Probing the rotor subunit interface of the ATP synthase from Ilyobacter tartaricus Denys Pogoryelov 1,2 , Yaroslav Nikolaev 3, *, Uwe Schlattner 4,5 , Konstantin Pervushin 3, , Peter Dimroth 1 and Thomas Meier 1,2 1 Institute of Microbiology, Eidgeno ¨ ssische Technische Hochschule, Zurich, Switzerland 2 Department of Structural Biology, Max-Planck Institute of Biophysics, Frankfurt am Main, Germany 3 Laboratory of Physical Chemistry, Eidgeno ¨ ssische Technische Hochschule, Zurich, Switzerland 4 Institute of Cell Biology, Eidgeno ¨ ssische Technische Hochschule, Zurich, Switzerland 5 Laboratory for Fundamental and Applied Bioenergetics, Inserm E0221, University Joseph Fourier, Grenoble, France F-ATP synthases convert the energy of an electro- chemical proton or sodium ion gradient into ATP, the universal chemical energy source of living cells. These enzymes are composed of a water-soluble F 1 domain (subunits a 3 b 3 cde), with the catalytic sites for ATP synthesis, and the membrane-embedded F o domain (bacterial subunits ab 2 c 10–15 ), with the sites for the translocation of the ions. In ATP synthesis mode, the F o motor converts the electrochemical ion gradient into torque to force the F 1 motor to act as an ATP generator, whereas, for ATP hydrolysis, F 1 converts the chemical energy of ATP hydrolysis into torque causing the F o motor to act as an ion pump (for reviews, see [1–4]). Rotation of the asymmetric Keywords c ring; F 1 F o ATP synthase; Ilyobacter tartaricus; rotor subunit interaction; surface plasmon resonance Correspondence T. Meier, Max-Planck Institute of Biophysics, Max-von-Laue Str. 3, 60438 Frankfurt am Main, Germany Fax: +49 69 63033002 Tel: +49 69 63033038 E-mail: thomas.meier@mpibp-frankfurt. mpg.de Present addresses *Biozentrum, University of Basel, Switzerland School of Biological Sciences, Nanyang Technological University, Singapore; Biozen- trum, University of Basel, Switzerland (Received 8 February 2008, revised 29 July 2008, accepted 1 August 2008) doi:10.1111/j.1742-4658.2008.06623.x The interaction between the c 11 ring and the ce complex, forming the rotor of the Ilyobacter tartaricus ATP synthase, was probed by surface plasmon resonance spectroscopy and in vitro reconstitution analysis. The results pro- vide, for the first time, a direct and quantitative assessment of the stability of the rotor. The data indicated very tight binding between the c 11 ring and the ce complex, with an apparent K d value of approximately 7.4 nm. The rotor assembly was primarily dependent on the interaction of the c ring with the c subunit, and binding of the c ring to the free e subunit was not observed. Mutagenesis of selected conserved amino acid residues of all three rotor components (cR45, cQ46, cE204, cF203 and eH38) severely affected rotor assembly. The interaction kinetics between the ce complex and c 11 ring mutants suggested that the assembly of the c 11 ce complex was governed by interactions of low and high affinity. Low-affinity binding was observed between the polar loops of the c ring subunits and the bottom part of the c subunit. High-affinity interactions, involving the two residues cE204 and eH38, stabilized the holo-c 11 ce complex. NMR experiments indicated the acquisition of conformational order in otherwise flexible C- and N-terminal regions of the c subunit on rotor assembly. The results of this study suggest that docking of the central stalk of the F 1 complex to the rotor ring of F o to form tight, but reversible, contacts provides an explanation for the relative ease of dissociation and reconstitution of F 1 F o complexes. Abbreviations DDM, n-dodecyl b- D-maltoside; DHPC, dihexanoylphosphatidylcholine; HSQC, heteronuclear single quantum correlation; OG, octyl b- D-glucoside; POPC, 1-palmitoyl-2-oleoylphosphatidylcholine; RU, response unit; SPR, surface plasmon resonance; TROSY, transverse relaxation-optimized NMR spectroscopy. 4850 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS c subunit within the hexameric assembly of alternating a and b subunits elicits conformational changes in the catalytic b subunit sites, resulting in ATP synthesis, consistent with the ‘binding change model’ [5], the crystal structure of F 1 [6] and single-molecule video microscopy [7]. In an ATP synthase at work, drag is imposed by the F 1 motor components; this has been proposed to cause elastic energy storage within the a-helical domain of the c subunit [8], the peripheral stalk [9] and the rotat- ing c ring [10]. To withstand the resulting strain of up to – 55 kJÆmol )1 [11], the c ring forms a tight rotor complex with the F 1 subunits c and e, similar to the tight binding between the stator components (ab 2 a 3 b 3 d)ofF 1 and F o [11–13]. Although high-resolu- tion structures exist for ce within the F 1 complex (e.g. [14], ce complex [15], isolated e subunit [16] and the c 11 ring from Ilyobacter tartaricus [10]), our knowledge about the F 1 –F o rotor interaction is restricted to a 3.9 A ˚ resolution F 1 c 10 structure from yeast ATP syn- thase [17]. On the basis of these structures, the lower part of the F 1 complex can be derived at a resolution suitable for the identification of