Mutations in Single Hairpin Units of Genetically Fused Subunit c Provide Support for a Rotary Catalytic Mechanism in F 0 F 1 ATP Synthase*

Previously, we generated genetically fused dimers and trimers of subunit c of the Escherichia coli ATP synthase based upon the precedent of naturally occur-ring dimers in V-type H 1 -transporting ATPases. The c 2 and c 3 oligomers have proven useful in testing hypoth- esis regarding the mechanism of energy coupling. In the first part of this paper, the uncoupling Q42E substitution has been introduced into the second loop of the c 2 dimer or the third loop of the c 3 trimer. Both mutant proteins proved to be as functional as the wild type c 2 dimer or wild type c 3 trimer. The results argue against an obligatory movement of the e subunit between loops of monomeric subunit c in the c 12 oligomer during ro- tary catalysis. Rather, the results support the hypothesis that the c - e connection remains fixed as the c -oli-gomer rotates. In the second section of this paper, we report on the effect of substitution of the proton translocating Asp 61 in every second helical hairpin of the c 2 dimer, or in every third hairpin of the c 3 trimer. Based upon the precedent of V-type ATPases, where the c 2 dimer occurs naturally with a single proton translocating carboxyl in every second hairpin, these modified versions of the E. coli c 2 and c 3 fused proteins were predicted to have a functional H 1 -transporting ATPase activity, with a reduced H 1 /ATP stoichiometry, but to be inactive as ATP synthases. A variety of Asp 61 -substi-tuted proteins proved to lack either activity indicating that the switch

H ϩ -transporting F 1 F 0 ATP synthases utilize the energy of a transmembrane electrochemical H ϩ gradient to catalyze formation of ATP. Closely related enzymes are found in the plasma membrane of eubacteria, the inner membrane of mitochondria, and the thylakoid membrane of chloroplasts (1). The enzyme is composed of distinct extramembranous and transmembrane sectors, termed F 1 and F 0 , respectively. Proton movement through F 0 is reversibly coupled to ATP synthesis or hydrolysis in catalytic sites of F 1 . Each sector of the enzyme is composed of multiple subunits with the simplest composition being ␣ 3 ␤ 3 ␥␦⑀ for F 1 and a 1 b 2 c 12 for F 0 in the case of the Escherichia coli enzyme (2)(3)(4). Homologous subunits are found in mitochondria and chloroplasts. An atomic resolution x-ray structure of the ␣ 3 ␤ 3 ␥ portion of bovine F 1 shows the three ␣ and three ␤ subunits alternating around a centrally located ␥ subunit, with the ␥ subunit interacting asymmetrically with the three ␤ catalytic subunits (5). Subunit ␥ was subsequently shown to rotate with respect to the three ␤ subunits during catalysis (6 -8). Rotation of ␥ is thought to change the binding affinities in alternating catalytic sites to promote tight substrate binding and product release during catalysis (9). During ATP synthesis, the rotation of ␥ must be driven by proton translocation through F 0 .
The mechanism by which H ϩ movement through the F 0 membrane sector drives rotation of the ␥ subunit during ATP synthesis remains to be established (10). The ␥ and ⑀ subunits are known to interact with each other and appear to rotate as a fixed unit (10 -13). Rotation of the ⑀ subunit during ATP hydrolysis by F 1 ATPase has been directly demonstrated (14,15). In addition, the ␥ and ⑀ subunits are known to interact directly with the c subunits of F 0 (16 -18). Subunit c is known to fold in the membrane as a hairpin of two hydrophobic ␣-helices connected by a polar loop on the F 1 binding side of the membrane (19). The conserved Asp 61 carboxyl, centered in the second transmembrane helix, is known to catalyze proton transport via interaction with subunit a (10,19). In a recently determined NMR structure of monomeric subunit c, the folding and interaction of the two transmembrane helices takes place as is predicted from genetic and chemical studies of F 0 in situ (20). The 12 c subunits are now known to be arranged in an oligomeric ring, or cylinder, with subunits a and b of F 0 placed at the periphery of the ring (2,10,(21)(22)(23)(24)(25)(26). A recent 4-Å resolution x-ray diffraction model of a yeast mitochondrial F 1 -c 10 subcomplex supports the previously deduced oligomeric ring structure and the c-⑀ and c-␥ connections between the surfaces of F 0 and F 1 (27). The sequential protonation-deprotonation of Asp 61 at the a-c interface was proposed to drive rotation of the c-oligomer in stepwise 30°increments (6, 28 -30), and in so doing drive rotation of subunit ␥ within F 1 . Direct evidence for rotation of the c-oligomeric ring was recently presented (31). In an alternative hypothesis, the ␥ and ⑀ subunits were proposed to move from one subunit c to the next as a consequence of sequential conformational changes at the c-␥⑀ interface brought about by proton binding and release through each subunit c (19,32).
