STRUCTURE AND FUNCTION OF THE F o COMPLEX OF THE ATP SYNTHASE FROM ESCHERICHIA COLI

Transcription

1 The Journal of Experimental Biology 203, (2000) Printed in Great Britain The Company of Biologists Limited 2000 JEB STRUCTURE AND FUNCTION OF THE F o COMPLEX OF THE ATP SYNTHASE FROM ESCHERICHIA COLI KARLHEINZ ALTENDORF, WOLF-DIETER STALZ, JÖRG-CHRISTIAN GREIE AND GABRIELE DECKERS-HEBESTREIT* Universität Osnabrück, Fachbereich Biologie/Chemie, Abteilung Mikrobiologie, D Osnabrück, Germany *Author for correspondence ( deckers_hebestreit@biologie.uni-osnabrueck.de) Accepted 18 October; published on WWW 13 December 1999 The membrane-bound ATP synthase (F 1 F o ) from mitochondria, chloroplasts and bacteria plays a crucial role in energy-transducing reactions. In the case of Escherichia coli, the reversible, proton-translocating ATPase complex consists of two different entities, F 1 and F o. The watersoluble F 1 part carries the catalytic sites for ATP synthesis and hydrolysis. It is associated with the membraneembedded F o complex, which functions as a proton channel and consists of subunits a, b and c present in a stoichiometry of 1:2:12. Subunit b was isolated by preparative gel electrophoresis, acetone-precipitated and renatured in a cholate-containing buffer. Reconstituted subunit b together with purified ac subcomplex is active in proton translocation and F 1 binding, thereby demonstrating that subunit b had recovered its native conformation. Circular dichroism spectroscopy of subunit b reconstituted into liposomes revealed a rather high degree of α-helical conformation of 80 %. After addition of a His 6 -tag to the N terminus of subunit a, a stable ab 2 subcomplex was purified instead of a single subunit a, arguing in favour of a direct interaction between these subunits. After addition of Summary subunit c and reconstitution into phospholipid vesicles, an F o complex was obtained exhibiting rates of proton translocation and F 1 binding comparable with those of wild-type F o. The epitopes of monoclonal antibodies against subunit c are located in the hydrophilic loop region (cl31 Q42) as mapped by enzyme-linked immunosorbent assay using overlapping synthetic heptapeptides. Binding studies revealed that all monoclonal antibodies (mabs) bind to everted membrane vesicles irrespective of the presence or absence of F 1. Although the hydrophilic region of subunit c, and especially the highly conserved residues ca40, cr41, cq42 and cp43, are known to interact with subunits γ and ε of the F 1 part, the mab molecules have no effect on the function of F o, either in proton translocation or in F 1 binding. However, the F 1 part and the mab molecule(s) are bound simultaneously to the F o complex, suggesting that not all c subunits are involved in the interaction with F 1. Key words: F 1F o-atpase, F o subunit, ab 2 subcomplex, monoclonal antibody, circular dichroism, Escherichia coli. Introduction The membrane-bound ATP synthases (F 1 F o -ATPases) of bacteria serve two important physiological functions. Utilizing the energy of an electrochemical ion gradient, the enzyme catalyses the synthesis of ATP from ADP and inorganic phosphate. Under conditions of low driving force, ATP synthases function as ATPases, thereby generating a transmembrane ion gradient at the expense of ATP. The high homology between all known F-ATPases indicates that the mechanism of ion translocation and catalysis and the mode of coupling are the same in all organisms. The simplest F 1 F o ATP synthase, e.g. that of Escherichia coli, is composed of eight types of subunit present in the stoichiometry α 3 β 3 γδε for the water-soluble F 1 part carrying the catalytic sites for ATP synthesis/hydrolysis and ab 2c 12 for the membrane-embedded F o complex functioning as a proton channel (Deckers- Hebestreit and Altendorf, 1996; Fillingame et al., 1998). For an overview of the structural data available for the ATP synthase, see Fig. 1. The F 1 part extends from the membrane with subunits α and β alternately arranged in a hexamer surrounding the centrally located subunit γ (Abrahams et al., 1994). ATP synthesis/hydrolysis occurs in the different β subunits according to the cooperative binding of ADP and inorganic phosphate at one catalytic site, which is coupled to the release of ATP at a second site, whereas the third site is empty (Boyer, 1997; Abrahams et al., 1994). There is consensus that subunits γ and ε rotate relative to the α 3 β 3 domain (Noji et al., 1997; Yasuda et al., 1998; Kato-Yamada et al., 1998; Sabbert et al., 1997; Zhou et al., 1997; Bulygin et al., 1998), thereby causing different binding affinities in the catalytic sites accomplished by a high cooperativity. To allow

2 20 K. ALTENDORF AND OTHERS Fig. 1. Structural data available for the F 1F o ATP synthase. The intersubunit arrangement is fitted into the electron microscopy density map (copyright permission for reproduction was kindly provided by Nature) and is therefore speculative. Subunits or subunit domains are shown in different colours. the energy-linked rotational movement of subunits γ and ε, a second structural link between F 1 and F o is necessary to stabilize the F 1 F o complex. This structural link is composed at least of subunit δ of F 1 and the two copies of subunit b of F o (Beckers et al., 1992; Lill et al., 1996; Rodgers et al., 1997; Dunn and Chandler, 1998; McLachlin et al., 1998). Recent electron micrographs indicate that this link may be located at the periphery at one side of the F 1 F o molecule (Wilkens and Capaldi, 1998; Böttcher et al., 1998), which is in good accord with the asymmetrical arrangement of the F o complex with subunits a and b located outside the ring-like subunit c oligomer (Birkenhäger et al., 1995; Singh et al., 1996; Takeyasu et al., 1996). Proton translocation through F o is mediated by subunit c, a hydrophobic protein with a hairpin-like structure comprising two transmembrane helices connected by a polar loop region (Girvin et al., 1998, and references therein). During proton translocation, proton binding and release take place via a carboxyl group (cd61 in E. coli) thought to be located within the centre of the second transmembrane helix. In this protonation/deprotonation step, subunit a (essentially the highly conserved ar210) is thought to play an important role, although the precise mechanism remains to be established (for reviews, see Deckers-Hebestreit and Altendorf, 1996; Fillingame, 1990). With regard to the energy coupling between F 1 and F o, the presence of 12 copies of subunit c within the F o complex (Jones and Fillingame, 1998; Jones et al., 1998) is in accordance with recent results obtained for chloroplast and cyanobacterial F 1 F o demonstrating that four protons are translocated per ATP synthesized (van Walraven et al., 1996). During ATP synthesis/hydrolysis, subunits γ and ε rotate by 120 for each ATP synthesized or hydrolyzed (Yasuda et al.,