possible amino acid residue candidates forming the interface between ce and the c ring, and these residues have been corrobo- rated by cross-linking experiments and EPR spectro- scopy of site-directed spin labels. Using these approaches, the e subunit residues 26–33 and 38 (Esc- herichia coli numbering) [18–20] and the c subunit resi- dues 200–210 [21,22] are localized in the direct vicinity of the hydrophilic loop units of the c ring. In this article, we have used surface plasmon reso- nance (SPR) [23] and NMR spectroscopy [transverse relaxation-optimized NMR spectroscopy (TROSY) and NOE-TROSY [24]] to obtain a greater understanding of the interaction sites and affinities between the ce complex and the c 11 ring during the assembly of the I. tartaricus ATP synthase. We report tight, but revers- ible, binding between the rotor parts of F 1 and F o , with a K d value in the nanomolar range, and identify indi- vidual contributions of important amino acid residues in rotor complex stability. Further, we were able to monitor the accretion of structural ordering within flex- ible domains of the c subunit on rotor assembly. Results In vitro rotor assembly from the c 11 ring and subunits c and e The aim of this study was to obtain a better under- standing of the binding processes of the rotor subunits (c 11 , c and e) from I. tartaricus ATP synthase. For this purpose, the membrane-embedded F o rotor part, the c 11 ring, was used from wild-type I. tartaricus cells and from I. tartaricus cells and from E. coli cells heterolo- gously expressing the I. tartaricus c 11 ring ([25,26] and Supporting information). The c and e subunits, form- ing the water-soluble F 1 rotor complex of the I. tar- taricus ATP synthase, were heterologously expressed in E. coli cells: we constructed appropriate expression vectors for the synthesis of His-tagged c and e sub- units, and purified individual c¢ (residues 12–253 [15]) and e subunits, and the c¢e pair, by Ni 2+ -nitrilotriace- tic acid affinity chromatography (Fig. 1A, lane 1). To assess rotor assembly, the c 11 ring was applied to the c¢e complex on the surface of the Ni 2+ -nitrilotriacetic acid resin of the column, and the c 11 c¢e complex (rotor) was eluted by increasing the imidazole concen- tration (Fig. 1A, lane 2). This method yielded stable rotor complexes in the presence of several non-ionic detergents, e.g. dihexanoylphosphatidylcholine (DHPC), octyl b-d-glucoside (OG) and n-dodecyl b-d-maltoside (DDM) (shown for DHPC in Fig. 1A, lane 2). The in vitro formation of these rotor assemblies was further corroborated by native gel electrophoresis and gel filtration experiments (data not shown). Binding characteristics of the c 11 ring to the c¢e complex studied by SPR The kinetic characteristics of the interaction between the c¢e complex (as a ligand) and the isolated c ring (as an analyte) were studied in detail by SPR spectros- copy with a Biacore instrument. A typical set of exp- erimental kinetic traces recorded at different concentrations of analyte is shown in Fig. 2A. At higher c ring concentrations, a minor systematic devia- tion was observed between the measured and fitted curves, indicating a slow, probably nonspecific, binding process (Fig. 2B). Association and dissociation rate constants (k on and k off , respectively; see Eqns (1) and (2) in Experimental procedures) were independent of c ring concentration in the range 1–300 nm. These data allowed us to calculate (K d = k off ⁄ k on ) a dissociation equilibrium constant (or affinity constant) K d of about 7nm based on 50 independent binding experiments under standard conditions, with individual experimen- tal values scattering in the range 4.1–10.7 nm (Table 1A). Thus, the high-affinity interaction between the c ring and the c¢e complex is characterized by a very slow dissociation. Such an affinity is comparable with that of a typical antigen–antibody complex [27], and consistent with that published for the E. coli F 1 F o complex [13]. The parameters of all the interac- tions that could be quantified are summarized in D. Pogoryelov et al. Rotor interactions of the F-ATP synthase FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4851 Table 1. In binding experiments using monomeric c subunits, we could not detect any interaction with the c¢e complex (data not shown). With respect to salt, binding of the c ring to an immobilized c¢e complex was weak at NaCl concentra- tions below 500 lm and strong at NaCl concentrations AB C D E Fig. 1. SDS-PAGE showing the purification and reconstitution experiments of rotor subunits (c 11 c¢e) from I. tartaricus ATP synthase. The rotor subunits c¢, e and the c 11 ring were purified as described in the Supporting information. Reconstitution was performed by binding the His-tagged subunits (either His-c¢ or e-His) to Ni 2+ -nitrilotriacetic acid agarose with subsequent application of the c ring. The eluates were col- lected and analysed by SDS-PAGE. The molecular masses and proteins used in this experiment are indicated on the left and right, respec- tively. (A) 1, purified c¢e complex; 2, elution of the rotor in 1.5 m M DHPC. (B) Reconstitution of the heterologously synthesized