Membrane topography and near-neighbor relationships of the mitochondrial ATP synthase subunits e, f, and g.

The well characterized subunits of the bovine ATP synthase complex are the α, β, γ, δ, and ϵ subunits of the catalytic sector, F1; the ATPase inhibitor protein; and subunits a, b, c, and d, OSCP (oligomycin sensitivity-conferring protein), F6, and A6L, which are present in the membrane sector, F0, and the 45-Å-long stalk that connects F1 to F0. It has been shown recently that bovine ATP synthase preparations also contain three small polypeptides, designated e, f, and g, with respective molecular masses of 8.2, 10.2, and 11.3 kDa. To ascertain their involvement as bona fide subunits of the ATP synthase and to investigate their membrane topography and proximity to the above ATP synthase subunits, polyclonal antipeptide antibodies were raised in the rabbit to the COOH-terminal amino acid residues 57-70 of e, 75-86 of f, and 91-102 of g. It was shown that (i) e, f, and g could be immunoprecipitated with anti-OSCP IgG from a fraction of bovine submitochondrial particles enriched in oligomycin-sensitive ATPase; (ii) the NH2 termini of f and g are exposed on the matrix side of the mitochondrial inner membrane and can be curtailed by proteolysis; (iii) the COOH termini of all three polypeptides are exposed on the cytosolic side of the inner membrane; and (iv) f cross-links to A6L and to g, and e cross-links to g and appears to form an e-e dimer. Thus, the bovine ATP synthase complex appears to have 16 unlike subunits, twice as many as its counterpart in Escherichia coli.

verse the membrane twice via two hydrophobic stretches of amino acids near its NH 2 terminus, with its NH 2 -terminal end (ϳ30 residues) emerging from the membrane on the F 1 side (6). The remainder of each of the two b subunits (ϳ130 residues) extends from F 0 to F 1 . To these extramembranous segments of the b subunits as well as to F 1 are attached OSCP (oligomycin sensitivity-conferring protein), d, and the two copies of F 6 . A6L is anchored to the membrane via its NH 2 -terminal 25-30 hydrophobic residues. The remainder of the molecule (total number of residues ϭ 66) is extramembranous on the F 1 side and resides near subunit d (5).
The newly characterized polypeptides that are found in preparations of bovine ATP synthase are three small molecules, designated e, f, and g, with respective molecular masses of 8.2, 10.2, and 11.3 kDa (7). Collinson et al. (2) have shown that an ATP synthase fraction prepared after treatment of bovine SMP 1 with 3.3 M guanidine HCl, followed by solubilization with n-dodecyl ␤-D-maltoside and multiple column chromatographies, is devoid of F 1 subunits and OSCP, but contains all the other F 0 and stalk subunits mentioned above, including e, f, and g. Indeed, F 1 subunits and OSCP are the only components of the bovine ATP synthase that are easily removed, as was shown previously by Racker and Horstman (8) with SMP (ASU particles) and by Galante et al. (1) with purified bovine ATP synthase. In addition, both of these preparations, i.e. the ASU particles and the (F 1 ϩ OSCP)-depleted ATP synthase, could be reconstituted with added F 1 and OSCP, indicating that the removal of these components could be achieved without damage to the remainder of the ATP synthase complex (1,9). Whether the preparation of Collinson et al. (2) is capable of a similarly functional reconstitution with F 1 and OSCP is not known. However, the fact that this purified preparation contains e, f, and g suggested that these polypeptides should be considered as possible subunits of the bovine ATP synthase complex.
This paper shows that e, f, and g are present in highly purified preparations of the ATP synthase complex with fully oligomycin-sensitive ATPase and uncoupler-sensitive ATP-32 P i exchange activities and can be immunoprecipitated from a SMP fraction enriched in ATP synthase with antibodies specific for OSCP. It also shows data regarding the membrane topography of these polypeptides in SMP as well as their proximity to one another and to other ATP synthase subunits.
