Factor B and the Mitochondrial ATP Synthase Complex*

Factor B is a subunit of the mammalian ATP synthase complex, whose existence has been controversial. This paper describes the molecular and functional properties of a recombinant human factor B, which when added to bovine submitochondrial particles depleted of their factor B restores the energy coupling activity of the ATP synthase complexes. The mature human factor B has 175 amino acids and a molecular mass of 20,341 Da. The preparation is water-soluble, monomeric, and is inacti-vated by monothiol- and especially dithiol-modifying reagents, probably reacting at its cysteine residues Cys-92 and Cys-94. A likely factor B gene composed of 5 exons has been identified on chromosome 14q21.3, and the functional role of factor B in the mammalian ATP synthase complex has been discussed. It is generally considered that the mammalian mitochondrial ATP synthase complex is composed of 16 unlike subunits (1–3). These subunits are (cid:1) , (cid:2) , (cid:3) , (cid:4) , and (cid:5) in the catalytic F 1 domain; OSCP, a , b , c , d , e , f , g , F 6 and A6L in F O and stator; and the ATPase inhibitor protein, IF 1 , which binds reversibly to F 1 to inhibit ATP hydrolysis. In 1967, Sanadi and co-workers (4) showed that submitochondrial particles prepared by sonication from bovine mitochondria suspended in 0.25 M sucrose and 0.6 m M EDTA and adjusted to pH 9.0 with ammonium hydroxide lost considerable activity for respiration-driven ATP synthesis and ATP hydrolysis-driven electron transfer from succinate

It is generally considered that the mammalian mitochondrial ATP synthase complex is composed of 16 unlike subunits (1)(2)(3). These subunits are ␣, ␤, ␥, ␦, and ⑀ in the catalytic F 1 domain; OSCP, a, b, c, d, e, f, g, F 6 and A6L in F O and stator; and the ATPase inhibitor protein, IF 1 , which binds reversibly to F 1 to inhibit ATP hydrolysis. In 1967, Sanadi and co-workers (4) showed that submitochondrial particles prepared by sonication from bovine mitochondria suspended in 0.25 M sucrose and 0.6 mM EDTA and adjusted to pH 9.0 with ammonium hydroxide lost considerable activity for respiration-driven ATP synthesis and ATP hydrolysis-driven electron transfer from succinate to NAD. Addition to the ammonia-EDTA-treated particles (AE-SMP) 1 of a partially purified soluble protein extracted from mitochondrial acetone powders partially restored these activities (4,5). Sucrose density gradient centrifugation suggested a molecular mass of 32 kDa for the active peak of the soluble preparation, which was designated factor B (4,5). By using bovine mitochondrial acetone powder extracts, we isolated a pure and monodisperse protein, which restored ATP synthasecoupled activities to AE-SMP (6). Its molecular mass as estimated from sedimentation equilibrium and gel filtration experiments was 11-12 kDa, and it was immunoprecipitated by Sanadi's anti-factor B antiserum in an Ouchterlony double diffusion experiment (7). Sanadi and co-workers (5,8) revised the molecular mass of their preparation to 29.2 kDa and suggested that it is a dimer of monomer molecular mass of 14.6 kDa. They also obtained preparations of relative molecular mass of 13-15 and 47 kDa, which exhibited factor B-like activity (5,9,10), and in 1990 (11) they published the sequence of the 55 amino-terminal amino acids of a factor B preparation with a relative molecular mass of 22 kDa.
The existence of factor B as a component of the ATP synthase complex remained controversial, however. Although Sanadi (5,12) claimed that the ATP synthase complex prepared in his laboratory contained factor B, a thorough analysis of the polypeptide composition of an ATPase complex prepared in Walker's laboratory demonstrated the existence of the 16 unlike polypeptides mentioned above but no factor B (1,2). The multiplicity of Sanadi's factor B-like preparations with relative molecular masses ranging from 13-15 to 47 kDa, the small yield of our preparation, which precluded antibody production and chemical analyses, and the compelling extensive data of Walker's laboratory discouraged further interest in the pursuit of factor B.
