Chemical Mechanism of ATP Synthase

The chemical mechanism by which ATP synthases catalyze the synthesis of ATP remains unknown despite the recent elucidation of the three-dimensional structures of two forms of the F1 catalytic sector (subunit stoichiometry, α3β3γδε). Lacking is critical information about the chemical events taking place at the catalytic site of each β-subunit in the transition state. In an earlier report (Ko, Y. H., Bianchet, M. A., Amzel, L. M., and Pedersen, P. L. (1997) J. Biol. Chem. 272, 18875–18881), we provided evidence for transition state formation in the presence of Mg2+, ADP, and orthovanadate (Vi), a photoreactive phosphate analog with a trigonal bipyramidal geometry resembling that of the γ-P of ATP in the transition state of enzymes like myosin. In the presence of ultraviolet light and O2,the MgADP·Vi-F1 complex was cleaved within the P-loop (GGAGVGKT) of a single β-subunit at alanine 158, implicating this residue as within contact distance of the γ-P of ATP in the transition state. Here, we report that ADP, although facilitating transition state formation, is not essential. In the presence of Mg2+ and Vi alone the catalytic activity of the resultant MgVi-F1 complex is inhibited to nearly the same extent as that observed for the MgADP·Vi-F1 complex. Inhibition is not observed with ADP, Mg2+, or Vi alone. Significantly, in the presence of ultraviolet light and O2,the MgVi-F1 complex is cleaved also within the P-loop of a single β-subunit at alanine 158 as confirmed by Western blot analyses with two different antibodies, by N-terminal sequence analyses, and by quantification of the amount of unreacted β-subunits. These novel findings indicate that Mg2+ plays a pivotal role in transition state formation during ATP synthesis catalyzed by ATP synthases, a role that involves both its preferential coordination with Pi and the repositioning of the P-loop to bring the nonpolar alanine 158 into the catalytic pocket. A reaction scheme for ATP synthases depicting a role for Mg2+ in transition state formation is proposed here for the first time.

ATP synthases are involved in the synthesis of ATP from ADP and P i by oxidative phosphorylation in both aerobic bacteria and in the mitochondria of eukaryotic cells (1)(2)(3)(4) and by photosynthetic phosphorylation in chloroplast of plant cells (5). In all cases, the ATP synthase involved is comprised of two basic units, a water-soluble catalytic moiety called F 1 , which binds ADP and P i and synthesizes ATP, and a detergent-soluble unit called F 0 , which delivers the energy from an electrochemical proton gradient to the F 1 -ATP complex to induce the release of ATP. The F 1 unit consists of five different subunit types in the stoichiometric ratio ␣ 3 ␤ 3 ␥␦⑀ (6, 7). Two recently derived three-dimensional structures of F 1 (8,9) show that it consists of a hexagonal array of alternating ␣and ␤-subunits with a centrally located ␥-subunit, which has been shown to rotate during catalysis (10 -12). Much evidence supports the view that ATP synthesis at the quaternary structural level of F 1 occurs by a binding change mechanism (13,14), whereby energy from the electrochemical proton gradient, transmitted via the rotating ␥-subunit, induces the release of tightly bound ATP on one of the three ␤-subunits while promoting ATP synthesis from ADP and P i on a second ␤-subunit and binding of ADP and P i to a third. Following rotation of the ␥-subunit a full 360°, each of the three ␤-subunits has bound ADP and P i, synthesized ATP, and released ATP.
Despite our extensive knowledge about the events involved in ATP synthesis at the quaternary structural level of F 1 , our knowledge (15)(16)(17) is limited about the chemical events occurring at the active site of each ␤-subunit in the transition state, where chemical bond formation/breakage occur between ADP and P i to produce ATP and H 2 O. As a major step in this direction, we recently showed, in the presence of Mg 2ϩ , ADP, and orthovanadate (V i ), 1 a photoreactive phosphate analog, that an inhibited MgADP⅐V i -F 1 transition state-like complex is formed (16), similar to that reported for myosin (18 -20) where the predicted trigonal bipyramidal geometry of V i has been visualized by x-ray analysis of the MgADP⅐V i -myosin complex (21). Significantly, photocleavage of the protein backbone in the presence of uv light and O 2 of both the MgADP⅐V i -F 1 and the MgADP⅐V i -myosin transition state complexes occurs at the third position in the P-loop region (GXXXXGKT) (16,22). In F 1 , a conserved alanine (158 in the animal enzymes) occupies this position (Fig. 1A). (The chemistry involved in the photocleavage events has been described (23).) Specifically, as these earlier results relate to the mechanism of the ATP synthase reaction, they indicate that alanine 158, which in the two reported crystal structures of F 1 (8,9) is near but not within the catalytic pocket (residing, respectively, Ͼ6.0 Å and Ͼ4.5 Å away from the ␤ and ␥-P atoms of MgADP and MgATP (8)), moves within the catalytic pocket in the transition state (Fig. 1A).