Preparation of Mitochondria and Mitoplasts-Heavy bovine heart mitochondria were prepared according to Hatefi and Lester (12), and mitoplasts according to Krebs et al. (13). The milligram ratio of digitonin to mitochondrial protein for the preparation of mitoplasts was 0.24. The activities of monoamine oxidase (outer membrane enzyme) and malate dehydrogenase (matrix enzyme) in the mitoplast preparations were Ͻ10 and Ͼ95%, respectively, of the total activities of the original sample of mitochondria.
Antipeptide Antibodies-Peptides corresponding to the COOH-terminal residues 57-70 of e (14), 75-86 of f, and 91-102 of g (2) plus an NH 2 -terminal Cys for coupling to keyhole limpet hemocyanin were synthesized, linked to keyhole limpet hemocyanin, and used to immunize rabbits as previously described (5). Positively reacting IgG appeared in the sera after the third injections. The antibodies were affinity-purified using the respective peptides coupled to CNBr-Sepharose (Pharmacia Biotech Inc.) or antigens immobilized on nitrocellulose filters as described before (5). The affinity-purified IgG were stored at Ϫ20°C in small aliquots. Their potencies were tested by enzyme-linked immunosorbent assays, and their specificities by immunoblotting. In each case, the immunospecificity of the affinity-purified IgG was confirmed by blotting mitochondria with the IgG preparation preincubated with the respective peptide antigen. Rabbit polyclonal antibodies against other ATP synthase subunits have been described elsewhere (5).
Isolation of the ATPase-rich Fraction of SMP and Immunoprecipitation-Bovine SMP were solubilized with deoxycholate in the presence of 1 M KCl, conditions that solubilize all the respiratory chain and ATP synthase complexes (15,16). The solubilized SMP were centrifuged for 20 min at 100,000 ϫ g, and the supernatant was applied to a Bio-Gel A-5m column (2.5 ϫ 65 cm; fractionation range of 1 ϫ 10 4 to 5 ϫ 10 6 Da) as described previously for the purification of the ATP synthase complex (10). Fractions of 3.8 ml were collected at a flow rate of 12 ml/h, assayed for oligomycin-sensitive ATPase activity (17), and analyzed by immunoblotting with antibodies against b, OSCP, c, e, f, and g. The column was calibrated using molecular mass standards from Sigma. Immunoprecipitation of the ATP synthase complex with anti-OSCP antibodies was performed essentially according to Robbins et al. (18), except that the concentration of Triton X-100 did not exceed 0.5% and 0.5 mM phenylmethylsulfonyl fluoride was included in all buffers. Extraction of SMP with 0.1 M Na 2 CO 3 or 6 M urea was performed essentially according to Fujiki et al. (19). The buffers contained 0.5 mM phenylmethylsulfonyl fluoride. After 30 min on ice, the solubilized proteins were separated by centrifugation and precipitated with 10% trichloroacetic acid. Then, the insoluble and solubilized proteins were prepared for SDS-PAGE and immunoblotting.
Cross-linking Conditions-Washed SMP were suspended in 0.25 M sucrose containing 50 mM triethanolamine, pH 8.0, and adjusted to 1 mg/ml. Freshly prepared DSS in dimethyl sulfoxide was added to a final concentration of 0.5 mM. After 40 min of incubation at room temperature, the reaction was terminated by the addition of ammonium acetate to 50 mM, and the samples were prepared for SDS-PAGE. Cross-linking conditions for DST, EDAC, sulfo-DST, and sulfo-DSS were the same as described before (5). For cross-linking with EEDQ, SMP were suspended at 1 mg/ml in 0.25 M sucrose containing 25 mM MOPS, pH 7.0. EEDQ dissolved in methanol was added to a final concentration of 10 mM, and after 40 min of incubation at room temperature, the reaction was stopped by the addition of a solution of lysine, pH 6.8, to 0.1 M. After 10 min, the samples were prepared for SDS-PAGE. For cross-linking with SANPAH or N-(5-azido-2-nitrobenzoyloxy)succinimide, SMP in the same buffer as indicated for DSS cross-linking were incubated for 20 min at room temperature in the dark with a 0.1 mM final concentration of the reagent freshly dissolved in dimethylformamide. Either excess reagent was removed by centrifugation and resuspension of SMP in buffer, or the modified particles were directly placed under a long wavelength UV lamp (8 watts) at a distance of 5 cm for 10 min on ice, with occasional mixing. Cross-linking of F 1 -depleted ASU particles and purified ATP synthase with the above reagents was done similarly. Then, the samples were prepared for SDS-PAGE, electrotransfer to nitrocellulose, and immunoblotting.