More recently, two nucleotide sequences corresponding in part to Sanadi's 55-amino acid sequence of the factor B amino terminus were detected by our colleagues in the human genome. The shorter frame corresponding to 96 amino acid residues was expressed with a histidine tag. The expressed protein was found in inclusion bodies, and when extracted with 8 M urea and dialyzed the protein exhibited no factor B-like activity. The longer frame corresponding to 175 amino acid residues was expressed once with a histidine tag and a second time fused to thioredoxin. Both were recognized after purification by polyclonal antibodies raised to the shorter polypeptide. The histidine-tagged preparation was also inactive, but the other, after removal of thioredoxin, exhibited all the functional features described previously by Sanadi and co-workers (4,5) and by ourselves (6, 7) for factor B. This paper describes the molecular and functional properties of this human factor B preparation.

EXPERIMENTAL PROCEDURES
Materials-NAD, NADH, ATP, oligomycin, and 2,4-dithiothreitol were obtained from Calbiochem; DCCD, venturicidin, FCCP, tributyltin chloride, NEM, p-chloromercuribenzoate, sodium succinate, and 2-mercaptoethanol were from Sigma; DEAE-Sepharose was from Amersham Biosciences; polyacrylamide was from Bio-Rad; Tris was from ICN; and oxonol VI was from Molecular Probes. All other chemicals were reagent grade. AE-SMP were prepared from heavy beef heart mitochondria (13) according to Ref. 14. After a final wash with 0.25 M sucrose containing 10 mM Tris-HCl, pH 7.8, the particles were suspended in the same medium at 17 mg of protein/ml and stored at Ϫ80°C in small aliquots. F 1 -ATPase was prepared according to Senior and Brooks (15) and stored in 50% saturated ammonium sulfate at 4°C. The catalytic activity of the enzyme was in the range of 90 -100 mol of ATP hydrolyzed (min⅐mg) Ϫ1 . OSCP was prepared by Dr. A. Matsuno-Yagi as described previously (16).
Assays-ATP-driven reverse electron transfer from succinate to NAD was assayed at 38°C essentially according to Joshi and Sanadi (14), except that after 1-2 min of incubation of AE-SMP with factor B, succinate, ATP, MgCl 2 , KCN, and dithiothreitol, the reaction was started by the addition of NAD. The progress of the reaction was monitored at 340 nm in a Milton Roy 1201 spectrophotometer, and the rate of NAD reduction was calculated using an extinction coefficient of 6.22 mM Ϫ1 . Inactivation of factor B with thiol-modifying reagents was performed by incubating factor B for 5 min on ice in 50 mM Tris-sulfate, pH 7.8, with increasing concentration of CdSO 4 dissolved in water, or of phenylarsine oxide, p-chloromercuribenzoate, or NEM dissolved in dimethylformamide. The incubation mixtures containing CdSO 4 and phenylarsine oxide also contained 0.25 mM 2-mercaptoethanol. The inhibitor-treated samples of factor B were assayed for reconstitution of the ATP-driven reverse electron transfer activity of AE-SMP in the absence of dithiothreitol. The ATPase activities of F 1 and AE-SMP were measured spectrophotometrically by the coupled pyruvate kinase/lactate dehydrogenase method as before (17).
SDS-polyacrylamide gel electrophoresis and immunoblotting were performed as described elsewhere (18). The immune complexes were visualized using the SuperSignal West Pico chemiluminescence detection kit from Pierce. Protein concentration was determined, using the BCA kit from Pierce. Standard molecular biology techniques, including cloning, restriction enzyme analysis, plasmid preparation, and agarose gel electrophoresis were as described in Ref. 19. Amino-terminal analysis of purified recombinant human factor B was performed at the Protein and Nucleic Acid Core Facility of this institution. Tryptic peptide mass fingerprinting and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry were performed at the Scripps Center for Mass Spectrometry.
