Role of α-Subunit VISIT-DG Sequence Residues Ser-347 and Gly-351 in the Catalytic Sites of Escherichia coli ATP Synthase*

This paper describes the role of α-subunit VISIT-DG sequence residues αSer-347 and αGly-351 in catalytic sites of Escherichia coli F1Fo ATP synthase. X-ray structures show the very highly conserved α-subunit VISIT-DG sequence in close proximity to the conserved phosphate-binding residues αArg-376, βArg-182, βLys-155, and βArg-246 in the phosphate-binding subdomain. Mutations αS347Q and αG351Q caused loss of oxidative phosphorylation and reduced ATPase activity of F1Fo in membranes by 100- and 150-fold, respectively, whereas αS347A mutation showed only a 13-fold loss of activity and also retained some oxidative phosphorylation activity. The ATPase of αS347Q mutant was not inhibited, and the αS347A mutant was slightly inhibited by MgADP-azide, MgADP-fluoroaluminate, or MgADP-fluoroscandium, in contrast to wild type and αG351Q mutant. Whereas 7-chloro-4-nitrobenzo-2-oxa-1, 3-diazole (NBD-Cl) inhibited wild type and αG351Q mutant ATPase essentially completely, ATPase in αS347A or αS347Q mutant was inhibited maximally by ∼80–90%, although reaction still occurred at residue βTyr-297, proximal to the α-subunit VISIT-DG sequence, near the phosphate-binding pocket. Inhibition characteristics supported the conclusion that NBD-Cl reacts inβE (empty) catalytic sites, as shown previously by x-ray structure analysis. Phosphate protected against NBD-Cl inhibition in wild type and αG351Q mutant but not in αS347Q or αS347A mutant. The results demonstrate that αSer-347 is an additional residue involved in phosphate-binding and transition state stabilization in ATP synthase catalytic sites. In contrast, αGly-351, although strongly conserved and clearly important for function, appears not to play a direct role.

F 1 F o -ATP synthase is the enzyme responsible for ATP synthesis by oxidative or photophosphorylation in membranes of bacteria, mitochondria, and chloroplasts. It is the fundamental means of cell energy production in animals, plants, and almost all microorganisms. It works like a nanomotor and is structurally similar in all species. In its simplest form, as in Escherichia coli, it contains eight different subunits distributed in the water-soluble F 1 sector (subunits ␣ 3 ␤ 3 ␥␦⑀) and the membraneassociated F o sector (subunits ab 2 c 10 ). The total molecular size is ϳ530 kDa. In chloroplasts there are two isoforms of subunit b. In mitochondria, there are 7-9 additional subunits, depending on the source, but in toto they contribute only a small fraction of additional mass and may have regulatory roles (1)(2)(3)(4).
ATP hydrolysis and synthesis occur in the F 1 sector. X-ray structures of bovine enzyme (5) established the presence of three catalytic sites at ␣/␤ subunit interfaces of the ␣ 3 ␤ 3 hexamer. Proton transport occurs through the membrane-embedded F o . The ␥-subunit contains three ␣-helices. Two of these helices form a coiled coil and are located in the central space of the ␣ 3 ␤ 3 hexamer. Proton gradient-driven clockwise rotation of ␥ (as viewed from the membrane) leads to ATP synthesis and anticlockwise rotation of ␥ results from ATP hydrolysis. In recent terminology, the rotor consists of ␥⑀c n , and the stator consists of b 2 ␦ (6, 7). The function of the stator is to prevent co-rotation of catalytic sites with the rotor. Detailed reviews of ATP synthase structure and function may be found in Refs. 8 -13. To better understand the reaction mechanism of ATP synthesis and hydrolysis and their relationship to mechanical rotation in this biological nanomotor, we have focused our efforts on determining the role of conserved residues in and around catalytic site P i -binding subdomain. Knowledge of P i -binding residues and residues surrounding the P i -binding subdomain is imperative for accomplishing (i) the molecular modulation of the catalytic site for the improved catalytic and motor function of this enzyme, (ii) an explanation of how