Treatment of SMP and Mitoplasts with Proteases-SMP and freshly prepared mitoplasts were suspended at 3 mg/ml in 0.25 M sucrose containing 20 mM Tris-HCl, pH 7.5. TPCK-treated trypsin or TLCKtreated chymotrypsin was added at a protease/particle protein concentration of 1:100 or 1:10 (w/w), and the suspension was incubated for 30 min at 37°C. Where indicated, the proteolysis mixture also contained 0.4% deoxycholate. Proteolysis was stopped by the addition of 2 mM final concentration of phenylmethylsulfonyl fluoride, and aliquots were prepared for SDS-PAGE and immunoblotting.
Gel Electrophoresis and Immunoblotting-SDS-PAGE was performed according to Laemmli (20), routinely using a separating gel containing 15% acrylamide. Gels were stained with Coomassie Brilliant Blue, or the protein bands were electrotransferred to nitrocellulose sheets according to Towbin et al. (21) and probed with the affinitypurified antibodies of interest as described before (5). The immunoblots were developed using enhanced chemiluminescence reagents.
Interaction of Antibodies with SMP and Mitoplasts-Particles (0.2-0.3 mg/ml) in 0.25 M sucrose containing 50 mM Tris acetate, pH 7.7, and 0.5% bovine serum albumin were incubated for 2 h at 24°C with 1:1000 to 1:10 diluted affinity-purified antibodies against e, f, and g or with immunoglobulin fractions prepared from the corresponding sera. The particles were then washed three times by centrifugation and resuspension in the same buffer. To 500 l of particle suspension was added an equal volume of 125 I-protein A (200,000 cpm), and the mixture was incubated for 1 h at 24°C. The particles were sedimented by centrifugation and washed three times with buffer, and their radioactivity was counted in a Packard Auto-Gamma Counter.
Protein Assay-Protein concentration was determined according to Lowry et al. (22), using bovine serum albumin as a standard. For mitochondria, mitoplasts, and SMP, protein concentration was estimated by the biuret method (23) in the presence of 0.1% potassium deoxycholate.

RESULTS AND DISCUSSION
Immunodetection of e, f, and g in SMP and the ATP Synthase Complex-Antipeptide antibodies were raised in the rabbit to keyhole limpet hemocyanin-linked synthetic peptides corresponding to the COOH-terminal 14 residues of e (Glu-57-Lys-70), 12 residues of f (Tyr-75-Tyr-86), and 12 residues of g (Ile-91-Val-102). Immunoblots of whole bovine heart mitochondria showed that each affinity-purified IgG was specific for the polypeptide it was intended and that each could be completely blocked by the synthetic peptide against which it was raised. Other experiments showed that e, f, and g were firmly bound to the mitochondrial inner membrane, and like subunits a and c of the ATP synthase complex, they could not be extracted with 6 M urea or at high pH (0.1 M Na 2 CO 3 ). By comparison, F 1 subunits, OSCP, and d were extracted with 6 M urea, and F 1 subunits, OSCP, d, and F 6 with 0.1 M Na 2 CO 3 . In these extractions, F 1 and OSCP were removed completely, and d and F 6 partially (data not shown). As is shown in Fig. 1, e, f, and g are present in highly purified preparations of the bovine ATP synthase complex. They are difficult to distinguish on the SDS gel (lane 1), but are clearly exhibited when the ATP synthase complex is blotted with each affinity-purified antiserum (lanes 