes upstream of the F₁ genes, it is natural that the analyzed fragment does not include a promoter region. The GC contents of codon usage for atpH, atpA, atpG, atpD, and atpC genes were 62.2, 63.7, 63.1, 62.6, and 64.0%, respectively. The GC contents of a total of 4,950 bp sequenced in this study were 62.8%, in contrast to 61.7% for cytochrome c₃ (16) and 74.4% for flavin mononucleotide-binding protein (14), 66.2% for [NiFe] hydrogenase (8), and 62.1% for cytochrome c₅₅₃ (15) from D. vulgaris Miyazaki F. Thus, the codon preference should be similar to that of other genes of D. vulgaris Miyazaki F.

FIG. 2.

The organization of atp genes of D. vulgaris Miyazaki F. ORFs are represented by boxes. The numbers in the atpH to atpC boxes indicate those of the amino acid residues. The horizontal arrows with numbers indicate the fragments used in various subclones designated pMK1, pMK2, pMK3, and pMK4, respectively (see Table 1). The black box of λMK27 genomic DNA indicates the cloned 17-kbp fragment.

Purification of a component with ATPase activity from D. vulgaris Miyazaki F.

A component with ATPase activity that was purified from D. vulgaris Miyazaki F membranes was described under Materials and Methods. As shown in Table 3, the specific activity increased approximately 100-fold during the three steps of purification. SDS-PAGE of the fractions with ATPase activity showed the presence of three major bands corresponding to the molecular masses of 60.5, 51.5, and 34.7 kDa (Fig. 3). To identify these proteins, each of these bands was transferred electrophoretically from a SDS–12% PAGE gel onto a nylon filter, cut separately, and analyzed by a protein sequencer. The sequence of the N terminus of the largest one among the three bands was determined to be Met-Gln-Ile-Lys-Ala-Glu-Glu-Ile-Ser-Lys-…, which is identical to that of the expected product of the atpA gene. The N-terminal sequence of the second largest one was Ser-Ala-Asn-Ile-Gly-Lys-Ile-Val-Gln-Val-Ile-Gly-Ala-Val-Val-Asp-Val-Glu-Phe-Pro-…, which is identical to that of the expected product of the atpD gene, with the initial methionine residue cleaved. The sequence of the third one was Pro-Ser-Leu-Lys-Asp-Val-Lys-Val-Lys-Ile-Ala-Gly-Val-Lys-Lys-Thr-Lys-Gln-Ile-Thr-Lys-Ala-Met-Asn-Met-Val-Ala-…, which is identical to that of the expected product of the atpG gene, with the initial methionine residue cleaved. Therefore, it can be concluded that the F₁-ATPase genes are actually expressed in D. vulgaris Miyazaki F cell membranes under physiological conditions.

TABLE 3.

Purification of D. vulgaris Miyazaki F F₁-ATPase

Purification step	Specific activity (μmol/ min/mg protein)	Total activity (μmol/min)a
Cytoplasmic membranes	0.067	48.6
Chloroform treatment	1.04	17.1
MonoQ	2.66	11.7
Phenyl Sepharose	6.56	5.37

^aTotal activity per 100 g of D. vulgaris Miyazaki F wet cells was determined. 

FIG. 3.

SDS–12% PAGE analysis of the relevant fraction at each purification step. The fractions with ATPase activity were analyzed by SDS–12% PAGE and stained with Coomassie brilliant blue. The molecular mass markers in lanes 1 and 7 are composed of phosphorylase b (97,400 Da), bovine serum albumin (66,200 Da), ovalbumin (45,000 Da), carbonic anhydrase (31,000 Da), soybean trypsin inhibitor (21,500 Da), and lysozyme (14,000 Da). Lane 2, the cytoplasmic membrane fraction; lane 3, the supernatant after chloroform treatment; lane 4, after MonoQ elution; lane 5, after phenyl Sepharose elution; lane 6, β subunit purified from thermophilic bacterium PS 3.

DISCUSSION

The whole genes of the α, β, γ, δ, and ɛ subunits of F₁-ATPase of a sulfate-reducing bacterium have been cloned and sequenced for the first time in this work. Furthermore, the actual presence of the gene products in the D. vulgaris Miyazaki F membranes was also confirmed. Therefore, this should be present as F₁F_o-ATPase in the cell, although its actual components and their stoichiometry are not yet established. To understand the biological role of the ATPase gene products in D. vulgaris Miyazaki F cells, its specific activity was compared with that of E. coli membranes. E. coli JM109 was cultured aerobically overnight at 37°C. The membrane fraction of the E. coli cells was prepared in the same way as that of D. vulgaris Miyazaki F. The specific ATPase activity of the membrane fraction of E. coli was 0.078 μmol/min/mg protein. This is similar to the specific activity of the D. vulgaris Miyazaki F membrane fraction shown in Table 3. Therefore, D. vulgaris Miyazaki F should have a similar amount of F₁F_o-ATPase in the cytoplasmic membrane as E. coli does under aerobic conditions. F₁F_o-ATPase in E. coli works for energy production, namely for ATP synthesis under aerobic conditions. The presence of a large amount of ATPase in the D. vulgaris Miyazaki F cell strongly suggests that the gene product is actually the F-type ATP synthase and is working to synthesize ATP in the D. vulgaris Miyazaki F cell even under anaerobic conditions. The amount is too much for the F₁F_o-ATPase to work as a proton pump at expense of ATP in the neutral pH region.