ATP synthase binds ADP and P i within its catalytic sites in the face of a relatively high ATP/ADP concentration ratio, and (iii) understanding the relationship between P i binding and subunit rotation (14 -16). Earlier attempts to measure P i binding in purified E. coli F 1 using [ 32 P]P i (15) or by competition with ATP or AMP-PNP 2 in fluorescence assays of nucleotide binding (18,19) failed to detect appreciable P i binding at physiological P i concentration. So, we turned to the assay devised by Perez et al. (20) in which the protection afforded by P i against inhibition of ATPase activity induced by covalent reaction with 7-chloro-4-nitrobenzo-2oxa-1, 3,-diazole (NBD-Cl) provides the measure of P i binding. Earlier Orriss et al. (21) showed by x-ray crystallography that the covalent adduct formed by NBD-Cl is specifically in the ␤E catalytic site (Fig. 1A); thus protection afforded by P i indicates that binding of P i occurs at the ␤E catalytic site. By modifying the above assay for use with E. coli purified F 1 or F 1 F o mem-* This work was partly supported by East Tennessee State University Major branes, we have previously investigated the relationship between P i binding and catalysis for six residues, namely ␤Arg-246, ␤Asn-243, ␣Arg-376, ␤Lys-155, ␤Arg-182, and ␣Phe-291. 3 All of these residues are positioned in proximity to the phosphate analogs AlF 3 or SO 4 2Ϫ in x-ray structures of catalytic sites (22,23). We found that four residues, namely ␤Arg-246, ␣Arg-376, ␤Lys-155, and ␤Arg-182, grouped in a triangular fashion are directly involved in P i binding (Fig. 1B) (24 -30).
It is interesting to note that Penefsky (31) detected [ 32 P]P i binding with a K d (P i ) in the range of 0.1 mM in mitochondrial membranes using a pressure ultrafiltration method, and the results are in agreement with data obtained from the NBD-Cl protection assay (20). However, Penefsky could not detect P i binding in E. coli F 1 F o , and thus it is evident that P i dissociates more rapidly from E. coli F 1 than it does from mitochondrial F 1 .
This unfortunately renders the potentially more convenient centrifuge column assay unsuitable with the E. coli enzyme.
A mechanism of condensation of P i with MgADP was proposed by Senior et al. (32). The x-ray crystallography structure of bovine ATP synthase by Menz et al. (23) shows the transition state analog MgADP-AlF 4 Ϫ trapped in catalytic sites (Fig.  1B). It is clear from the geometry of this complex that the fluoroaluminate group occupies the position of the ATP-␥-phosphate in the predicted transition state. Similarly, Pedersen and co-workers (33) reported the first transition state-like structure of F 1 using enzyme obtained from rat liver and crystallized with the P i analog vanadate (V i ). This work further demonstrated that ADP was not essential, suggesting that the MgVi-F 1 complex inhibited the catalytic activity to the same extent as that observed for the MgADP-Vi-F 1 complex. Unfortunately, neither MgVi nor MgADP-Vi inhibits the E. coli enzyme (24). Thus we have relied on inhibition of ATPase activity by fluoroaluminate (or fluoroscandium) to assess the potential to stabilize a transition state complex (24 -26, 28, 30). Through mutagenesis and by employing the NBD-Cl protection assay as well as ATPase inhibition by transition state analogs, we can probe the direct or indirect role of residues in P i binding. In this manuscript, we explore the possible role played by ␣Ser-347 and ␣Gly-351 residues in the highly conserved ␣-subunit VISIT-DG sequence. Fig. 1B shows the location of ␣Ser-347 and ␣Gly-351 residues. Notably, ␣Ser-347 appears to occupy a strategic position in the P i -binding subdomain. Fig. 2 shows the evolutionarily conserved ␣-subunit VISIT-DG sequence along with surrounding residues of ␣-subunit from a variety of species. The basic questions we asked were: what role does ␣Ser-347 or ␣Gly-351 play? Do the mutations ␣S347A, ␣S347Q, or ␣G351Q have any effect on P i binding or transition state formation?