In a previous study, we constructed genetically fused functional dimers and trimers of subunit c based upon the precedent of the subunit c in vacuolar (V-type) ATPases. The functional fusion proteins provided a novel experimental system to test whether a functional polar loop was required in each subunit c of the oligomer, as would be expected if the ␥⑀ subunits sequentially moved from one c subunit to the next in the process of rotation. The cQ42E polar loop substitution had been shown to "uncouple" H ϩ translocation from ATP synthesis or hydrolysis (33), a phenotype that could be suppressed by second site substitutions in Glu 31 of subunit ⑀ (34). In this study we show the cQ42E substitution can be introduced into every second loop of the c 2 dimer or every third loop of the c 3 trimer without altering function. The result is clearly consistent with the suggestion that the c-␥⑀ interface remains fixed during rotary catalysis (6, 16, 28 -30). The c subunits of V-ATPases are composed of four transmembrane helices and seem to have evolved by gene duplication of an F 0 type progenitor gene (35). They have a single H ϩ transporting carboxylate in the fourth transmembrane helix or second helical hairpin. In this study, we have studied the effect of removal of one of the essential aspartates in the F-type c 2 dimer or c 3 trimer to test the hypothesis that the enzyme might still function but as a V-type, H ϩ -pumping ATPase with a reduced H ϩ /ATP stoichiometry (35). We were unable to replace any of the transmembrane Asp in a variety of substituted dimers or trimers without complete loss of function.
Generation of cQ42E-substituted Plasmids-Plasmid pC1Q42E was generated by subcloning a PstI/BstEII fragment containing the cQ42E substitution from the whole unc operon plasmid pYZ186 (34) into the respective sites of plasmid pDF163 and the cloning was confirmed by sequencing. Plasmid pC1Q42E was subsequently used to generate plasmids pC2Q42E, pC3Q42, and pC4Q42E, carrying, respectively, genes for a subunit c dimer (c 2 ) and trimer (c 3 ) and tetramer (c 4 ) harboring in each case the Q42E substitution in the last polar loop. A single AvaI site is present in the plasmid pC1Q42E at bases 1976 -1981 in the uncE gene in codons for amino acid residues 31 and 32 of subunit c. In plasmid pPJC2 encoding the fused c 2 dimer, the AvaI fragment cloned between two AvaI sites encodes the C-terminal segment from residues 31 and 32 of the first subunit, the linking peptide, and the N-terminal segment to residues 31 and 32 of the second subunit. This AvaI fragment was cloned into the AvaI site of the cut and alkaline phosphatase-treated plasmid pC1Q42E to generate plasmid pC2Q42E. Similarly, plasmids pPJC3 or pJC4 were partially digested with AvaI and the appropriate, large AvaI fragment purified from an agarose gel and cloned into the AvaI site of pC1Q42E to generate pC3Q42E and pC4Q42E. All constructs were confirmed by restriction mapping and sequencing.
Generation of Plasmids with Single Substitutions for Asp 61 -The c 2 dimeric constructs with Asp 61 substitutions in the first or N-terminal unit were constructed using the PCR 2 strategy described in Jones and Fillingame (4) wherein a plasmid encoding the substitution (D61N, D61G, or D61S) was used to generate the PCR product B. The substitutions generated by this method were c 2 (D61N), c 2 (D61G), and c 2 (D61S). 3 Plasmids bearing the c 3 (D61ЈN), c 3 (D61ЈG), and c 3 (D61ЈS) substitutions were generated by cloning the AvaI fragment from the aforementioned substituted c 2 plasmids into an incomplete AvaI digest of plasmid pPJC2. Plasmids bearing the c 2 (D61ЈN), c 3 (D61ЈЈN, and c 4 (D61ЈЈЈN) were generated by the previously cited PCR strategy (4) using plasmid encoding the D61N substitution to generate PCR product A. Wild type AvaI fragments from pPJC2 were cloned into partial AvaI digest of the c 2 (D61ЈN) plasmid product to generate plasmids bearing c 3 (D61ЈЈN) and c 4 (D61ЈЈЈN). Each of the constructs was confirmed by DNA sequencing.

