TSpace Research Repository tspace.library.utoronto.ca

Sub-nanometer resolution structure of the intact T. thermophilus H+-driven ATP

synthase

Wilson C. Y. Lau and John L. Rubinstein

Version Post-print/accepted manuscript

Citation

(published version)

Lau. W. C. Y., Rubinstein, J. Sub-nanometer resolution structure of the

intact T. thermophilus H+-driven ATP synthase. Nature 481, 214-218 (2012). Doi: 10.1038/nature10699.

How to cite TSpace items

Always cite the published version, so the author(s) will receive recognition through services that track

citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace because you cannot access the published version, then cite the TSpace version in addition to the published

version using the permanent URI (handle) found on the record page.

This article was made openly accessible by U of T Faculty.

Please tell us how this access benefits you. Your story matters.

1

Sub-nanometer resolution structure of the intact T. thermophilus H+-driven ATP synthase Wilson C. Y. Lau1,2 and John L. Rubinstein1,2,3,*

1 Molecular Structure and Function Program, The Hospital for Sick Children Departments of 2 Biochemistry and 3 Medical Biophysics, The University of Toronto

* To Whom Correspondence Should be Addressed: john.rubinstein@utoronto.ca Ion translocating rotary ATPases serve either as adenosine triphosphate (ATP) synthases, using energy from a transmembrane ion motive force to create the cell’s supply of ATP, or as transmembrane ion pumps that are powered by ATP hydrolysis 1. The members of this family of enzymes each contain two rotary motors: one that couples ion translocation to rotation and one that couples rotation to ATP synthesis or hydrolysis. During ATP synthesis, ion-translocation through the membrane-bound region of the complex causes rotation of a central rotor that drives conformational changes and ATP synthesis in the catalytic region of the complex. There are no structural models available for the intact membrane region of any ion translocating rotary ATPase. Here we present a 9.7 Å resolution map of the H+-driven ATP synthase from Thermus thermophilus obtained by electron cryomicroscopy (cryo-EM) of single particles in ice. The 600 kDa complex has an overall subunit composition of A3B3CDE2FG2IL12. The membrane-bound motor consists of a ring of L-subunits and the C-terminal region of subunit I, which are equivalent to the c- and a-subunits of most other rotary ATPases, respectively. The map shows that the ring contains twelve L-subunits 2 and that the I-subunit has eight transmembrane helices 3. The L12-ring and I-subunit have a surprisingly small contact area in the middle of the membrane, with helices from the I-subunit making contacts with two different L-subunits. The transmembrane helices of subunit I form bundles that could serve as half-channels across the membrane, with the first half-channel conducting protons from the periplasm to the L12-ring and the second half-channel conducting protons from the L12-ring to the cytoplasm. This structure therefore suggests the mechanism by which a transmembrane proton motive force is converted to rotation in rotary ATPases. In order to investigate how the structure of the T. thermophilus ATP synthase allows its rotary mechanism, we imaged the detergent solubilized intact enzyme by cryo-EM and calculated a 3D map to ~10 Å resolution (Supplemental Fig. 1). This resolution was sufficient to observe α-helices in the structure. Achieving this resolution required optimization of the specimen preparation and imaging conditions, and development of a new map refinement algorithm (Methods). Fig. 1a shows the refined 3D map. Available crystal structures of subunits and subcomplexes of the enzyme were fitted into the map with remarkably good agreement (Fig. 1b and c). A cross-section through the soluble region of the map shows that the α-helices that make up each of the E- and G-subunits of the two peripheral stalks can be resolved (Fig. 1d, purple and beige arrows, respectively), as can densities that correspond to α-helices in subunits with more complex folds, such as the A-subunits (Fig. 1d, e.g. circled in yellow). Map segments corresponding to subunits matched available crystal structures with high fidelity (Fig. 1e-g). If the protein particles imaged occupied all possible rotational states of the central rotor, the 3D map would show the average conformation of each A and B subunit and the density corresponding to the DF-subcomplex would appear as a rotational average of the structure of these

2

subunits. However, docking an A3B3DF crystal structure 4 into the map (Fig. 1b and c, Supplemental Fig. 2) and fitting the DF-rotor subcomplex into its corresponding map segment (Fig. 1g) revealed that the enzyme was arrested in a single rotational state 5. This homogeneity explains why we were able to extract nanometer-resolution information when averaging many different particles to calculate the map. The crystal structure of the A3B3DF subcomplex showed that two of the AB-subunit pairs took on two similar closed or narrow conformations, designated ANBN and AN′BN′, while the third adopted a wide-open conformation, designated AWBW 4. From the position of the D- and F-subunits in the cryo-EM map, it is apparent that the ANBN pair, which corresponds to the βTPαTP pair in mitochondrial F1-ATPase 6, is positioned at the A/B interface between the two peripheral stalks (Supplemental Fig. 2). Because we did not take any measures to stop the rotor in one rotational state, the existence of this unique position suggests that in the intact complex the observed state has a lower energy than other positions of the rotor. For the membrane-bound subunits I and L, where no crystal structures are available, segmented densities are shown in Fig. 1b and c. The extramembranous portion of the map contains a wealth of information on how the different subunits interact in the intact enzyme. The two peripheral stalks of the complex have the same structures in the intact assembly, both adopting conformations that match the available crystal structure 7. Contacts between the catalytic A3B3 hexamer and the peripheral stalks exclusively involve the B-subunits and E-subunits of the stalks, and are mostly between the N-terminal β-barrel domains of the B-subunits and the C-terminal α/β-domains of the E-subunits. The N-terminal region of both subunits E and G in both of the peripheral stalks interact with the N-terminal region of subunit I. The funnel-shaped C subunit, which links the DF portion of the central rotor to the L12-ring and has been crystallized in isolation 8, is significantly more open in the intact enzyme than in the crystal structure so that it can accept the DF subunits into its central cavity (Fig. 2). The three pseudo-symmetric domains of the C-subunit can be differentiated in the map showing that interaction of the C-subunit and the L12-ring strongly involves the N-terminal α-helix of the first domain of subunit C. The C-subunit sits asymmetrically on the L12-ring and does not penetrate significantly into the central pore of the ring. The cryo-EM map shows that the N-terminal region of subunit I consists of an elongated helical bundle flanked by two domains, consistent with a crystal structure of a homologous protein that was published after submission of this manuscript 9 (Supplemental Fig. 3). The map provides the first detailed insight into how the I-subunit and L12-ring fit together to allow the generation of rotation (Fig. 3a). Cross sections through the detergent-embedded region of the map show two concentric rings of densities, with the outer ring consisting of twelve well-resolved densities (Fig. 3b-e). These twelve densities undoubtedly correspond to the outer helices of the L12-ring, where each L-subunit consists of a helical hairpin. The N-terminal inner helices of the L12-ring could not be resolved and appear as a nearly continuous density. Our ability to resolve the outer helices is consistent with the loose packing and gaps seen between the outer helices in crystral structures of related ring subcomplexes 10-15. Although composed of similarly sized residues as the outer helices, the crystallographic studies suggest that the inner