Construction of Wild Type and Mutant Strains of E. coli-
The wild type strain was pBWU13.4/DK8 (34). Mutagenesis was by the method of Vandeyar et al. (35). The template for oligonucleotide-directed mutagenesis was M13mp18 containing the HindIII-XbaI fragment from pSN6. pSN6 is a plasmid 3 E. coli residue numbers are used throughout. Ϫ -inhibited enzyme (23). E. coli residue numbering is shown. The triangle shows the residues ␤Lys-155, ␤Arg-182, ␤Arg-246, ␣Arg-376, and ␣Ser-347 forming a triangular P i -binding site. Rasmol software was used to generate these figures. containing the ␤Y331W mutation from plasmid pSWM4 (36) introduced on a SacI-EagI fragment into pBWU13.4 (34), which expresses all the ATP synthase genes. pSWM67/AN888 strain was used for ␣S347A mutant (37). The mutagenic oligonucleotide for ␣S347Q was CCAACGTAATCCAGATTAC-CGATGG, where the underlined bases introduce the mutation and a new XcmI restriction site, and that for ␣G351Q was CCATTACCGATCAGCAAATCTTCCTGGAAACC, where the underlined bases introduce the mutation and a silent mutation removes BglII restriction site. DNA sequencing was performed to confirm the presence of mutations and absence of undesired changes in sequence, and the mutations were transferred to pSN6 on a Csp451 (an isoschizomer of BstBI) and Pml1 fragment generating the new plasmids pZA13 (␣S347Q/ ␤Y331W) and pZA14 (␣G351Q/␤Y331W). Each plasmid was transformed into strain DK8 (38) containing a deletion of ATP synthase genes for expression of the mutant enzymes. It may be noted that both mutant strains contained the ␤Y331W mutation, which is valuable for measurement of nucleotide binding parameters (36) and does not affect function significantly on its own. Although the presence of ␤Y331W mutation was not utilized in this work, the Trp mutation was included for possible future use.
Preparation of E. coli Membranes, Measurement of Growth Yield in Limiting Glucose Medium, and Assay of ATPase Activity of Membranes-E. coli membranes were prepared as in Ref. 39. It should be noted that this procedure involves three washes of the initial membrane pellets. The first wash is performed in buffer containing 50 mM TES, pH 7.0, 15% glycerol, 40 mM 6-aminohexanoic acid, 5 mM p-aminobenzamidine. The following two washes are performed in buffer containing 5 mM TES, pH 7.0, 15% glycerol, 40 mM 6-aminohexanoic acid, 5 mM p-aminobenzamidine, 0.5 mM DTT, 0.5 mM EDTA. Prior to the experiments, the membranes were washed twice more by resuspension and ultracentrifugation in 50 mM TrisSO 4 , pH 8.0, 2.5 mM MgSO 4 . Growth yield in limiting glucose was measured as in Ref. 40. ATPase activity was measured in 1 ml of assay buffer containing 10 mM NaATP, 4 mM MgCl 2 , 50 mM TrisSO 4 , pH 8.5, at 37°C. The reactions were started by the addition of membranes and stopped by the addition of SDS to 3.3% final concentration. P i released was assayed as in Ref. 41. For wild type membranes (20 -30 g of protein), reaction times were 5-10 min. For mutant membranes (40 -60 g of protein), reaction times were 30 -50 min. All of the reactions were shown to be linear with time and protein concentration. SDS gel electrophoresis on 10% acrylamide gels was as in Ref. 42. Immunoblotting with rabbit polyclonal anti-F 1 -␣ and anti-F 1 -␤ antibodies was as in Ref. 43.
Inhibition of ATPase Activity by NBD-Cl and Protection by MgADP or P i -NBD-Cl was prepared as a stock solution in dimethyl sulfoxide and protected from light. The membranes (0.2-0.5 mg/ml) were reacted with NBD-Cl for 60 min in the dark at room temperature in 50 mM TrisSO 4 , pH 8.0, 2.5 mM MgSO 4 , and then 50-l aliquots were transferred to 1 ml of ATPase assay buffer to determine ATPase activity. Where protection from NBD-Cl inhibition by ADP or P i was determined, the membranes were preincubated 60 min with protecting agent at room temperature before the addition of NBD-Cl. MgSO 4 was present, equimolar with ADP or P i . Control samples containing the ligand without added NBD-Cl were included. Neither P i (up to 50 mM) nor MgADP (up to 10 mM) had any inhibitory effect alone.