Effect on Function of the Q42E Substitution in the Last Loop of Subunits c 2 and c 3 -A Q42E mutation was introduced into
the last polar loop in the c 2 dimer, c 3 trimer, and c 4 tetramer to generate subunits c 2 (Q42ЈE), c 3 (Q42ЈЈE), and c 4 (Q42ЈЈЈE), respectively. The modified c subunits were expressed from a plasmid that also encoded the a, b, and ␦ subunits of F 0 F 1 in the recA, ⌬uncBEFH chromosomal deletion strain JJ001. Subunits ␣, ␤, ␥, and ⑀ are expressed from the chromosome of this strain. Growth of transformants was tested on succinate minimal medium, where growth depends upon a functional oxidative phosphorylation system. As described previously, the cQ42E mutant was unable to grow on succinate. However, mutants expressing the c 2 (Q42ЈE)-and c 3 (Q42ЈЈE)-substituted proteins grew similarly to strains expressing the wild type version of the c 2 dimer or c 3 trimer (Table I). Strains expressing the c 4 (wild type) or c 4 (Q42ЈЈЈE) protein did not grow on succinate minimal medium. Wild type and mutant proteins were expressed at roughly comparable levels, as indicated by the immunoblot shown in Fig. 1A. The anti-c serum used had previously been shown to be directed to residues in the polar loop (41), and the lighter band for c 1 Q42E membranes may reflect reduced antibody binding to the substituted polar loop. Although several of the bands run as doublets for reasons that we do not under- 1 The unc DNA numbering system corresponds to that used by Walker et al. (37). 2 The abbreviations used are: PCR, polymerase chain reaction; ACMA, 9-amino-6-chloro-2-methoxyacridine; LDAO, lauryldimethylamine oxide; Tricine, N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine; DCCD, dicyclohexylcarbodiimide. 3 In the text, the repeated residues are designated by Ј, ЈЈ, and ЈЈЈ to indicate the second, third, or fourth hairpin repeat. For example, c 2 dimers with an Asp 3 Asn substitution in the first or second hairpin are designated c 2 (D61N) or c 2 (D61ЈN), respectively.
a Subunit c derivatives or fusion proteins were expressed from a pNOC-like plasmid in ⌬uncBEFH(a,c,b,␦) chromosomal background strain JJ001 (recA), except as noted.
stand, the immunoblots give no indication that the fused proteins are degraded to smaller units, and seem to rule out proteolysis as a possible explanation for the observed function. The ATPase activity of membranes prepared from these cells is shown in Table II. The LDAO-activated ATPase activity is the best indicator of the total amount of F 1 bound to the membrane (42). In wild type membranes, LDAO results in a 3-4-fold activation of membrane-bound F 1 ATPase. The low LDAO-activated ATPase activity of the c 1 Q42E membranes may indicate significant loss of F 1 ATPase from the membrane during cell rupture and membrane preparation, although the amount of F 0 in these membranes could also be somewhat reduced based upon the immunoblotting (Fig. 1). The extent of loss of F 1 ATPase from the chromosomal Q42E mutant had previously been shown to be strain-dependent (36). The c 2 and c 3 membranes (wild type or Q42E) show similar extents of activation by LDAO. Less F 1 ATPase activity seems to be lost during preparation of the c 2 (Q42ЈE) and c 3 (Q42ЈЈE) membranes than from c 1 Q42E membranes. In the c 4 mutant membranes (wild type or Q42E), the membrane-bound ATPase is more extensively inhibited and, when treated with LDAO, activated approximately 8-fold.
The ATP-driven ACMA quenching response of these membrane vesicles is shown in Fig. 2. As reported previously, the c 1 (Q42E) mutant shows virtually no activity, whereas quenching activity comparable to that seen for the wild type dimer and trimer was observed for c 2 (Q42ЈE) and c 3 (Q42ЈЈE) membranes. As we have reported previously (4), the c 3 trimeric protein promotes better ATP-driven proton pumping, as measured by ACMA quenching, than does the c 2 dimer. This is clearly unusual since strains expressing the c 2 dimer grow better by oxidative phosphorylation on succinate minimal medium (Table I). ATP-driven ACMA quenching had not previously been tested with the membranes from the c 4 tetramer (4). As shown here (Fig. 2), mutant membranes from either the c 4 (wild type) or c 4 (Q42ЈЈЈE) show a significant quenching response even though the activity is insufficient to support growth by oxidative phosphorylation (Table I).
