Identification of two segments of the γ subunit of ATP synthase responsible for the different affinities of the catalytic nucleotide-binding sites

ATP synthase uses a rotary mechanism to couple transmembrane proton translocation to ATP synthesis and hydrolysis, which occur at the catalytic sites in the β subunits. In the presence of Mg2+, the three catalytic sites of ATP synthase have vastly different affinities for nucleotides, and the position of the central γ subunit determines which site has high, medium, or low affinity. Affinity differences and their changes as rotation progresses underpin the ATP synthase catalytic mechanism. Here, we used a series of variants with up to 45- and 60-residue-long truncations of the N- and C-terminal helices of the γ subunit, respectively, to identify the segment(s) responsible for the affinity differences of the catalytic sites. We found that each helix carries an affinity-determining segment of ∼10 residues. Our findings suggest that the affinity regulation by these segments is transmitted to the catalytic sites by the DELSEED loop in the C-terminal domain of the β subunits. For the N-terminal truncation variants, presence of the affinity-determining segment and therefore emergence of a high-affinity binding site resulted in WT-like catalytic activity. At the C terminus, additional residues outside of the affinity-determining segment were required for optimal enzymatic activity. Alanine substitutions revealed that the affinity changes of the catalytic sites required no specific interactions between amino acid side chains in the γ and α3β3 subunits but were caused by the presence of the helices themselves. Our findings help unravel the molecular basis for the affinity changes of the catalytic sites during ATP synthase rotation.

F 1 F o -ATP synthase catalyzes the final step of oxidative phosphorylation and photophosphorylation, the synthesis of ATP from ADP and P i . F 1 F o -ATP synthase consists of the membrane-embedded F o subcomplex with, in most bacteria, a subunit composition of ab 2 c 10 and the peripheral F 1 subcomplex with a subunit composition of ␣ 3 ␤ 3 ␥␦⑀. The energy necessary for ATP synthesis is derived from an electrochemical transmembrane proton (or, in some organisms, sodium ion) gradient. Proton flow, down the gradient, through F o is coupled to ATP synthesis on F 1 by a unique rotary mechanism. The protons flow through (half) channels at the interface of a and c subunits, which drives rotation of the ring of c subunits. The c 10 ring, together with F 1 subunits ␥ and ⑀, forms the rotor. Rotation of ␥ leads to conformational changes in the catalytic nucleotide-binding sites on the ␤ subunits where ADP and P i are bound. The conformational changes result in formation and release of ATP. Thus, ATP synthase converts electrochemical energy, the proton gradient, into mechanical energy in the form of subunit rotation and back into chemical energy as ATP. In bacteria, under certain physiological conditions, the process runs in reverse. ATP is hydrolyzed to generate a transmembrane proton gradient, which the bacterium requires for such functions as nutrient import and locomotion (1-6). F 1 (or "F 1 -ATPase") has three catalytic nucleotide-binding sites, located on the three ␤ subunits, at the interface to the adjacent ␣ subunit. The catalytic sites have pronounced differences in their affinity for Mg 2ϩ -nucleotide. In Escherichia coli, K d1 for MgATP is in the nanomolar range ("high-affinity site"), K d2 is Ϸ1 M ("medium-affinity site"), and K d3 is Ϸ30 -100 M ("low-affinity site") (7,8). The affinity of a catalytic site at any given point of time is determined by the position of the central ␥ subunit. This implies that during rotational catalysis the affinities change. After rotation of ␥ by 120°, the sites have swapped their affinities. Experimental evidence for the crucial role of ␥ in determining the affinities of the catalytic sites comes from observations of the dependence of substrate binding and product release on the rotational angle of ␥ (9 -13). Further support is provided by the discovery of mutations in ␤ at the ␤/␥ interface that affect nucleotide binding affinities despite the fact that these mutations are located in the DELSEED loop in the C-terminal domain (14,15), Ն30 Å away from the catalytic binding site.
The affinity differences and changes seem to be of central importance for the enzymatic mechanism with respect to coupling between rotation and catalysis as well as catalysis itself. ATP synthesis and hydrolysis occur only on the high-affinity site (16). According to most catalytic models, in ATP synthesis proton translocation-driven rotation of ␥ forces the high-affinity site open, thereby reducing the affinity to "low" so that the newly formed ATP can be released (2,(17)(18)(19). In ATP hydrolysis, closing of the low-affinity site around the newly bound ATP, accompanied by conversion of the site to high-affinity, is widely believed to push ␥ and make it rotate for the experimentally observed 80°substep. The subsequent 40°substep appears to be energetically linked to the release of P i and/or ADP (2, 12, 15, 17, 20 -22).
Although a coherent picture of the general chemomechanical coupling mechanism is emerging, many aspects of the mechanism on the molecular level are still unresolved. For one, knowledge of the interactions between ␥ and ␤ (and/or ␣) that are responsible for the different affinities of the catalytic binding sites is fragmentary. The majority of contacts between ␥ and the ␣ 3 ␤ 3 ring involve the long N-and C-terminal helices of ␥.
Because of the large number of interactions that might possibly contribute, the first goal of the present study was to identify regions of ␥ that play a role in the assignment of affinities. In search of a screening method that allows looking at multiple residues at the same time, our approach was based on the observation that it is possible to generate mutants in ATP synthase from Geobacillus stearothermophilus (formerly known as Bacillus PS3) that have portions of the N-and/or C-terminal helices removed (23)(24)(25)(26). We measured binding of MgATP to a series of N-and C-terminal truncation mutants in the G. stearothermophilus ␣ 3 ␤ 3 ␥ subcomplex. For both helices, increasing the length of the truncation converted the MgATP binding behavior from asymmetric and WT-like, with a clearly present high-affinity site, to (nearly) symmetric, as also observed in the complete absence of ␥. This approach allowed us to narrow down the region of ␥ causing the differences in affinity of the catalytic sites to two short segments of ϳ10 residues, one in each helix.
To possibly identify individual residues responsible for conferring the nucleotide binding asymmetry, we replaced the amino acids in these two segments in groups of 5-6 residues by alanine. MgATP binding experiments with the alanine-replacement mutants all gave the WT-like asymmetric binding pattern. These results suggest that there is no specific residue in ␥ that causes the pronounced affinity differences of the three catalytic nucleotide-binding sites. Instead, the ␣-helices of ␥ themselves in the identified region appear to be the cause.