c rings with the c¢e complex: 1, elution fraction with the wild-type c rings; flow through and elution fractions with the two c ring mutants cR45A (lanes 2 and 3, respectively) and cQ46E (lanes 4 and 5, respectively). (C) Purification of c¢F203A ⁄ e: 1, flow through; 2, wash; 3, elution. (D) Reconsti- tution of the c rings with the c¢e complexes harbouring two point mutations: flow through and elution fractions of the experiments with the mutants c¢E204A ⁄ e (lanes 1 and 2, respectively) and c¢eH38A (lanes 3 and 4, respectively). (E) Reconstitution of c rings with separate sub- units c¢ and e: flow through and elution fractions of the experiments with the e-His (lanes 1 and 2, respectively) and His-c¢ (lanes 3 and 4, respectively) subunits. Time (s) 0 100 200 300 Response (RU) 0 500 1000 1500 2000 2500 3000 A B (1) 500 n M (2) 300 n M (3) 100 n M (4) 10 n M (5) 1 n M Fit k on Fit k off (1) (2) (3) (4) (5) Time (s) 020 6040 80 100 120 140 160 Response (RU) –100 –50 0 50 100 (1) (2) (3) (4) (5) (1) 500 n M (2) 300 nM (3) 100 nM (4) 10 nM (5) 1 nM Fig. 2. SPR binding and dissociation kinetics of detergent (DHPC)-solubilized c ring to the c¢e complex immobilized on an Ni 2+ -nitrilotri- acetic acid surface. (A) Overlay plot showing the concentration-dependent interaction kinetics of the c ring at 500, 300, 100, 10 and 1 n M, and the single exponential fitting curves (bold) for association (black) and dis- sociation (grey). (B) Representative residual plot showing the deviation of the mathe- matical fit relative to the data points. Rotor interactions of the F-ATP synthase D. Pogoryelov et al. 4852 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS above 10 mm (Fig. 3A). A strong interaction was also observed in the presence of Mg 2+ at concentrations above 10 mm, and this c 11 –ce interaction could not be distinguished from effects caused by other ions (K + , Mg 2+ ,Ca 2+ ,Cl ) and SO 4 2) ) at concentrations above 10 mm, indicating that the binding strength was depen- dent on the ionic strength of the buffer and not on a specific ion (e.g. Mg 2+ ). Therefore, the specific require- ment of Mg 2+ for the assembly of F 1 and F o into a functional entity could not be attributed to these con- tact sites at the rotor interface. The pH of the solution, however, had a significant impact on the rate constants k on and k off of the inter- acting partners (Fig. 3B). A low pH (5.5) favoured fast dissociation of the c ring from the c¢e complex (high k off ), which was partially compensated for by a fast association rate k on . At a higher pH, dissociation stabilized at a slower rate, but, above pH 8.5, the association rate was drastically reduced. Taken together, the affinity constant remained relatively unchanged over a wide range of pH, but decreased significantly (higher K d ) at values above pH 8.5. If a high pH (9.5) was combined with a low salt condi- tion, the association of the c ring with the c¢e com- plex was impeded (not shown). This combination essentially reproduced the well-documented F 1 –F o separation (or stripping) condition, which seems to be caused by an impaired reassociation of F 1 with F o at the rotor interface. Mutations affecting the interaction of c 11 with c¢e In order to assess the interaction of the c¢e complex with selected amino acids in the loop region of the iso- lated c ring [amino acids RQPE(D)], we introduced point mutations at position cR45 or cQ46 and isolated the corresponding c rings (Fig. 1B, lanes 2 and 4). The interactions of the stable c rings with the c¢e complex are shown in Fig. 4A. Strong binding was observed for the heterologously synthesized wild-type c rings, with rate constants (k on and k off ) and derived dissociation equilibrium constants (K d ) almost identical to those obtained with the c ring isolated from I. tartaricus cells (Table 1B). Mutant c rings (R45A, Q, Y and E; Q46A, Y and E) did not bind to the c¢e complex, as revealed by SDS-PAGE (Fig. 1B) and SPR kinetic analysis (Fig. 4A). The mutants cP47A and cE48A did not form c ring complexes sufficiently stable for isolation (not shown). The contact region of the c¢e complex to the polar loop of the c subunit can be allocated to the E. coli c subunit residues 200–210 [21,22]. An amino acid sequence alignment of this c subunit region (Fig. 5A) shows low sequence conservation, but some acidic resi- dues are abundant. We replaced each of these residues (c¢E197, c¢E204, c¢E208 and c¢D209, I. tartaricus num- bering) individually by Ala and determined the SPR kinetics of c ring binding to the mutant c¢e complexes (Fig. 6A). The rate constants k on and k off and the Table 1. Summary of binding parameters. Binding parameters ⁄ constants for the interaction of c rings with immobilized wild-type (wt) c¢e complex (A, B), c¢e complex with c¢ mutants (C), c¢e complex with e mutants (D) and individual c¢ subunits (E). Rate constants and K d were calculated as described in Experimental procedures. Data in A represent the mean ± standard deviation (SD) of 10 independent experi- ments at five different ligand concentrations each; data in B–E represent the mean of two to three independent experiments at three differ- ent