The rotary mechanism of ATP synthase Daniela Stock*, Clyde Gibbons*, Ignacio Arechaga*, Andrew GW Leslie and John E Walker*

Transcription

1 672 The rotary mechanism of ATP synthase Daniela Stock*, Clyde Gibbons*, Ignacio Arechaga*, Andrew GW Leslie and John E Walker* Since the chemiosmotic theory was proposed by Peter Mitchell in the 1960s, a major objective has been to elucidate the mechanism of coupling of the transmembrane proton motive force, created by respiration or photosynthesis, to the synthesis of ATP from ADP and inorganic phosphate. Recently, significant progress has been made towards establishing the complete structure of ATP synthase and revealing its mechanism. The X-ray structure of the F 1 catalytic domain has been completed and an electron density map of the F 1 c 10 subcomplex has provided a glimpse of the motor in the membrane domain. Direct microscopic observation of rotation has been extended to F 1 -ATPase and F 1 F o -ATPase complexes. Addresses *The Medical Research Council Dunn Human Nutrition Unit, Hills Road, Cambridge CB2 2XY, UK The Medical Research Council Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK walker@mrc-dunn.cam.ac.uk Current Opinion in Structural Biology 2000, 10: X/00/$ see front matter Published by Elsevier Science Ltd. Abbreviations EM electron microscopy F 1 factor 1 F o factor oligomycin OSCP oligomycin sensitivity conferring protein Pi inorganic phosphate pmf proton motive force Introduction ATP synthase contains a rotary motor involved in biological energy conversion. Respiratory complexes in mitochondria and eubacteria, and photosynthetic complexes in chloroplasts and photosynthetic eubacteria use energy derived from the oxidation of nutrients and from light, respectively, to generate a transmembrane proton motive force (pmf) [1 3]. ATP synthase uses the pmf to make ATP from ADP and inorganic phosphate (Pi). As summarised in Figure 1a,b, the enzyme has two major structural domains, known as F 1 (factor 1) and F o (factor oligomycin). The globular F 1 catalytic domain in the mitochondrial enzyme is an assembly of five subunits with the stoichiometry 3 3 γ 1 δ 1 ε 1. Subunits γ, δ and ε form a central stalk linking the () 3 subcomplex of F 1 to the membrane domain, F o. The () 3 subcomplex and F o are also linked by a peripheral stalk, sometimes called the stator [2]. In the F 1 domain, the three subunits and the three subunits are arranged alternately around a central -helical coiled coil in the γ subunit [4]. This arrangement suggested that the enzyme works by a mechanism involving the cyclic modulation of nucleotide affinity in catalytic subunits, as required by the binding-change mechanism [1], by rotation of the asymmetrical γ subunit. During ATP synthesis, the rotation would be generated in F o and fuelled by the pmf. During ATP hydrolysis in F 1 F o (or in F 1 alone), the energy released by hydrolysis would drive rotation in the opposite direction and reverse the direction of proton translocation. Subsequently, the rotation of the γ subunit in an () 3 γ complex was observed directly by microscopy and was shown to depend on ATP hydrolysis [5]. Recent structural results have provided additional insight into the nature of the central stalk [63 ]. This feature links the F 1 and F o domains, and forms part of the rotor in the ATP synthase molecular motor. The way in which the central stalk is linked to a ring of c subunits in the F o domain has been suggested from a low-resolution electron density map of a subcomplex of the yeast enzyme. As yet, no structural information is available on other key subunits in the F o domain, but a number of models have been proposed for torque generation. A much clearer picture of the molecular mechanism of the motor in ATP synthase is slowly emerging. The central stalk Until recently, the protruding part of the central stalk was disordered in crystals of bovine F 1 -ATPase [4], although the () 3 domain and the penetrating -helical coiled-coil part of the central stalk were resolved in the same crystals. By modification of the cryoprotection conditions, the crystal lattice of bovine F 1 -ATPase (covalently inhibited with dicyclohexylcarbodiimide) has been shrunk, thereby ordering the protruding central stalk region and allowing the entire structure to be resolved to 2.4 Å (Figure 1c) (C Gibbons, MG Montgomery, AGW Leslie, JE Walker, unpublished data; see [63 ]). This analysis