The sulfate-reducing bacteria contain abundant cytochromes although they are strict anaerobes (20). These cytochromes should be involved in electron transport systems in sulfate respiration. Some sulfate-reducing bacteria can grow well by using molecular hydrogen as a sole energy source (3). Furthermore, an increase of the H⁺ concentration outside of the cytoplasmic membrane coupled with hydrogen consumption and sulfite reduction had been reported for D. vulgaris Miyazaki (17). This observation indicates generation of the H⁺ concentration gradient across the cytoplasmic membrane by the electron transport from hydrogen to sulfite. Therefore, it can be concluded that the products of the F₁-ATPase genes are involved in ATP synthesis, utilizing the H⁺ gradient generated by the electron transport system in the sulfate respiration. This is the first evidence showing the presence of the F-type ATP synthase associated with sulfate respiration. In 1981, Odom and Peck proposed a chemiosmotic hydrogen cycling model for energy coupling in Desulfovibrio species. This model has predicted that Desulfovibrio can produce ATP by oxidative phosphorylation and should have a H⁺-ATP synthase. Our conclusion is consistent with their model. However, the possibility for the presence of proton pumps in the electron transport system cannot be removed. The ATP synthesis coupled to nitrite respiration in Desulfovibrio gigas membrane vesicles was reported in connection with generation of the proton gradient (4). Our conclusion can explain this observation as well.

The role of F-type H⁺-ATP synthase in strict anaerobes is not yet well understood. It is inferred to work as an ATPase coupled with proton pumping in some strict anaerobes. On the other hand, there are some examples functioning as an ATP synthase. An Na⁺-translocating F₁F_o-ATPase in P. modestum is known to synthesize ATP, utilizing the Na⁺ gradient across the cytoplasmic membrane (11). A decarboxylation reaction is directly coupled with Na⁺ pumping in the methylmalonyl-CoA decarboxylase. The fermentation of succinate to propionate and CO₂ operates at a total free energy change of only −20 kJ/mol (10). This amount of energy is not sufficient to synthesize 1 mol of ATP from ADP and inorganic phosphate. The free energy expense for ATP synthesis under in vivo conditions is about 70 to 80 kJ/mol (31). Thus, three or four decarboxylation reactions have to couple with the synthesis of one ATP molecule. This was indicated as the major reason for the presence of ATP synthase in P. modestum (11). A H⁺-translocating F₁F_o-ATPase in M. thermoacetica was also reported to synthesize ATP, utilizing the H⁺ gradient generated by an electron transport chain which contains a menaquinone and two b-type cytochromes (13). The detailed mechanism and role of this energy conversion are not yet clear. It is also reported that the structure of the F_o domain of M. thermoautotrophica ATPase is different from that of the common F-type ATP synthase (5, 6).

In the case of sulfate-reducing bacteria, the oxidation of lactate to acetate results in the generation of 1 mol of ATP and 4 mol of electrons (or 2 mol of hydrogen molecules) at the substrate level. The ATP synthesized at the substrate level is used to activate the sulfate ion for reduction (20). The electrons are also used to reduce sulfate ion. The total reaction can be written as follows (26): CH₃CH(OH)COOH + (1/2)SO₄²⁻ = CH₃COOH + CO₂ +H₂O + (1/2)S²⁻ ΔG° = −94 kJ

Since the free energy released from this oxidation-reduction reaction is much larger than that of P. modestum, the energy conversion system in D. vulgaris Miyazaki F should be different from that in P. modestum. The free energy change of −94 kJ/mol is converted to ATP through electron transport coupled with the H⁺ gradient generation. The generation of H⁺ gradient is not directly coupled with the oxidation of organic compounds, in contrast to P. modestum. This process is a prototype of oxygen respiration.

There would be three types of current strict anaerobes in terms of the physiological role of F₁F_o-ATPase. The most primitive types use F₁F_o-ATPase as an ATP synthase without involvement of electron transport systems. The proton or Na⁺ gradient is used for efficient accumulation of small energy. The second type does not heavily rely on F₁F_o-ATP synthase for the energy production. ATP synthesis at the substrate level in the fermentation should be the major source of the energy in these anaerobes. In this case, F₁F_o-ATPase must be mainly used for the regulation of the cytoplasmic pH. The third type of strict anaerobes heavily relies on F₁F_o-ATP synthase in association with electron transport systems for the energy production. The energy conversion system in the third type of strict anaerobes can be placed between the fermentation and oxygen respiration from the evolutional point of view. The sulfate-reducing bacteria are one of the major groups of the third type.