concentrations each. Rate and affinity constants of the mutants are considered to be different from wt when differing by more than 1SD (wt). Analyte Ligand k off · 10 )3 (s )1 ) k on · 10 4 (M )1 Æs )1 ) K d (nM) (A) wt c ring wt c¢e 1.1 ± 0.1 14.9 ± 3.2 7.4 ± 3.3 (B) wt c ring, recombinant wt c¢e 1.1 9.9 11.1 (C) wt c ring c¢D209A ⁄ e 0.8 8.6 10.1 c¢E208A ⁄ e 0.9 9.7 10.5 c¢E197A ⁄ e 1.1 8.6 14.3 c¢Y201A ⁄ e 2.0 8.6 21.9 c¢E204A ⁄ e 76.7 0.5 16300 c¢E204Q ⁄ e 78.5 0.6 12800 (D) wt c ring c¢eD31A 2.0 7.2 27.8 c¢eD31K 5.5 7.5 66.4 c¢eE29K 5.5 5.8 94.1 c¢eE29A 6.5 8.1 80.3 c¢eH38A 69.8 1.1 6600 (E) wt c ring c¢WT 1.5 7.5 19.7 a c¢E204A 59.9 0.3 20800 a This interaction is not entirely well described by a single exponential fit. D. Pogoryelov et al. Rotor interactions of the F-ATP synthase FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4853 derived dissociation equilibrium constants (K d ) of the c¢E197A, c¢E208A and c¢D209A mutants were in the range of those determined for the wild-type c¢e com- plex (Table 1C). In contrast, the two c¢E204 (A or Q) mutations affected both k on and k off significantly. The k on values of both mutants decreased approximately 10-fold, and the k off values increased by two orders of magnitude (Table 1C). Consequently, their dissociation equilibrium constants (K d ) were at least three orders of magnitude higher than those obtained with the wild- type c¢e complex. Therefore, the formation of rotor complexes harbouring the c¢E204A mutant with a reduced stability can be corroborated by the in vitro reconstitution method (Fig. 1D, lanes 1 and 2). The c¢E204K mutant showed only weak residual binding, with k on and k off values at the detection limits, suggest- ing that c¢E204 is the most critical of these acidic resi- dues for the formation of a stable rotor complex. This observation is corroborated by earlier work which showed that replacement of the homologous residue in E. coli (cE208) by K or C decreased the enzyme’s cou- pling and proton pumping efficiency [22,28]. However, the second site mutations in the hydrophilic loop of the c subunits suppressed the uncoupling effect of cE208K [28]. In addition to the negatively charged residues, the flexible loop at the bottom of the c subunit also con- tains two aromatic residues (cY201 and cF203), which are conserved in bacterial ATP synthases (Fig. 5A). The results of SPR analysis of the complex formation for the c¢Y201A mutant (Fig. 6A, Table 1C) showed only minor changes in the affinity, but the c¢F203A mutant prevented the formation of a stable c¢e com- plex (Fig. 1C) and only weak binding between the c ring and c¢F203A was detected (Fig. 4B), in agree- ment with functional studies made with the homolo- gous amino acid residue Y205 in the c subunit of E. coli [29,30]. It may be noteworthy that in vitro A B Fig. 3. Effect of salt and pH on the binding of the c ring (100 n M) to immobilized c¢e complex. (A) Dependence of the equilib- rium response (R eq ) on the salt (NaCl) con- centration in the binding buffer. The values for R eq were derived from the contact phase fit of the corresponding experimental kinetic traces. The binding experiments were per- formed in BisTrisPropane-HCl buffer (2 m M, pH 7). (B) pH dependence of the binding association (k on ) and dissociation (k off ) rate constants determined in 10 m M BisTrisPro- pane-HCl buffer (pH 5.5–9.5) in the pres- ence of 300 m M NaCl and 2 mM MgCl 2 . Rotor interactions of the F-ATP synthase D. Pogoryelov et al. 4854 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS reconstitution experiments (Fig. 1E, lanes 3 and 4), as well as SPR analyses (Fig. 4B), indicate that the c ring binds to the separate c¢ subunit with an approximately five-fold lower SPR response when compared with the c¢e complex (Fig. 2A, Table 1E). This probably occurs as a result of improper folding of the protein, in line with the observed slight deviation of the binding and dissociation kinetics. Binding of the c¢E204A mutant, with or without complex formation, with the e subunit to the c ring also shows a similar low range of K d (compare Fig. 4B, Table 1E with Fig. 6A, Table 1C). The latter observation suggests that interaction studies with the isolated c subunits may represent a feasible approach for selected cases. Influence of the e subunit on the stability of the rotor In contrast with the separate c subunit, a specific inter- action of the c ring with a separate e subunit could not be observed by SPR analysis (Fig. 4B) or in vitro reconstitution (Fig. 1E, lanes 1 and 2). To investigate whether the e subunit has an auxiliary role in rotor assembly, interaction kinetics with the e subunit mutants were recorded. The results in Fig. 6B and Table 1D show that the replacement of eE29 or eD31 with A or K (numbering is equivalent in E. coli and I. tartaricus) increased the dissociation rate of the c ring from the c¢e complex by about two- to six-fold, but the association rates remained largely unchanged. The resulting increased K d value (i.e. lower affinity) indicates a contribution of residues eE29 and eD31 to rotor stability, and is in good agreement with previous work, which showed partial uncoupling