Preparation of Recombinant Human Factor B-The I.M.A.G.E. clone 2663774 containing EST AW173068 was purchased from Research Genetics. The sequence of the insert was confirmed by DNA sequence analysis. The forward primer P1, 5Ј-TTCTGGGGCTGGTTGAATGCA-3Ј, and reverse primers P2, 5Ј-TTACTTCAATTGTAATTTTAGTTC-3Ј, and P3, 5Ј-CTTCAATTGTAATTTTAGTTCCAG-3Ј, with and without termination codon, respectively, were designed to amplify the full length of mature factor B by PCR using Taq DNA polymerase (Qiagen). The amplicons were subcloned into pBAD/Thio-TOPO plasmid (Invitrogen) according to manufacturer's recommendations, and clones containing plasmids, designated pBAD-FB1 and pBAD-FB2, with the correct sequence of the inserts were selected after DNA sequencing. In both pBAD-FB plasmids, the factor B nucleotide sequence was located downstream of the nucleotide sequence coding for His-Patch thioredoxin under regulation of araBAD promoter.
In the present study, a single colony of TOP10 cells transformed with pBAD-FB1 plasmid from a fresh LB plate containing 100 g of ampicillin/ml was used to inoculate 5 ml of 2ϫ YT liquid medium in the presence of 100 g of ampicillin/ml and grown overnight in a shaker at 37°C. Then 0.5 liter of the 2ϫ YT liquid medium plus 100 g of ampicillin/ml were placed in a 2-liter flask and inoculated with 3 ml of the overnight culture and grown in a shaker (250 rpm) at 37°C until A 600 reached 0.5-0.6. The temperature in the shaker was shifted to 30°C, and the protein expression was induced by adding 20% arabinose to 0.02% final concentration. The incubation was continued for 4 h, and the culture was harvested by centrifugation and stored at Ϫ80°C until further processing.
The bacterial cells were disrupted by 4 cycles of freeze-thawing, followed by sonication on ice slurry, using a Branson sonifier. The soluble fraction was recovered by centrifugation for 1 h at 15,000 rpm in a Beckman SS-34 rotor, and affinity-purified on His-Bind resin (Novagen) preloaded with NiCl 2 . After washing with a pH 8.0 buffer containing 50 mM sodium phosphate, 0.3 M NaCl, and 20 mM imidazole, the fusion protein was eluted with 50 mM imidazole in the above buffer. The fractions containing the fusion protein were concentrated using the Centriprep YM-10 centrifugal filter devise (Millipore) and subjected to size-seiving chromatography on Ultrogel AcA 44 (column size 1.5 ϫ 90 cm) equilibrated in a pH 8.0 buffer containing 50 mM Tris-HCl, 0.1 M NaCl, and 5 mM 2-mercaptoethanol.
To remove the thioredoxin moiety, the recombinant protein was cleaved overnight at room temperature with recombinant enterokinase (Novagen) and purified by chromatography on DEAE-Sepharose Fast Flow, using a pH 8.0 buffer containing 50 mM Tris-HCl and 5 mM 2-mercaptoethanol. The fractions were assayed for restoration of ATPdriven reverse electron transfer from succinate to NAD as catalyzed by AE-SMP and combined based on activity and purity after analysis by SDS-PAGE. The protein was concentrated on YM-10 and stored in small aliquots at Ϫ80°C.
Preparation of cDNA for the Open Reading Frame of Human Factor B-To obtain a cDNA encoding the full-length open reading frame of human factor B, the oligonucleotide primers P1 and P2 (see above) as well as the forward primer P4, 5Ј-CCTGCTCTGTACTCGACCCG-3Ј, and the reverse primer P5, 5Ј-AAGAGTCGGTGGCGTCGAT-3Ј, were used to amplify by PCR the nucleotide sequences encoded by inserts in I.M.A.G.E. clones 2663774 (GenBank TM accession number AW173068) and 24431 (GenBank TM accession number AF052186) obtained from Research Genetics. The PCR amplifications were performed in 50-l volumes, each containing 5 l of High Fidelity PCR buffer, 2 mM MgSO 4 , 0.2 mM dNTPs, 0.2 M of each primer, ϳ10 ng of template, and 1 unit of PLATINUM Taq High Fidelity DNA polymerase (Invitrogen). The PCR conditions used are as follows: initial denaturation at 94°C for 2 min followed by 30 cycles each of 94°C for 30 s, 55°C for 1 min, 72°C for 1 min, and a final extension at 72°C for 7 min. In a second round, 5 l from each of the PCR mixtures was combined and subjected to amplification with primers P2 and P4 as above. The final PCR product was purified on agarose gel and subcloned into the pDrive vector (Qiagen). The nucleotide sequence of the insert was verified in both directions by ABI 327 sequencer using BigDye terminator sequencing.