has revealed a new / domain in the γ subunit, containing a Rossmann fold, that does not bind nucleotides. It appears to be a buttress, stabilising the lower section of the coiled-coil shaft. There is little agreement between the structure of the bovine γ subunit in the Rossmann fold region of the central stalk and a model of the same region of the Escherichia coli γ subunit, deduced from a 4.4 Å resolution electron density map of bacterial F 1 -ATPase [6]. The bovine structure confirms the structural homology between the mitochondrial δ and bacterial (and chloroplast) ε subunits. Similar to the bacterial ε subunit [7], the bovine δ subunit has two domains, an N-terminal sandwich with 10 strands (residues 15 98) and a C-terminal -helical hairpin (residues ). The 50 amino acid bovine ε subunit has no counterpart in bacteria or chloroplasts. It has a helix-loop-helix structure and appears to

2 The rotary mechanism of ATP synthase Stock et al. 673 Figure 1 Structure of ATP synthase. (a,b) Summary of current knowledge of the structure of ATP synthase from mitochondria and eubacteria. (a) Mitochondrial ATP synthase. The model is based on EM studies of single particles [17 ]. It incorporates the structure of bovine F 1 -ATPase [4,62 ] and information from the electron density map of the F 1 c 10 complex from S. cerevisiae [36 ]. The composition, stoichiometry and arrangement of the subunits in the peripheral stalk (subunits OSCP, F 6, b and d) come from biochemical and reconstitution studies [20,33]. The position of subunit a relative to the c 10 ring was deduced from studies of the bacterial enzyme [28]. Minor subunits (e, f, g, A6L) in the F o domain are not shown. They have no known functions in the enzyme s mechanism. (b) Eubacterial ATP synthase. The overall model is also based on EM studies [14,15 ]. The core structure of the central F 1 c ring was deduced by homology with the mitochondrial enzyme. However, the c ring may contain 12 c subunits, not 10 [45]. The positions of the subunits in the peripheral stalk (subunits b and δ) are supported by biochemical and EM studies [14,18,26 ]. The δ subunit (structure determined by NMR studies [27]) appears, from EM work, to sit on top of the () 3 domain [26 ]. The structure of the E. coli ε subunit was also determined independently [7,11]. The general structure of ATP synthase from chloroplasts is very similar to that of the bacterial enzyme. The main differences are that the c ring may contain 14 c protomers [46 ] and that the two identical b subunits in some eubacterial enzymes are replaced by homologous, but not identical, subunits b and b. A similar arrangement of b and b subunits is also found in other eubacterial species. (c) The complete structure of bovine F 1 -ATPase shown in stereo (C Gibbons, MG Montgomery, AGW Leslie, JE Walker, unpublished data; see [63 ]). The and subunits (red and yellow, respectively) are arranged alternately around an -helical coiled coil in the γ subunit (blue). Regions of the γ subunit present in the original F 1 structure [4] are shown in sky blue, those regions determined in the latest structure [63 ] are in dark blue. The central stalk consists of the γ, δ and ε subunits (blue, green and magenta, respectively). (a) (c) OSCP γ δ ε c 10 a b F 6 d (b) δ γ ε b 2 c 9 12 a Current Opinion in Structural Biology stabilise the foot of the central stalk, where the γ, δ and ε subunits all interact extensively. It is probable that all three subunits contact the F o domain. In E. coli F 1 -ATPase, interactions between and within subunits have been examined by the introduction of cysteine residues at specific sites and formation of disulfide crosslinks by oxidation. Cross-links observed within the bacterial ε subunit [8 ] and ε γ cross-links [9,10] are consistent with the bovine model, but the ε and ε cross-links [11 13] are not, as they are between 40 and 60 Å apart in the bovine structure. One possible interpretation is that the bacterial ε subunit detaches wholly or partially from the foot during the catalytic cycle, so that it can interact with the lower surface of the () 3 domain. However, the functional significance of such a rearrangement is obscure. A critical re-examination of the formation of the ε and ε cross-links is warranted. The peripheral stalk There is general agreement that the F 1 and F o domains are also connected by a second, peripheral, stalk [2]. This has been observed by single-particle analysis using electron microscopy (EM) in negative stain of bacterial [14,15 ],