Reversal of NBD-Cl Inhibited ATPase Activity by DTT-For reversal of NBD-Cl inhibition by DTT, the membranes were first reacted with NBD-Cl (150 M) for 1 h at room temperature, and then DTT (final ϭ 4 mM) was added, and incubation continued for 1 h at room temperature before ATPase assay. Control samples without NBD-Cl and/or DTT were incubated for the same times.
Inhibition of ATPase Activity by Azide, Fluoroaluminate, or Fluoroscandium-Azide inhibition was measured by preincubating membranes with varied concentrations of sodium azide for 30 min. Then 1 ml of ATPase assay buffer was added to measure the activity. For measurements of fluoroaluminate or fluoroscandium inhibition, the membranes were incubated for 60 min at room temperature in 50 mM TrisSO 4 , 2.5 mM MgSO 4 , 1 mM NaADP, and 10 mM NaF at a protein concentration of 0.2-0.5 mg/ml in presence of AlCl 3 or ScCl 3 added at varied concentration (see "Results"). 50-l aliquots were then added to 1 ml of ATPase assay buffer, and activity was measured as above. It was confirmed in control experiments that no inhibition was seen if MgSO 4 , NaADP, or NaF was omitted.
Inhibition of ATPase Activity by Dicyclohexylcarbodiimide (DCCD)-Covalent modification of wild type and mutant membrane was performed as described by Weber et al. (44). ATPase activity was measured by adding 1 ml of ATPase assay buffer containing 10 mM NaATP, 4 mM MgCl 2 , 50 mM TrisSO 4 , pH 8.5, at 37°C to the 100-l aliquots of 16 h DCCD-modified ATP synthase.

RESULTS
Growth Properties of ␣S347Q, ␣S347A, and ␣G351Q Mutants of E. coli ATP Synthase-Three new mutants, ␣S347Q, ␣S347A, and ␣G351Q, were generated. These two residues were chosen for mutagenesis because of their strong conservation in the ␣-subunit VISIT-DG sequence and their close proximity to the P i -binding pocket. The ␣S347A mutant was used to appreciate the role of Ser-OH side chain in P i binding and transition state. ␣S347Q and ␣G351Q mutants were designed to understand the impact of larger side chain of Gln on ␣Ser-347 and ␣Gly-351. Table 1 shows that introduction of Gln as ␣S347Q and ␣G351Q resulted in the loss of oxidative phosphorylation. Both mutations prevented growth on succinate-containing medium, and growth yields in limiting glucose medium were reduced close to those of the ATP synthase null control. ␣S347A mutant, on the other hand, resulted in partial loss of oxidative phosphorylation. Specific ATPase activities of membrane preparations containing mutant enzymes were compared with wild type and null control, and the values are shown in Table 1. ␣S347Q and ␣G351Q reduced the ATPase activity by 100 -150-fold, whereas ATPase activity was reduced only 13-fold by ␣S347A. Membranes prepared from the mutants contained the same amount of ␣ and ␤ subunits as wild type, as determined by SDS gel electrophoresis and immunoblotting (26) (data not shown); therefore, reduced ATPase is not due to impaired assembly of ATP synthase or loss of F 1 during membrane preparation.