3

helices are more tightly packed and resolving them would require a resolution better than 9.7 Å. As shown previously 5, the detergent used to keep the enzyme soluble, dodecylmaltoside, has a density higher than that of ice and is visible surrounding the transmembrane region of the complex (Fig. 3b-i, white bars). Features in the detergent micelle, which we expect to be mostly unstructured, may represent noise that is enhanced by construction of the map to nanometer resolution. Density is visible within the center of the L12-ring (Figure 3b-i, yellow circle) and probably corresponds to a detergent or lipid plug observed by atomic force microscopy of isolated rings 16. The map presents the first determination of the ring structure within an intact rotary ATPase, providing insight into how the I-subunit affects the L12-ring. This information is necessary to assess the likelihood of proposed catalytic models in which the L-subunits undergo conformational changes when they contact the I-subunit 17. From the cryo-EM map it is evident that the twelve-fold symmetry of the L12-ring is not broken significantly, even where the L-subunits contact the I-subunit (Fig. 3c and d). This preserved symmetry of the ring is inconsistent with models of proton translocation that require major conformational changes in the rotor, but could still be consistent with swiveling of the outer helices of the L-subunits if this motion did not distort the ring significantly 17. Proton translocation facilitated by the I-subunit drives rotation of the L12-ring, but until now there was no structural information available for subunit I or its equivalent subunit from any rotary ATPase. Here we observe eight transmembrane densities that can be attributed to α-helices in the C-terminal region of subunit I (Fig 3b-e). This number of transmembrane helices is consistent with an experimentally tested topology map of the S. cerevisiae V-ATPase a-subunit 3, with which the I-subunit sequence aligns well (Supplemental Fig. 4). Although we can trace the complete trajectory of the eight transmembrane densities of subunit I, our inability to resolve the helices in the tightly packed inner ring of the L12-oligomer means that we cannot rule out the possibility that we have missed a transmembrane helix in our analysis of subunit I. At the cytoplasmic surface of the membrane region, the densities from the N- and C-terminal regions of subunit I appear to be connected in more than one place (Fig. 3a, b-ii, blue arrows). These connections show that there must be protein-protein interactions between the N- and C-terminal regions of subunit I involved in keeping the two regions rigidly attached. Within the membrane, subunit I divides into two clusters of helices: one that is mostly perpendicular to the membrane, and another that contains tilted helices adjacent to the L12-ring. The first cluster contacts a single L-subunit near the middle of the membrane region (Fig. 3c-ii, circled in red) while the second cluster contacts the adjacent L-subunit a small distance further towards the periplasm (Fig. 3d-ii, circled in blue). The maximal separation of the I-subunit helices from the outer helices of the L12-ring is of around the same distance, ~ 15 Å between helical-axes, as is the maximal separation between adjacent outer helices of the L12-ring. Proton translocation in rotary ATPases is thought to involve protonation and deprotonation of conserved mid-membrane glutamic acid or aspartic acid residues in the outer helices of the ring-forming subunits. In the L-subunit from the T. thermophilus ATP synthase, this conserved protonatable residue is Glu63. The contact of the I-subunit with two different L-subunits (Fig. 3c and d) places the two L-subunits in distinct