3 674 Proteins Figure 2 50 Å 83 Å (a) γ 58 Å δ C (b) 55 Å δ Current Opinion in Structural Biology Stereo views of an electron density map of the F1 c10 complex from S. cerevisiae at 3.9 Å resolution [36 ]. (a) Side view. (b) End-on view, rotated 90 with respect to (a). Two rings, an inner ring and an outer ring, composed of 10 c protomers are visible. The inserts indicate the locations of subunits. chloroplast [16] and mitochondrial [17 ] F1Fo-ATPases. Its function has not been demonstrated, but it may act as a stator to counter the tendency of the ()3 domain to follow the rotation of the central stalk [2]. In E. coli, it contains the δ subunit and the extrinsic membrane domains of two identical b subunits that form a parallel -helical coiled coil [18 ] (see Figure 1b). The membrane domains of the b subunits (one transmembrane helix each) also interact and form part of Fo [19 ]. In some other bacterial species and in chloroplasts, the two identical b subunits are replaced by single copies of homologous subunits b and b. The bovine peripheral stalk contains one copy each of the OSCP (oligomycin sensitivity conferring protein) subunit (the equivalent of bacterial δ), the extrinsic domain of subunit b and the d and F6 subunits [2] (see Figure 1a). It has been assembled in vitro and interacting regions have been

4 The rotary mechanism of ATP synthase Stock et al. 675 Figure 3 (c) (a) (b) Actin filament F o c ring γ Actin filament ε Actin filament F 1 ATP ATP ATP ADP + Pi ADP + Pi ADP + Pi Current Opinion in Structural Biology Observations of rotation in ATP synthase. The direct observation of rotation using fluorescently labelled actin filaments attached to (a) the γ subunit in the () 3 γ complex [5,50,52 ], (b) F 1 -ATPase [51,53 ] and (c) F 1 F o -ATP synthase [55,56 ]. The N termini of subunits in the () 3 domain are associated with a nickel-coated glass surface. Counterclockwise rotation dependent on ATP hydrolysis was observed in a fluorescence microscope. defined [20]. In Saccharomyces cerevisiae, cross-links have been observed between the b subunit and subunits, OSCP and d (in agreement with the bovine findings), and also to the membrane subunit a (and other minor F o subunits) [21,22]. The peripheral stalk subunits are poorly conserved (relative to F 1 components, for example) and subunits b can be shortened and lengthened without having a major effect on the enzyme s activity [23,24]. For many years, it has been known that the δ and OSCP subunits in the E. coli and bovine enzymes, respectively, interact with the N-terminal regions of the subunits, which protrude from the crown at the top of F 1. This arrangement has been confirmed by cross-linking experiments [25] and EM [26 ]. The structure of the N-terminal domain of the E. coli δ subunit has been established by NMR studies [27]. The F o domain In E. coli, the F o domain is composed of three subunits with the stoichiometry a 1 b 2 c 9 12 (Figure 1b). The a and c subunits are in contact and protons are thought to be translocated through the interface between them [28,29]. Both subunits are conserved in all F-ATPases. The E. coli a subunit is hydrophobic and is probably folded into five transmembrane helices [30,31]. It contains basic and acidic residues (Arg210, His245, Glu196, Glu219) that are essential for proton translocation. The c subunit is also hydrophobic. The protomer structure, determined by NMR spectroscopy in organic solvents, has two transmembrane helices linked by a polar loop [32]. The C-terminal helix contains a carboxyl group (Asp61) that is also essential for proton translocation. The conservation and arrangement of the b subunits was discussed above. The F o domains of mitochondrial enzymes contain a number of small subunits that appear to have no direct role in catalysis [33 35]. They are absent from bacterial and chloroplast enzymes. The first view of the structure of the F o domain came from an electron density map of F 1 -ATPase associated with a ring of 10 c subunits from S. cerevisiae [36 ] (see Figure 2). This F 1 c 10 complex was formed from ATP synthase during the crystallisation process, when other subunits dissociated. The electron density map contains a number of important features. First, the 10 c protomers appear to have secondary structure similar to the c protomer structure determined by NMR. The map also shows that the C-terminal helices form an outer ring, with the N-terminal helices in a second inner ring. Second, the map shows that the extensive footprint of the central stalk sits asymmetrically on the polar loop regions of six c subunits. This arrangement is consistent with the rotation of the central stalk and the c ring as an ensemble, as are covalent crosslinks between the E. coli ε and c subunits that do not affect the enzyme s activity [37,38,39 ]. Third, 10 c subunits are found in the ring and not 12, as was widely anticipated. Therefore, there is a symmetry mismatch between the