of the E. coli ATP synthase by the mutations eE29, eD31 and eH38 [18–20,31]. A substantial alteration in the assembly of the rotor was observed in the mutant eH38A (Fig. 6B), resulting in an approximately 10-fold decrease in k on and increase in k off by almost two orders of magnitude (K d : micromolar range; Table 1D). Moreover, the A B Fig. 4. SPR binding and dissociation kinetics of the c rings to the immobilized c¢e com- plex and the separate subunits c and e. (A) Heterologously expressed wild-type c ring (broken line) and cR45 or cQ46 mutant (full lines) at 100 n M. (B) Kinetic traces of wild- type c ring (500 n M) to immobilized separate subunits c (1–3) and e (4). The single expo- nential fitting curves are depicted in bold for association (black) and dissociation (grey) phases. The c ring does not interact with the c¢F203A mutant and the separate e sub- unit. Binding of the c ring to the separate c¢ unit seems to be more complex, involving a more pronounced slow component. D. Pogoryelov et al. Rotor interactions of the F-ATP synthase FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4855 replacement of eH38 with K or D affected rotor assembly so severely that the binding kinetics were clearly too slow for quantitative analysis in our time window. NMR analysis of the interaction between the c¢e complex and the c 11 ring To investigate the interaction between the isolated c¢e complex and the detergent-solubilized c 11 ring by NMR spectroscopy, we employed conventional stable isotope ( 2 H ⁄ 15 N ⁄ 13 C) labelling techniques, as well as TROSY-heteronuclear single quantum correlation (HSQC) and three-dimensional (3D)-TROSY- HNCA ⁄ HNCACB pulse schemes [32] (for assignment and data interpretation, see Supporting information). Titration experiments were performed using the 2 H, 15 N-labelled c¢e complex with an unlabelled c ring solubilized in DHPC micelles. All changes in the 1 H, 15 N-TROSY-HSQC spectra were attributed solely to the interaction of the c¢e complex with the c oligo- mer, as no changes in the c¢e spectra were observed when adding detergent micelles without protein. Figure 7 shows that 18 of the 28 cross-peaks assigned to both N- and C-terminal loop regions of the c sub- unit (residues 59–70 and 198–207, Fig. S1) were broad- ened beyond detection when titrating the isolated c¢e complex with the c oligomer. The maximum effect of resonance broadening was observed at c 11 : c¢e molar ratios of 1 : 1 and above, confirming a single binding site between the interacting components, as predicted by SPR analysis. The observed broadening of the c¢ subunit resonances indicates that flexible loops of the c¢ subunit become structured on binding, reaching the spin relaxation rates of the entire com- plex. Although the contribution of the c subunit could be clearly shown by both techniques used in this work, an involvement of the e subunit in rotor assembly could only be pinpointed by SPR analysis (eE29, eD31and eH38) because of the lack of signals in TRO- SY. Discussion Two F 1 –F o binding affinities during rotor assembly We have shown that the interaction between the c ring and the central stalk subunits c and e of the rotor of I. tartaricus ATP synthase comprises high-affinity A B Fig. 5. Protein sequence alignments of amino acid stretches structurally located at the F 1 –F o interface of the central stalk domain of F-ATP synthases. The sequence alignments of subunit c (A) and subunit e (bacteria) ⁄ d (eukaryotes) (B) include species for which high-resolution structures are available (comprising the amino acid stretches of interest). Secondary structures are shown on top of the alignments (bacteria, full line; eukaryotes, broken line). The numbering is according to the sequence of I. tartaricus. Conserved amino acids [57] are in bold. Resi- dues which have been characterized by F 1 –F o cross-links (for references, see Introduction) are underlined. The conserved charged and aro- matic amino acid residues attributed to the rotor interface are highlighted (in black or grey, respectively). The critical residues for the interaction of the c¢e complex with the c ring are marked by an asterisk. Rotor interactions of the F-ATP synthase D. Pogoryelov et al. 4856 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS binding (K d % 7.4 nm). This value is similar to the binding affinity determined in the stator complex (ab 2 F 1 ) of the E. coli ATP synthase [13]. Hence, rotor and stator appear to contribute equally to the intrinsic binding energy of complex assembly. The assembly of the c 11 ring and the c¢e complex can be distinguished experimentally by a high-affinity interaction (nm) that can be shifted to low-affinity binding (lm) by mutation of the c¢E204 or eH38 residues to Ala. These residues appear to be responsible for specific high-affinity con- tacts with cR45 and cQ46; consequently, mutation of c¢E204 or eH38 results in a fast dissociation of the c ring from the c¢e complex. The isolated c¢ subunit can achieve a high-affinity interaction with the c 11 ring in the absence of the e subunit, although less robust than with the c¢e complex. However, in the mutant c¢E204A, only low-affinity binding is