Synthesis of adenosine triphosphate (ATP) by the F1F0 ATP synthase involves a membrane‐embedded rotary engine, the F0 domain, which drives the extra‐membranous catalytic F1 domain. The F0 domain consists of subunits a1b2 and a cylindrical rotor assembled from 9–14 α‐helical hairpin‐shaped c‐subunits. According to structural analyses, rotors contain 10 c‐subunits in yeast and 14 in chloroplast ATP synthases. We determined the rotor stoichiometry of Ilyobacter tartaricus ATP synthase by atomic force microscopy and cryo‐electron microscopy, and show the cylindrical sodium‐driven rotor to comprise 11 c‐subunits.
In most forms of life, ATP synthesis is driven by a transmembrane electrochemical gradient, generated by light or oxidative reactions. The flow of protons or sodium ions down the gradient propels the smallest existing rotary motor, the membrane resident F0 part of the ATP synthase. Rotation of the 5–7 nm large cylindrical c‐oligomer in F0 is transmitted to a long rod (Gibbons et al., 2000), which induces the conformational changes required for production of ATP in the globular extra‐membranous F1 part of the enzyme. The latter exhibits a highly conserved structure comprising three catalytic sites arranged around a 3‐fold axis (Abrahams et al., 1994), which strongly suggests that three ATPs are synthesized for each full rotation of the long rod. Therefore, the proton/ATP ratio appears to be directly linked to the stoichiometry of the rotor, which is thought to rotate by 2π/n per translocated cation, where n is the number of c‐subunits in the ring. Since such central processes have been tuned to maximum performance during evolution, variation in the ring stoichiometry (n = 10 in yeast mitochondria and n = 14 in spinach chloroplasts) came as a surprise.
Much uncertainty surrounds the proton/ATP ratio. Early results suggested a value of two for mitochondria, where the situation is complicated by the transport systems for ADP, Pi and ATP. Currently four protons/ATP is the accepted overall value for mitochondria (Ferguson, 2000) as well as chloroplasts (Van Walraven et al., 1996; Pänke and Rumberg, 1997). The prediction of 12 c‐subunits in the F0 ring in Escherichia coli, based on genetic fusion of c‐subunits and cross‐linking experiments (Jones and Fillingame, 1998), is in perfect agreement with this proton/ATP ratio (12 protons would cause a 360° rotation of the c‐ring and release three ATP molecules from F1). Therefore, other stoichiometries for the rotary motor from different sources appeared to be counterintuitive. A recent discussion sheds light on the puzzle: even if the overall proton/ATP ratio is close to four, approximately three protons are available per ATP in mitochondria, because one proton is consumed during the transport of ATP into the cytosol and the transport of ADP as well as Pi back into the mitochondrial matrix (Ferguson, 2000). The values, calculated from the ring stoichiometries, are 3.3 protons/ATP in mitochondria (Stock et al., 1999) and 4.7 protons/ATP in chloroplasts (Seelert et al., 2000). Although the latter value is significantly higher than recent experimental results (Pänke and Rumberg, 1997), the suggested relative proton translocation stoichiometry of 3:4 in mitochondria versus chloroplasts (Ferguson, 2000) is quite close to the relative number of subunits per ring, 10:14. However, accurate measurements of the mitochondrial P:O ratio when electrons pass from NADH to oxygen (P:O ≥2.5) set an upper limit for the proton/ATP ratio of four, provided that this oxidative process translocates 10 protons outwards as commonly accepted (Ferguson, 2000). The most precise proton/ATP ratios currently available are based on structural investigations. These data, however, do not provide clues about the restrictions that apply for the number of c‐subunits to assemble into functional rotors in different organisms. We have, therefore, elucidated the stoichiometry of the c‐oligomer of the Na+‐ATPase of Ilyobacter tartaricus, presenting for the first time data for a prokaryotic organism and for a sodium‐dependent ATP synthase.