5 676 Proteins Figure 4 (a) H + (b) (c) Na + a subunit c subunits a subunit c subunits 140 a c H + Na + Current Opinion in Structural Biology Models of the generation of rotation by movement of ions through the F o domain of ATP synthase. (a) A two-channel model proposed by Junge [2,58]. Two half channels across the interface between the a subunit and the c ring are linked by rotation of the c ring. (b) A single-channel model [60 ] for the Na + -motive ATP synthase in P. modestum. Sodium ions enter via a channel in the interface between the a subunit and the c ring, and bind to c protomers near to the cytoplasmic surface where they are released. (c) A model based on ph-induced structural changes observed by NMR of the c protomer in organic solvents [61 ]. Deprotonation of Asp61 and release of the proton triggers a 140 rotation of the c protomer C-terminal (outer) helix and concomitant movement of the c ring. The observed direction of rotation in Figure 3 is counterclockwise, as viewed from the membrane towards F 1, and driven by ATP hydrolysis. In Figure 4, the direction of rotation during ATP synthesis is counterclockwise, as viewed from F 1 towards the membrane. F 1 and F o domains, which may help to facilitate rotation by avoiding the deeper energy minima that would accompany matching symmetries. Symmetry mismatch has been discussed in relation to other macromolecular assemblies that contain rotating elements [40 43]. The number of c subunits in the c ring Based on metabolic labelling and mechanistic models of the generation of rotation, the notion has grown up that E. coli F o contains 12 c subunits arranged in a ring and, by implication, that mitochondrial and chloroplast F o domains also contain 12 c subunits similarly arranged. Cross-linking experiments and genetic fusions [44,45] have been interpreted as supporting this view. This notion has been challenged by the F 1 c 10 structure (above) [36 ] and by the observation of 14-fold symmetry in rings of c subunits from spinach chloroplasts [46 ]. At the present time, the possibility that subunits were lost from the S. cerevisiae c ring during crystallisation cannot be excluded, unlikely as this proposal seems. However, there are now clear indications that the c-ring symmetry may differ among species. The c-ring symmetry may also vary within a single species according to physiological conditions [47]. If the concept of symmetry mismatch is an important general feature of ATP synthases, it would argue against c-ring stoichiometries divisible by three. It also implies that the number of protons that transverse the membrane for each ATP synthesised is nonintegral, possibly between three and four in mitochondria. As the generation of each ATP requires a 120 rotation of the central stalk, an elastic element, possibly in the γ subunit, may be needed to store energy and release it in quanta, as required by a stepping motor mechanism [48,49 ] (see below). Direct observation of rotation By attachment of fluorescent actin filaments to either the γ or ε subunit, rotation of the central stalk driven by ATP hydrolysis has been observed by microscopy of tethered 3 3 γ [5,50 ] complexes and of F 1 itself [51,52,53 ] (see Figure 3a,b). The main characteristics of this rotation are that it is highly efficient in energy usage, that it proceeds in 120 steps [54] and that the rotation is counterclockwise as viewed from the tip of the central stalk protrusion. Attempts have also been made to observe the rotation in F 1 F o -ATPase preparations by attaching actin filaments to the c ring on the surface distal from F 1 [55,56 ] (Figure 3c). Although technical objections have been voiced concerning these experiments [57 ], they can be reasonably interpreted as showing that the F 1 c ring rotates as an ensemble in response