4

chemical environments and establishes the conditions necessary for a two half-channel model for proton translocation 13,18, with one L-subunit exchanging protons with the periplasm and another L-subunit exchanging protons with the cytoplasm. One cluster of transmembrane helices in the I-subunit could conduct protons from the periplasm to the mid-membrane Glu63 residue of an L-subunit (Fig. 4a, blue circle). The other cluster of transmembrane helices in subunit I could conduct protons from the mid-membrane Glu63 residue of the adjacent L-subunit to the cytoplasm (Fig. 4a, red circle). Crystal structures of other membrane proteins indicate that four transmembrane helices are sufficient for forming a proton pore, with available examples showing proton 19 and sodium 20 conducting pores composed of four and three transmembrane helices, respectively. The structure suggests that protons flow from the periplasmic half-channel to the Glu63 residue of the L-subunit in contact with the periplasmic half-channel (the L-subunit with an outer helix labeled “1” in Fig. 3c-e). Protonation of this Glu63 neutralizes the negative charge of the residue and allows the clockwise rotation (viewed from the cytoplasm) that places the Glu63 in the hydrophobic environment of the lipid bilayer (Fig. 4b). With the negative charge neutralized, Glu63 can assume the proton-locked conformation seen in a crystal structure of the Spirulina platensis c15-ring 13. In this conformation, both oxygen atoms in Glu63 are involved in hydrogen bonding with other residues in the L12-ring and the neutral Glu63 residue is tucked into the crevice between L-subunits. The directionality of the rotation is explained by a Brownian ratchet mechanism where random thermal rotational fluctuations are biased to go in the correct direction by the direction of the proton motive force across the membrane 18. Rotation of the L12-ring brings an L-subunit bearing a protonated Glu63 (on the helix labeled “2” in Fig. 3c-e) out of the lipid environment and into contact with the cytoplasmic half-channel of subunit I. Subunit I contains a conserved and essential arginine residue (Arg563) of unknown function 21,22. It has been postulated that Arg563 causes a decrease in the pKa of the Glu63 residue on the L-subunit in contact with the cytoplasmic half-channel, thereby causing it to lose its proton to the channel 23. Arg563 may also stabilize the deprotonated Glu63 by forming a salt bridge 12. This arrangement of half-channels is consistent with a model where L-ring rotation could be driven in either direction, depending only on the direction of the proton motive force across the membrane. However, the specific arrangements of amino acids in the structure, which cannot be resolved in this map, could be such that proton translocation from the cytoplasm to the periplasm is not able to induce rotation. At 9.7 Å resolution, it is not possible to unambiguously relate the sequence of the C-terminal region of subunit I and predicted transmembrane helices (shown in Supplemental Fig. 4) to the transmembrane densities in the map. Nonetheless, Arg563 has been proposed to reside on transmembrane helix 7 and form part of the cytoplasmic half-channel 3, which would identify transmembrane helix 7 as one of the two helices in contact with the L-subunit labeled “2” in Fig. 3. At both the cytoplasmic and periplasmic limits of the membrane-embedded region of the map (Fig 3b and e) all helices from subunit I are well separated from the helices of the L12-ring, making it highly unlikely that an entire half-channel could be formed by the interface of helices from the I- and L-subunits 24. F-type ATP synthases have been proposed to have five transmembrane helices 25,26. If true, this topology would suggest that in those enzymes the two half-channels consist of three helices, that one or more

5

helices contributes to both half-channels simultaneously, or that helices from the rotor are involved in formation of the half-channels. The observation that the two L-subunits in contact with the I-subunit are immediately adjacent to each other indicates that after a Glu63 residue is deprotonated by the cytoplasmic half-channel it is immediately reprotonated by the periplasmic half-channel with a single 30° rotational step of the rotor. The small contact area between the I-subunit and L12-ring suggests that these membrane-bound components rely on the two peripheral stalks to hold them together in the precise arrangement necessary for the complex’s biological activity. This minimal contact is consistent with an important but fragile interaction that might easily be disrupted by non-physiological conditions such as those needed to make 3D crystals, thus helping to explain why the membrane region of this class of enzyme has not been crystallized, despite significant efforts by many research groups. Knowledge of the precise residues involved in proton translocation will require construction of cryo-EM maps to higher resolution or formation of well-ordered crystals of this hitherto refractory protein complex. At subnanometer resolution, the complete structure of the T. thermophilus H+-driven ATP synthase suggests the mechanism by which the energy stored in a transmembrane proton motive force is converted into rotation in rotary ATPases. Methods Summary Thermus thermophilus HB8 was grown and the H+-driven ATP synthase purified as described previously 5. Specimens were prepared for cryo-EM with a Vitrobot (FEI) and imaged with a FEI Tecnai F20 electron microscope operating at 200 kV and equipped with a field emission gun. An electron exposure of ~18 electrons/Å2 was used during imaging to optimize the signal-to-noise ratio at relevant spatial frequencies 27. Images were recorded on Kodak SO-163 film and digitized with a Photoscan densitometer (Intergraph). The previously published cryo-EM map of the T. thermophilus V-ATPase 5 was filtered to 30 Å resolution and used as an initial model for refinement of the entirely new data set of 46,105 particle images. Initial particle orientations were determined by projection matching with Frealign 28 using information out to 20 Å resolution. The accuracy of particle orientations was refined further with a new program (Refine_fspace), ultimately using information out to 11.2 Å resolution. Acknowledgements: We thank Drs. Voula Kanelis, Frank Sicheri, Peter Rosenthal, Lewis Kay, and Richard Henderson for discussions and critical reading of this manuscript and Drs. John Walker and Régis Pomès for discussions. Computations were performed on the general purpose cluster supercomputer at the SciNet HPC Consortium. WCYL was supported by an Ontario Graduate Scholarship. JLR was supported by a New Investigator Award from the CIHR and an Early Researcher Award from the Ontario Ministry of Research and Innovation. This research was funded by operating grant MOP 81294 from the CIHR. Author Contributions JLR and WCYL designed the experiments and JLR supervised the research. WCYL performed protein purification and cryo-EM. JLR wrote new computer programs.