3 F o complex of Escherichia coli ATP synthase ), with the rotation being coupled to the deprotonation/ protonation of four c subunits. The coupling between ATP synthesis/hydrolysis and proton translocation is thought to occur via interactions between subunits γ and ε in F 1 and subunit c in F o. However, the mechanism by which the protonation/deprotonation step is coupled to the rotation of γ and ε is still unknown. Whether the subunit c oligomer co-rotates with subunits γ and ε, which are in a fixed position relative to the c subunits, or whether the γε subcomplex alone rotates along the surface of the loop regions of the subunit c oligomer, is controversial (Jones et al., 1995; Fillingame, 1997; Junge et al., 1997; Elston et al., 1998; Oster and Wang, 1999; Junge, 1999; Dimroth et al., 1999). Interaction of subunit c with F 1 A ring-like structure has been proposed for the substructure of the subunit c oligomer within F o (Schneider and Altendorf, 1987; Fillingame, 1990; Groth and Walker, 1997), which correlates well with recent cross-linking studies (Jones et al., 1998), with tryptophan substitutions (Groth et al., 1998) and with electron and atomic force microscopy studies (Birkenhäger et al., 1995; Singh et al., 1996; Takeyasu et al., 1996). However, which helix packs at the interior and which at the periphery of the ring is controversial (Jones et al., 1998; Groth et al., 1998). Several lines of evidence exist for the requirement of subunit c in the coupling of F o to F 1. Mutations in the conserved, hydrophilic loop region, mainly at positions cr41, cq42 and cp43, exhibit effects on the coupling of F 1 that are dependent on the mutation introduced and on the genetic background used for characterization (for reviews, see Fillingame, 1990; Deckers-Hebestreit and Altendorf, 1996). Furthermore, suppressor mutants to the cq42e mutation have been isolated, in which ATP-driven proton translocation is recoupled again by a second mutation in subunit ε, changing εe31 to G, V or K (Zhang et al., 1994). Cross-linking studies with double mutants carrying cysteine residues at position εe31 and at positions ca40, cq42 or cp43 have revealed a possible direct interaction between the two subunits (Zhang and Fillingame, 1995). Furthermore, it could be demonstrated that cysteine residues present in a continuous stretch reaching from εt26 to εg33 of subunit ε can be cross-linked to cysteine residues introduced at positions ca40, cq42 and cd44 of subunit c (Hermolin et al., 1999). A possible interaction between the hydrophilic loop of subunit c (cq42, cp43, cd44) and the tyrosine residue γy205 of subunit γ has also been demonstrated using cysteine mutants (Watts et al., 1995, 1996). Epitope mapping of monoclonal antibodies (mabs) prepared against subunit c revealed that the epitopes of all mabs are located within the hydrophilic loop region and, furthermore, that all epitopes (comprising residues c35-flegaarq-42) except one (c31-lggkfle-37) are located within the highly conserved region involving c39-aarqp-43. Binding studies with membrane vesicles of different orientation demonstrated strong binding of all mabs to everted membrane vesicles. Thus, the mab studies clearly demonstrate that at least region c31-lggkflegaarq-42 is accessible from the cytoplasm (Birkenhäger et al., 1999). These results indicate that the polar loop surface of subunit c exposed to the water phase is more expanded, as proposed from cysteine mutant analysis by crosslinking and chemical modification (Zhang and Fillingame, 1995; Watts et al., 1995, 1996; Watts and Capaldi, 1997) and from nuclear magnetic resonance (NMR) analysis (Girvin et al., 1998, and references therein; see also Fillingame, 1997). In addition, cross-linking between subunits a and c clearly demonstrates that helix 2 of subunit c starts near residue cf54 (Jiang and Fillingame, 1998). In summary, the results indicate that the surface loop region of subunit c comprises at least residues cl31 F53, which may be mainly α-helical with a few randomly organized residues around cp43, as deduced from NMR analysis of subunit c dissolved in chloroform/methanol/ water (Girvin et al., 1998). NMR studies of subunit c of the Na + -translocating ATP synthase from Propionigenium modestum dissolved in dodecyl sulphate micelles are also interpreted to show that at least this part of the polypeptide chain of subunit c is exposed to the cytoplasm (Matthey et al., 1999). Furthermore, the epitope mapping of the mabs within region cl31 Q42 suggests that helix 1 of subunit c packs at the periphery of the oligomeric ring, as proposed by Groth et al. (1998), being accessible to mab molecules also in the presence of F 1. Almost all anti-c mabs used are of the IgG1 subclass, which have a head-to-tail distance of 7.5 nm on average (Zheng et al., 1992). Therefore, despite the fact that only one or two mabs may bind to F o and that only part of the surface of the subunit c ring is shielded by F 1, a localization of helix 1 at the periphery must be favoured to overcome steric hindrance between bound F 1 and simultaneously bound mab molecule(s). The mab molecules with their high specificity, alone or in a mixture, have no influence on the function of F o. Because of steric hindrance, it is to be expected that only one or two antic mab molecules with a relative molecular mass of are bound per subunit c oligomer (relative molecular mass approximately ). However, from N-ethylmaleimide (NEM)-labelling studies with cysteine mutants, it is known that an effect on the function of F o could only be obtained after modification of more than 60 % of the c subunits (Watts and Capaldi, 1997). Binding of all mabs to everted membrane vesicles occurred independently of the presence or absence of F 1. Furthermore, the results obtained clearly demonstrate that F 1 and the mab molecule(s) are bound simultaneously to the F o complex and that the amino acid residues involved in antibody recognition (c31-lggkflegaarq-42) overlap with those known to interact with subunits γ and ε of the F 1 complex (residues c40-arqpd-44) (Fraga et al., 1994; Zhang et al., 1994; Zhang and Fillingame, 1995; Watts et al., 1995, 1996). Therefore, it can be concluded that not all the c subunits are directly involved in interaction with F 1, since at least one of the subunit c loop regions is occupied by an mab molecule (Birkenhäger et al., 1999). In addition, a different NEMlabelling behaviour for the 12 copies of subunit c in F o could