2-4). It can be seen that e (lane 2), f (lane 3), and F 6 (lane 5), with respective molecular masses of 8.2, 10.2, and 9.0 kDa, move very close to one another on the Laemmli-type SDS gel (15% acrylamide) and that g (lane 4) and A6L (lane 6), with respective molecular masses of 11.3 and 8.0 kDa, band together. That the relative mobilities of e, f, and g on the SDS gel do not correspond to their molecular masses has been noted previously (2). Fig. 2 shows the cross-reactivity of antipeptide antibodies to bovine e, f, and g with polypeptides in rat liver SMP. It can be seen in lanes 4 and 6 of Fig. 2 that antigens recognized by our antipeptide antibodies to bovine g and f (lanes 3 and 5), respectively, are also present in rat liver submitochondrial particles, with the mobility of rat liver f being clearly different from that of bovine heart. Assuming that the reactive epitope of rat liver f is also at the COOH terminus and that this polypeptide has not suffered proteolysis during preparation of rat liver SMP, the faster mobility of rat liver f may mean that its matrix-side extramembranous segment is shorter than its bovine counterpart (see below). The COOHterminal peptide of bovine e against which our anti-e antibodies were raised has the amino acid sequence ERELAEAQEDTILK. The corresponding segment of e from rat liver has the sequence ERELAEAEDVSIFK (24). As seen in Fig. 2, our anti-e IgG did not recognize the rat protein, which appears to be referable to the differences in their COOH-terminal heptapeptides. In addition, our antipeptide antibodies to e, f, and g failed to recognize any polypeptides in mitochondria from S. cerevisiae.
Although e, f, and g are present in our highly purified bovine ATP synthase and also in the (F 1 ϩ OSCP)-depleted ATP synthase preparation of Collinson et al. (2), it was still desirable to see whether they can be immunoprecipitated together with other ATP synthase subunits, using antibodies to a well established ATP synthase subunit. For this purpose, bovine heart SMP were solubilized with deoxycholate (0.3 mg/mg of protein) in the presence of 1 M KCl, a procedure that solubilizes all five enzyme complexes of the mitochondrial electron transport/oxidative phosphorylation system (15,16). Then, after a brief centrifugation at 100,000 ϫ g, the supernatant was placed on a column of Bio-Gel A-5m as described under "Experimental Procedures," and the fraction with the highest oligomycin-sensitive ATPase activity was collected. Using antibodies specific for OSCP, the ATP synthase complex was immunoprecipitated from this fraction and was shown by immunoblotting to contain e, f, and g (Fig. 3, lanes 1-3, respectively). Lane 4 of Fig. 3 is a control showing that these antigens were absent when preimmune serum was used for immunoprecipitation. The results of Fig. 3 demonstrate that e, f, and g occur in a fraction of submitochondrial particles enriched in oligomycin-sensitive ATPase activity and can be immunoprecipitated therefrom with antibodies specific for OSCP. Hence, it could be concluded that e, f, and g are closely associated with the ATP synthase complex and may be considered as subunits of this enzyme complex. If so, this would mean that the mammalian ATP synthase complex is composed of 16 unlike subunits, twice as many as the unlike subunits of the bacterial ATP synthase complex.