Selection of enzyme source
The enzymes from E. coli and G. stearothermophilus (formerly Bacillus PS3) are arguably the best-characterized bacterial ATP synthases. They are sufficiently closely related so that the G. stearothermophilus F 1 subcomplex can be reconstituted with E. coli F o to give a functional ATP synthase and vice versa (27). For the present study, we chose the G. stearothermophilus enzyme because of its higher oligomeric stability. Specifically, it could be shown that the G. stearothermophilus ATP synthase can form an ␣ 3 ␤ 3 complex in the absence of ␥ (28) that is stable enough to be crystallized (29) or monitored in high-speed atomic force microscopy (30). Under appropriate storage conditions, as ammonium sulfate precipitate at 4°C, even after several years about two-thirds of the complex was still in ␣ 3 ␤ 3 form with about 20% isolated ␣ and ␤ subunits and the remainder unidentified degradation products, which are absent directly after preparation of the enzyme (Fig. S1). Thus, truncations of ␥ should not affect the stability of the ␣ 3 ␤ 3 complex.

N-truncation mutants of ␥: an overview
The following N-terminal truncation mutants of the ␥ subunit of G. stearothermophilus ATP synthase ␣ 3 (␤Y341W) 3 ␥ subcomplex were generated: ␥⌬N4, missing the first 4 residues of ␥; ␥⌬N9; ␥⌬N13; ␥⌬N29; and ␥⌬N45 (the numbering system assumes that E. coli, used to express the G. stearothermophilus enzyme, removes the Met encoded by the start codon as is observed for the native E. coli ATP synthase). Subcomplexes with full-length ␥, and ␥-less subcomplex were included as controls. The ␥⌬N45 mutant eliminates all contacts between the N-terminal helix of ␥ and the ␣ 3 ␤ 3 cylinder (31)(32)(33). The crystal structures of mitochondrial (31,32) and E. coli F 1 (33) indicate that the N-terminal helix of ␥ starts immediately at the N terminus. Secondary structure predictions suggest that the same applies to G. stearothermophilus WT ␥ and to the truncation mutants investigated here ( Fig. 1A) with the possible exception of the ␥⌬N9 mutant. For the ␥⌬N9 mutant, the 5th residue is the first that reaches a probability for formation of an ␣-helix of Ն50%, whereas in all other cases it is the 2nd or 3rd residue; thus, for a few residues at the extreme N terminus of ␥⌬N9, the helix might be unwound. For ␥⌬N4, the predicted probability for formation of an ␣-helix by the initial residues is lower than in the other cases but still Ն50% (as compared with 90 -100% for the other truncation mutants).