4 22 K. ALTENDORF AND OTHERS be observed with membranes of the subunit c mutant cq42c in the presence of F 1, indicating that only four or five c subunits participate directly in the F 1 interaction (Watts and Capaldi, 1997). Cross-linking between region εt26 G33 of subunit ε and residues ca40, cq42 and cd44 of subunit c revealed that two loop regions of subunit c are possibly occupied by interaction with subunit ε (Hermolin et al., 1999). Since residue γy205 of subunit γ is also able to form disulphide bridges to residues cq42, cp43 and cd44 of subunit c (Watts et al., 1995, 1996), it seems likely that subunit γ interacts with a different set of c subunits from those interacting with subunit ε, explaining very well the observations of Watts and Capaldi (1997) that four or five c subunits participate in interaction with F 1. Rotation versus conformational changes in the subunit c oligomer The coupling between proton translocation and ATP synthesis/hydrolysis is thought to take place by rotation of subunits γ and ε relative to the αβ hexamer, leading to the proposal that the subunit c oligomer co-rotates past a static, peripherally located ab 2 subcomplex in the membrane (Vik and Antonio, 1994; Engelbrecht and Junge, 1997; Elston et al., 1998; Junge, 1999), with subunits γ and ε being fixed to a few of the 12 c subunits (Engelbrecht and Junge, 1997; Junge et al., 1997). It is worthwhile mentioning that this model fits very well with recent cross-linking studies, which showed that region εt26 G33 of subunit ε interacts with residues ca40, cq42 and cd44 of subunit c. Furthermore, a structural interaction model additionally based either on the NMR and X-ray diffraction structures of monomeric subunits (Girvin et al., 1998; Wilkens et al., 1995; Uhlin et al., 1997) or on the packing arrangement of oligomeric subunit c (Jones and Fillingame, 1998; Jones et al., 1998) revealed that region εt26 G33 protrudes as a well-defined lobe that essentially fills the space between two loops of subunit c (Hermolin et al., 1999). Alternatively, it has been suggested that the F o complex remains fixed relative to subunits α and β in F 1, and that subunits γ and ε move along the polar loop surface of the subunit c oligomer (Jones et al., 1995; Fillingame, 1997). Conformational changes would then be relayed from the site of proton binding/release to the polar loop region of subunit c (Assadi-Porter and Fillingame, 1995; Penefsky, 1985), which then drive movement of subunits γ and ε from one copy of subunit c to the next (Fillingame, 1997). The studies with mabs against subunit c, revealing no influence on the function of F o, although F 1 and the mab molecule(s) are bound simultaneously to the F o complex, indicate that a complete rotation of the subunit c oligomer together with subunits γ and ε (Junge et al., 1997; Elston et al., 1998; Junge, 1999) along the stator seems to be unlikely because of steric hindrance generated by the large mab molecule(s) bound to subunit(s) c. However, a movement of the γε subcomplex from loop to loop of subunit c in a circular fashion during catalysis is also difficult to understand with the results obtained, since at least one of the subunit c loop regions should be occupied by an mab molecule (Birkenhäger et al., 1999). However, in a highly α-helical subunit c oligomer with the N-terminal helix oriented towards the outside, the mab epitopes (cl31 Q42) are proposed to be located laterally at the upper periphery of the oligomer. This location makes an interaction between the polar loop surfaces (ca40 D44) and rotating subunits γ and ε at the top of the complex more likely, even in the presence of bound mab(s). Studies for or against the argument that, despite the high affinity of antibody binding, the mab(s) bound to subunits c are probably sheared off by the torque generated at large driving forces during catalysis are under way. Structure of subunit b Subunit b is anchored in the membrane by its hydrophobic N-terminal region, whereas the remainder of the protein, except for a short stretch of hydrophobic amino acids near the C terminus (residues bv124 A132), is hydrophilic and highly charged. Proteolysis studies have shown that the hydrophilic part of subunit b is essential for the association of F 1 with F o (Steffens et al., 1987, and references therein). Removal of two residues from the C terminus of subunit b disrupts assembly of the ATP synthase (Takeyama et al., 1988), as do mutations bv124d, ba128d (Howitt et al., 1996) and bg131d (Jans et al., 1985). Studies with the hydrophilic portion of subunit b (b sol and b syn ) expressed as a separate protein showed that it forms a homodimer with an elongated shape (Dunn, 1992), which can be disrupted to monomers by mutation of b sol A128 to aspartate (Howitt et al., 1996; Rodgers et al., 1997). In addition, the introduction of cysteine residues followed by disulphide bond formation confirms the dimeric conformation of b sol as well as of the intact protein in the membrane (for an overview, see Fig. 2). Cysteines at positions bv124, ba128, bg131, ba132, bs139, ba144 and bs146 showed particularly strong tendencies to form disulphide bonds with their counterparts in the dimer (McLachlin and Dunn, 1997; Rodgers et al., 1997; Rodgers and Capaldi, 1998) without affecting ATPase activity (Rodgers et al., 1997; Rodgers and Capaldi, 1998). For the b syn protein, it has been shown that residues bd53-k66 are also essential for dimerization, whereas residues by24-k52 are not necessary (McLachlin and Dunn, 1997). In addition, cross-linking of the b subunits could also be observed for cysteine substitutions bs60c (T. Sieck and K. Altendorf, unpublished results), bs84c and bl156c (Rodgers and Capaldi, 1998), but with low yield or only after treatment with large amounts of CuCl 2. The distances between the α carbon atoms usually found for natural disulphide bridges in proteins are constrained to nm (Richardson and Richardson, 1989) and, therefore, such a distance is also expected for disulphide formation by cross-linking, indicating the proximity between the participating residues. Nevertheless, the distances between the α carbons of the two polypeptide chains are expected to be larger for cross-links formed with lower yield, since cross-linking may be due to swivelling or thermal motion of the molecules.