Molecular Properties of Factor B-
The predicted amino acid sequence of human factor B is shown in Fig. 1. In the aminoterminal 55 residues, there are three differences between the mature human factor B shown in Fig. 1 and the sequence published by Sanadi and co-workers for the bovine protein (11). The human amino acid residues Tyr-14, Cys-33, and Glu-43 are, respectively, His, Gly, and Gln in the Sanadi partial sequence (11). 2 Fig. 2 shows on SDS-polyacrylamide gel the polypeptide pattern of Escherichia coli total cell lysate before (lane A) and after (lane B) induction of fused thioredoxin-factor B synthesis. Lane C is the pattern of the soluble proteins of the induced cells, and lane D is the purified thioredoxin-factor B fused preparation. Lane E shows two protein bands due to factor B (upper) and thioredoxin (lower) after treatment of the fused protein with enterokinase, and lane F shows a single protein band due to purified factor B. As determined by aminoterminal sequence analysis, this preparation of factor B contained at its amino terminus an extra tripeptide, LAL, which had been retained after cleavage of thioredoxin. Otherwise, mass spectrometric analysis of tryptic peptide masses of our recombinant factor B matched the calculated masses of the corresponding peptides expected from its tryptic degradation (data not shown). Thus, the mature factor B shown in Fig. 1 is composed of 175 amino acids with a calculated molecular mass of 20,341 Da. The molecular mass derived from matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of the factor B preparation containing the amino-terminal LAL tripeptide was 20,607 Da (calculated value, 20,638 Da), and gel exclusion suggested a mass of ϳ22 kDa, which indicated that the recombinant human factor B is monomeric (data not shown). Hydropathy analysis of the human factor B sequence revealed no hydrophobic clusters of amino acids for possible membrane anchorage.
Functional Properties of Factor B-The term factor B has been applied to several protein preparations derived from bovine heart mitochondria with molecular masses of 11-12 to 47 kDa, which partially or largely restored to AE-SMP preparations ATP synthesis and energy-linked reactions driven by ATP hydrolysis. Therefore, it was assumed that extraction of mitochondria with ammonia-EDTA partially removed one or more factor B-like molecules from the mitochondrial ATP synthase complexes. However, the effectiveness and the specificity of this extraction procedure for factor B removal were unclear. Fig. 3 depicts immunoblots of mitochondria (column A), SMP (column B), and AE-SMP (column C) showing their relative contents of factor B (row 1), F 1 ␣ subunit (row 2), OSCP (row 3), and IF 1 (row 4). The results clearly show that ammonia-EDTA extraction is highly effective for the efficient removal of factor B. 3 The levels of F 1 as represented by its ␣ subunit, and OSCP, which is not anchored to the membrane, were essentially unchanged by the extraction. The level of IF 1 was decreased in AE-SMP (Fig. 3, bottom row) which was understandable, because of the alkaline pH of the extraction medium (24). Consistent with these results, the ATPase activity of AE-SMP was shown to be about 40% higher than that of SMP. Fig. 4 shows on the ordinate the activity of AE-SMP for ATP hydrolysis-driven electron transfer from succinate to NAD as affected by addition of increasing amounts of recombinant human factor B. The activity of AE-SMP in the absence of factor B was 4 nmol of NADH formed (min⅐mg of protein) Ϫ1 . Addition of factor B increased this activity to 225, which was 87% of the activity of the SMP before ammonia-EDTA extraction. The  (25), the factor B half-saturation amount of 0.33 nmol/mg of AE-SMP indicated that this watersoluble recombinant human protein bound specifically and with high affinity to the site from which its bovine counterpart had been removed. Addition to AE-SMP of F 1 plus OSCP, instead of factor B, resulted in a marginal increase of reverse electron transfer activity (10 -12% as compared with 87% when factor B was added), indicating in agreement with the data of Fig. 3 that the poor coupling activity of AE-SMP was not due to the loss of F 1 and/or OSCP.