to ATP hydrolysis in F 1. However, because the detergents used to isolate the complex destabilise interactions of the c ring with the a subunit, these experiments should not be taken as definitive proof of the rotation of the F 1 c ring in an intact F 1 F o complex that is capable of synthesising, as well as hydrolysing, ATP. Definitive proof may require rotation to be observed under conditions in which ATP is being synthesised. Generation of torque A hypothetical model of how rotation might be generated was developed by Junge et al. [58], based upon models of bacterial flagellar rotation (see [3,49,59] for a detailed description and further discussion of this model) (Figure 4a). A related model has been described to explain the generation of rotation by the Na + -motive F 1 F o -ATPase from the bacterium Propionigenium modestum [60 ]

6 The rotary mechanism of ATP synthase Stock et al. 677 (Figure 4b). In this model, the carboxyl sidechains of the essential residue Glu65 in subunit c are negatively charged when they enter the interface between the c ring and subunit a. The positive charge of Arg227 in subunit a attracts the negative charge of the essential carboxylate in subunit c and also prevents ion leakage. Once this carboxylate has been neutralised by a Na + ion from the periplasm, it will move by thermal vibrations, bringing the next negatively charged carboxylate into the channel. Electrostatic forces strongly bias the rotation, making it effectively unidirectional. As in the Junge model, the central stalk is attached to the c ring, which drives its rotation directly. A radically different model for the generation of rotation of the central stalk has been advanced on the basis of NMR studies in organic solvents of the c protomer from E. coli, in which reduction in ph and protonation of Asp61 cause the C-terminal helix to rotate by 140 about its helix axis. It is proposed that this rotation either drives the rotation of the c ring (Figure 4c) or, alternatively, generates rotation of the central stalk without the c ring itself turning [61 ]. Conclusions The rather extensive current knowledge of how ATP synthase works is based largely upon accurate and novel structures of subcomplexes of the enzyme [4,36,62,63 ]; striking progress had been made using this approach in the past six years. However, current models for explaining the generation of rotation in F o are tentative and require further experimental validation. It is unlikely that the mechanism of rotation in ATP synthase will be understood fully until accurate molecular models of the entire enzyme complex in different conformational states have been established. Determination of these structures requires either the crystallisation of the intact ATP synthase complex or the establishment of an accurate low-resolution model by EM of single complexes, which can then be used as a framework for building a molecular model from structures of subcomplexes and individual subunits. References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest 1. Boyer PD: The ATP synthase a splendid molecular machine. Annu Rev Biochem 1997, 66: Walker JE: ATP synthesis by rotary catalysis (Nobel Lecture). Angew Chem Int Ed Engl 1998, 37: Nakamoto RK, Ketchum CJ, Alshawi MK: Rotational coupling in the F o F 1 ATP synthase. Annu Rev Biophys Biomol Struct 1999, 28: Abrahams JP, Leslie AGW, Lutter R, Walker JE: Structure at 2.8 Å resolution of F 1 -ATPase from bovine heart mitochondria. Nature 1994, 370: Noji H, Yasuda R, Yoshida M, Kinosita K: Direct observation of the rotation of F 1 -ATPase. Nature 1997, 386: Hausrath AC, Gruber G, Matthews BW, Capaldi RA: Structural features of the gamma subunit of the Escherichia coli F 1 - ATPase revealed by a 4.4 Å resolution map obtained by X-ray crystallography. Proc Natl Acad Sci USA 1999, 96: Uhlin U, Cox GB, Guss JM: Crystal structure of the epsilon subunit of the proton-translocating ATP synthase from Escherichia coli. Structure 1997, 5: Schulenberg B, Capaldi RA: The epsilon subunit of the F 1 F o complex of Escherichia coli cross-linking studies show the same structure in situ as when isolated. J Biol Chem 1999, 274: The authors provide evidence that the isolated ε subunit has a similar conformation [11] as when it is associated with F 1 -ATPase. 