Topography of e, f, and g in SMP-Although our affinitypurified antipeptide antibodies reacted strongly with e, f, and g in immunoblots (Figs. 1-3) as well as with the corresponding free peptides in enzyme-linked immunosorbent assays (data not shown), they exhibited no reactivity with their respective antigens in SMP and mitoplasts. We had encountered this situation with antibodies specific for other ATP synthase subunits and had shown that the membrane-bound antigens unreactive with antibodies were nevertheless accessible to proteases (4). With e, f, and g also, proteases provided the means for studying their membrane topography, as summarized in Fig. 4. In Fig. 4, each vertical lane is a separate experiment, the conditions for which are given above the immunoblots. Thus, bovine SMP (lanes 1-7) or mitoplasts (lanes 8 -11) were treated with a low (L) or high (H) concentration of trypsin or chymotrypsin in the absence or presence of deoxycholate, as specified under "Experimental Procedures," and then subjected to SDS gel electrophoresis and immunoblotting with affinity-purified anti-e, anti-f, or anti-g antipeptide antibodies. Lane 1 is a control showing immunoblots of e, f, and g in untreated SMP. Lanes 2 and 3 show that treatment of SMP with a low or high concentration of trypsin had no effect on e and f. Subunit g was converted to slightly smaller antigenic polypeptides, which is consistent with the presence of arginine at position 5 (25) and lysine at position 11 of native g. However, as seen in lane 4, when SMP were solubilized with deoxycholate, a low trypsin concentration was sufficient to degrade e, f, and g to small fragments that were either nonantigenic or not retained on the SDS gels. Lane 5 shows that treatment of SMP with a low concentration of chymotrypsin partially degraded g to a slightly smaller antigenic fragment, and lane 6 shows that a high concentration of chymotrypsin partially degraded f and completely degraded g to smaller antigenic fragments. The fragment of g was stable and that of f disappeared upon further incubation of the particles with chymotrypsin. As seen in these lanes, e was unaffected by chymotrypsin. However, when deoxycholate-solubilized SMP were treated with a low concentration of chymotrypsin (lane 7), e, f, and g were again completely degraded, as was the case in lane 4 with trypsin.
Lanes 8 -11 of Fig. 4 show the effect of trypsin and chymotrypsin on e, f, and g in mitoplasts. It can be seen that e was highly sensitive to either trypsin or chymotrypsin and that f was partially degraded by the low concentration of trypsin and completely by the high concentration of this protease. The effect of chymotrypsin on f was small, but was much greater on g (lanes 10 and 11). Trypsin did not appear to degrade g in mitoplasts (see, however, the legend to Fig. 4).
It is clear from the results of Fig. 4 that the carboxyl termini of e, f, and g are all exposed on the cytosolic side of the mitochondrial inner membrane and can be cleaved off by proteolysis (lanes 8 -11). Furthermore, f and g appear to be transmembranous, with segments on the matrix side of the inner membrane that can be curtailed by chymotrypsin (lanes 5 and 6; see also the effect of trypsin on g in lanes 2 and 3). However, there is no indication from these results whether e is also transmembranous and exposed on the matrix side of the inner membrane. We also checked the effect of proteinase K on e, f, and g, and the results were essentially the same as those shown in Fig. 4 for chymotrypsin. Proteinase K hydrolyzed f and g in SMP, resulting in the formation of antigenic fragments, but had no effect on e from the matrix side of the inner membrane. Removal of F 1 from SMP changed the kinetics of proteolysis of the remaining ATP synthase subunits, but not the final proteolysis results.
Near Neighbors of e, f, and g- Fig. 5 shows the results of attempts by cross-linking experiments to identify the near neighbors of e, f, and g in SMP. The cross-linking reagents used were EEDQ in Fig. 5A, DSS in Fig. 5B, and SANPAH in Fig.  5C, with respective cross-linking distances of 0, 11.4, and 18.2 Å. SMP were treated with the cross-linking reagents and then subjected to SDS-polyacrylamide gel electrophoresis, transfer to nitrocellulose sheets, and immunoblotting as described under "Experimental Procedures" and in the legend to Fig. 5. The affinity-purified antipeptide antibodies with which the nitrocellulose strips were blotted are indicated in Fig. 5 at the top of each pair of lanes, of which the left lane (odd-numbered lane) is the control SMP not subjected to cross-linking. Treatment of SMP (as well as (F 1 ϩ OSCP)-depleted particles and purified ATP synthase) with EEDQ resulted in