6

WCYL and JLR performed the image analysis, interpreted the data, and wrote the manuscript. The cryo-EM map of the T. thermophilus H+-driven ATP synthase has been deposited in the Electron Microscopy Data Bank with accession number EMD-5335 and the docked atomic models in the PDB with PDB ID 3J0J. The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to john.rubinstein@utoronto.ca References: 1 Muench, S. P., Trinick, J., and Harrison, M. A., Structural divergence of the

rotary ATPases. Q Rev Biophys 44, 311 (2011). 2 Toei, M. et al., Dodecamer rotor ring defines H+/ATP ratio for ATP synthesis of

prokaryotic V-ATPase from Thermus thermophilus. Proc Natl Acad Sci U S A 104, 20256 (2007).

3 Toei, M., Toei, S., and Forgac, M., Definition of membrane topology and identification of residues important for transport in subunit a of the vacuolar ATPase. J Biol Chem 286, 35176 (2011).

4 Numoto, N., Hasegawa, Y., Takeda, K., and Miki, K., Inter-subunit interaction and quaternary rearrangement defined by the central stalk of prokaryotic V1-ATPase. EMBO Rep 10, 1228 (2009).

5 Lau, W. C. and Rubinstein, J. L., Structure of intact Thermus thermophilus V-ATPase by cryo-EM reveals organization of the membrane-bound V(O) motor. Proc Natl Acad Sci U S A 107, 1367 (2010).

6 Abrahams, J. P., Leslie, A. G., Lutter, R., and Walker, J. E., Structure at 2.8 A resolution of F1-ATPase from bovine heart mitochondria. Nature 370, 621 (1994).

7 Lee, L. K. et al., The structure of the peripheral stalk of Thermus thermophilus H+-ATPase/synthase. Nat Struct Mol Biol 17, 373 (2010).

8 Iwata, M. et al., Crystal structure of a central stalk subunit C and reversible association/dissociation of vacuole-type ATPase. Proc Natl Acad Sci U S A 101, 59 (2004).

9 Srinivasan, S., Vyas, N. K., Baker, M. L., and Quiocho, F. A., Crystal structure of the cytoplasmic N-terminal domain of subunit I, a homolog of subunit a, of V-ATPase. J Mol Biol 412, 14 (2011).

10 Murata, T. et al., Structure of the rotor of the V-Type Na+-ATPase from Enterococcus hirae. Science 308, 654 (2005).

11 Stock, D., Leslie, A. G., and Walker, J. E., Molecular architecture of the rotary motor in ATP synthase. Science 286, 1700 (1999).

12 Meier, T. et al., Structure of the rotor ring of F-Type Na+-ATPase from Ilyobacter tartaricus. Science 308, 659 (2005).

7

13 Pogoryelov, D., Yildiz, O., Faraldo-Gomez, J. D., and Meier, T., High-resolution structure of the rotor ring of a proton-dependent ATP synthase. Nat Struct Mol Biol 16, 1068 (2009).

14 Watt, I. N. et al., Bioenergetic cost of making an adenosine triphosphate molecule in animal mitochondria. Proc Natl Acad Sci U S A 107, 16823 (2010).

15 Preiss, L. et al., A new type of proton coordination in an F(1)F(o)-ATP synthase rotor ring. PLoS Biol 8, e1000443 (2010).

16 Meier, T. et al., The central plug in the reconstituted undecameric c cylinder of a bacterial ATP synthase consists of phospholipids. FEBS Lett 505, 353 (2001).

17 Fillingame, R. H., Angevine, C. M., and Dmitriev, O. Y., Mechanics of coupling proton movements to c-ring rotation in ATP synthase. FEBS Lett 555, 29 (2003).

18 Junge, W., Lill, H., and Engelbrecht, S., ATP synthase: an electrochemical transducer with rotatory mechanics. Trends Biochem Sci 22, 420 (1997).

19 Stouffer, A. L. et al., Structural basis for the function and inhibition of an influenza virus proton channel. Nature 451, 596 (2008).

20 Gonzales, E. B., Kawate, T., and Gouaux, E., Pore architecture and ion sites in acid-sensing ion channels and P2X receptors. Nature 460, 599 (2009).

21 Cain, B. D. and Simoni, R. D., Proton translocation by the F1F0ATPase of Escherichia coli. Mutagenic analysis of the a subunit. J Biol Chem 264, 3292 (1989).

22 Kawasaki-Nishi, S., Nishi, T., and Forgac, M., Arg-735 of the 100-kDa subunit a of the yeast V-ATPase is essential for proton translocation. Proc Natl Acad Sci U S A 98, 12397 (2001).

23 Pogoryelov, D. et al., Microscopic rotary mechanism of ion translocation in the F(o) complex of ATP synthases. Nat Chem Biol 6, 891 (2010).

24 Steed, P. R. and Fillingame, R. H., Aqueous accessibility to the transmembrane regions of subunit c of the Escherichia coli F1F0 ATP synthase. J Biol Chem 284, 23243 (2009).

25 Long, J. C., Wang, S., and Vik, S. B., Membrane topology of subunit a of the F1F0 ATP synthase as determined by labeling of unique cysteine residues. J Biol Chem 273, 16235 (1998).

26 Valiyaveetil, F. I. and Fillingame, R. H., Transmembrane topography of subunit a in the Escherichia coli F1F0 ATP synthase. J Biol Chem 273, 16241 (1998).

27 Baker, L. A., Smith, E. A., Bueler, S. A., and Rubinstein, J. L., The resolution dependence of optimal exposures in liquid nitrogen temperature electron cryomicroscopy of catalase crystals. J Struct Biol 169, 431 (2010).