The cQ42E mutant was described as being "uncoupled" because its plasma membrane vesicles, containing normal amounts of bound F 1 , were permeable to protons (34). This was reflected by a reduction in the NADH-driven quinacrine quenching response, caused by partial collapse of the H ϩ electrochemical potential due to proton transport through F 0 . Membrane vesicles of the c 1 (Q42E) transformant strain exhibit this uncoupled phenotype as illustrated in Fig. 3. In compari-  a Activity Ϯ S.D. of n determinations on a single membrane preparation. Q42E and wild type c 1 , c 2 , c 3 , and c 4 subunits. Membranes were diluted to 0.5 mg/ml in HMK assay buffer (10 mM HEPES, pH 7.5, 5 mM MgCl 2 , 300 mM KCl) and ACMA added to 0.3 g/ml. ATP was added to 0.94 mM and the uncoupler SF6847 added to 0.3 M at the times indicated. Comparison of quenching response: A, wild type versus Q42E c 1 ; B, wild type versus Q42ЈE c 2 ; C, wild type versus Q42ЈЈE c 3 ; D, wild type versus Q42ЈЈЈE c 4 . son to c 1 (wild type), the c 1 (Q42E) membrane vesicles exhibit a significantly reduced NADH-driven quinacrine quenching response, which on DCCD treatment to block proton permeation via F 0 was restored to wild type levels. In contrast, this uncoupled phenotype was not apparent with the c 2 (Q42ЈE) and c 3 (Q42ЈЈE) mutant membranes.

FIG. 2. Comparison of ATP-driven quenching of ACMA fluorescence by membranes expressing
The passive proton permeability of the mutant membrane vesicles was also examined after removal of F 1 from the membrane (Fig. 4). The c 1 membrane vesicles (wild type and Q42E) exhibited a high proton permeability, as indicated by the small NADH-driven quinacrine quenching response, which was greatly enhanced by DCCD treatment to block proton transport by F 0 . In contrast, stripping of F 1 from the c 2 membranes (wild type or Q42E) only marginally increased the proton permeability, suggesting perhaps that the c 2 dimeric proton channel was less stable. The response of c 3 membrane vesicles (wild type or Q42E) was close to that seen with c 1 membranes. This more active passive transport capacity may relate to the greater ATP-driven Hϩ transport activity seen in c 3 versus c 2 membrane vesicles.
Effect of Substitution of Asp 61 in Fused Dimers, Trimers, and Tetramers of Subunit c-Based upon the precedent of eucaryotic V-type ATPases, one might expect to be able to substitute one of the multiple Asp 61 carboxylates of the E. coli fused subunit c and retain ATP-driven proton pumping activity with a reduced H ϩ /ATP ratio. We generated a variety of Asp 61substituted fused proteins to test this hypothesis (Table II). Each of the mutated fused proteins was expressed, as shown by immunoblotting (see Fig. 1B for examples). None of the transformant strains expressing these proteins proved capable of growth on succinate minimal medium, which indicated a loss of oxidative phosphorylation activity. The membrane ATPase activity of all of the Asp 61 -substituted strains was very low (Table  II), and the very low activity can be attributed to inhibition of the F 1 ATPase in the membrane-bound state. This is indicated by the 15-35-fold activation caused by LDAO treatment (Table  II).