5 F o complex of Escherichia coli ATP synthase 23 c 12 β ε a δ α γ b 2 β b156/α90 b156 b155/δ b146 b144 b131 b128 b104 b84 b60 b10 b6 b2 Dimer formation b sol 155/δ b sol 139 b sol 132 b sol 128 b sol 124 b sol 60 b sol 59 Fig. 2. Assembly of the subunit b dimer within the F 1F o ATP synthase. Cross-linking products obtained for b sol or assembled subunit b are indicated by stars. The structure of the peptide encompassing residues bm1 I33 of subunit b is derived from nuclear magnetic resonance data (Dmitriev et al., 1999). The structural content and the length of α helices within the hydrophilic part of subunit b are derived from circular dichroism spectrocopic data, whereas the positioning of the helices is speculative. Dimerization could also be observed for the hydrophobic N- terminal part of subunit b after the introduction of cysteine residues into region bn2 C21 (see Fig. 2). Cysteine residues that formed disulphide cross-links were found with a periodicity indicative of one face of an α helix over the span of residues bn2 V18, where bn2c, bt6c and bq10c formed dimers in high yield, whereas the naturally occurring bc21 showed no cross-linking with its counterpart (Dmitriev et al., 1999). The last observation is in good agreement with mutant analyses, which revealed a high tolerance towards amino acid substitutions in the region bc21 W26, indicating that this region is probably not a protein protein contact area (Kauffer et al., 1991; Cox et al., 1986; Jans et al., 1984). A model for the dimeric transmembrane helix of subunit b presented by Dmitriev et al. (1999), which is based on the NMR structure (discussed below) and the distance constraints derived from cross-linking, positioned the transmembrane helices at 23 to each other with the side chains of bt6, bq10, bf14 and bf17 at the interface between the subunits. The transmembrane helices would then cross at an angle typical for helix helix packing in membrane proteins (Bowie, 1997a,b). In general, dimerization interactions between the two copies of subunit b depend on residues distributed throughout the polypeptide chain, indicating that the two subunits are adjacent to each other. Such an arrangement has been demonstrated by cross-linking and is valid not only for the hydrophilic domain (McLachlin and Dunn, 1997; Rodgers et al., 1997; Rodgers and Capaldi, 1998) but also for the N-terminal transmembrane helix (Dmitriev et al., 1999). However, one has to take into consideration that disulphide bridges between counterparts can only be formed when the two copies of subunit b are positioned in a face-to-face orientation to allow direct contact between the corresponding residues of the two polypeptide chains. Recently, the structure of the hydrophobic N-terminal stretch of subunit b (synthetic peptide of sequence bm1 I33) was determined by NMR using chloroform/methanol/water as solvent (see Fig. 2). Residues bn4 M22 form a continuous α helix, which is interrupted by a rigid bend in the region of residues bk23 W26 with the α-helical structure resuming at residue bp27 at an angle offset by 20 from the main helix (Dmitriev et al., 1999). The distance between the α carbons of bn2 and bw26 was determined to be 3.4 nm in the structure, which is close to the distance of 3.2 nm predicted between fatty acyl carbonyls in opposing leaflets of a dioleoylphosphatidylcholine bilayer (Dmitriev et al., 1999; Wiener and White, 1992; Yau et al., 1998). Therefore, residues bn4 M22 are thought to span the hydrophobic domain of the lipid bilayer, anchoring the largely hydrophilic subunit b in the membrane, whereas the change in direction of helical packing after the rigid bend (region bk23 W26) may be necessary to redirect the following α helix at an angle more perpendicular to the membrane as it emerges into the cytoplasm, which may be critical for the dimerization of the cytoplasmic domain (Dmitriev et al., 1999). The secondary structure of b sol has been determined by circular dichroism (CD) to be highly α-helical (Dunn, 1992; Rodgers et al., 1997), whereas for the mutant protein b sol A128D, present in monomeric form, a loss of α-helical structure was reported, which is largely compensated by an increase in parallel β sheet and random coil structure (Rodgers et al., 1997). Recently, we determined the secondary structure composition of full-length subunit b using CD spectroscopy, thereby overcoming the difficulties in the spectroscopic analysis of membrane proteins due to artificial CD signals arising from the use of detergents or from light-scattering effects in the presence of membranes. For this purpose, subunit b of the F o complex was purified to homogeneity by preparative SDS gel electrophoresis, precipitated with acetone and, after ion pair extraction, redissolved in a cholatecontaining buffer. That subunit b recovered its native conformation was demonstrated by reconstitution into proteoliposomes together with an ac subcomplex; the resulting F o complex was functional both in proton translocation and in F 1 binding. The light scattering of membrane systems during CD spectroscopic analysis could be minimized by reducing the

6 24 K. ALTENDORF AND OTHERS diameter of the proteoliposomes to approximately 50 nm. The resulting spectra revealed a secondary structure composition for subunit b of 80 % α helix (J.-C. Greie, G. Deckers- Hebestreit and K. Altendorf, unpublished results), which is in agreement with the results reported for b sol (Dunn, 1992; Rodgers et al., 1997), although there are some differences in the corresponding CD spectra. A structure determination for subunit b incorporated into phospholipid vesicles mimicking its native environment seems to be of great importance, since the structure of the hydrophilic part of subunit b was strongly dependent on the detergent used to replace the phospholipid environment. This was shown by identical trypsin digestion patterns for subunit b and for subunit b present in F o, which was strongly influenced by a change in the detergent used for solubilization (J.