Here, we report the novel finding that transition state formation in the ATP synthase reaction does not require the presence of ADP and occurs in the presence of only Mg 2ϩ and V i . As V i alone is without effect on transition state formation, these findings have rather profound implications for the role of Mg 2ϩ in the reaction mechanism of ATP synthases as described below.

Materials
The source of rats (Harlan Sprague-Dawley, white males) for the preparation of the F 1 moiety of the ATP synthase was Charles River Breeding Laboratories. Two different polyclonal antibodies against the F 1 -␤-subunit were raised in rabbits using the synthetic peptides KIGLFGGAGVGKCT (antibody 1) and YVPADDLTDPAPATTFAHL-DAC (antibody 2). Antibody 1 has been shown in our laboratory to recognize only the epitope KIGLFGG. Western reagents, orthovanadate (V i ), and PVDF membranes were obtained, respectively, from Amersham Pharmacia Biotech, Sigma, and Millipore.

Methods
Purification and Assay of the F 1 Moiety of Mitochondrial ATP Synthase-F 1 was prepared from rat liver mitochondria by the procedure of Catterall and Pedersen (6) with the modification described by Pedersen et al. (24) and then stored and processed for the studies described here exactly as outlined by Ko et al. (16). ATPase activity was assayed by a spectrophotometric procedure in which ADP formed was coupled to the pyruvate kinase and lactic dehydrogenase reactions (6).
Prior Treatment of F 1 with Orthovanadate (V i ) and F 1 Ligands-The V i solution used for these studies was carefully prepared exactly as described previously (16) to minimize the presence of polymeric species and forms other than V i . The V i concentration was determined by measuring the optical density at 265 nm and using the extinction coefficient 2925 M Ϫ1 cm Ϫ1 . F 1 (50 g) was prior incubated for the times indicated in the figure legends to Figs SDS-PAGE and Western Blot Analysis-SDS-PAGE was carried out according to the procedure of Laemmli (25) in 15% acrylamide using a Bio-Rad Mini-Protean dual slab cell. The Coomassie-stained bands were subjected to densitometric analysis using a Fujifilm Bas-1500 phosphorimager and MacBas (version 2.31) software. For Western blot analysis, proteins, after SDS-PAGE, were transferred electrophoretically onto a PVDF membrane and then probed with two different F 1 antibodies (see "Materials") exactly as described previously (16) using horseradish peroxidase-conjugated anti-rabbit IgG as a secondary antibody and the enhanced chemiluminescence (ECL) system for detection.
N-terminal Sequence Analysis-The 17-and 34-kDa peptide fragments of the F 1 -␤-subunit were transferred onto PVDF membranes as described previously (16) and then subjected to N-terminal sequencing using an Applied Biosystems 475A Protein Sequencing System (26).
Protein Determination-Protein was determined by Pierce's Coomassie dye binding assay protocol using bovine serum albumin as standard.

RESULTS AND DISCUSSION
The F 1 Moiety of Mitochondrial ATP Synthase Forms an Inhibitory Complex in the Presence of Only Mg 2ϩ and V i -In a previous report (16), we showed that when incubated with Mg 2ϩ , ADP, and V i , the F 1 moiety of mitochondrial ATP synthase forms a MgADP⅐V i -F 1 inhibitory complex, and that under the same incubation conditions, V i , Mg 2ϩ , ADP, MgADP, and V i ϩ ADP have little or no effect on catalytic activity. These findings, when taken together with the additional findings made, i.e. that inhibition of F 1 by MgADP⅐V i occurs under turnover conditions, is reversible and results in specific cleavage at alanine 158 of the P-loop region (Fig. 1A) of a single ␤-subunit (catalytic subunit) in the presence of uv light and O 2 , provided evidence that the MgADP⅐V i -F 1 complex represents a transition state-like intermediate. These findings also implicated movement of alanine 158 into the catalytic pocket (Fig.  1A). Significantly, results presented in Fig. 1B show that when F 1 is prior incubated in the presence of Mg 2ϩ and V i alone, the catalytic activity of F 1 is inhibited to nearly the same extent as that observed in the presence of Mg 2ϩ , ADP, and V i , although the formation of the inhibited complex is faster when ADP is present. In control studies, prior incubation of F 1 with Mg 2ϩ alone (16), V i alone, ADP ϩ V i , or ADP ϩ Mg 2ϩ had little or no inhibitory capacity (Fig. 1C). These findings indicated that Mg 2ϩ plays an important role in transition state formation in the ATP synthase reaction. Therefore, the MgV i -F 1 complex was studied in greater depth.