maintained, and this is influenced by changes in the ionic strength and pH. This is completely abolished by mutating selected residues [cR45 (A,Q,Y,E), cQ46 (A,Y,E) and cF203A], suggesting that the c subunit also contributes to the establishment of the low-affinity contacts with the c ring. Furthermore, our data suggest that the sepa- rately synthesized e-His subunit does not interact with the c ring by itself; only when complexed with the c¢ subunit can the conserved eH38 establish a high- affinity interaction. This is in agreement with data from chloroplast and yeast mitochondrial ATP synthase [33,34], where the rotor could be assembled only from the c subunit and the c ring. However, in contrast with E. coli and Bacillus PS3 ATP synthases, the e subunit is essential for functional reconstitution of F 1 with F o [20,35–38], but the partial contribution of the e subunit to the stability of the rotor in these cases is not yet clear. Does the interaction of the c ring with the c and e subunits have anything to do with the regulation of enzyme activity? Potentially, this may be so. The low and high affinities within the c 11 ring and ce complex demonstrate not only a high stability, but also a high A B Fig. 6. SPR kinetic traces of the interaction between the wild-type c ring and c¢e com- plexes carrying mutations in the c subunit (A) and e subunit (B). Overlay plot showing the SPR kinetics together with the single exponential fitting curves (bold) for associa- tion (black) and dissociation (grey). The c ring concentration was varied from 10 to 500 n M; only the SPR kinetics recorded at 300 n M of the c ring are shown. Mutations mainly affect the dissociation kinetics. No binding was observed between the c ring and c¢E204K (7) (A) or eH38K ⁄ D(7⁄ 8) (B). D. Pogoryelov et al. Rotor interactions of the F-ATP synthase FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4857 plasticity, between the F 1 and F o complexes, and the switching between tight and weak affinity may play a role in the coupling activity of the enzyme in some ATP synthases. Presently, there is no experimental evidence for this hypothesis, but it is well established for chloroplast ATP synthases that the c subunit con- tains a unique, 40-amino-acid regulatory domain at the bottom of the c subunit (Fig. 5A), which is involved in coupling of the enzyme via redox-thiol modulation [33,39,40]. The approach of probing the rotor’s interface, as established in this work, could represent a feasible method to study further the cou- pling and regulation between the F 1 and F o complexes in these plant-type ATP synthases. Structural considerations in the assembly and interaction of rotor components Figure S1 shows a model for the location of the resi- dues involved in the high-affinity interaction (cE204 and eH38), which is based on the structure of the cor- responding complex from E. coli [15]. Both residues are at the bottom of the ce complex and in close prox- imity to each other. In the available structures of c subunits from different organisms [14,15,38,41–44], the amino acid stretch (residues 198–207, I. tartaricus numbering) of the putative F 1 –F o interface falls into the flexible region of the c subunit loop including resi- dues cE(D)204 and cF(Y)203. These are the only con- served residues in this stretch (Fig. 5A), and are critical for the rotor stability as shown in this work. According to our NMR spectroscopy data (Fig. 7), this flexible region of the c subunit undergoes struc- tural rearrangements in concert with the stretch of residues 59–70, and they both become stabilized on high-affinity interaction with the DHPC-solubilized c 11 ring. The involvement of residues 59–70 from the c subunit for complex formation with the c 11 ring has not been detected previously [21] and, according to the available structures of the c¢e complex, this loop is not located at the predicted interface region. Therefore, a possible involvement of this region in complex forma- tion requires further research. In contrast with the c and e subunits, which have a considerably high variation of amino acid residues in the contact region with the c ring, multiple amino acid sequence comparisons of c subunits from F-ATP syn- thases show very high conservation of the loop amino acids [R(K), Q, P, E(D)]. The surface structures of these c ring loop regions, and their local charge distri- bution [10] in particular, indicate that the contact sites of all c rings comprise inner and outer rings with posi- tive and negative charges, respectively (Fig. S1). This A B Fig. 7. Solution NMR of the c¢e complex. 