It has long been known that addition of low concentrations of oligomycin to SMP treated with EDTA or ammonia-EDTA partially restores their energy-coupled activities. Apparently, when added at low concentrations oligomycin preferentially reacts with and seals the proton leakiness of the fraction of the ATP synthase complexes that have been rendered defective, thereby allowing the remaining intact ATP synthases to function normally (26). Further addition of oligomycin would then begin to inhibit the intact ATP synthases, resulting in a decrease of the ATP synthase-coupled activity tested. Similar results are shown in Fig. 5. It is seen that addition to AE-SMP of oligomycin up to about 150 ng (0.3 nmol)/mg protein increased the ATP-driven reverse electron transfer activity of the particles from about 5 to about 54 nmol of NADH formed (min⅐mg protein) Ϫ1 , and that further increase in oligomycin concentration resulted in inhibition. As was shown previously (6), addition of low concentrations of oligomycin (Ͻ120 ng/mg protein) to factor B-replenished AE-SMP caused no activation but only partial inhibition of ATP-driven reverse electron transfer activity. We have examined the effect of other F O inhibitors with regard to activation of AE-SMP for ATP-driven electron transfer from succinate to NAD. At concentrations that caused inhibition of ATPase activity of AE-SMP from 10 to 70%, venturicidin and tributyltin chloride had no stimulating effect on its reverse electron transfer activity. Incubation of AE-SMP at 0°C with 6 M DCCD caused a time-dependent inhibition of its ATPase activity, as expected. At 10 and 20 min of incubation, when ATPase activity was inhibited by 30 and 43%, respectively, ATP-driven reverse electron transfer activity of the particles was increased from 4 nmol of NADH formed (min⅐mg) Ϫ1 to 18 at 10 min and to 27 at 20 min of incubation. Further incubation of AE-SMP with 6 M DCCD lowered this stimulated activity to 22 at 30 min and to 7 at 50 min.
Previous preparations of bovine factor B were shown to be inhibited by thiol and dithiol modifiers (5-7). The effects of these modifiers on recombinant human factor B are shown in Fig. 6. It is seen that factor B was inhibited when treated with the compounds shown, with the dithiol modifiers phenylarsine oxide and Cd 2ϩ causing 50% inhibition at Ͻ10 M concentration. As seen in Fig. 1, human factor B contains 6 cysteine residues, two of which are located at positions 92 and 94. The strong inhibitory effects of phenylarsine oxide and Cd 2ϩ suggest that these dithiol modifiers bind to cysteine residues 92 and 94 of factor B.
Effect of Factor B on Membrane Potential- Fig. 7 shows the formation of a membrane potential in SMP (traces A and B), AE-SMP (traces C and D), and AE-SMP plus factor B (traces E and F) as monitored by the absorbance change of oxonol VI at 630 minus 603 nm. Traces A and B are controls with SMP, showing membrane energization upon addition of ATP or NADH and deenergization by inhibition of ATP hydrolysis by oligomycin, inhibition of NADH oxidation by rotenone, or by uncoupling with FCCP. Traces C and D show that AE-SMP could not develop a high membrane potential upon addition of either ATP or NADH but that addition of oligomycin repaired this defect and allowed high membrane potential formation upon subsequent addition of NADH. Similar results were obtained when oligomycin was replaced with venturicidin or tributyltin chloride or when the AE-SMP was treated for 60 min at 0°C with 10 M DCCD. Like oligomycin, these reagents inhibited ATP hydrolysis but allowed membrane potential formation, as in Fig. 7C, upon addition of NADH (data not shown). These results suggested, therefore, that removal of factor B created a proton leak in the F O of AE-SMP, which could be blocked by the specific F O inhibitors mentioned. Traces E and F show that the addition of factor B to AE-SMP repaired the defect shown in traces C and D and made it possible for the particles to develop a high membrane potential upon addition of either ATP or NADH. As seen in traces E and F, the development of the membrane potential to a static head level in

Factor B
factor B-supplemented AE-SMP was a slow process as compared with the results shown in traces A and B for SMP. It is possible that factor B binding to AE-SMP requires an energized conformation of F O -F 1 , apparently best achieved with the addition of ATP. This possibility would be somewhat analogous to the binding of IF 1 to F 1 , which occurs during ATP hydrolysis by the enzyme (24).