Inhibition of ATPase Activity of ATP Synthase in Membranes by NBD-Cl and Reversal by Dithiothreitol-We previously established that P i binding by mutant or wild type ATP synthase can be assayed using either membrane preparations or purified F 1 , with equivalent results (24,26). In this work we used membrane preparations that are more convenient. Fig. 3 shows NBD-Cl inhibition of wild type and mutant membranes in the presence of varied concentrations of NBD-Cl. In wild type potent inhibition occurred with no residual activity, and this is consistent with previous studies (24 -28, 30). ␣G351Q mutant was also completely inhibited, and ␣S347Q or ␣S347A mutant were inhibited by ϳ80 -90% with ϳ20 -10% residual activity. In previous studies (24 -28, 30), we have noted several instances where mutant ATP synthases were incompletely inhibited by NBD-Cl. To be sure that maximal reaction with NBD-Cl had been reached, we incubated each membrane preparation with 150 M NBD-Cl for 1 h as in Fig. 3, followed by an additional pulse of 200 M NBD-Cl, continuing the incubation for an additional hour before assaying ATPase activity. Very little or no additional inhibition occurred (Fig. 4, left panel). This shows that the reaction of NBD-Cl was complete and that fully reacted ␣S347Q mutant membranes retained residual activity. Next, we checked whether inactivation by NBD-Cl could be reversed by the addition of the reducing agent DTT because reversibility by DTT was indicative of specificity of reaction in previous studies. Wild type and mutant enzymes were preincubated with 150 M NBD-Cl as in Fig. 3, and then 4 mM DTT was added, and incubation continued for 1 h before ATPase assay. It was seen that DTT completely restored full activity in all cases (Fig. 4, right panel). This indicates that NBD-Cl reacts specifically with residue ␤Tyr-297 in the wild type as well as in both mutants (45,46).
Protection against NBD-Cl Inhibition of ATPase Activity by MgADP or P i - Fig. 5 shows the data for MgADP protection in membrane enzymes, and it is seen that wild type and mutants were protected against NBD-Cl inhibition. Previously we have shown that MgADP protects against NBD-Cl inhibition of wild type soluble F 1 as well as membrane preparations of F 1 F o ; how-ever, protection occurred only at high concentrations (EC 50 ϭ ϳ4.5 mM MgADP). In this study the EC 50 values were 5.1, 2.9, 3.0, and 4.0 mM for wild type, ␣S347Q, ␣S347A, and ␣G351Q, respectively. We reason that high concentrations are required to effectively keep the ␤E site occupied by MgADP in time average and thus impede the access to NBD-Cl by sterically obstructing the site (24 -30). This idea is consistent with the conclusion of Orriss et al. (21), who provided evidence that NBD-Cl reacts specifically in the ␤E catalytic site by x-ray crystallographic studies. We conclude that NBD-Cl is reacting in ␤E in the mutants and that the ATPase activities measured in the mutants are referable to ATP synthase enzyme and not caused by a contaminant.
MgP i protection against NBD-Cl reaction is presented in Fig.  6. It is evident that P i protected well against NBD-Cl inhibition of ATPase activity in wild type and ␣G351Q mutant but did not protect at all against inactivation in ␣S347Q or ␣S347A mutants.
Inhibition of ATPase Activity by Fluoroaluminate, Fluoroscandium, and Azide-We next examined the effects of transition state and ground state analogs. Fig. 7 (A and B) show inhibition of wild type and mutant enzymes by MgADP-fluoroaluminate and MgADP-fluoroscandium, respectively. Wild type and ␣G351Q were completely inhibited. ␣S347A showed only ϳ25% inhibition. In contrast, the mutant ␣S347Q was remarkably resistant to inhibition. Azide is also a potent inhibitor of ATPase in ATP synthase. Fig. 7C shows that although wild type is strongly inhibited by azide, the mutants showed  varied resistance with ϳ70% inhibition in ␣G351Q and only ϳ20 -25% inhibition in ␣S347Q and ␣S347A mutants. Inhibition of ATPase Activity by DCCD- Fig. 8 shows the wild type and ␣S347Q, ␣S347A, and ␣G351Q mutant enzymes inactivated by DCCD. Although wild type is completely inhibited by 200 M DCCD after16 h of incubation at room temperature, mutants show varied degrees of inhibition. ␣G351Q is inhibited about 10%, ␣S347A is inhibited only ϳ30%, and ␣S347Q is not inhibited at all. In a similar series of experiments, carried out with the same range of DCCD concentrations and reaction conditions, but for only 2-or 5-h incubations, we found that wild type still became fully inhibited, ␣G351Q and ␣S347Q both showed zero inhibition, and ␣S347A was inhibited maximally by 6% (2 h) and 15% (5 h).