9. Watts SD, Tang CL, Capaldi RA: The stalk region of the Escherichia coli ATP synthase tyrosine 205 of the gamma-subunit is in the interface between the F 1 and F o parts and can interact with both the epsilon and c oligomer. J Biol Chem 1996, 271: Tang CL, Capaldi RA: Characterization of the interface between gamma and epsilon subunits of Escherichia coli F 1 -ATPase. J Biol Chem 1996, 271: Wilkens S, Capaldi RA: Solution structure of the epsilon subunit of the F 1 -ATPase from Escherichia coli and interactions of this subunit with beta subunits in the complex. J Biol Chem 1998, 273: Aggeler R, Haughton MA, Capaldi RA: Disulfide bond formation between the COOH-terminal domain of the beta subunits and the gamma and epsilon subunits of the Escherichia coli F 1 -ATPase. Structural implications and functional consequences. J Biol Chem 1995, 270: Aggeler R, Weinreich F, Capaldi RA: Arrangement of the epsilon subunit in the Escherichia coli ATP synthase from the reactivity of cysteine residues introduced at different positions in this subunit. Biochim Biophys Acta 1995, 1230: Wilkens S, Capaldi RA: ATP synthase s second stalk comes into focus. Nature 1998, 393: Bottcher B, Bertsche I, Reuter R, Graber P: Direct visualisation of conformational changes in EF o F 1 by electron microscopy. J Mol Biol 2000, 296: The authors describe the first three-dimensional reconstruction of E. coli ATP synthase. 16. Bottcher B, Schwarz L, Graber P: Direct indication for the existence of a double stalk in CF 0 F 1. J Mol Biol 1998, 281: Karrasch S, Walker JE: Novel features in the structure of bovine ATP synthase. J Mol Biol 1999, 290: Single-particle analysis of electron micrographs of bovine F 1 F o -ATPase. Evidence is revealed for a peripheral stalk and for formerly unseen features. 18. Revington M, McLachlin DT, Shaw GS, Dunn SD: The dimerization domain of the b subunit of the Escherichia coli F 1 F o -ATPase. J Biol Chem 1999, 274: A biochemical demonstration of the dimerisation of the bacterial b subunit, showing that the b subunits form a single stator that interacts extensively with the and subunits in F Dmitriev O, Jones PC, Jiang WP, Fillingame RH: Structure of the membrane domain of subunit b of the Escherichia coli F o F 1 ATP synthase. J Biol Chem 1999, 274: NMR studies of the membrane sector of the E. coli F 1 F o -ATP synthase subunit b, showing its -helical nature. 20. Collinson IR, van Raaij MJ, Runswick MJ, Fearnley IM, Skehel JM, Orriss GL, Miroux B, Walker JE: ATP synthase from bovine heart mitochondria in vitro assembly of a stalk complex in the presence of F 1 -ATPase and in its absence. J Mol Biol 1994, 242: Soubannier V, Rusconi F, Vaillier J, Arselin G, Chaignepain S, Graves PV, Schmitter JM, Zhang JL, Mueller D, Velours J: The second stalk of the yeast ATP synthase complex: identification of subunits showing cross-links with known positions of subunit 4 (subunit b). Biochemistry 1999, 38: Velours J, Paumard P, Soubannier V, Spannagel C, Vaillier J, Arselin G, Graves PV: Organisation of the yeast ATP synthase F o : a study based on cysteine mutants, thiol modification and cross-linking reagents. Biochim Biophys Acta 2000, 1458: Sorgen PL, Caviston TL, Perry RC, Cain BD: Deletions in the second stalk of F 1 F o -ATP synthase in Escherichia coli. J Biol Chem 1998, 273:

7 678 Proteins 24. Sorgen PL, Bubb MR, Cain BD: Lengthening the second stalk of F 1 F o -ATP synthase in Escherichia coli. J Biol Chem 1999, 274: Ogilvie I, Aggeler R, Capaldi RA: Cross-linking of the delta subunit to one of the three alpha subunits has no effect on functioning, as expected if delta is a part of the stator that links the F 1 and F o parts of the Escherichia coli ATP synthase. J Biol Chem 1997, 272: Wilkens S, Zhou J, Nakayama R, Dunn SD, Capaldi RA: Localization of the delta subunit in the Escherichia coli F 1 F o -ATP synthase by immune electron microscopy: the delta subunit binds on top of the F 1. J Mol Biol 2000, 295: Confirmation of the position of the N-terminal domain of the δ subunit at the top of F Wilkens S, Dunn SD, Chandler J, Dahlquist FW, Capaldi RA: Solution structure of the N-terminal domain of the delta subunit of the E. coli ATP synthase. Nat Struct Biol 1997, 4: Jiang WP, Fillingame RH: Interacting helical faces of subunits a and c in the F 1 F o -ATP synthase of Escherichia coli defined by disulfide cross-linking. Proc Natl Acad Sci USA 1998, 95: Vik SB, Long JC, Wada T, Zhang D: A model for the structure of subunit a of the Escherichia coli ATP synthase and its role in proton translocation. Biochim