The Asp 61 substitutions made all seem to totally abolish ATP-driven proton pumping as assayed in the ACMA quenching assay (Fig. 5). The following combinations were tested. In direct analogy to the vacuolar ATPase proteolipid, the Asp in the first hairpin of the dimer was replaced by Asn, Ser, and Gly (Fig. 5B). An Asp 3 Asn substitution was also tried in the second hairpin of the c 2 dimer (Fig. 5B). Gly and Ser were substituted in the second (middle) hairpin of the c 3 trimer and Asn substituted in the third hairpin (Fig. 5C). The final substitution attempted was Asp 3 Asn in the fourth hairpin of the c 4 tetramer (Fig. 5D). DISCUSSION Substitution of Q42E into the last polar loop of the c 2 dimer and c 3 trimer of subunit c resulted in functional enzymes that permitted growth on succinate minimal medium. Relative to the wild type c 2 and c 3 , the substitution of Q42E in every second or every third loop had little effect on in vitro function, i.e. the extent of ATP-driven quenching of ACMA fluorescence was equivalent on comparing wild type and Q42ЈE c 2 dimers or wild type and Q42ЈЈE c 3 trimers. Further, in contrast to the monomeric Q42E c 1 mutant, membranes of the c 2 and c 3 Q42 mutants did not exhibit an obvious proton leak. The facile genetic complementation of the Q42E c 1 mutant supports the idea that an F 0 F 1 with a mixture of wild type and Q42E subunits can be functional (33,43). The results are most easily interpreted by a rotational model, where the ⑀ subunit remains fixed in linkage with a single c subunit of the the oligomeric ring, as suggested by others (16). Cross-linking analysis suggests that the ⑀Glu 31 residue packs near the Gln 42 residue at the interface of two subunit c, which are packed in a front-toback manner, and subunit ␥ is predicted to pack at neighboring dimeric c-interfaces (18). The observed uncoupling may require a substitution of Q42E in each subunit c to disrupt a specific, fixed c-⑀ linkage.
Based upon accessibility of Cys in the cQ42C-substituted mutant, Watts and Capaldi (45) have suggested that four or five c subunits may be involved in F 1 binding via the ⑀ and ␥ subunits. The results presented here indicate that portions of the ⑀ and ␥ subunits are able to pack at the interface of Q42Esubstituted subunit c without disruption of function. The dimensions of subunit ⑀ indicate that it is likely to have contacts with at least two subunit c within the oligomeric ring. The finding that all second site suppressor mutations isolated in the cQ42E mutant mapped to ⑀Glu 31 (34) suggest a very specific interaction between these residues, and that this portion of subunit ⑀ is likely packed with wild type loops of the c 2 and c 3 substituted proteins. Ketchum and Nakamoto (44) have recently shown that second site suppressors to the ␥E208K mutant map to residues in the loop region of subunit c that fall on both faces of subunit c in the NMR structure. These results are consistent with the region surrounding residue 208 of subunit ␥ also packing between c subunits, as is proposed for the Glu31 region of subunit ⑀ (18).
The results discussed above support a model where the ⑀ subunit remains fixed to a special pair, or pairs, of subunits as the c-oligomer rotates to drive the rotation of the ␥ subunit. The question had been debated, based upon the differing effects of Cys-Cys cross-linking with cQ42C mutants. Based upon the inhibitory effect of cross-linking between ⑀E31C and cQ42C on ATP-driven quenching of ACMA fluorescence, Zhang and Fillingame (17) had suggested that the ⑀ subunit might move from one subunit c to the next in response to conformational changes linked to the protonation and deprotonation of Asp 61 . We have subsequently examined ⑀E31C/cQ42C membranes after crosslinking and conclude that at least a portion of the diminution of FIG. 3. Comparison of proton leakiness of Q42E membranes using NADH-driven quinacrine quenching response. Membranes were diluted to 0.5 mg/ml in HMK assay buffer (10 mM HEPES, pH 7.5, 5 mM MgCl 2 , 300 mM KCl) containing 0.375 g/ml quinacrine, and NADH was added to 50 M. The reversal of quenching is due to consumption of NADH in the cuvette. DCCD-treated membranes (ϩDCCD) were diluted into HMK assay buffer and incubated with 20 M DCCD for 15 min at room temperature prior to addition of NADH. A, wild type c 1 ; the ATP-driven ACMA quenching response can be attributed to increased proton leakiness. 4 In contrast, Watts et al. (16) noted minimal effects of cross-linking of the ␥Y205C and cQ42C subunits on membrane ATPase activity and suggested that these results were most consistent with a fixed ␥⑀-c linkage. The ␥-c cross-link did abolish ATP-driven ACMA quenching, but this was attributed to a proton leak induced by the cross-linking reaction (16). The situation has recently been clarified in the study of Schulenberg et al. (46), who introduced the cQ42C mutation into second loop of a c 2 dimer. Near quantitative cross-linking of c 2 Q42ЈC to ⑀E31C resulted in minimal disrup-tion of ATP synthesis or ATP-driven ACMA quenching and strongly supports the fixed ␥⑀-c rotor hypothesis. Ironically, cross-linking of c 2 Q42ЈC to ␥Y207C led to uncoupling, as was the case in cross-linking of ␥Y205C to monomeric cQ42C (16).