DISCUSSION
Molecular Properties of Factor B-According to the data reported here, mature human factor B is composed of 175 amino acid residues with a molecular mass of 20,341 Da and the amino acid sequence shown in Fig. 1. The active, recombinant human factor B purified and used in the studies reported here contained an extra LAL tripeptide at its amino terminus. It was water-soluble, monomeric, and stable when stored at Ϫ80°C in a pH 8.0 buffer containing 5 mM 2-mercaptoethanol. In its amino-terminal 55 residues and molecular mass, this recombinant human factor B is similar to the latest of a number of bovine mitochondrial factor B preparations reported previously by Sanadi and co-workers (11). Because rigorous data on the purity and the amino-terminal sequences of the other factor B-like preparations of Sanadi and co-workers are not available, it is not possible to discuss them further, nor can we compare our pure protein of molecular mass 11-12 kDa, which was isolated from bovine heart mitochondria and exhibited a factor B-like activity (6,7), with the recombinant factor B described here. Its low yield precluded in 1976 any attempts at amino-terminal sequencing, but its sensitivity to thiol and dithiol modifiers suggests that it might have been an active proteolytic fragment of the bovine factor B. This possibility is supported by the fact that immunoblots of bovine heart mitochondria with our polyclonal antibody used in the present work recognized only a single protein band corresponding in mobility on SDS gels to that of our recombinant human factor B.
Factor B Gene- Fig. 8 shows a segment of the human genome on chromosome 14q21.3. This segment spans ϳ13 kb and contains 5 exons, of which the nucleotide sequences of the areas depicted in Fig. 8 by the filled boxes from ATG in exon 1 to TAA in exon 5 correspond to the sequence of the human factor B cDNA shown in Fig. 1, except that in the draft sequence of the Human Genome Project (27) the nucleotide T-168 is given as C, which would result in the substitution of Pro for Leu in the corresponding position of the presequence of factor B. As seen in Fig. 8, there are 2 stop codons and 2 polyadenylation signals in this DNA segment. An mRNA transcript arising from the processing of pre-mRNA at poly(A) site ATTAAA would encode a hypothetical protein HSU79253 (Locus identification number 27109). Reverse transcription PCR, using multiple tissue cDNA panels I and II (CLONTECH), showed that mRNAs corresponding to both the short (exons 1-3) and the long (exons 1-5) sequences were present in 16 different human tissues that we examined. As mentioned above, the short polypeptide con- taining a His tag was expressed in our laboratory as inclusion bodies and was used to raise polyclonal antibodies. This antibody preparation recognized in bovine heart mitochondria a single protein with a mass of 22 kDa, which could be extracted from SMP by the procedures used to remove factor B (Fig. 3). The short polypeptide, expected to have a mass of ϳ11 kDa, was not detected in bovine heart mitochondria. This consideration suggests that this short "isoform" of factor B, if translated, either does not enter mitochondria or is degraded after entering this organelle.
A second human homologue, the hypothetical protein FLJ10241 (Locus identification number 55101) encoded by a gene located on chromosome 19q13.2, with sequence similarity of 49% to human factor B, was identified by BLASTP (28) search of the GenBank TM data base. This search also revealed the existence of two mouse orthologues (GenBank TM accession numbers NP_080812 and BAB26107) with sequence similarities of 83 and 47%, respectively, as well as two gene products in Drosophila melanogaster (GenBank TM accession numbers AAF58055 and AAF51634) and Caenorhabditis elegans (Gen-Bank TM accession numbers AAK95868 and AAF59541) exhibiting 44 -52% sequence similarity to human factor B. Neither a prokaryotic homologue nor a counterpart in Saccharomyces cerevisiae has been identified.
Role of Factor B-The results presented here clearly show that factor B is necessary for the energy transduction activity of the ATP synthase complex. Treatment of well coupled SMP with ammonia-EDTA at pH 8.8 resulted in the specific removal of a protein of 22 kDa and in the inability of the AE-SMP to develop and maintain a membrane potential as a result of ATP hydrolysis or NADH oxidation (Fig. 7). This defect could be repaired by addition to the AE-SMP of a recombinant human protein of 22 kDa at a molar concentration nearly stoichiometric to the ATP synthases of the particles. The recombinant human protein was homologous in molecular mass and aminoterminal sequence to the purified bovine factor B of Sanadi and co-workers (11), and like all factor B preparations was sensitive to treatment with monothiol and especially dithiol modifiers.