C-truncation mutants of ␥: an overview
The C-terminal ␥ truncation mutants generated for this study were as follows: ␥⌬C14, eliminating 14 residues from the C terminus; ␥⌬C20; ␥⌬C27; ␥⌬C36; and ␥⌬C60. In the ␥⌬C60 mutant, the C terminus is outside of the ␣ 3 ␤ 3 cylinder. According to the crystal structures of the mitochondrial and E. coli enzymes (31)(32)(33), the C-terminal helix starts 73 and 76 residues, respectively, upstream of the C terminus and runs uninterrupted to the C terminus. Secondary structure predictions (Fig. 1B) suggest similar behavior for the C-terminal helix of the G. stearothermophilus enzyme, although for the terminal 6 -7 residues the probability to form a helix is only about 50%. According to the predictions, also in the truncation mutants the helix reaches close to the respective C terminus.

Stability and ATPase activity of the ␣ 3 ␤ 3 ␥ subcomplexes
An ␣ 3 ␤ 3 ␥ subcomplex could be isolated from each of the five N-terminal truncation mutants and from four of five C-terminal truncation mutants. Of the N-terminal truncation mutants, based on Western blotting using an antibody against the globular portion of ␥, the ␥⌬N4 and ␥⌬N9 mutants contained a full complement of ␥. In the mutants with longer N-terminal truncations, the amount of ␥ was substoichiometric. Obviously, these truncations reduce the stability of the interactions of ␥ with the ␣ 3 ␤ 3 cylinder. Still, even for the ␥⌬N45 mutant in Affinity of catalytic sites of ATP synthase nearly a half of the enzyme complexes ␥ was still present (Table  1 and Fig. S2A). As to the C-terminal mutants, for the longest truncation, ␥⌬C60, an ␣ 3 ␤ 3 ␥ subcomplex could not be isolated. No ␥ was found in the preparation; instead, an ␣ 3 ␤ 3 subcomplex was obtained (data not shown). Thus, this mutant was not pursued any further. All shorter C-terminal truncations contained stoichiometric amounts of ␥ (Table 1 and Fig. S2B).
Of the ␣ 3 ␤ 3 ␥ subcomplexes with N-terminal truncations, only ␥⌬N4 showed an ATPase activity that was similar to that of the WT enzyme. In the case of the ␥⌬N9 and ␥⌬N13 ␣ 3 ␤ 3 ␥ subcomplexes, the ATPase activity was about 10% of the WT value. For ␥⌬N29 and ␥⌬N45 ␣ 3 ␤ 3 ␥ it was even less but still clearly higher than for a ␥-less ␣ 3 ␤ 3 subcomplex (Table 1). Of the C-terminal truncation mutants, again only the shortest one, ␥⌬C14, had an ATPase activity in the same range of that of enzyme with full-length ␥. The activity of ␥⌬C20 was about 10% of WT; those of ␥⌬C27 and ␥⌬C36 were 2-3%. Again, even the C-terminal deletion mutants with the largest functional impairment still had ATPase activities significantly higher than the ␥-less enzyme (Table 1). A general decrease of the ATPase activity with Figure 1. Secondary structure predictions for the ␥ truncation mutants. All secondary structure predictions were obtained using the PredictProtein server (49). The amino acid sequence is given on top of the figure. The figure plots the probability of formation of an ␣-helix. The top panels (panels 1; black) show the prediction for the WT G. stearothermophilus enzyme with full-length ␥. It should be noted that the prediction for ␥ of mitochondrial ATP synthase gave very similar results (not shown) with a lower probability of ␣-helix formation between residues ␥10 and ␥15 and a disruption of the helix between ␥40 and ␥45. The crystal structures (31,32) show the ␥10 -15 segment as helical and confirm the helix break around ␥40 -45. A, N-terminal truncations. Panels 2-6 show the predictions for the N-terminal truncation mutants ␥⌬N4 (green), ␥⌬N9 (red), ␥⌬N13 (blue), ␥⌬N29 (pink), and ␥⌬N45 (cyan). B, C-terminal truncations. Panels 2-6 show the predictions for the C-terminal mutations ␥⌬C14 (green), ␥⌬C20 (red), ␥⌬C27 (blue), ␥⌬C36 (pink), and ␥⌬C60 (cyan).