Protein purification and crystallization
Sodium ions propel the rotary motor of the ATP synthase from I. tartaricus by a mechanism that is similar to the proton movement and torque generation in E. coli ATP synthase (Junge et al., 1997; Elston et al., 1998; Dimroth et al., 1999). The sodium‐translocating ATP synthase of I. tartaricus is one of a few examples comprising cx‐oligomers of extreme stability (Neumann et al., 1998). This enabled us to isolate the pure cx‐oligomer, which moves on SDS–PAGE exactly as cx in the purified ATP synthase (Figure 1A, lanes 1 and 2). Such behaviour is similar to that of ring‐shaped oligomers of chloroplast subunit III (Seelert et al., 2000). After acidification, cx dissociates completely into monomers (Figure 1A, lane 3). Upon reconstitution of cx‐oligomers into lipid bilayers at a low lipid‐to‐protein ratio, rings assembled into two‐dimensional (2D) crystalline arrays. Matrix‐assisted laser desorption ionization (MALDI) mass spectrometry of such samples yields a single monomeric mass of 8790 ± 5 Da (Figure 1B), corresponding well with the calculated monoisotopic mass of 8790.71 Da for subunit c (our unpublished data). Two additional signals can be assigned to c‐dimers and double‐charged c‐monomers. No additional masses were detected (Figure 1B), documenting that subunit c is the only protein present in the crystals.
Atomic force microscopy
Atomic force microscopy (AFM) (Müller et al., 1997, 1999) reveals loosely packed disordered membranes and highly ordered crystalline arrays (Figure 2A). The thickness of the latter is 10.5 ± 0.4 nm, whereas that of the surrounding lipid bilayer is 4.1 ± 0.3 nm. Higher magnification scans show a crystalline arrangement of ring‐like cx‐oligomers, with adjacent rotors having different heights (Figure 2B). This suggests that they are oppositely oriented within the bilayer, analogous to the crystalline packing arrangement of the subunit III rings of chloroplast ATP synthase (Seelert et al., 2000). Thus, both the cytoplasmic surface and the extracellular surface of the cx‐oligomer were imaged in one topograph. Since the higher ends of the cx complexes protrude by 1.0 ± 0.2 nm above the lower ends, the AFM stylus cannot resolve the subunits of the lower rings (Figure 2B). However, densely packed, non‐crystalline patches allowed the lower ends to be contoured with the stylus at high resolution (Figure 2C). As seen directly in unprocessed AFM topographs, individual rotors with low ends have 11 subunits (Figure 2D), whereas the subunits of cx complexes exposing their higher end are not resolved (Figure 2E). Therefore, individual rotors with high or low ends were selected, aligned, classified and averaged, revealing the 11‐fold symmetry of the cx‐oligomers (Figure 2F and G). The lower end protrudes 1.5 ± 0.3 nm from the bilayer and shows a central depression with a diameter of 1.5 ± 0.3 nm (Figure 2F). In contrast, the higher end exhibits a central plug that is 0.7 ± 0.3 nm higher than the ring, which itself protrudes 2.5 ± 0.3 nm from the bilayer (Figure 2G).
Transmission cryo‐electron microscopy
Transmission cryo‐electron microscopy (cryo‐EM; Dubochet et al., 1988; Henderson et al., 1990) documents the order of the 2D crystals (Figure 3A). The non‐symmetrized projection structure (symmetry group P1) at 6.9 Å resolution (Figure 3B; Table 1) shows a pseudo‐hexagonal arrangement of ring‐like structures with 11 subunits. The rings have an outer diameter of 5.0 ± 0.2 nm and an inner diameter of 1.7 ± 0.2 nm. This outer diameter is slightly larger than that of the c10‐complexes of yeast ATP synthase, but is smaller than that measured for the rotor of the chloroplast ATP synthase comprising 14 subunits. Each subunit of the 11‐fold symmetrized projection map (Figure 3C) exhibits two densities with a slight vorticity, which is also visible in the averaged AFM topographs (Figure 2F and G), resembling the arrangement of the two α‐helices of subunit c of the yeast ATP synthase (Stock et al., 1999). The cryo‐EM projection map (Figure 3B) shows a much lower density in the interior of the rings than for the surrounding lipid bilayer. This implies that the plug observed in the AFM topographs (Figure 2D) is of a lower molecular weight than a lipid bilayer. As shown by biochemical analyses (Figure 1), no additional protein is present in the 2D crystals. This indicates that the central protrusion represents either a part of the c‐subunit, or a monolayer of lipids, consistent with the density observed in the central cavity of the yeast ATP synthase c‐oligomer (Stock et al., 1999).