Biophys Acta 2000, 1458: Valiyaveetil FI, Fillingame RH: Transmembrane topography of subunit a in the Escherichia coli F 1 F o -ATP synthase. J Biol Chem 1998, 273: Wada T, Long JC, Zhang D, Vik SB: A novel labeling approach supports the five-transmembrane model of subunit a of the Escherichia coli ATP synthase. J Biol Chem 1999, 274: Girvin ME, Rastogi VK, Abildgaard F, Markley JL, Fillingame RH: Solution structure of the transmembrane H + -transporting subunit c of the F 1 F o -ATP synthase. Biochemistry 1998, 37: Collinson IR, Runswick MJ, Buchanan SK, Fearnley IM, Skehel JM, van Raaij MJ, Griffiths DE, Walker JE: F o membrane domain of ATP synthase from bovine heart mitochondria: purification, subunit composition, and reconstitution with F 1 -ATPase. Biochemistry 1994, 33: Arnold I, Bauer MF, Brunner M, Neupert W, Stuart RA: Yeast mitochondrial F 1 F o -ATPase: the novel subunit e is identical to Tim11. FEBS Lett 1997, 411: Arnold I, Pfeiffer K, Neupert W, Stuart RA, Schagger H: ATP synthase of yeast mitochondria. Isolation of subunit j and disruption of the ATP18 gene. J Biol Chem 1999, 274: Stock D, Leslie AGW, Walker JE: Molecular architecture of the rotary motor in ATP synthase. Science 1999, 286: The 3.9 Å resolution structure of the yeast F 1 c 10 complex has provided the first insight into the arrangement of the c ring and its interactions with stalk subunits γ, δ and ε. The close contact between these subunits and the c ring supports the idea that the γ, δ and ε subunits and the c ring rotate as an ensemble. The unexpected finding of 10 subunit c protomers in the ring has profound implications for the mechanism of coupling and for the number of protons translocated through F o for each ATP molecule synthesised in F Hermolin J, Dmitriev OY, Zhang Y, Fillingame RH: Defining the domain of binding of F 1 subunit epsilon with the polar loop of F o subunit c in the Escherichia coli ATP synthase. J Biol Chem 1999, 274: Further confirmation of the interaction between bacterial ε and c subunits, providing evidence that the c ring rotates together with the central stalk. 38. Watts SD, Capaldi RA: Interactions between the F 1 and F o parts in the Escherichia coli ATP synthase. Associations involving the loop region of c subunits. J Biol Chem 1997, 272: Schulenberg B, Aggeler R, Murray J, Capaldi RA: The gamma- epsilon-c subunit interface in the ATP synthase of Escherichia coli. Cross-linking of the epsilon subunit to the c subunit ring does not impair enzyme function, that of gamma to c subunits leads to uncoupling. J Biol Chem 1999, 274: The authors infer that the ε subunit and c ring rotate as an ensemble during catalysis and that conformational changes in the γ subunit might occur. 40. Hendrix RW: Bacteriophage DNA packaging: RNA gears in a DNA transport machine. Cell 1998, 94: Valpuesta JM, Fernandez JJ, Carazo JM, Carrascosa JL: The threedimensional structure of a DNA translocating machine at 10 Å resolution. Structure 1999, 7: Thomas DR, Morgan DG, DeRosier DJ: Rotational symmetry of the c ring and a mechanism for the flagellar rotary motor. Proc Natl Acad Sci USA 1999, 96: Beuron F, Maurizi MR, Belnap DM, Kocsis E, Booy FP, Kessel M, Steven AC: At sixes and sevens: characterization of the symmetry mismatch of the ClpAP chaperone-assisted protease. J Struct Biol 1998, 123: Jones PC, Fillingame RH: Genetic fusions of subunit c in the F o sector of H + -transporting ATP synthase. Functional dimers and trimers and determination of stoichiometry by cross-linking analysis. J Biol Chem 1998, 273: Jones PC, Jiang WP, Fillingame RH: Arrangement of the multicopy H + -translocating subunit c in the membrane sector of the Escherichia coli F 1 F o -ATP synthase. J Biol Chem 1998, 273: Seelert H, Poetsch A, Dencher NA, Engel A, Stahlberg H, Müller DJ: Proton-powered turbine of a plant motor. Nature 2000, 405: Atomic force microscopy images of the c-subunit ring from chloroplast ATP synthase show 14 protomers in the ring. Therefore, the number of c subunits in ATP synthases may differ from species to species. 47. Schemidt RA, Qu J, Williams JR, Brusilow WSA: Effects of carbon source on expression of F o genes and on the stoichiometry of the c subunit in the F 1 F o ATPase of Escherichia coli. J Bacteriol 1998, 180: Cherepanov DA, Mulkidjanian AY, Junge W: Transient accumulation of elastic energy in proton translocating ATP synthase. FEBS Lett 1999, 449:1-6. A theoretical model is proposed for the generation of torque, involving an elastic element. 