Affinity of catalytic sites of ATP synthase
the length of the deletion had been described before for N-terminal as well as C-terminal truncations (23,25,26).

Effect of truncations on MgATP binding
The fluorescence of residue ␤Trp 341 in the ␣ 3 ␤ 3 ␥ subcomplex was used to determine the MgATP binding properties of the truncation mutants. The results are compiled in Fig. 2, A (N-terminal truncations) and B (C-terminal truncations), and in Table 1. MgATP binding data for the WT ␣ 3 ␤ 3 ␥ subcomplex and the ␥-less subcomplex, ␣ 3 ␤ 3 , are shown as controls. ␣ 3 ␤ 3 ␥ shows the well-established MgATP binding asymmetry with a K d1 below 10 nM, a K d2 of about 4 M, and a K d3 of about 30 M. For the ␣ 3 ␤ 3 subcomplex, in the absence of ␥, it had been shown previously that it has a highly symmetrical structure (29) and has lost its ability to bind MgTNP-ATP 5 and MgTNP-ADP with high affinity (34). Here, we confirmed the "functional symmetry" of the ␣ 3 ␤ 3 subcomplex. High-affinity binding was not observed, and the three catalytic binding sites had very similar affinities for MgATP. An optimal fit was obtained using a model with three different sites; the fit suggested that one of the three sites might still have a slightly higher affinity than the remaining two sites. However, as can be seen from Fig. 2 (best visible in C), the difference between this fit (solid black line through the open circles) and the fit for a model with three identical sites (dotted black line) is minimal. In the latter case, K d123 ϭ 6.1 M.
In the nucleotide binding experiments, both series of truncation mutants showed the same tendencies, from a WT-like asymmetrical pattern with a clearly expressed high-affinity binding site for the shorter truncations to a symmetrical binding behavior with a loss of the high-affinity site for the longer truncations. Of the N-terminal truncation mutants, only ␥⌬N4 displayed strong asymmetric behavior ( Fig. 2A); of the C-terminal truncations, both ␥⌬C14 and ␥⌬C20 did (Fig. 2B). In con- Table 1 ATPase activities and ATP-binding properties of ␥ truncation mutants The content of ␥ in the subcomplex preparations was measured by Western blotting using an antibody against the globular portion of ␥. Each Western blot contained an ␣ 3 ␤ 3 ␥ control, set as 100%, and an ␣ 3 ␤ 3 control, set as 0%. ATPase activities were determined in duplicate at 42°C and pH 8.0 by the amount of P i released. Nucleotide binding to the three catalytic sites was measured at 23°C and pH 8.0 using the fluorescence of residue ␤Trp 341 as signal. This technique does not allow resolving K d values below 0.01 M with confidence. All values given in this table represent the average from at least two independent experiments with two different enzyme preparations with standard deviations in parentheses.