In the Presence of UV Light and O 2 , the MgV i -F 1 -inhibited Complex Is Cleaved in the Third Position (Alanine 158) of the P-loop Region within the ␤-Subunit-Our previous study showed that when the MgADP⅐V i -F 1 complex is subjected to uv light (320 nm) and O 2 (atmospheric conditions), the ATP synthase ␤-subunit is cleaved at alanine 158 of the P-loop into a 17-kDa fragment and a 34-kDa fragment (16). It is expected that if the MgV i -F 1 inhibitory complex mimics a transition state-like intermediate in the ATP synthase reaction pathway, then the cleavage site in the presence of uv light and O 2 will be identical to that observed previously for the MgADP⅐V i -F 1 com- FIG. 1. A, transition state formation at the active site of the F 1 moiety of ATP synthase. The diagram summarizes conclusions made from an earlier study (16) in which F 1 was trapped in an inhibitory MgADP⅐V i -F 1 transition state-like complex and then subjected to photocleavage. The studies implicated a movement of the ␤-subunit P-loop region such that alanine 158, which resides at a nonbonding distance from the ␤-P atom of ADP and the P atom of P i in the substrate bound state (8,9), is brought within contact distance in the transition state. B, demonstration that the catalytic activity of F 1 is inactivated to nearly the same extent with Mg 2ϩ and V i as with Mg 2ϩ , ADP, and V i in the absence of uv light. F 1 was incubated prior to assay for the times indicated in the absence (no ligand) and presence of 200 M each of the indicated components exactly as described under "Methods." One-hundred percent (100%) refers to the catalytic activity of F 1 in the absence of added ligand. C, experimental controls. The data presented demonstrate that, under the conditions described in B, V i alone, ADP ϩ V i , and ADP ϩ Mg 2ϩ have little or no effect on the catalytic capacity of F 1 . (Note: in the presence of uv light and O 2 (atmospheric conditions), the formation of both the MgADP⅐V i -F 1 and MgV i -F 1 inhibitory complexes are facilitated such that maximal inhibition occurs within 1 h under the conditions described above.) plex (16). Results presented in Fig. 2 show that this is indeed the case. Thus, Fig. 2A, which compares the SDS-PAGE patterns obtained after the two different inhibitory complexes had been subjected to uv light and O 2 for the time period ranging from 0 min to 1 h, clearly shows in both cases (lanes 3-7 for the MgADP⅐V i -F 1 complex and lanes 8 -12 for the MgV i -F 1 complex) the appearance of two new bands, 34 and 17 kDa, relative to control F 1 (lanes 1 and 2) depicting the five different F 1 subunits (␣, ␤, ␥, ␦, and ⑀). When the SDS-PAGE gels were subjected to Western blot analysis using one ␤-subunit antibody with its epitope (KIGLFGG, residues 151-157) overlapping with the P-loop region and a second ␤-subunit antibody with its epitope (YVPADDLTDPAPATTFAHLDA, residues 311-331) in the C-terminal region (Fig. 2B), the appearance of the 17-kDa band and the 34-kDa band could be observed throughout the course of the cleavage reaction as shown in Fig. 2, C and D, respectively. Consistent with the finding presented in Fig. l showing that the MgADP⅐V i -F 1 inhibitory complex is formed faster than the MgV i -F 1 complex, results presented in Fig. 2, C and D, show that the former complex is also cleaved faster in the presence of uv light and O 2 than the latter complex (compare lanes 2-6, MgADP⅐V i -F 1 , with lanes 7-11, MgV i -F 1 ).
Results obtained from N-terminal sequence analyses of the 17-and 34-kDa fragments (Fig. 3, A and B, respectively), together with the data obtained with the antibodies, respectively, recognizing epitope 1 (KIGLFGG) and epitope 2, confirm that the MgV i -photoinduced cleavage site in the F 1 -␤-subunit lies at alanine 158 within the P-loop (GGAGVGKT). Thus, data in Fig.  3A clearly identify the 17-kDa fragment as commencing from the N terminus of the ␤-subunit (known sequence ϭ APKAGTA for isolated rat liver F 1 ), and data in Fig. 3B indicate that this fragment must extend as far as glycine 157, the second amino acid residue within the P-loop. The N-terminal sequence (GVGKTVL) of the 34-kDa fragment (Fig. 3B) commences at glycine 159, one residue after alanine 158 within the P-loop. Thus, alanine 158 must be the amino acid residue that was photo-oxidized and cleaved at the junction between the end of the 17-kDa fragment and the beginning of the 34-kDa fragment.