the appearance of a product with a relative molecular mass of 22 kDa, which was recognized by our anti-e antibodies only (Fig. 5A, lane 2). Because this product was not recognized by antibodies to other ATP synthase subunits (we do not have antibodies to the ⑀ subunit of F 1 ), we assume that it is the result of cross-linking between e molecules, possibly an e-e dimer. EEDQ treatment of SMP did not produce any cross-linked products involving f and FIG. 4. Effects of trypsin and chymotrypsin on e, f, and g in SMP and mitoplasts. SMP or mitoplasts (3 mg/ml) were incubated with a low (L; protease/particle protein concentration of 1:100 (w/w)) or a high (H; protease/particle protein concentration of 1:10 (w/w)) concentration of TPCK-treated trypsin or TLCK-treated chymotrypsin for 30 min at 37°C. Where indicated at the top, the sample also contained 0.4% deoxycholate (DOC). Proteolysis was terminated by the addition of 2 mM phenylmethylsulfonyl fluoride, and the samples were electrophoresed on three identical 15% SDS gels. The separated proteins were then electrotransferred to nitrocellulose and blotted with affinity-purified antibodies to e, f, and g. Conditions for each lane are shown at the top, where Ϫ and ϩ indicate the absence and presence, respectively, of the material indicated on the left. The three sets of blots marked e, f, and g show only the appropriate immunoreactive molecular mass regions, which were cut from the remainder of each blot that was devoid of any immunoreactive bands. The broadening of the bands in lane 9 for g and in lane 10 for f may mean partial proteolysis of the COOH termini of these subunits without removal of their epitopes for antibody recognition.
FIG . 5. Cross-linking of e, f, and g to  g. Treatment of SMP with DSS resulted in the appearance of cross-linked products that were recognized by antibodies to e (Fig. 5B, lane 2), e and g (lanes 2 and 6), and f and g (lanes 4 and  6). The cross-linked products e-g and f-g reacted weakly with anti-e and anti-f antibodies, as seen in Fig. 5B (lanes 2 and 4,  respectively). For this reason, the amount of SMP applied to each lane of the SDS gels in the experiments shown in panel B was twice the amount used in each lane of the experiments shown in panels A and C. Fig. 5C shows the results of crosslinking experiments with the heterobifunctional photoreactive reagent SANPAH, of which the succinimidyl moiety reacts with free amino groups, and the substituted azidonitrophenyl yields a nitrene upon photoactivation, capable of covalent modification of protein residues in its vicinity. It can be seen in Fig. 5C that the use of this reagent resulted in the formation of a cross-linked product of f and A6L (lanes 2 and 4). A similar band, but in much lower yield, was also produced when N-(5azido-2-nitrobenzoyloxy)succinimide (cross-linking distance of 7.7 Å) was used instead of SANPAH. Photoirradiation of rat liver SMP after treatment with SANPAH did not cross-link f to A6L. This may be related to the possibility discussed above that the matrix-side extramembranous segment of rat liver f may be shorter than its bovine counterpart.
In previous studies (5), it was found that DST (cross-linking distance of 6.4 Å) was highly effective in producing in SMP dimeric cross-linked products involving ␣, ␤, OSCP, b, d, F 6 , and A6L. In SMP, EDAC cross-linked b to F 6 , and in F 1depleted SMP, it also cross-linked d to F 6 . These reagents were ineffective in producing cross-linked products of e, f, and g in SMP. However, in the purified ATP synthase complex, DST produced e-b, f-b, and g-b cross-links. These protein bands exhibited relative molecular masses around 36 and 48 kDa, and the former reacted with anti-F 6 antibodies as well (data not shown). Treatment of SMP with EDAC did not produce any cross-linked products involving f and g. In the purified ATP synthase complex, SANPAH produced cross-linked products involving e, g, and b. However, none of the reagents used in this study resulted in any cross-links involving e, f, or g with the F 1 subunits ␣, ␤, ␥, and ␦.
Because the results of Fig. 4 had indicated that the COOH termini of e, f, and g as well as most (if not all) of the bulk of e are on the cytosolic side of the mitochondrial inner membrane, cross-linking experiments were also performed with mitoplasts (mitochondria denuded of outer membrane). The cross-linking reagents used were the water-soluble compounds sulfo-DST (cross-linking distance of 6.5 Å), sulfo-DSS (cross-linking distance of 11.4 Å), and EDAC (EEDQ was also used; cross-linking distances of 0). The only cross-linked product of e, f, and g detected was a band similar to that in Fig. 5 (A, lane 2; and B, lane 2), i.e. a possible e-e dimer.