Methods (detailed) Specimen preparation and imaging Specimens were prepared on glow-discharged Quantifoil R2/2 carbon coated TEM grids (Quantifoil Microtools GmbH). Images were recorded with defocus values between 2.5 and 4.5 µm at 50,000 x magnification and film was developed in D19 for 8 min. Images were screened and only those that showed high contrast from thin ice layers, no noticeable drift, and oscillations of the contrast transfer function (CTF) beyond 10 Å

8

resolution were selected for further analysis. From these micrographs, CTF parameters were determined 29 and particle images were selected interactively with Ximdisp 30. 3D map construction and segmenting The new dataset, obtained with a lower electron exposure from specimens in a thinner ice layer than used previously 5, showed CTF oscillations to a higher resolution than the original dataset. The new program used to refine particle orientation parameters, Refine_fspace, performs projection matching in Fourier space while allowing continuous constrained optimization of the Euler angles and shifts with a simplex minimization algorithm 31,32. Prior to projection matching with a normalized correlation coefficient, image Fourier transforms were multiplied by the CTF and the map projections by the square of the CTF. To avoid influencing the measurement of resolution with noise bias, the highest spatial frequency used during refinement was kept below the resolution limit of the map. From the data set of particle images, the top ~90% (42075) with the best correlation coefficients at their determined orientations were selected and the final 3D map was calculated by sinc function interpolation in Fourier space 28. The resolution of the final map was assessed by Fourier shell correlation (FSC) and Fourier neighbor correlation 33 with the 0.143 34 and 0.5 thresholds. Fourier components were sharpened with an inverse B-factor of 750 Å2 and weighted for the signal-to-noise ratio with a Cref filter 35. Segmentation was performed automatically using Segger 36, semiautomatically using Wateredge 37, and manually using EMAN qsegment 38. Model building and fitting The PDB-IDs for atomic models used to interpret the 3D cryo-EM map were 3A5C (A3B3DF complex 4), 3K5B (EG complex 7), 1R5Z (subunit C 8), and 3RRK (the N-terminal domain of subunit I from M. ruber 9). A comparative model of the L-subunit in its proton-locked conformation was built with Phyre2 39 using the two C-terminal transmembrane helices of the NtpK subunit of the sodium-driven V-ATPase from Enterococcus hirae (PDI-ID 2BL2 10) as the template. This template was identified automatically by Phyre2 with 99.9% confidence and represents 90% coverage of subunit L. An atomic model of the L12-ring was constructed in Situs 40. Rigid-body fitting of atomic models into the cryo-EM map was done with UCSF Chimera 41 and flexible fitting of subunit C into the map was performed with IMODFIT (http://chaconlab.org/imodfit/index.html). All figures were rendered with UCSF Chimera 41. Methods References 28 Grigorieff, N., FREALIGN: high-resolution refinement of single particle

structures. J Struct Biol 157, 117 (2007). 29 Mindell, J. A. and Grigorieff, N., Accurate determination of local defocus and

specimen tilt in electron microscopy. J Struct Biol 142, 334 (2003).

9

30 Crowther, R. A., Henderson, R., and Smith, J. M., MRC image processing programs. J Struct Biol 116, 9 (1996).

31 Nelder, J. A. and Mead, R., A simplex method for function minimization. Comput J 7, 308 (1965).

32 Press, W. H., Teukolsky, S. A., Vetterlin, W. T., and Flannery, B. P., Numerical recipes in Fortran 77, 2nd ed. (Cambridge University Press, Cambridge, 2003).

33 Sousa, D. and Grigorieff, N., Ab initio resolution measurement for single particle structures. J Struct Biol 157, 201 (2007).

34 Rosenthal, P. B., Crowther, R. A., and Henderson, R., An Objective Criterion for Resolution Assessment in Single-particle Electron Microscopy (appendix). J Mol Biol 333, 743 (2003).

35 Rosenthal, P. B. and Henderson, R., Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. J Mol Biol 333, 721 (2003).

36 Pintilie, G. D. et al., Quantitative analysis of cryo-EM density map segmentation by watershed and scale-space filtering, and fitting of structures by alignment to regions. J Struct Biol 170, 427 (2010).

37 Baker, L. A. and Rubinstein, J. L., Edged Watershed Segmentation: A semi-interactive algorithm for segmentation of low-resolution maps from electron cryomicroscopy. J Struct Biol 176, 127 (2011).

38 Ludtke, S. J., Baldwin, P. R., and Chiu, W., EMAN: semiautomated software for high-resolution single-particle reconstructions. J Struct Biol 128, 82 (1999).

39 Kelley, L. A. and Sternberg, M. J., Protein structure prediction on the Web: a case study using the Phyre server. Nat Protoc 4, 363 (2009).

40 Chacon, P. and Wriggers, W., Multi-resolution contour-based fitting of macromolecular structures. J Mol Biol 317, 375 (2002).

41 Goddard, T. D., Huang, C. C., and Ferrin, T. E., Visualizing density maps with UCSF Chimera. J Struct Biol 157, 281 (2007).