-C. Greie, G. Deckers-Hebestreit and K. Altendorf, unpublished results). A model in which a pair of long α helices extends from the membrane towards the F 1 part is very attractive at first sight. In addition, a cross-link between the C terminus of subunit b (bl156c) and residue αc90 of subunit α (see below), which is located in the N-terminal domain of subunit α, the portion of the protein most distant from the membrane (Rodgers and Capaldi, 1998), fits very well with this model. Calculations based on the X-ray structure of F 1 and electron microscopy of the central stalk suggest that subunit b must extend at least 11 nm from the membrane to reach the top of the molecule, which means that at least 80 amino acid residues are required to traverse the expected distance in a continuous α helix (Rodgers and Capaldi, 1998). However, the hydrophilic region of subunit b comprises some 125 residues and there are several indications that subunit b is not such a rigid rod-like structure. Secondary structure predictions indicate that two extraordinarily long α-helical segments are interrupted by a β turn at positions br82 Q85 (Walker et al., 1982; Senior, 1983), and mutagenesis of this region revealed that it is necessary for the stability of the F 1 F o -ATPase (McCormick and Cain, 1991). The structure of subunit b is predicted to be 80 % α-helical, with the helices not amounting to more than 25 amino acid residues in length (J.-C. Greie, G. Deckers- Hebestreit and K. Altendorf, unpublished results). The results indicate that, in addition to a high α-helical content, other structures are present (predominantly β turns, as derived from the CD spectra; J.-C. Greie, G. Deckers-Hebestreit and K. Altendorf, unpublished results), which may allow a conformational flexibility within the subunit b dimer possibly required for coupled catalysis, as deduced from the following observations. A deletion analysis revealed that subunit b can tolerate the removal of up to 11 amino acid residues within region ba50 I75, which shortens subunit b by approximately 1.6 nm, while maintaining coupled enzymatic activity (Sorgen et al., 1998b), indicating an inherent stretching flexibility. Cross-linking between subunits b and α produced an uncoupling between ATPase activity in F 1 and proton translocation through F o (Rodgers and Capaldi, 1998), possibly caused by blocking a flexible motion between both subunits. Furthermore, the model of an elastic ATP synthase (Junge, 1999; Cherepanov et al., 1999), which may account for the wealth of phenomena attributed to the activation transitions in both F o and F 1, argues for strong flexibility within the hydrophilic regions of subunit b to allow an elastic deformation during catalysis. The model reflects the elastic coupling, namely the energy transmission between the 12 protons translocated in F o and the 3 ATP molecules synthesized in F 1. Thereby, the sequential loading of four protons causes an accumulation of elastic energy until its eventual discharge for the release of tightly bound ATP. In this context, it has been suggested that the intertwined helices of subunit γ serve as a torsional spring and that the α-helical segments of the b subunits constitute a parallelogram-like elastic counterbearing, which is held together by interactions between subunits b and subunits δ and α at the top and with subunit a at the basis (see below). Furthermore, the 12-to-3 transmission implies that the first proton in a series operates against a lower torque than the fourth (Cherepanov et al., 1999; Junge, 1999). Taken together, the α- helical content of 80 % within subunit b and the calculation that some 80 amino acid residues are necessary to reach the top of the F 1 molecule clearly indicate that in the hydrophilic region of subunit b, containing approximately 125 residues, there are enough residues for β turns or random-coil structures, which are essential to raise the conformational flexibility of an otherwise highly α-helical, rod-like structure. Although the data obtained suggest additional structural elements other than α helices, the exact positions of helices and linker regions within the hydrophilic part of subunit b are not yet known, and the model presented in Fig. 2 is therefore speculative. Interactions of subunit b as part of the stator Dividing the subunits of the ATP synthase complex into structural elements of rotor and stator, there is strong evidence that subunits γ and ε are involved in rotational catalysis and, despite the lack of strong experimental evidence, the affiliation of the subunit c oligomer to the rotor is accommodated by recently published models. In addition, there is general agreement that the α 3 β 3 complex and subunit δ of F 1 as well as subunits b and a of F o belong to the stator (compare Junge et al., 1997; Nakamoto, 1999; Junge, 1999). The interactions between the subunit b dimer and other components of the stator will now be discussed in detail. As described above, the two copies of subunit b are adjacent to each other, both in the hydrophilic domain and in the transmembrane helix, and the formation of a subunit b dimer seems to be a necessary prerequisite for the binding of the F 1 part (Sorgen et al., 1998a; Dunn and Chandler, 1998). However, subunit δ is also required for the binding of F 1 to F o. Membranes of mutant strains expressing truncated forms of subunit δ showed little ATPase activity, indicating that F 1 cannot bind to the membrane in the absence of intact subunit δ (Jounouchi et al., 1992). Several independent studies suggested that subunit δ is located at the periphery in the upper third of the α 3 β 3 hexamer (Dunn et al., 1980; Maggio et al., 1988; Lill et al., 1996; Ogilvie et al., 1997). Recently,