49. Oster G, Wang H: Reverse engineering a protein: the mechanochemistry of ATP synthase. Biochim Biophys Acta 2000, 1458: Simplified physical models were developed for both the F 1 and F o sectors. The solutions of the resulting equations reproduce many of the empirical measurements. 50. Hisabori T, Kondoh A, Yoshida M: The gamma subunit in chloroplast F 1 -ATPase can rotate in a unidirectional and counter-clockwise manner. FEBS Lett 1999, 463: The authors demonstrate that rotation of the γ subunit driven by ATP hydrolysis, as first observed in the bacterial enzyme, also occurs in the chloroplast enzyme. 51. Kato-Yamada Y, Noji H, Yasuda R, Kinosita K, Yoshida M: Direct observation of the rotation of epsilon subunit in F 1 -ATPase. J Biol Chem 1998, 273: Noji H, Hasler K, Junge W, Kinosita K, Yoshida M, Engelbrecht S: Rotation of Escherichia coli F 1 -ATPase. Biochem Biophys Res Comm 1999, 260: The authors provide evidence of rotation in intact F 1 -ATPase. 53. Omote H, Sambonmatsu N, Sambongi Y, Iwamato-Kihara A, Yanagida T, Wada Y, Futai M: The gamma-subunit rotation and torque generation in F1-ATPase from wild-type or uncoupled mutant Escherichia coli. Proc Natl Acad Sci USA 1999, 96: Further demonstration of rotation of the γ subunit in the E. coli F 1 domain. Significantly, a mutation of the γ subunit, known to cause uncoupling, had no effect on torque generation. 54. Yasuda R, Noji H, Kinosita K, Yoshida M: F 1 -ATPase is a highly efficient molecular motor that rotates with discrete 120 o steps. Cell 1998, 93: Sambongi Y, Iko Y, Tanabe M, Omote H, Iwamoto-Kihara A, Ueda I, Yanagida T, Wada Y, Futai M: Mechanical rotation of the c subunit oligomer in ATP synthase (F 0 F 1 ): direct observation. Science 1999, 286: This paper presents the first direct evidence for rotation of the c ring in an ATP synthase complex. Objections have been raised concerning the interpretation of these experiments [57 ]. The major remaining concern is whether the ATP synthase is intact. 56. Panke O, Gumbiowski K, Junge W, Engelbrecht S: F-ATPase: specific observation of the rotating c subunit oligomer of EF o EF 1. FEBS Lett 2000, 472: This paper describes one specific experimental approach to the direct observation of the rotation of the c ring. The specificity of the attachment of the

8 The rotary mechanism of ATP synthase Stock et al. 679 actin filament to subunit c was ensured by the introduction of a strep-tag sequence in the C-terminal region of subunit c. 57. Tsunoda SP, Aggeler R, Noji H, Kinosita K, Yoshida M, Capaldi RA: Observations of rotation within the F o F l -ATP synthase: deciding between rotation of the F o c-subunit ring and artifact. FEBS Lett 2000, 470: A critique of experiments directly demonstrating rotation in ATP synthase. 58. Junge W, Lill H, Engelbrecht E: ATP synthase: an electrochemical transducer with rotary mechanics. Trends Biol Sci 1997, 22: Elston T, Wang HY, Oster G: Energy transduction in ATP synthase. Nature 1998, 391: Dimroth P, Wang H, Grabe M, Oster G: Energy transduction in the sodium F-ATPase of Propionigenium modestum. Proc Natl Acad Sci USA 1999, 96: A novel mechanochemical model for the generation of rotation in sodiumdependent ATP synthase involving a single channel, rather than two half channels, as proposed for the proton-dependent ATP synthase. 61. Rastogi VK, Girvin ME: Structural changes linked to proton translocation by subunit c of the ATP synthase. Nature 1999, 402: The authors describe the structural changes undergone by the c protomer that accompany deprotonation of essential residue Asp61, as determined by NMR in organic solvents. A novel model is proposed for the rotation of subunit c and for its interactions with subunit a. 62. Braig K, Menz IR, Montgomery MG, Leslie AGW, Walker JE: Structure of bovine F 1 -ATPase inhibited by Mg 2+ ADP and aluminium fluoride. Structure 2000, 8: A description of a transition state in the catalytic cycle of F 1 -ATPase. Now published The work referred to in the text as (C Gibbons, MG Montgomery, AGW Leslie, JE Walker, unpublished data) is now published: 63. Gibbons C, Montgomery MG, Leslie AGW, Walker JE: The structure of the central stalk at 2.4 Å resolution. Nat Struct Biol 2000, 7: A description of the structure of the central stalk of F 1 F o -ATPase determined in the context of an intact F 1 -ATPase assembly.