Figure captions: Figure 1. 3D map of the T. thermophilus ATP synthase. (a), A surface view of the 3D map. (b) and (c), The overall map (semi-transparent grey) with fitted crystal structures and segments corresponding to individual subunits. Segments of the cryo-EM map are shown for the L12-ring, subunit I, and residues of subunit D missing from its crystal structure. (d), A cross section through the soluble region of the map shows that α-helices from the two E-subunits (purple arrows) and two G-subunits (beige arrows) can be resolved. An A-subunit density (circled in yellow) shows that α-helices can be resolved. The map segments agree with crystal structures, such as (e) subunit A, (f) the EG-subcomplex, and (g) the DF-subcomplex. Density corresponding to missing residues from the crystal structure of the D-subunit is indicated with blue arrows. (Scale bars, 25 Å) Figure 2. Fitting of the C-subunit crystal structure. (a), Side and (b), top views comparing rigid body fitting of the C-subunit crystal structure (red) to flexible fitting of

10

the crystal structure (cyan) into the corresponding map segment. Arrows indicate regions of major difference between the two atomic models. (c), Side and (d), top views showing that flexible fitting of the C-subunit gives it a more open conformation that allows it to accommodate the segment for the DF-subcomplex (dark blue) into its central cavity. (e), A side view of the C-subunit, and the segment corresponding to the L12-ring (magenta), show that subunit C sits asymmetrically on the ring with its N-terminal α-helix (indicated by a cyan arrow) mediating most of the contact with the L-subunits. (Scale bar, 25 Å) Figure 3. The membrane-bound region of the enzyme. (a), Map segments of the L12-ring (magenta) and subunit I (green) showing multiple contacts between the N- and C-terminal regions of subunit I (blue arrows). (b), Cross sections through the map (i) and map segments truncated at the same height (ii) show subunit I separated from the L12-ring near the cytoplasm. Outer helices of the L12-ring are indicated (red arrows) and the transmembrane helices of subunit I are outlined (green). Cross sections show the detergent micelle (white bars) and detergent or lipid in the centre of the L12-ring (yellow circle). (c), Near the middle of the membrane, subunit I contacts an L-subunit, likely forming the mid-membrane end of the cytoplasmic half channel (circle in red in ii). (d), ~6 Å further towards the periplasm subunit I contacts a different L-subunit, likely forming the mid-membrane end of the periplasmic half-channel (circled in blue in ii). (e), Subunit I is separated from the L12-ring near the periplasm. (Scale bars, 25 Å) Figure 4. Model for proton translocation. (a), Subunit I (green) is shown parallel to the membrane and the L12-ring is indicated by a semi-transparent magenta rectangle. During ATP synthesis, protons enter the periplasmic half-channel in subunit I (dashed blue oval) and are conducted to the center of the lipid bilayer where they neutralize the Glu63 residues of an L-subunit. The ring rotates to bring a protonated Glu63 residue into contact with the cytoplasmic half-channel (dashed red oval). (b), Viewed from the cytoplasm, two half-channels in subunit I are depicted as clusters of green cylinders. Deprotonated Glu63 residues are shown with two red oxygen atoms while protonated Glu63 residues are shown with one red and one yellow oxygen atom. Protonation of L-subunit “1” by the periplasmic half-channel (i) and deprotonation of L-subunit “2” by the cytoplasmic half-channel (ii) leads to rotation of the ring, with L-subunit “1” assuming a proton-locked conformation as it enters the lipid bilayer. Rotation brings the proton-locked L-subunit “3” to the cytoplasmic half-channel (iii) where it assumes an unlocked conformation, allowing the sequence of events to repeat.

d

e

f ga b c

90°

I

L-ring

EG

C

DF

AB

A

E

GD

F

a

b

c

d

e

12

3

4

5

6 78

9

10

1112

12

4

56 7

8

9

10

1112

3

12

3

4

5

6 78

9

10

1112

N-term. region

C-term.region

Cytoplasm

Periplasm

a b (ii)(ii)

(ii) (ii)

c

e d

(i) (i)

(i)(i)

To cytoplasm+

-12

3

4

12

11

10

9

876

5

a

periplasm

cytoplasm

+H

+H

periplasmichalf-channel

cytoplasmichalf-channel

+

12

3

4

12

11

10

9

876

5

-

bi

ii

From periplasm

1

23

412

11

10

987

6

5

-

iii

1

Supplemental figure captions Supplemental Figure 1. Resolution assessment. The Fourier shell correlation curve (solid line) indicates a resolution of 9.7 Å with the 0.143 criterion 1 and 11 Å with the 0.5 criterion. The accuracy of this resolution assessment is supported by an independent measurement with Fourier neighbor correlation 2 (dashed line), which indicates a resolution of 10.3 Å with the 0.143 criterion and 11.7 Å with the 0.5 criterion. All of the features described in this manuscript were apparent at all of these resolutions, but most distinctly so at 9.7 Å resolution. Therefore, the map was filtered to 9.7 Å before rendering figures. Supplemental Figure 2. Fitting of an A3B3DF crystal structure 3 into the cryo-EM map reveals the orientations of the three different types of catalytic A/B subunit interfaces. (a) A view of the catalytic region of the map with the fitted crystal structure showing subunits A (yellow), B (red), D (blue), F (orange), E (purple), and G (beige). The dashed line indicates the position of the thin section in part b. (b) A thin section through the map with the fitted crystal structure shows the positions of the AB-pairs designated in the crystal structure as Wide (AWBW), Narrow (ANBN), and Narrow′ (AN′BN′). As indicated in parentheses, these conformations correspond with the βEmptyαEmpty , βTPαTP, and βDPαDP conformations seen in crystal structures of the bovine F1-ATPase, respectively. The scale bar corresponds to 25 Å. Supplemental Figure 3. The map segment corresponding to the N-terminal domain of subunit I showed a remarkable similarity to the crystal structure of the Saccharomyces cerevisiae subunit C, which is not found in the T. thermophilus enzyme. Following submission of this manuscript, a crystal structure of the N-terminal domain of subunit I from the eubacterium Meiothermus ruber confirmed this similarity 4. A homology model of the T. thermophilus N-terminal domain of subunit I was constructed with Phyer2 5 based on the M. ruber crystal structure. As viewed along the membrane surface (a) and from the catalytic region towards the membrane (b) the homology model (green ribbon diagram) fit into the map segment (semitransparent grey surface) with excellent agreement. The 43 C-terminal residues were disordered in the crystal structure, contributing to the unoccupied cryo-EM density in the domain on the left side of the structure. The scale bar corresponds to 25 Å. Supplemental Figure 4. Sequence alignment between T. thermophilus ATP synthase subunit I, and S. cerevisiae V-type ATPase subunit a. The transmembrane helices identified in Ref. 6 are outlined in black. The soluble N-terminal region of the subunits are shown in blue while the membrane-bound C-terminal region is shown in black. The critical conserved arginine residue (Arg563) is indicated with a red arrow. Symbols indicate residues that are identical (*), strongly similar (:), or weakly similar (.). Supplemental movie 1. The 3-D map with docked crystals structures of subunits A (yellow), B (red), C (cyan), D (blue), E (purple), F (orange), and G (beige) subunits and segments for density from subunits I (green), L (magenta), and the missing density from subunit D (blue). The scale bar corresponds to 25 Å.