7 F o complex of Escherichia coli ATP synthase 25 interaction between subunit b (or b sol ) and subunit δ in the absence as well as in the presence of other F 1 F o subunits was observed, suggesting an extended structure for subunit b (Dunn and Chandler, 1998; Rodgers et al., 1997; Sawada et al., 1997). Furthermore, cross-linking studies clearly demonstrated that both subunits interact via their C-terminal regions (Beckers et al., 1992; Joshi and Burrows, 1990; Rodgers et al., 1997; McLachlin et al., 1998). Preferred sites of cross-linking of the mutant b subunits b sol E155C and b sol C158 (two residues beyond the normal C terminus of subunit b) to subunit δ were shown to be C-terminal to δm148 (McLachlin et al., 1998). Interaction between subunit b and subunits α and β has been observed by cross-linking (Aris and Simoni, 1983). Cryoelectron microscopy using b sol and F 1 revealed that subunit b interacts with a β subunit of F 1 different from that interacting with subunit ε (Wilkens et al., 1994). In addition, the sequence homology between subunits ε and b involving residues εt82 K100/bN80 E97 and εs107 Q127/bR49 E71 (Walker et al., 1984; Wilkens et al., 1994) implies a possible binding of subunit b to the C-terminal region of subunit β, since εs108 is also involved in binding to that region (Dallmann et al., 1992; Wilkens et al., 1994). However, the introduction of a cysteine residue at position bl156 allowed cross-linking to subunit α via the naturally occurring αc90 in high yields in the presence of low concentrations of CuCl 2, indicating a direct interaction between the two partners. Whereas the ATPase activity of this cross-linked F 1 F o complex is unaffected, ATP-dependent proton translocation is markedly reduced probably because of uncoupling, suggesting that an inherent flexibility is also necessary for the stator elements (Rodgers and Capaldi, 1998). Because of the interaction between subunit b and both subunits α and β, it is tempting to speculate that one b subunit interacts with subunit α, whereas the other makes contact with subunit β. The linkage proposed between the subunit b dimer and subunit a as part of the stator in F o has now been verified by purification of a stable ab 2 subcomplex. The addition of a His 6 -tag to the N terminus of subunit a allowed specific binding to Ni NTA columns. Surprisingly, instead of a single subunit a, a stable ab 2 subcomplex was purified. Furthermore, after co-reconstitution with purified subunit c, an F o complex was formed that was functional in proton translocation and F 1 binding at rates comparable with those for wild-type F o (W.-D. Stalz, J.-C. Greie, G. Deckers-Hebestreit and K. Altendorf, unpublished results). However, the contact sites between subunit a and the two b subunits, which must be different because of the stoichiometry, still have to be elucidated. Topology of subunit a Mutational analyses have revealed that only residue ar210 of subunit a is essential for ATP-driven proton translocation within F o, possibly by forming a transient salt bridge with cd61. This view is supported by the observation that even the most conservative substitution to lysine abolishes ATPdependent proton translocation (for reviews, see Fillingame, 1990; Deckers-Hebestreit and Altendorf, 1996). However, recent data have revealed that ar210 is not essential for passive proton transport mediated by F o after removal of F 1 (Valiyaveetil and Fillingame, 1997). To understand the function of subunit a in the proton translocation process during ATP synthesis/hydrolysis, information about its topology is a prerequisite. Several independent approaches have been applied to determine the topological organization of subunit a, always using the reciprocal pattern of reactivity or accessibility of hydrophilic regions at the cytoplasmic and periplasmic sides of the membrane. Whereas the localization of the N-terminal part of subunit a is still controversial, a detailed picture of the orientation of the C-terminal two-thirds of the protein has been developed. However, the existing discrepancies, as discussed in detail by Deckers-Hebestreit et al. (1999), have led to two models, one with six transmembrane helices (Yamada et al., 1996; Jäger et al., 1998; Deckers-Hebestreit et al., 1999) and a second with five such helices (Hatch et al., 1995; Valiyaveetil and Fillingame, 1998; Long et al., 1998; Wada et al., 1999). For the N-terminal part of subunit a, the chemical accessibility of genetically introduced cysteine residues in membrane vesicles of different orientation indicates that the N terminus is exposed to the periplasm and that only one transmembrane helix is present up to residue ak66, in accord with a five-helix model (Valiyaveetil and Fillingame, 1998; Long et al., 1998; Wada et al., 1999). However, the reciprocal pattern of reactivity has a very narrow range, with a factor of only for right-side-out vesicles versus inside-out vesicles (Valiyaveetil and Fillingame, 1998). In contrast, the binding behaviour of monoclonal and peptide-specific antibodies to membrane vesicles of different orientation always revealed the accessibility of the N-terminal region at the cytoplasmic side of the membrane with a factor of at least 10 for the range of the reciprocal pattern of reactivity. These results require the presence of two transmembrane helices up to residue ak66, in accord with a six-helix model (Jäger et al., 1998; Deckers- Hebestreit et al., 1999; Yamada et al., 1996). In general, for binding studies with antibodies raised against F o subunits in our laboratory, membrane vesicles were always prepared from a wild-type strain carrying the atp operon on the chromosome. As a consequence, neither genetic manipulation nor overexpression of single subunits or protein complexes, which might alter the properties of the membrane or the assembly of the ATP synthase complex, have to be taken into consideration. For the topology of the C-terminal three-quarters of the polypeptide chain of subunit a, agreement has been achieved using the different methods, suggesting the presence of a cytoplasmic loop region around ak66, the presence of four transmembrane helices and the location of the C terminus in the cytoplasm (see Deckers-Hebestreit et al., 1999). Work from the authors laboratory was supported by the Deutsche Forschungsgemeinschaft (SFB431/D3), by Human Frontiers (HFSP, RG0571/1996-M), by the Fonds der Chemischen Industrie (fellowship to J.-C.G.) and by the