DISCUSSION
The goal of this study was to examine the functional role(s) of residue ␣Ser-347 and ␣Gly-351 of E. coli ATP synthase. These residues are part of the strongly conserved ␣-subunit VISIT-DG sequence. The VISIT-DG sequence residues are located in close proximity to the ␣/␤ interface flanking the P i -binding pocket (Fig. 1B). X-ray crystal structures of the AlF 3 -inhibited enzyme (22) as well as the AlF 4 Ϫ -inhibited enzyme (which also contained SO 4 2Ϫ in a second catalytic site) (23) show that the side chain of residue ␣Ser-347 is very close to these bound P i analogs (Fig. 1) and that ␣Gly-351 is also close. P i binding is a primary step in ATP synthesis by ATP synthase, thus exploring the molecular basis of P i binding is an important way to examine and understand the functional role of residues in the catalytic site.
Earlier studies established that mutagenesis combined with the use of the P i protection assay against NBD-Cl inhibition, as well as the use of inhibitory analogs, enabled characterization of functional role(s) of residues suspected to be involved in P i binding (24 -30). From analysis of six such catalytic site residues, we determined that four residues, namely, ␣Arg-376, ␤Arg-182, ␤Arg-246, and ␤Lys-155, are critical for P i binding and form a triangular subdomain within the catalytic site (24 -30) (Fig. 1B). We also established that introduction of a negative or positive charge in this location resulted in drastic modulation of P i binding (25,26,30), indicating that negative charge within the triangular subdomain was an important determinant of P i binding. Here we used the same approaches to study residues ␣Ser-347 and ␣Gly-351.
We introduced the mutations ␣S347Q, ␣S347A, and ␣G351Q, none of which affected assembly nor structural integrity of the membrane ATP synthase. Membranes showed similar content of F 1 -␣ and ␤ subunits as compared with wild type. Both ␣S347Q or ␣G351Q mutations had severely inhibitory effects on oxidative phosphorylation as judged by growth on  succinate or limiting glucose medium, and both strongly inhibited ATPase activity. On the other hand the ␣S347A mutation showed small residual oxidative phosphorylation and ATPase activity ( Table 1). The results with the ␣S347Q and ␣S347A mutants showed that they abolished P i binding (Fig. 6). Although based on Table 1 data for ␣S347A mutant, it can be argued that there could be a small amount of P i binding in cells but not significant enough to be measurable in the P i binding assay of membranes (Fig. 6). Fluoroaluminate and fluoroscandium in combination with MgADP potently inhibit wild type E. coli ATP synthase (24 -27, 30, 47, 48), and both are believed to mimic the chemical transition state. Transition state-like structures involving bound MgADP-AlF 4 Ϫ complex have been seen in catalytic sites in ATP synthase by x-ray crystallography (23). It was evident that the ␣S347Q mutant strongly destabilized the transition state (Fig. 7, A and B), because no inhibition by MgADP-fluoroaluminate or MgADP-fluoroscandium was apparent. Clearly, therefore, residue ␣Ser-347 is involved directly and to an important degree in catalysis and may be added as a fifth member of the group of P i -binding residues that make up the triangular P i -binding pocket. ␣S347A mutant did show some residual inhibition (ϳ25%) with both MgADPfluoroaluminate and MgADP-fluoroscandium, which is in agreement with the partial oxidative phosphorylation and ATPase activity found in this mutant. In contrast, the ␣G351Q mutation did not prevent P i binding (Fig. 6) and had lesser effects in destabilizing the transition state as judged by fluoroaluminate and fluoroscandium inhibition of ATPase (Fig. 7, A and B). Its effects on catalysis are therefore more indirect.