Affinity of catalytic sites of ATP synthase
trast, ␥⌬N13, ␥⌬N29, and ␥⌬N45 exhibited the (nearly) symmetrical binding pattern of the ␥-less enzyme as did ␥⌬C27 and ␥⌬C36. Whereas for the C-terminal truncations the binding pattern went directly from WT-like in ␥⌬C20 to ␥-less-like in ␥⌬C27, among the N-terminal truncations ␥⌬N9 presented a transition case. The affinity of site 1 in ␥⌬N9 appeared to be slightly higher than in the mutants with longer truncations. As noted above, in ␥⌬N13, ␥⌬N29, and ␥⌬N45, the amount of ␥ was substoichiometric. Nevertheless, as these mutants display binding characteristics that are virtually indistinguishable from those of ␣ 3 ␤ 3 , correction of the binding curves for "contamination" by ␥-less enzyme did not change the results significantly. However, this correction assumes that the amount of ␥ in the ␣ 3 ␤ 3 ␥ subcomplex remains constant during the binding assay. To test whether this assumption is valid, we concentrated the assay mixture after the binding assay using a Centricon centrifugal filter device with an exclusion limit of 50,000 Da, which would let isolated ␥ pass through. Immunoblot analysis of the concentrate indicated that for ␥⌬N29 and ␥⌬N45 the amount of ␥ in the ␣ 3 ␤ 3 ␥ subcomplex was indeed further reduced by 30 -50%. However, the ␥⌬N13 ␣ 3 ␤ 3 ␥ subcomplex showed the same amount of ␥ as before the binding experiment (Fig. S2C). Thus, there is no doubt that ␥⌬N13 displays a symmetrical binding pattern, just as ␥-less enzyme. Although the precise amount of ␥ in the ␣ 3 ␤ 3 ␥ subcomplex of ␥⌬N29 and ␥⌬N45 at the exact moment of the binding assay is not known and therefore a correction for the contamination by ␥-less enzyme is not straightforward, it appears highly unlikely that these truncation mutants would revert to an asymmetric pattern.

The segments of ␥ responsible for the nucleotide binding asymmetry of the catalytic sites: N terminus
At the N terminus, truncation of just 4 residues preserved the asymmetric WT binding pattern. Truncation of 13 or more residues resulted in the symmetric pattern seen in the absence of ␥. Truncation of 9 residues showed intermediate characteristics with a higher degree of asymmetry than observed for the longer truncations but no pronounced high-affinity binding site. Based on the results, one might conclude that the segment most important for determination of the affinities of the three catalytic sites consists of residues ␥5-9. However, it is possible that the failure of the ␥⌬N9 truncation mutant to give a WTlike binding pattern (like ␥⌬N4) could be due to failure to form an ␣-helix immediately at its N terminus (see "Results" and Fig.  1). In contrast to ␥⌬N9, secondary predictions for ␥⌬N13 indicate that this mutant reaches ␣-helical conformation within 2-3 residues. Thus, the affinity-determining segment on the N-terminal helix should not extend beyond residue ␥15, giving it the sequence ␥ 5 DIKTRINATKK 15 .
None of the residues of the ␥ 5 DIKTRINATKK 15 segment is completely conserved; however, several display exclusively conservative substitutions (Table S1). In general, positions ␥7-␥15 show an accumulation of arginine and lysine residues, between 2 and 5 in the examples shown; the net charge of this stretch is between ϩ1 and ϩ5. The ␥5-15 segment comes close to three subunits of the ␣ 3 ␤ 3 cylinder, ␤ TP , 6 ␤ DP , and ␣ E (Figs. 3A and S3A). In the structure of the E. coli enzyme (33), ␥5-15 also has contacts with the C-terminal helix of the ⑀ subunit. Possible interactions with ␤ TP and ␤ DP occur with the DELSEED loop of these subunits. The multiple negatively charged side chains of the DELSEED loop appear well suited for interactions with the positive charges of the ␥5-15 6 The nomenclature of ␣ and ␤ subunits is based on the nucleotide occupancy of the catalytic sites in the original structure of the mitochondrial enzyme (31). Subunits ␣ TP and ␤ TP contribute to formation of the catalytic site occupied by a non-hydrolyzable ATP analog, AMP-PNP (5Ј-adenylyl-␤,␥-imidodiphosphate); ␣ DP and ␤ DP contribute to formation of the catalytic site occupied by ADP; and ␣ E and ␤ E contribute to formation of the empty catalytic site.

Affinity of catalytic sites of ATP synthase
segment. However, it has been shown that the negative charges of the DELSEED motif can be removed without affecting enzymatic activity, MgATP binding asymmetry, and rotational torque (35)(36)(37)(38).