The bovine ATP synthase complex contains at least 7 subunits of totally unknown function, namely subunits ⑀, F 6 , A6L, d, e, f, and g. This complexity makes it difficult to employ the E. coli ATP synthase with only 8 unlike subunits (29) as a model into which to build a role for factor B. However, the basic operational design of the E. coli ATP synthase, which is composed of a catalytic, a rotor, and a stator domain (30,31), applies to the more complicated ATP synthases (32,33), and we will discuss the role of factor B in the light of this basic design. The fact that factor B is not a component of the catalytic domain is clear and needs no further consideration. It cannot be a necessary component of the stator either, because unlike the systems lacking OSCP the ATPase activity of AE-SMP (5 mol/min/mg) is completely inhibited by F O inhibitors, including DCCD whose mechanism of inhibition is well known. The sensitivity of the ATPase activity of AE-SMP to F O inhibitors also indicates that in these particles the rotating part of F 1 , i.e. ␥ and ␦, is not disengaged from F O . These considerations allow the conclusion, therefore, that factor B is a component of F O .
As seen in Fig. 7, AE-SMP is uncoupled and incapable of forming a membrane potential as a result of respiration or ATP hydrolysis. This defect can be repaired by addition of factor B or by addition of F O inhibitors at concentrations that these reagents inhibit ATP hydrolysis. Repair by factor B allows membrane potential formation as a result of ATP hydrolysis or respiration. Repair by F O inhibitors allows only respiration-dependent membrane potential formation, which indicates that the proton leak of AE-SMP involves, at least in part, the normal proton channel of F O . This leads to the following important question. Because the proton leak of AE-SMP can be blocked by F O inhibitors, especially DCCD, does the proton leak from the cytosolic (positive) side to the matrix (negative) side of the membranes in respiring AE-SMP require the rotation of the c ring? Based on the generally accepted basic design and operation of the ATP synthase complex, the answer to this question is no. This is because in respiring AE-SMP the proton leak, which can be blocked by DCCD, occurs in the absence of F 1 substrates. These considerations allow the conclusion that the F O of mammalian ATP synthase is capable of uncoupled transmembrane proton translocation via a second path that involves, at least in part, its normal proton channel, possibly including Glu-58 of subunit c. Proton translocation via this second path does not require the operation of the rotor of the ATP synthase and can be blocked by the water-soluble factor B that appears to bind to F O on the matrix side.
Whether the F O subunits F 6 , A6L, d, e, f, and g, which like factor B do not have prokaryotic counterparts, are involved in this second proton path remains to be seen. Another mammalian F O -F 1 subunit that does not have a prokaryotic counterpart is the ATPase inhibitor protein, IF 1 , that binds to F 1 ␤ subunits and prevents futile ATP hydrolysis when the protonmotive force is low (34 -36). The fact that factor B can also be easily and reversibly removed from F O -F 1 and the fact that its displacement results in dissipation of membrane potential may be indicative of a regulatory function for factor B as well. It is generally assumed that the state 4 rate of respiration is due to a slow proton leak through the mitochondrial inner membrane at high proton-motive force. However, it is well known that proteoliposomes of sonicated phospholipids plus a proton pump such as cytochrome oxidase or nicotinamide nucleotide transhydrogenase are not as proton leaky at high proton-motive force as SMP. Therefore, considering that over-reduction of the respiratory chain results in an increased rate of superoxide anion production, which leads to the formation of toxic H 2 O 2 and hydroxyl radicals (37,38), it is possible that factor B acts as a pressure valve for maintaining the proton-motive force (hence the reduced level of the respiratory chain) below a damaging threshold (see also the concept of "mild uncoupling" in Ref. 38).
Finally, another point that deserves consideration is the designation "factor B" for the protein discussed here. More than 2 decades ago, the word "factor" was correctly replaced by "subunit," and already there are among the ATP synthase subunits a ␤ subunit and a b subunit. Continuing the alphabetical sequence beyond e, f, and g would result in confusion with the yeast ATP synthase subunit designations. Therefore, it may be appropriate to designate the protein under consideration here as subunit s, after its original discoverer.