2

References 1 Rosenthal, P. B., Crowther, R. A., and Henderson, R., An Objective Criterion for

Resolution Assessment in Single-particle Electron Microscopy (appendix). J Mol Biol 333, 743 (2003).

2 Sousa, D. and Grigorieff, N., Ab initio resolution measurement for single particle structures. J Struct Biol 157, 201 (2007).

3 Numoto, N., Hasegawa, Y., Takeda, K., and Miki, K., Inter-subunit interaction and quaternary rearrangement defined by the central stalk of prokaryotic V1-ATPase. EMBO Rep 10, 1228 (2009).

4 Srinivasan, S., Vyas, N. K., Baker, M. L., and Quiocho, F. A., Crystal structure of the cytoplasmic N-terminal domain of subunit I, a homolog of subunit a, of V-ATPase. J Mol Biol 412, 14 (2011).

5 Kelley, L. A. and Sternberg, M. J., Protein structure prediction on the Web: a case study using the Phyre server. Nat Protoc 4, 363 (2009).

6 Toei, M., Toei, S., and Forgac, M., Definition of membrane topology and identification of residues important for transport in subunit a of the vacuolar ATPase. J Biol Chem 286, 35176 (2011).

Resolution (Å)

0.0

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1Fou

rier

She

ll C

orre

latio

n

59.7 9.0

44.8

35.8

29.9

25.6

22.4

19.9

17.9

16.3

14.9

13.8

12.8

12.0

10.0 8.5

8.2

11.2

10.5 7.5

6.2

5.6

6.9

7.2

6.6

6.4

9.4

7.8

5.8

6.0

a bAN´BN´(βDPαDP)

AWBW(βEαE)

ANBN(βTPαTP)

a

b

Thermus VatI 1 MIAPMEKL---------VLAGPKGRAKELLQSLQQAGVVHLETLRPEALSAYQLSPEERAELRRWE 57 Sacch. Vph1 1 MAEKEEAIFRSAEMALVQFYIPQEISRDSAYTLGQLGLVQFRDLNSK-VRAFQRTF--VNEIRRLD 63 * * : : *: ::: :* * *:*::. *..: : *:* : *:** : Thermus VatI 58 AVSAGAEHTLSLLGLEAE-----PARPFPEGLEAAEKALSPIQAHAEGLTRQKQELEEELALAQAY 118 Sacch. Vph1 64 NVERQYRYFYSLLKKHDIKLYEGDTDKYLDG--SGELYVPPSGSVIDDYVRNASYLEERLIQMEDA 127 *. .: *** . : : :* :.* :.* : :. .*: . ***.* : Thermus VatI 119 LEPLERLAALAHGLDKSPFLRVIPFLLT-EKELPLVEE---ALRKALEDRYLLAHEAYAGGVAAL- 179 Sacch. Vph1 128 TDQIE---VQKNDLEQYRFI------LQSGDEFFLKGDNTDSTSYMDEDMIDANGENIAAAIGASV 184 : :* . :.*:: *: * .*: * : : ** * *..:.* Thermus VatI 180 ----VVVHRKEVDQAKAALSRAGVAELRLPGALGELPLSEAAR--------------------RLK 221 Sacch. Vph1 185 NYVTGVIARDKVATLEQILWRVLRGNLFFKTVEIEQPVYDVKTREYKHKNAFIVFSHGDLIIKRIR 250 *: *.:* : * *. .:* : . * *: :. *:: Thermus VatI 222 E--------------RAEAAPRELSEVRQHLAKLA---RESASTLQS-LWTRAQ------DEVARL 263 Sacch. Vph1 251 KIAESLDANLYDVDSSNEGRSQQLAKVNKNLSDLYTVLKTTSTTLESELYAIAKELDSWFQDVTRE 316 : *. .::*::*.::*:.* : :::**:* *:: *: ::*:* Thermus VatI 264 KALEELA------SGRFGFALLGYVPVK----AKPKVEEALARHKESVVYAFEPVDEHHEADRIPV 319 Sacch. Vph1 317 KAIFEILNKSNYDTNRKILIAEGWIPRDELATLQARLGEMIARLGIDVPSIIQVLDTNHTP---PT 379 **: *: :.* : *::* . :.:: * :** .* :: :* :* . *. Thermus VatI 320 VLDNPPWAKPFELLVSFLNTPKYGTFDPTPVVPVFFPFWFGMIVGDIGYALLFYLVGRWLSGY-VK 384 Sacch. Vph1 380 FHRTNKFTAGFQSICDCYGIAQYREINAGLPTIVTFPFMFAIMFGDMGHGFLMTLAALSLVLNEKK 445 . . :: *: : . . .:* ::. . * *** *.::.**:*:.:*: *.. * * Thermus VatI 385 RNEPLVIDLFALKLKPQVIGKLVHILNWMVFWTVVWGVIYGEFFGTFLEHL-GVFGTPEH------ 443 Sacch. Vph1 446 INKMKRGEIFDMAFTGR------YIILLMGVFSMYTGFLYNDIFSKTMTIFKSGWKWPDHWKKGES 505 *: ::* : :. : :*: * .::: *.:*.::*.. : : . : *:* Thermus VatI 444 -----PGLIPILIHR----ID------TAKTANLLILLSVAFGVVLVFFGLALRAYLGLKHRHMAH 494 Sacch. Vph1 506 ITATSVGTYPIGLDWAWHGTENALLFSNSYKMKLSILMGFIHMTYSYFFSLANHLYFNSMIDIIGN 571 * ** :. : .: . :* **:.. . . **.** : *:. :.: Thermus VatI 495 FWEGVGYLGGLVGVLALAASYLG--------NLQAGWLQG-------------LMYL---GFGVFL 536 Sacch. Vph1 572 FIPGLLFMQGIFGYLSVCIVYKWAVDWVKDGKPAPGLLNMLINMFLSPGTIDDELYPHQAKVQVFL 637 * *: :: *:.* *::. * .* *: :* . *** Thermus VatI 537 -LAVLMSRIWLMIPE--------------------------------------------------- 550 Sacch. Vph1 638 LLMALVCIPWLLLVKPLHFKFTHKKKSHEPLPSTEADASSEDLEAQQLISAMDADDAEEEEVGSGS 703 * .*:. **:: : Thermus VatI 551 -------------------IFTQAGHILSHIRIYAVGAAGGILAGLLTDVGFALAER-LGLLGVLL 596 Sacch. Vph1 704 HGEDFGDIMIHQVIHTIEFCLNCVSHTASYLRLWALSLAHAQLSSVLWTMTIQIAFGFRGFVGVFM 769 :. ..* *::*::*:. * . *:.:* : : :* *::**:: Thermus VatI 597 GLLV-----AGVLHLLILLLTTLGHMLQPIRLLWVEFFTKFGFYEENGRPYRPFK----------- 646 Sacch. Vph1 770 TVALFAMWFALTCAV-LVLMEGTSAMLHSLRLHWVESMSKFF--VGEGLPYEPFAFEYKDMEVAVA 832 : : * . : ::*: . **:.:** *** ::** :* **.** Thermus VatI 647 --SVREAQ 652 Sacch. Vph1 833 SASSSASS 840 * :.