The segments of ␥ responsible for the nucleotide binding asymmetry of the catalytic sites: C terminus
At the C terminus, truncation of up to 20 residues gave the asymmetric, WT-like binding pattern, whereas removal of 27 or more residues resulted in the symmetric pattern, lacking a high-affinity binding site, also observed in the absence of ␥. Thus, the affinity-determining segment of the C-terminal helix should, at the minimum, consist of residues ␥ 259 TLSYNRA 265 , which are present in ␥⌬C20 but absent in ␥⌬C27. Taking into account that the terminal residues of ␥⌬C27 might not be in quite the same helical confirmation as in full-length ␥, a reasonable estimate for the C-terminal affinity-determining segment appears to be ␥ 256 RTLTLSYNRA 265 . Of these residues, ␥Asn 263 is conserved; several other positions contain exclusively conservative substitutions (Table S2).

Is it possible to identify the residue(s) responsible for the nucleotide binding asymmetry?
To answer this question, all residues in the two identified segments were replaced by alanine. We made two alanine mutants per segment by replacing 5-6 residues at a time by alanine because making longer alanine mutants was challenging. Thus, we generated the mutations ␥A 5-10 , ␥A 10 -15 , ␥A 256 -260 , and ␥A 260 -265 for N-and C-terminal segments, respectively.
The results of the functional assays are shown in Table 1 and Fig. 2C. All four mutant enzymes had WT-like ATPase activity and a WT-like asymmetric MgATP binding pattern. Thus, it seems that no specific interaction between amino acid side chains in ␥ and ␤ (or ␣) causes the binding asymmetry. Instead, it appears as if the ␣-helices themselves in this region are responsible.

Discussion
One major goal of the research on ATP synthase in our lab at this time is to explore the molecular basis for the changes in affinity of the catalytic sites during rotation and to analyze the role of these affinity changes in coupling of catalysis and rotation. The affinity of a catalytic site for substrate(s) and product(s) at any given point of time is determined by the position of the central ␥ subunit. In the absence of ␥, all three sites have the same, rather low affinity. Several residues at the ␤/␥ interface that affect binding affinities of the catalytic sites have been described before (14,15,39). In the present study, we aimed to identify the part(s) of ␥ that included all residues that are responsible for the affinity differences. For this purpose, we measured nucleotide binding to the catalytic sites of two series of ␥ truncation mutants, one that had between 4 and 45 residues from the N terminus of ␥ removed and the other between 14 and 60 residues from the C terminus. The truncations were done using the enzyme from G. stearothermophilus because of its superior oligomeric stability. In the absence of ␥, it can form a stable ␣ 3 ␤ 3 complex (28), which has not yet been described for the E. coli enzyme. However, both enzymes are functionally so similar, specifically in their nucleotide binding pattern (for comparison, see the K d (MgATP) values for E. coli given in the Introduction and the G. stearothermophilus values in Table 1), that there is no reason to assume that these patterns were achieved by different mechanisms.
The truncation approach yielded two segments of about 10 residues, one on the N-terminal helix, ␥5-15, and the other on the C-terminal helix, ␥256 -265. Both segments are required to give the asymmetric, WT-like nucleotide binding pattern with pronounced affinity differences and a clearly expressed highaffinity site. If one of them is missing, the enzyme shows the symmetric binding behavior with a relatively low overall affinity that is also observed in the complete absence of ␥.
Over both segments, about half of the residues are conserved or conservatively substituted (see "Results" and Tables S1 and S2), and a number of these residues are able to form hydrogen bonds and/or salt bridges with residues on the ␣ or ␤ subunits (Fig. S3). However, experiments where the residues in both segments were replaced by alanine gave a WT-like asymmetric nucleotide binding pattern and WT-like catalytic activity. Thus, no individual amino acid side chains appear to be responsible for asymmetry and activity. This finding is reminiscent of the ␤DELSEED loop, which is essential for driving ␥ rotation and enzymatic activity due to its overall bulk shape, without individual interactions between amino acid side chains required for its function (35)(36)(37)(38).
The affinity-determining segment of the N-terminal helix, ␥5-15, makes contact with the DELSEED loop of the ␤ TP and ␤ DP subunits, which carry the high-and medium-affinity catalytic site, respectively (18). This observation confirms the importance of the DELSEED loop in affinity regulation of the catalytic sites as suggested by previous studies (14,15). Absence of the ␥5-15 segment apparently fails to bring the respective ␤ subunit into the high-or medium-affinity conformation. The affinity-determining segment of the C-terminal helix, ␥256 -265, is located at a level similar to its N-terminal counterpart, ␥5-15, offset by approximately one helical turn in direction away from the membrane (Fig. 3). A portion of the ␥256 -265 segment approaches the "catch loop" (31,40) of the ␤ E subunit, residues ␤309 -316. The ␥⌬C27 truncation removes any interaction of the C-terminal helix with the catch loop. Considering that the affinity of the low-affinity catalytic site on ␤ E is not affected by ␥, it seems possible that the role of this segment in determination of the affinity of the catalytic sites on ␤ TP and ␤ DP is indirect, by stabilizing the N-terminal ␥5-15 segment and its interactions with ␣ 3 ␤ 3 . Theoretically, it would also be possible that lack of this portion of the C-terminal helix might cause failure to form the N-terminal helix (41). However, we could show that a ␥ subunit consisting just of the N-terminal 35 or 42 residues was able to sustain a certain degree of ATP synthesis activity (19), which would be highly unlikely if the segment were unstructured. Furthermore, it has been shown that truncation of the N terminus by 50 residues (26) or of the C terminus by 36 residues (25) still allows production of torque close to 50% of WT, whereas the torque generated by an enzyme without both helices is significantly lower (24). Taken together, these findings suggest that each of the helices can