The structural data presented here are from isolated cx‐oligomers that had the same molecular mass as in the complete F1F0‐ATP synthase and that remained intact during the course of the sample preparation (Figure 1A, lanes 1–3). The undecameric state of the c‐oligomer is documented by AFM and by electron crystallography; invariably, all c‐subunit rings observed exhibited the c11 stoichiometry. According to this observation and the extreme stability of the c‐subunit ring, our results likely reflect the native structure of this assembly. The novel c‐oligomer stoichiometry is the first determined for a prokaryotic organism and represents the first analysis for a sodium‐powered ATP synthase.
None of the c‐subunit assemblies studied so far by structural methods exhibits a stoichiometry that can be divided by three. A symmetry mismatch is expected to facilitate the rotary mechanism (Stock et al., 2000; see also Simpson et al., 2000). It also implicates an elastic connection between F1 and F0 ATP synthase, possibly the γ‐subunit (W. Junge, personal communication; Cherepanov et al., 1999; Feniouk et al., 1999). While for yeast mitochondria, which need one proton for ATP translocation, a value of three protons/ATP for the ATP synthase is expected from biochemical measurements (Ferguson, 2000), similar experiments gave a value of four protons/ATP in chloroplasts (Van Walraven et al., 1996; Pänke and Rumberg, 1997). This varying ratio might be correlated with different ion motive forces that drive the enzymes in the different organisms. Although slightly larger, the structurally determined values of 3.3 for yeast and 4.7 for chloroplasts are compatible with these biochemical measurements. No biochemical data are available to be compared with the newly determined stoichiometry for the I. tartaricus ATP synthase of 3.7 sodium ions/ATP. Further structural analyses at the atomic level and accurate measurements of the cation/ATP ratio are now required to improve our understanding of the smallest existing rotary motors.
Purification and 2D crystallization.
ATP synthase from I. tartaricus was purified as described (Neumann et al., 1998). cx‐oligomers were purified from disrupted ATP synthase complexes by sucrose density gradient centrifugation to a final concentration of 0.8 mg/ml. For crystallization cx‐oligomers solubilized in 2.4% β‐octyl‐glucoside were mixed with palmitoyl‐oleoyl phosphatidylcholine (POPC) at a lipid‐to‐protein ratio of 1.5 (w/w), and dialysed in a temperature‐controlled dialysis apparatus for 60 h against detergent‐free buffer (200 mM NaCl, 10 mM Tris–HCl pH 7.0).
Atomic force microscopy.
The samples were diluted to a concentration of ∼10 μg/ml in 300 mM KCl, 10 mM Tris–HCl pH 7.8 and adsorbed to freshly cleaved mica. Contact mode AFM topographs were recorded in the same buffer, at room temperature, at a stylus loading force of <100 pN and a line frequency of typically 4–6 Hz. The AFM used was a Nanoscope III (Digital Instruments, Santa Barbara, CA) equipped with a J‐scanner (scan size 120 μm) and a fluid cell. Cantilevers (Olympus, Tokyo, Japan) had oxide‐sharpened Si3N4 tips and a spring constant of 0.09 N/m. No differences between topographs recorded simultaneously in trace and in re‐trace direction were observed, indicating that the scanning process did not influence the appearance of the biological sample. Three‐dimensional surface structures of the AFM topographs are displayed in a perspective view (5° tilt).
The 2D crystals were embedded in 1% trehalose on carbon‐coated copper grids. Grids were blotted, quick‐frozen and transferred with a Gatan 626 cryo holder into a Hitachi H8000 transmission electron microscope, operated at 200 kV. Images were recorded at a nominal magnification of 50 000× on Kodak SO 163 film, applying low‐dose conditions (5 e−/Å2). Digitized negatives were processed with the MRC program suite (Henderson et al., 1990). Amplitudes and phases from two images were corrected for the contrast transfer function and merged.
We thank Wolfgang Junge, Karl‐Heinz Altendorf, Nathan Nelson and Norbert Dencher for critical reading of the manuscript. We thank Paul Jenoe for the possibility of using the mass spectrometer. The work was supported by Swiss National Foundation for Scientific Research and the M.E. Müller Foundation of Switzerland.
- Copyright © 2001 European Molecular Biology Organization