The results of Fig. 4 on the membrane topography of e and f agree with their hydropathy profiles shown in Fig. 6. The hydropathy profile of f clearly shows a hydrophobic cluster of ϳ20 residues near its COOH terminus, which could traverse the mitochondrial inner membrane. In that case, the NH 2terminal 45-50 residues of the molecule could be on the matrix side of the inner membrane, and the COOH-terminal 10 -15 residues on the cytosolic side. The amino acid sequence of f shows the presence of a tryptophan at position 22, a leucine at position 24, a methionine at position 25, and a phenylalanine at position 28, the carboxyl-terminal side of any one of which could be the site of chymotrypsin cleavage. These cleavages would result in an antigenic fragment ϳ7-7.5 kDa, which agrees with the position of the cleavage product of f seen in Fig.  4 (lane 6). The hydropathy profile of e also agrees with the data of Fig. 4. Thus, the hydrophilic COOH-terminal 45-50 residues of e could be on the cytosolic side of the inner membrane and could be susceptible in mitoplasts to proteolysis by trypsin or chymotrypsin (Fig. 4, lanes 8 -11). As seen in Fig. 6, the NH 2terminal 25-30 residues of e are hydrophobic. This segment could anchor e to the inner membrane without protruding much from the matrix side. In contrast to e and f, the hydropathy profile of g does not show any hydrophilic segments. However, it is possible that the first 60 -65 NH 2 -terminal residues are extramembranous on the matrix side. Then, the protein enters the membrane and protrudes from the cytosolic side. In that case, multiple trypsin and chymotrypsin cleavage sites are available near the NH 2 terminus of g to produce the products seen in lanes 2, 3, 5, and 6 of Fig. 4. Fig. 7 summarizes in graphic form the results of Figs. 4 and 5. It can be seen that f is near A6L, which was shown earlier to cross-link with d (5), and that g is close to f and e. Because a cross-link between e and f was not detected in our studies, the two subunits have been placed on opposite sides of g in Fig. 7. As mentioned above, DST produced e-b, f-b, and g-b cross-links in ATP synthase preparations. These cross-linked products also reacted with anti-F 6 antibodies. These data have not been FIG. 6. Hydropathy plots of e, f, and g. Hydropathy scores were calculated according to Kyte and Doolittle (26) using a setting of 11 residues. The horizontal line at Ϫ0.4 on the ordinate denotes the average hydropathy of 84 fully sequenced soluble proteins. The areas above and below this line indicate relative hydrophobic and hydrophilic regions, respectively. FIG. 7. Scheme showing the proximities of d, A6L, e, f, and g. The hatched area represents the mitochondrial inner membrane. M and C are the matrix and the cytosolic sides, respectively, of the inner membrane. The areas of A6L, e, f, and g have been drawn relative to d in molecular mass. N and C represent the NH 2 and COOH termini, respectively, of each subunit. The membrane intercalation of e is an assumption based on the tight binding of e to the membrane and the hydrophobic nature of its NH 2 -terminal 25-30 residues. included in Fig. 7, however, because the cross-linked products mentioned were not seen with SMP.
Finally, one other point may be considered here. The amino acid sequence of the COOH-terminal half of e has the potential for a coiled-coil structure formation 2 (possibly with another molecule of e), as does the amino acid sequence of the ATPase inhibitor protein (27). The latter has been reported to undergo a ␤-strand to ␣-helix conformation change as the medium pH is changed from 6.5 to 8.0 (28). This conformation change has functional significance because at pH Ͻ7.0, the inhibitor protein binds to F 1 in the presence of MgATP and inhibits ATP hydrolysis, and at pH Ͼ7.0, it is released (29). Whether e undergoes functionally significant conformation changes with pH also remains to be seen.