Affinity of catalytic sites of ATP synthase
exist without support of the other. Thus, the structural integrity of ␥ is preserved in the truncation mutants.
Interestingly, despite the presence of a high-affinity site, the ␥⌬C20 truncation mutant showed only low ATPase activity; adding back 6 more residues, in the ␥⌬C14 mutant, restored WT-like activities. This finding is in contrast to the situation at the N terminus where the presence of a high-affinity site went hand in hand with normal ATPase activities as observed with ␥⌬N4; the ␥⌬N9 mutant, which had largely lost high-affinity binding capability, also had substantially reduced enzymatic activity. Thus, as far as the mutants under investigation here are concerned, the presence of a high-affinity site appears to be a necessary, but not sufficient, condition for high activity.
The 2 residues immediately downstream of the identified segment, ␥Arg 266 and ␥Gln 267 are strictly conserved. However, from the results presented here, it is clear that they have no influence on the binding affinity of the catalytic sites as these residues are missing in ␥⌬C20, which nevertheless has a WTlike binding pattern. Given the importance of ␥Arg 266 and ␥Gln 267 for the catalytic function (40,42), 7 it is likely that their absence in ␥⌬C20 is responsible for the low enzymatic activity of this truncation mutant. The lack of residue ␥Thr 271 in ␥⌬C20 (which is present in ␥⌬C14) could be another factor for its catalytic impairment (42,43). It is interesting to note that loss of hydrogen-bonding capability of ␥Thr 271 resulted in reduced nucleotide-binding affinity of the catalytic site(s) in the transition state (measured as rate of formation of the transition-state analog MgADP-fluoroaluminate complex (43)). However, as the results presented here show, ␥Thr 271 has no role in determining the affinity in the ground state.

Bacterial strains and plasmids
For generation of the ␥ truncation mutants in G. stearothermophilus ␣ 3 ␤ 3 ␥ and the ␥-less ␣ 3 ␤ 3 subcomplexes, the background plasmid was pNM2. Plasmid pNM2 is a derivative of plasmid pKAGB1 (44). pKAGB1 is used to express a Cys-and Trp-less form of the ␣ 3 ␤ 3 ␥ subcomplex of G. stearothermophilus ATP synthase. pNM2 contains an additional mutation to generate an ␣ 3 (␤Y341W) 3 ␥ subcomplex, which allows monitoring nucleotide binding to the three catalytic sites, and a His 10 tag at the N terminus of the ␤ subunits to facilitate purification. Site-directed mutagenesis was performed using the QuikChange II XL kit. To generate the N-terminal truncations, downstream of the ␥ start codon ATG, the codons for the next 4 -45 amino acids were eliminated. To obtain the C-terminal truncations, stop codons were inserted at the desired positions. For generation of the ␥-less enzyme, a stop codon was introduced at the ␥7 position. The insertion of an NheI site downstream of this stop codon allowed us to remove the remainder of the gene for ␥ on an NheI-NheI fragment as there is a natural NheI site downstream of the ␥ gene. Removal of segments was confirmed by DNA sequencing of the plasmid product. For expression of the mutant proteins, the plasmids were transformed into E. coli strain JM103⌬(uncB-uncD).