8 26 K. ALTENDORF AND OTHERS Graduiertenförderung of the State of Lower Saxony (fellowship to W.-D.S.). References Abrahams, J. P., Leslie, A. G. W., Lutter, R. and Walker, J. E. (1994). Structure at 2.8 Å resolution of F 1-ATPase from bovine heart mitochondria. Nature 370, Aris, J. P. and Simoni, R. D. (1983). Cross-linking and labeling of the Escherichia coli F 1F o-atp synthase reveal a compact hydrophilic portion of F o close to an F 1 catalytic subunit. J. Biol. Chem. 258, Assadi-Porter, F. M. and Fillingame, R. H. (1995). Protontranslocating carboxyl of subunit c of F 1F o H + -ATP synthase: The unique environment suggested by the pk a determined by 1 H NMR. Biochemistry 34, Beckers, G., Berzborn, R. J. and Strotmann, H. (1992). Zerolength crosslinking between subunits δ and I of the H + - translocating ATPase of chloroplasts. Biochim. Biophys. Acta 1101, Birkenhäger, R., Greie, J.-C., Altendorf, K. and Deckers- Hebestreit, G. (1999). F o complex of the Escherichia coli ATP synthase: Not all monomers of the subunit c oligomer are involved in F 1 interaction. Eur. J. Biochem. 264, Birkenhäger, R., Hoppert, M., Deckers-Hebestreit, G., Mayer, F. and Altendorf, K. (1995). The F o complex of the Escherichia coli ATP synthase. Investigation by electron spectroscopic imaging and immunoelectron microscopy. Eur. J. Biochem. 230, Böttcher, B., Schwarz, L. and Gräber, P. (1998). Direct indication for the existence of a double stalk in CF of 1. J. Mol. Biol. 281, Bowie, J. U. (1997a). Helix packing in membrane proteins. J. Mol. Biol. 272, Bowie, J. U. (1997b). Helix packing angle preferences. Nature Struct. Biol. 4, Boyer, P. D. (1997). The ATP synthase a splendid molecular machine. Annu. Rev. Biochem. 66, Bulygin, V. V., Duncan, T. M. and Cross, R. L. (1998). Rotation of the ε subunit during catalysis by Escherichia coli F of 1-ATP synthase. J. Biol. Chem. 273, Cherepanov, D. A., Mulkidjanian, A. Y. and Junge, W. (1999). Hypothesis. Transient accumulation of elastic energy in proton translocating ATP synthase. FEBS Lett. 449, 1 6. Cox, G. B., Fimmel, A. L., Gibson, F. and Hatch, L. (1986). The mechanism of ATP synthase: A reassessment of the functions of the b and a subunits. Biochim. Biophys. Acta 849, Dallmann, H. G., Flynn, T. G. and Dunn, S. D. (1992). Determination of the 1-ethyl-3-[(3-dimethylamino)propyl]- carbodiimide-induced cross-link between the β and ε subunits of Escherichia coli F 1-ATPase. J. Biol. Chem. 267, Deckers-Hebestreit, G. and Altendorf, K. (1996). The F of 1-type ATP synthases of bacteria: Structure and function of the F o complex. Annu. Rev. Microbiol. 50, Deckers-Hebestreit, G., Greie, J.-C., Stalz, W.-D. and Altendorf, K. (1999). The ATP synthase of Escherichia coli Structure and function of F o subunits. Biochim. Biophys. Acta (in press). Dimroth, P., Wang, H., Grabe, M. and Oster, G. (1999). Energy transduction in the sodium F-ATPase of Propionigenium modestum. Proc. Natl. Acad. Sci. USA 96, Dmitriev, O., Jones, P. C., Jiang, W. and Fillingame, R. H. (1999). Structure of the membrane domain of subunit b of the Escherichia coli F of 1 ATP synthase. J. Biol. Chem. 274, Dunn, S. D. (1992). The polar domain of the b subunit of Escherichia coli F 1F o-atpase forms an elongated dimer that interacts with the F 1 sector. J. Biol. Chem. 267, Dunn, S. D. and Chandler, J. (1998). Characterization of a b 2δ complex from Escherichia coli ATP synthase. J. Biol. Chem. 273, Dunn, S. D., Heppel, L. A. and Fullmer, C. S. (1980). The NH 2- terminal portion of the α subunit of Escherichia coli F 1 ATPase is required for binding the δ subunit. J. Biol. Chem. 255, Elston, T., Wang, H. and Oster, G. (1998). Energy transduction in ATP synthase. Nature 391, Engelbrecht, S. and Junge, W. (1997). Hypothesis. ATP synthase: A tentative structural model. FEBS Lett. 414, Fillingame, R. H. (1990). Molecular mechanics of ATP synthesis by F 1F o-type H + -translocating ATP synthases. In The Bacteria. A Treatise on Structure and Function, vol. 12 (ed. T. A. Krulwich), pp New York: Academic Press. Fillingame, R. H. (1997). Coupling H + transport and ATP synthesis in F 1F o-atp synthases: glimpses of interacting parts in a dynamic molecular machine. J. Exp. Biol. 200, Fillingame, R. H., Jones, P. C., Jiang, W., Valiyaveetil, F. I. and Dmitriev, O. Y. (1998). Subunit organization and structure in the F o sector of Escherichia coli F 1F o ATP synthase. Biochim. Biophys. Acta 1365, Fraga, D., Hermolin, J., Oldenburg, M., Miller, M. J. and Fillingame, R. H. (1994). Arginine 41 of subunit c of Escherichia coli H + -ATP synthase is essential in binding and coupling of F 1 to F o. J. Biol. Chem. 269, Girvin, M. E., Rastogi, V. K., Abildgaard, F., Markley, J. L. and Fillingame, R. H. (1998). Solution structure of the transmembrane H + -transporting subunit c of the F 1F o ATP synthase. Biochemistry 37, Groth, G., Tilg, Y. and Schirwitz, K. (1998). Molecular architecture of the c-subunit oligomer in the membrane domain of F-ATPases probed by tryptophan substitution mutagenesis. J. Mol. Biol. 281, Groth, G. and Walker, J. E. (1997). Hypothesis. Model of the c- subunit oligomer in the membrane domain of F-ATPases. FEBS Lett. 410, Hatch, L. P., Cox, G. B. and Howitt, S. M. (1995). The essential arginine residue at position 210 in the a subunit of the Escherichia coli ATP synthase can be transferred to position 252 with partial retention of activity. J. Biol. Chem. 270, Hermolin, J., Dmitriev, O. Y., Zhang, Y. and Fillingame, R. H. (1999). Defining the domain of binding of F 1 subunit ε with the polar loop of F o subunit c in the Escherichia coli ATP synthase. J. Biol. Chem. 274, Howitt, S. M., Rodgers, A. J. W., Jeffrey, P. D. and Cox, G. B. (1996). A mutation in which alanine 128 is replaced by aspartic acid abolishes dimerization of the b-subunit of the F of 1-ATPase from Escherichia coli. J. Biol. Chem. 271, Jäger, H., Birkenhäger, R., Stalz, W.-D., Altendorf, K. and Deckers-Hebestreit, G. (1998). Topology of subunit a of the Escherichia coli ATP synthase. Eur. J. Biochem. 251, Jans, D. A., Hatch, L., Fimmel, A. L., Gibson, F. and Cox, G. B. (1984). An acidic or basic amino acid at position 26 of the b subunit of Escherichia coli F 1F o-atpase impairs membrane proton permeability: Suppression of the uncf469 nonsense mutation. J. Bacteriol. 160,