1 H, 15 N-TROSY-HSQC spectra of the 2 H, 15 N-uniformly labelled c¢e complex in 3 mM DHPC, 50 mM K 2 HPO 4 ⁄ KH 2 PO 4 pH 7.0, 300 mM NaCl, 2 mM MgCl 2 and 10% D 2 O, recorded at 5 °C and 600 MHz for 12 h. (A) HSQC spectra of the c¢e complex (30 l M). (B) HSQC spectra of c¢e on addition of equimolar amounts of unlabelled c 11 ring. Numbering corresponds to the resonances attributed to the individual amino acid residues stemming from the c¢ subunit. Assignment (according to the numbering of the I. tartaricus c subunit): 1, cG59; 2, cG70; 8, cE191; 9, cI190; 17, cE204; 21, cR192; 28, cV193. Inset in (A) indicates the changes in the HSQC spectrum of the c¢e complex by mutating the cE204 residue to Gln. Inset in (B) indicates the changes in the selected areas of the HSQC spectrum of the c¢e complex imposed by the addition of unlabelled c 11 ring at differ- ent molar ratios. Rotor interactions of the F-ATP synthase D. Pogoryelov et al. 4858 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS appears to be a common feature in all F-ATP synthas- es, and this arrangement seems to be mandatory for the formation of stable hairpin folding of the two heli- ces of the c subunit [45]. Moreover, the c¢e complex is able to bind not only to c rings from its native ATPase, but also to larger c rings from other species (D. Pogoryelov and T. Meier, unpublished data; [46]). This work demonstrates clearly that the c ring residues cR45 and cQ46, which are part of the c ring loop region, are both obligatory for high- and low-affinity contacts with the c¢e complex, and that only c rings, but not monomeric c subunits, can form a complex with c¢e. The binding studies between the c¢e complexes and the c 11 ring presented here represent an in vitro simula- tion of complex assembly, and the limitations must be emphasized. The conclusions drawn in this article are based entirely on measurements with isolated subunits; hence, in principle, they may not represent the situa- tion during in vivo ATP synthase assembly. However, our results are in accordance with the reported affinity constants measured for the ATP synthase, contact sites, cross-links and effects of critical point mutations on the rotor assembly in the functional enzyme ([13,18–22,28–30,34,47–53] and references therein]. Hence, the rotor assembly observed in vitro in this study could provide a glimpse into the in vivo forma- tion of the native ATP synthase rotor and hence F 1 F o assembly. In our view, docking of the central stalk of the F 1 complex to the rotor ring of F o to form tight, but reversible, contacts must be one of the last steps in the assembly of the ATP synthase complex, and can explain the relative ease of dissociation and reconstitu- tion of F 1 F o complexes observed more than four dec- ades ago [54], and well documented ever since. Experimental procedures The construction of the plasmids, the synthesis and purifi- cation of the subunits (c¢, e and c rings) and NMR meth- ods are described in Supporting information. In vitro reconstitution of the rotor complex The whole reconstitution procedure was performed at 20 °C. The imidazole concentration of the c¢e sample was first decreased to 40 mm by diluting the purified protein (see above) 10 times with buffer containing 50 mm potas- sium phosphate (pH 7.0), 300 mm NaCl and 2 mm MgCl 2 . Then, 1 nmol of the material was immobilized on a 1 mL Ni 2+ -nitrilotriacetic acid agarose column and washed with three column volumes of 50 mm potassium phosphate buf- fer (pH 7.0) containing 300 mm NaCl, 50 mm imidazole and 2 mm MgCl 2 [buffer (1)]. The material on the column was then equilibrated with buffer (1) containing one of the selected detergents (1.5 mm DHPC, 0.02% DDM or 1% OG), and 12 mL of the purified c ring sample [0.1 lm in buffer (1) containing the same detergent] were applied. Unbound c ring was removed by washing with three col- umn volumes of buffer (1) containing the selected detergent, and elution of the reconstituted rotor complex was per- formed by the addition of two column volumes of elution buffer containing 50 mm potassium phosphate (pH 7.0), 300 mm NaCl, 400 mm imidazole and 2 mm MgCl 2 and the same detergent. The same procedure was used to check complex formation of the c ring with the single isolated subunits c¢ and e. All eluted rotor complexes were analysed by SDS-PAGE. SPR binding assays Binding of the c ring to immobilized His-tagged c¢e com- plexes, or to separate c¢ and e subunits, was studied quanti- tatively using a BIACORE 2000 and nitrilotriacetic acid sensor chip from Biacore AB (Uppsala, Sweden). The sur- face was Ni 2+ coated with a 3 min injection of 1 mm NiSO 4 at a flow rate of 10 lLÆmin )1 . About 1000 response units (RUs) of ligand (purified His-tagged proteins diluted in running buffer to 200 nm) were immobilized on the nitrilotriacetic acid chip. This binding capacity gave an optimal ratio between the specific signal (protein binding to loaded chip) and nonspecific binding signal (protein and detergent binding to empty chip), allowing the elimination of the latter by baseline correction (see below). As a result of the location of the His tag on the very top of the c¢ sub- unit, the immobilized c¢e complexes were oriented upside- down on the nitrilotriacetic acid surface of the chip, with the bottom part of the c¢e complex exposed to the bulk. Contaminating metal ions in the running buffer and ligand buffer can influence the binding of the ligand to the Ni 2+ -nitrilotriacetic acid surface. To increase the assay sta- bility without influencing the dissociation rate of the ligand from the surface, 50 lm of EDTA was added to all buffers [55]. Association kinetic traces were recorded when c rings in detergent containing buffer or reconstituted