Isolation of ␣ 3 ␤ 3 (␥) subcomplex and quantification of truncated ␥ subunit
The purification method of ␣ 3 ␤ 3 ␥ or ␣ 3 ␤ 3 subcomplex was modified from a previously described procedure (45). Cells were grown aerobically at 37°C in terrific broth medium containing 100 g/ml ampicillin. After cell lysis by French press, the cell debris was removed by centrifugation at 35,000 rpm for 30 min. The supernatant containing the complex was applied to a Ni 2ϩ -nitrilotriacetic acid column (Qiagen) equilibrated with 20 mM imidazole and 100 mM NaCl, pH 7.0. The column was washed with 50 mM imidazole and 100 mM NaCl, pH 7.0, and the enzyme was eluted with 500 mM imidazole and 100 mM NaCl, pH 7.0. The subcomplex was stored as precipitate in 70% saturated ammonium sulfate at 4°C. The amount of truncated ␥ subunit was determined via Western blotting using antibodies raised against a peptide corresponding to a part of the globular portion of ␥; the antibodies were a kind gift from Drs. Toshiharu Suzuki and Masasuke Yoshida (Japan Science and Technology Agency, Tokyo, Japan). ␣ 3 ␤ 3 ␥ subcomplex with full-length ␥ served as a standard. Blots were then developed with enhanced chemoluminescence substrate (Thermo Scientific). The expression level of ␥ subunit was quantified by measuring the integrated density of bands using a Photodyne imaging system and NIH ImageJ acquisition software. Control experiments with an anti-␤ antibody (Agrisera, Vännäs, Sweden) showed that the amount of ␤ was not affected by the mutations (Fig. S2, A and B).

Functional analysis of mutant enzymes
ATPase activities were assayed in a buffer containing 50 mM Tris/H 2 SO 4 , 10 mM ATP, and 4 mM MgSO 4 , pH 8.0, at 42°C. The reaction was started by addition of 10 -20 g/ml enzyme and stopped after 1 or 2 min (depending on the activity) by addition of SDS (final concentration, 5%, w/v). The released P i was measured as described (46). 1 unit of enzymatic activity corresponds to 1 mol of ATP hydrolyzed (equivalent to 1 mol of P i produced)/min.
Binding of MgATP to the catalytic sites of the purified ␣ 3 ␤ 3 or ␣ 3 ␤ 3 ␥ subcomplex was measured using the fluorescence of the inserted Trp residue ␤Trp 341 (7,8). Before use, the ␣ 3 ␤ 3 ␥ ammonium sulfate precipitate was pelleted by centrifugation and redissolved in a buffer containing 50 mM Tris/HCl and 10 mM CDTA, pH 8.0. After 1-h incubation at 23°C, the ␣ 3 ␤ 3 ␥ subcomplex was passed through two subsequent centrifuge columns containing Sephadex G-50 equilibrated with 50 mM Tris/HCl and 0.1 mM EDTA, pH 8.0. After this treatment, the 7 The cited references use the E. coli enzyme. Although at the N terminus of ␥ the amino acid numbers for G. stearothermophilus and E. coli are the same, at the C terminus adding 2 to the G. stearothermophilus numbers gives those for E. coli.

Affinity of catalytic sites of ATP synthase
enzyme subcomplex is essentially nucleotide-free (47); using the luciferin/luciferase method after heat denaturation of the enzyme complex, we found Ͻ0.1 mol of nucleotide (ATP plus ADP)/mol of subcomplex. To measure MgATP binding, fluorescence titrations were performed in a buffer containing 50 mM Tris/H 2 SO 4 and 2.5 mM MgSO 4 , pH 8.0, with ATP added to the desired concentration. K d values were determined by fitting of theoretical curves to the experimental data points by nonlinear least-squares analysis. All functional assays were performed within 2-3 days after preparation of the enzyme.

Miscellaneous
Protein concentrations were determined by the method of Bradford (48) using BSA as a standard. Secondary structure predictions were performed using the PredictProtein server (49). Figs. 3 and S3 were created using PyMOL (Schrödinger, Portland, OR) using Protein Data Bank code 1E79 (32) as starting material.