9 F o complex of Escherichia coli ATP synthase 27 Jans, D. A., Hatch, L., Fimmel, A. L., Gibson, F. and Cox, G. B. (1985). Complementation between uncf alleles affecting assembly of the F 1F o-atpase complex of Escherichia coli. J. Bacteriol. 162, Jiang, W. and Fillingame, R. H. (1998). Interacting helical faces of subunits a and c in the F 1F o ATP synthase of Escherichia coli defined by disulfide cross-linking. Proc. Natl. Acad. Sci. USA 95, Jones, P. C. and Fillingame, R. H. (1998). Genetic fusions of subunit c in the F o sector of H + -transporting ATP synthase. Functional dimers and trimers and determination of stoichiometry by crosslinking analysis. J. Biol. Chem. 273, Jones, P. C., Harrison, M. A., Kim, Y.-I., Finbow, M. E. and Findlay, J. B. C. (1995). The first putative transmembrane helix of the 16 kda proteolipid lines a pore in the V o sector of the vacuolar H + -ATPase. Biochem. J. 312, Jones, P. C., Jiang, W. and Fillingame, R. H. (1998). Arrangement of the multicopy H + -translocating subunit c in the membrane sector of the Escherichia coli F 1F o ATP synthase. J. Biol. Chem. 273, Joshi, S. and Burrows, R. (1990). ATP synthase complex from bovine heart mitochondria. Subunit arrangement as revealed by nearest neighbor analysis and susceptibility to trypsin. J. Biol. Chem. 265, Jounouchi, M., Takeyama, M., Chaiprasert, P., Noumi, T., Moriyama, Y., Maeda, M. and Futai, M. (1992). Escherichia coli H + -ATPase: Role of the δ subunit in binding F 1 to the F o sector. Arch. Biochem. Biophys. 292, Junge, W. (1999). ATP synthase and other motor proteins. Proc. Natl. Acad. Sci. USA 96, Junge, W., Lill, H. and Engelbrecht, S. (1997). ATP synthase: An electrochemical transducer with rotatory mechanics. Trends Biochem. Sci. 22, Kato-Yamada, Y., Noji, H., Yasuda, R., Kinosita, K., Jr. and Yoshida, M. (1998). Direct observation of the rotation of ε subunit in F 1-ATPase. J. Biol. Chem. 273, Kauffer, S., Deckers-Hebestreit, G. and Altendorf, K. (1991). Substitution of the cysteinyl residue (Cys21) of subunit b of the ATP synthase from Escherichia coli. Eur. J. Biochem. 202, Lill, H., Hensel, F., Junge, W. and Engelbrecht, S. (1996). Crosslinking of engineered subunit δ to (αβ) 3 in chloroplast F-ATPase. J. Biol. Chem. 271, Long, J. C., Wang, S. and Vik, S. B. (1998). Membrane topology of subunit a of the F 1F o ATP synthase as determined by labeling of unique cysteine residues. J. Biol. Chem. 273, Maggio, M. B., Parsonage, D. and Senior, A. E. (1988). A mutation in the α-subunit of F 1-ATPase from Escherichia coli affects the binding of F 1 to the membrane. J. Biol. Chem. 263, Matthey, U., Kaim, G., Braun, D., Wüthrich, K. and Dimroth, P. (1999). NMR studies of subunit c of the ATP synthase from Propionigenium modestum in dodecylsulphate micelles. Eur. J. Biochem. 261, McCormick, K. A. and Cain, B. D. (1991). Targeted mutagenesis of the b subunit of F 1F o ATP synthase in Escherichia coli: Glu-77 through Gln-85. J. Bacteriol. 173, McLachlin, D. T., Bestard, J. A. and Dunn, S. D. (1998). The b and δ subunits of the Escherichia coli ATP synthase interact via residues in their C-terminal regions. J. Biol. Chem. 273, McLachlin, D. T. and Dunn, S. D. (1997). Dimerization interactions of the b subunit of the Escherichia coli F 1F o-atpase. J. Biol. Chem. 272, Nakamoto, R. K. (1999). Molecular features of energy coupling in the F of 1 ATP synthase. News Physiol. Sci. 14, Noji, H., Yasuda, R., Yoshida, M. and Kinosita, K., Jr (1997). Direct observation of the rotation of F 1-ATPase. Nature 386, Ogilvie, I., Aggeler, R. and Capaldi, R. A. (1997). Cross-linking of the δ subunit to one of the three α subunits has no effect on functioning, as expected if δ is a part of the stator that links the F 1 and F o parts of the Escherichia coli ATP synthase. J. Biol. Chem. 272, Oster, G. and Wang, H. Y. (1999). ATP synthase: two motors, two fuels. Structure 7, R67 R72. Penefsky, H. S. (1985). Mechanism of inhibition of mitochondrial adenosine triphosphatase by dicyclohexylcarbodiimide and oligomycin: Relationship to ATP synthesis. Proc. Natl. Acad. Sci. USA 82, Richardson, J. S. and Richardson, D. C. (1989). Principles and patterns of protein conformation. In Prediction of Protein Structure and the Principles of Protein Conformation (ed. G. D. Fasman), pp New York: Plenum Press. Rodgers, A. J. W. and Capaldi, R. A. (1998). The second stalk composed of the b- and δ-subunits connects F o to F 1 via an α- subunit in the Escherichia coli ATP synthase. J. Biol. Chem. 273, Rodgers, A. J. W., Wilkens, S., Aggeler, R., Morris, M. B., Howitt, S. M. and Capaldi, R. A. (1997). The subunit δ-subunit b domain of the Escherichia coli F 1F o ATPase. The b subunits interact with F 1 as a dimer and through the δ subunit. J. Biol. Chem. 272, Sabbert, D., Engelbrecht, S. and Junge, W. (1997). Functional and idling rotatory motion within F 1-ATPase. Proc. Natl. Acad. Sci. USA 94, Sawada, K., Kuroda, N., Watanabe, H., Moritani-Otsuka, C. and Kanazawa, H. (1997). Interaction of the δ and b subunits contributes to F 1 and F o interaction in the Escherichia coli F 1F o- ATPase. J. Biol. Chem. 272, Schneider, E. and Altendorf, K. (1987). Bacterial adenosine 5 - triphosphate synthase (F 1F o): Purification and reconstitution of F o complexes and biochemical and functional characterization of their subunits. Microbiol. Rev. 51, Senior, A. E. (1983). Secondary and tertiary structure of membrane proteins involved in proton translocation. Biochim. Biophys. Acta 726, Singh, S., Turina, P., Bustamante, C. J., Keller, D. J. and Capaldi, R. A. (1996). Topographical structure of membane-bound Escherichia coli F 1F o ATP synthase in aqueous buffer. FEBS Lett. 397, Sorgen, P. L., Bubb, M. R., McCormick, K. A., Edison, A. S. and Cain, B. D. (1998a). Formation of the b dimer is necessary for interaction with F 1-ATPase. Biochemistry 37, Sorgen, P. L., Caviston, T. L., Perry, R. C. and Cain, B. D. (1998b). Deletions in the second stalk of F 1F o-atp synthase in Escherichia coli. J. Biol. Chem. 273, Steffens, K., Schneider, E., Deckers-Hebestreit, G. and Altendorf, K. (1987). F o portion of Escherichia coli ATP synthase. Further resolution of trypsin-generated fragments from subunit b. J. Biol. Chem. 262, Takeyama, M., Noumi, T., Maeda, M. and Futai, M. (1988). F o portion of Escherichia coli H + -ATPase. Carboxyl-terminal region