What then is the molecular basis for an uncoupled phenotype, where ATP hydrolyis is uncoupled from proton translocation and the junction between F 0 and F 1 perturbed such that membrane is leaky to protons? Presumably, the ␥ subunit of F 1 is permitted to rotate at the top of the ring of c subunits while F 1 is still held in place. To explain the high proton permeability of whole membranes, the mutations at the ␥-c or ⑀-c interface must result in structural changes that lead to opening of proton channels through F 0 , where the opening would normally only take place during coupled ATP hydrolysis or synthesis in F 1 . The question of whether the oligomeric ring of subunit c is rotating during uncoupled proton transport remains to be answered. It is possible that the c subunits oscillate back and forth between exit and entrance half-channels on the two sides of the membrane to promote the bidirectional, passive H ϩ translocation. Such an oscillation might require a structural change in subunit a that alters the proximity of aArg 210 to cAsp 61 , which is postulated to promote unidirectional rotor motion (30). In support of this idea, the aR210A substitution itself leads to an uncoupled, H ϩ -leaky phenotype (47). The passive H ϩ conductance of whole or stripped aArg 210 membranes actually exceeds that of wild type stripped membranes.
Eight different versions of c 2 c 3 and c 4 fusion proteins were created wherein one of the multiple Asp 61 residues was replaced by Asn, Gly, or Ser. None of the strains expressing these subunits exhibited a detectable ATP-driven quenching response measured with inverted membrane vesicles. The lack of a quenching response may result at least partially from the severe inhibition of the membrane ATPase activity in each strain (Table II). 5 The inhibition observed suggests that F 1 is bound to the membrane in a coupled fashion and that replacement of one Asp 61 in a fused multimer of two, three, or four subunits is sufficient to abolish H ϩ translocation by F o . The inhibitory effect of the single Asp 61 substitutions is consistent with the results from previous studies where reaction of DCCD with approximately one of the 12 c subunits in the oligomer was shown to abolish ATPase activity (49), and where reconstitution of F o with one cD61G subunit or two cD61N subunits per c-oligomer was shown to abolish passive H ϩ translocation activity (50). Does H ϩ movement require the interaction of two adjacent carboxylates in F-type ATPases? Based upon a new NMR structure of subunit c in its deprotonated form, Rastogi and Girvin (51) have recently suggested that there may be 4 J. Hermolin and R. H. Fillingame, unpublished results. 5 The severe inhibition of membrane-associated F 1 ATPase seen with the Asp 61 substitutions in Table II proves to be unexpected, in that chromosomal cD61N and cD61G strains show high rates of ATPase activity even though proton translocation through F 0 is blocked (48), i.e. in the chromosomal strains, the F 1 appears to assemble with F 0 in an uncoupled fashion.
FIG. 4. Comparison of proton leakiness of Q42E stripped membranes using NADH-driven quinacrine quenching response. Stripped membranes were diluted to 0.3 mg/ml in HMK assay buffer (10 mM HEPES, pH 7.5, 5 mM MgCl 2 , 300 mM KCl) containing 0.375 g/ml quinacrine, and NADH was added to 50 M. The reversal of quenching is due to consumption of NADH in the cuvette. DCCD-treated membranes (ϩDCCD) were diluted into HMK assay buffer and incubated with 20 M DCCD for 15 min at room temperature prior to addition of NADH. large movements around Asp 61 at the subunit a interface and proposed a possible interaction of protonated and deprotonated carboxylates during H ϩ transport. Assuming that the c-oligomer of V-type (vacuolar) ATPase rotates past a site of protonation-deprotonation within a subunit a equivalent, with the H ϩ translocating carboxylate present on every other hairpin unit, then the mechanism may differ substantially from that in F o F 1 ATP synthases where rotation seems to be designed to occur in 30°increments. The conformational interactions and electrostatics taking place between V 0 subunits of V-type AT-Pases may be designed to facilitate the movement of the rotor in 60°or 120°increments since these enzymes seem to function unidirectionally as ATPases (35,52).