All of the mutations affected the degree of inhibition by azide, with ␣S347Q reducing it substantially (by ϳ80%), ␣S347A reducing it substantially (ϳ75%), and ␣G351Q reducing it by ϳ30% (Fig. 7C). A recent x-ray crystallography study (49) showed that azide inhibits ATP synthase by forming a tightly binding MgADP-azide complex in ␤DP catalytic sites, which resembles that formed by MgADP-beryllium fluoride and may therefore be considered an analog of the MgATP ground state. In the MgADP-azide complex, azide occupies a . Protection by P i of ATPase activity in wild type (WT) and mutant membranes from inactivation by NBD-Cl. The membranes were preincubated with MgP i at 0, 2.5, 5, or 10 mM concentration as shown, for 60 min at room temperature. Then NBD-Cl (150 M) was added, and aliquots were withdrawn for assay at time intervals as shown. ATPase activity remaining is plotted against time of incubation with NBD-Cl. E, no P i added; , 2.5 mM P i ; Ⅺ, 5 mM P i ; ‚, 10 mM P i . Each data point represents the average of four different experiments using two independent membrane preparations of each mutant. The membranes were preincubated for 60 min at room temperature with 1 mM MgADP, 10 mM NaF, and the indicated concentrations of AlCl 3 (A) or ScCl 3 (B). Then aliquots were added to 1 ml of assay buffer, and ATPase activity was determined. Sodium azide was added directly to the membranes and incubated for 30 min before assay (C), for details see "Materials and Methods." F, wild type; E,␣S347A; Ⅺ, ␣S347Q; ‚, ␣G351Q. All of the data points are the means of at least quadruplicate experiments. The variation was Ϯ10% between different experiments. position equivalent to that of the ␥-phosphate of MgATP. Thus mutants also had effects on substrate binding by virtue of an effect at the ␥-phosphate position.
DCCD inhibits wild type E. coli F 1 by reacting with residue ␤Glu-192 (50) and/or cAsp-61 (51), with the latter predominating at lower DCCD concentration and/or shorter incubation time. As expected, wild type ATP synthase was inhibited almost 100%. ␣S347Q mutant was not inhibited at all, ␣G351Q was inhibited to ϳ10%, and mutant ␣S347A is inhibited ϳ30% (Fig.  8). Notably, at shorter incubation times, ␣S347A showed even less inhibition (see "Results"). The data therefore indicate that in the ␣S347A mutant, ATPase activity on F 1 is only partly coupled to proton translocation in F o , which explains why ␣S347A mutant retains some growth on succinate and in limiting glucose (Table 1). It is interesting to note here that P i binding and release events have been shown to be directly linked to rotation of the central stalk in single molecule experiments (52). Perturbation of the P i -binding site might well be anticipated to perturb the integrity of the link between P i binding and rotation and be manifested as uncoupling. The overall data on ␣S347A mutant strongly suggests that the Ser-OH group is needed for transition state stabilization and P i binding.
The availability of x-ray structures allows one to discuss in detail the roles of residues ␣Ser-347 and ␣Gly-351. ␣Ser-347 is positioned close to bound AlF 4 Ϫ in catalytic sites (Fig. 1B) (23). The Ser-OH lies 5.0 Å from the F1 and F3 atoms in AlF 4 Ϫ and thus may contribute to transition state stabilization by direct interaction. It may be remarked that a similar conclusion was reached regarding the Ser-OH of the highly conserved LSGGQ ABC signature sequence in P-glycoprotein (17). Considering how P i binding is affected, ␣Ser-347-OH lies 6.1 Å from atom O 2 in SO 4 2Ϫ (23) and 4.6 Å from F1 of AlF 3 in the respective catalytic sites (22). Thus some direct interaction may be operative. However, more important than the above may be the fact that the Ser-OH lies 3.2 Å from the NH 2 of ␤Arg-246 (in the AlF 4 Ϫ site) and 3.0 and 4.1 Å, respectively from NH 2 and NH1 of ␤Arg-246 in the AlF 3 -occupied site. ␤Arg-246 is strongly conserved and critical for P i binding and transition state stabilization (24). Further, the carbonyl-O of ␣Ser-347 lies 3.2 Å from NH 2 of ␤Arg-182, another P i -binding residue. The likely H-bond interaction between ␣Ser-347 and ␤Arg-246 (and ␤Arg-182) suggests these residues act together to support P i binding and transition state stabilization. ␣Gly-351 is located at a distance of 7.7 Å from AlF 4 Ϫ and 8.7 Å from SO 4 2Ϫ . A more indirect role in catalysis is therefore indicated, likely predominantly structural in nature.
In summary, both ␣Ser-347 and ␣Gly-351 of the conserved VISIT-DG sequence in ATP synthase ␣-subunit are required for catalysis. ␣Ser-347 plays the more important role and is required for P i binding and transition state stabilization.