into 1-palmi- toyl-2-oleoylphosphatidylcholine (POPC) liposomes were passed over the loaded chip surface. In pilot SPR binding studies, c rings reconstituted into POPC liposomes and c rings solubilized in several detergents suitable for in vitro reconstitution experiments were tested. DHPC was found to cause negligible nonspecific binding to the immobilized c¢e complex and good reproducibility of the SPR binding traces, and was therefore selected for further studies. The running buffer was 20 mm Tris ⁄ HCl pH 7.0, 300 mm NaCl, 50 lm Na 2 EDTA and 1.5 mm DHPC. This composition was modified to account for specific experi- mental needs, as otherwise specified (detergent, salt or pH). D. Pogoryelov et al. Rotor interactions of the F-ATP synthase FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4859 [...]... architecture of the rotary motor in ATP synthase Science 286, 1700–1705 18 Zhang Y & Fillingame RH (1995) Subunits coupling H+ transport and ATP synthesis in the Escherichia coli ATP synthase Cys–Cys cross-linking of F1 subunit e to the polar loop of F0 subunit c J Biol Chem 270, 24609–24614 19 Hermolin J, Dmitriev OY, Zhang Y & Fillingame RH (1999) Defining the domain of binding of F1 subunit e with the polar... region of the Escherichia coli ATP synthase Tyrosine 205 of the c subunit is in the interface between the F1 and F0 parts and can interact with both the e and c oligomer J Biol Chem 271, 28341–28347 30 Peskova YB & Nakamoto RK (2000) Catalytic control and coupling efficiency of the Escherichia coli FoF1 ATP synthase: influence of the Fo sector and e subunit on the catalytic transition state Biochemistry 39,... loop of F0 subunit c in the Escherichia coli ATP synthase J Biol Chem 274, 17011–17016 20 Skakoon EN & Dunn SD (1993) Orientation of the e subunit in Escherichia coli ATP synthase Arch Biochem Biophys 302, 279–284 21 Watts SD, Zhang Y, Fillingame RH & Capaldi RA (1995) The c subunit in the Escherichia coli ATP synthase complex (ECF1F0) extends through the stalk and contacts the c subunits of the F0... [15N,1H]-TROSY J Biomol NMR 12, 345–348 25 Meier T & Dimroth P (2002) Intersubunit bridging by Na+ ions as a rationale for the unusual stability of the c-rings of Na+-translocating F1F0 ATP synthases EMBO Rep 3, 1094–1098 Rotor interactions of the F -ATP synthase 26 Meier T, Yu J, Raschle T, Henzen F, Dimroth P & Muller DJ (2005) Structural evidence for a constant ¨ c11 ring stoichiometry in the sodium F -ATP synthase. .. concentrations of bound analyte [56] The dissociation constant Kd was resolved by the equation Kd = koff ⁄ kon For the wild-type interaction of the c ring with the c¢e complex, this was independently cross-checked by analysis of equilibrium data (Scatchard plot), yielding a Kd value of the same order of magnitude Acknowledgements We thank Gregory Cook for reading the manuscript This work was supported by the. .. hairpin folding of the two helices of the ATP synthase subunit c Protein Sci 16, 2118–2122 Suzuki T, Ozaki Y, Sone N, Feniouk BA & Yoshida M (2007) The product of uncI gene in F1Fo -ATP synthase operon plays a chaperone-like role to assist c-ring assembly Proc Natl Acad Sci USA 104, 20776–20781 Fraga D & Fillingame RH (1989) Conserved polar loop region of Escherichia coli subunit c of the F1F0 H+-ATPase Glutamine... Walker JE (2000) The structure of the central stalk in bovine ˚ F1-ATPase at 2.4 A resolution Nat Struct Biol 7, 1055–1061 15 Rodgers AJ & Wilce MC (2000) Structure of the c– e complex of ATP synthase Nat Struct Biol 7, 1051– 1054 16 Wilkens S & Capaldi RA (1998) Solution structure of the e subunit of the F1-ATPase from Escherichia coli and interactions of this subunit with b subunits in the complex J... 38 Shirakihara Y, Leslie AG, Abrahams JP, Walker JE, Ueda T, Sekimoto Y, Kambara M, Saika K, Kagawa Y & Yoshida M (1997) The crystal structure of the nucleotide-free a3b3 subcomplex of F1-ATPase from the FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS 4861 Rotor interactions of the F -ATP synthase 39 40 41 42 43 44 45 46 47 48 49 50 D Pogoryelov et al thermophilic... & Dimroth P (2005) Structure of the rotor ring of F-Type Na+ATPase from Ilyobacter tartaricus Science 308, 659– 662 11 Diez M, Borsch M, Zimmermann B, Turina P, Dunn SD & Graber P (2004) Binding of the b -subunit in the ATP synthase from Escherichia coli Biochemistry 43, 1054–1064 FEBS Journal 275 (2008) 4850–4862 ª 2008 The Authors Journal compilation ª 2008 FEBS D Pogoryelov et al 12 Panke O, Cherepanov... the ATP synthase of Escherichia coli Cross-linking of the e subunit to the c subunit ring does not impair enzyme function, that of c to c subunits leads to uncoupling J Biol Chem 274, 34233–34237 Tsunoda SP, Aggeler R, Yoshida M & Capaldi RA (2001) Rotation of the c subunit oligomer in fully functional F1Fo ATP synthase Proc Natl Acad Sci USA 98, 898–902 Bulygin VV, Duncan TM & Cross RL (2004) Rotor . Probing the rotor subunit interface of the ATP synthase from Ilyobacter tartaricus Denys Pogoryelov 1,2 , Yaroslav Nikolaev 3, *,. Binding of the b -subunit in the ATP synthase from Escherichia coli. Biochemistry 43, 1054–1064. Rotor interactions of the F -ATP synthase D. Pogoryelov et