Only One of Three ␤-Subunits of the MgV i -F 1 Fig. 3C summarizes results obtained by quantifying both the staining intensities of the 17-and 34-kDa bands and the ␤-subunit bands following SDS-PAGE of the untreated MgV i -F 1 inhibitory complex and the uv light treated complex (Fig. 2A). These data show that 67% of the F 1 ␤-subunit band remains, while the lost 33% is accounted for by the appearance of the 17-and 34-kDa cleavage products (Fig. 3C, right column). Thus, only one of the three ␤-subunits has undergone cleavage, a result  Fig. 2A. N-terminal sequence analyses were carried out exactly as described under "Methods." C, quantification of the relative amounts of total protein attributable to the three ␤-subunits in Mg 2ϩ , ADP, V i , and Mg 2ϩ , V i -treated F 1 that have been converted to the 17-kDa and 34-kDa cleavage products following treatment with UV light. The Coomassie-stained bands were subjected to densitometric analysis using a Fujifilm Bas-1500 phosphorimager and MacBas (version 2.31) software.

-inhibited Complex Is Cleaved in the Presence of UV Light and O 2 -
identical to that obtained previously for the MgADP⅐V i -F 1 complex (16) and confirmed as a control in this study (Fig. 3C, left  column). These results are consistent with the binding change mechanism for ATP synthesis, which views ATP as being synthesized on only one ␤-subunit at a time (13,14).
The novel studies reported here indicate that Mg 2ϩ plays a pivotal role in the formation of the transition state in the ATP synthase catalyzed reaction. Speculation as to what this role may be is greatly aided and justified by the recent availability of crystal structures of two different states of F 1 (8,9), one which has ␤-subunits with ADP and P i bound at the active site but no Mg 2ϩ (9), referred to here as "␤ DP,Pi ," and the other, which has the nonhydrolyzable MgATP analog, MgAMP-PNP, bound at the active site of one ␤-subunit (8), previously called "␤ TP ." If it is assumed that the ␤ DP,Pi and the ␤ TP -subunits are representative, respectively, of the substrate and product bound states during ATP synthesis, the role of Mg 2ϩ in transition state formation based on work described here is best depicted as shown in Fig. 4. Thus, considering the fact that within ␤ DP,Pi (Fig. 4, top panel) the ␤-carbon atom of alanine 158 in the P-loop lies at a nonbonding distance of Ͼ7 Å from both the ␤-P atom of ADP and the P atom of P i (9) but in the transition-like state (Fig. 4, center panel) induced by Mg 2ϩ , this amino acid lies sufficiently close to V i to be oxidatively cleaved, the implication seems clear that a local remodeling of the active site has occurred. It is suggested here that during ATP synthe-sis Mg 2ϩ induces a conformational change in the ␤ DP,Pi -subunit such that the P-loop region containing alanine 158 moves into the active site pocket, and because of its nonpolar nature, displaces a nearby water molecule known to be present (9). Simultaneously, the conformational change induced by the bound Mg 2ϩ results in the proper alignment of ADP and P i with a catalytic base, most likely glutamic acid 188 (8,9). While remaining bound to P i , Mg 2ϩ is depicted as facilitating the departure of water, thus resulting in ATP formation. Subsequently, the ␤-subunit involved relaxes to the ␤ TP form (product bound state) in which the ␤-carbon atom of alanine 158 now lies again at nonbonding distances from both the ␤-P atom of ATP and from the newly formed ␥-P atom (Fig. 4, lower panel). In ␤ TP , the Mg 2ϩ is known to be coordinated to the ␤ and ␥ phosphate oxygens, whereas, in the transition state depicted here, it preferentially coordinates to P i (Fig. 4, center panel), thus implying a positional shift after ATP is formed (Fig. 4, lower panel). It will be noted also that in the final product bound state that two water molecules are known to be present (8). One is likely that derived from the formation of ATP, the other may be the water molecule found originally in the ␤ DP,Pi substrate bound state (Fig. 4, top panel), which returns following the exit of alanine 158 from the active site pocket after ATP synthesis. FIG. 4. Diagram illustrating the pivotal role that Mg 2؉ may play in transition state formation in the ATP synthase-catalyzed reaction. See text for description. Please note that to emphasize the importance of Mg 2ϩ in formation of the transition state, we have depicted its entry following that of ADP and P i . However, the order in which Mg 2ϩ enters the reaction remains to be elucidated.