Subject: Re: the revised version of your Nature manuscript Date: Thu, 27 Oct 2011 21:38:50 -0400 From: j.finkelstein@us.nature.com To: john.rubinstein@utoronto.ca Dear Professor Rubinstein, Thank you for submitting the revised version of your manuscript entitled "Sub-nanometer resolution structure of the intact T. thermophilus H+-driven ATP synthase." I am pleased to say that the work is nearly ready to formally accept. However, it looks like there was a file-conversion problem for a few of the files that were uploaded - I am sorry for the inconvenience, but could you email the manuscript checklist, the AOP form, the license to publish form, and the supplementary movie to jmfinkelstein@gmail.com Thanks in advance for your assistance with this matter. Best wishes, Joshua Joshua Finkelstein, Ph.D. Senior Editor Nature Subject: Re: Decision on Nature manuscript 2011-07-08889A Date: Fri, 30 Sep 2011 14:13:57 -0400 From: j.finkelstein@us.nature.com To: john.rubinstein@utoronto.ca Dear Professor Rubinstein, The revised version of your manuscript entitled "Sub-nanometer resolution structure of the intact T. thermophilus H+-driven ATP synthase reveals the arrangement of its trans-membrane helices" has been seen by two of the original referees, whose comments appear at the end of this letter. In light of their advice, I am delighted to say that we will be able to publish your manuscript as a Letter. First, however, we would like you to revise your paper to make sure that the manuscript is as brief as possible and in Nature format. Please carefully read through this email, as there are several changes you will need to make before we can formally accept your manuscript. [formatting instructions removed] We hope to hear from you within three weeks; please let me know if you would like to discuss anything prior to re-submission. Sincerely, Joshua

Joshua Finkelstein, Ph.D. Senior Editor Nature Reviewers' comments: Referee #1 (Remarks to the Author): The authors have done a great job at addressing the points of the reviewers. The significance of this work is very high and the manuscript is now majorly improved. I believe it is ready for publication in Nature. ------------------------------------------------------- Referee #2 (Remarks to the Author): The revised manuscript has answered all questions of the previous reviewers. It is now ready for publication in my opinion. Subject: Re: Decision on Nature manuscript 2011-07-08889 Date: Tue, 9 Aug 2011 13:05:47 -0400 From: j.finkelstein@us.nature.com To: john.rubinstein@utoronto.ca Dear Professor Rubinstein, Your manuscript entitled "Sub-nanometer resolution structure of the intact T. thermophilus H+-driven ATP synthase reveals the arrangement of its trans-membrane helices" has been seen by our referees, whose comments appear at the end of this email. You will see that the referees found the work to be of great interest; however, they have also made comments and/or suggestions that will need to be addressed in the form of a revised manuscript. Please also include a point-by-point response to the referees' comments. Please use the link below to submit a revised manuscript, which should remain formatted as a Letter: http://mts-nature.nature.com/cgibin/main.plex?el=XXXXXXXXXXXXXXXXXXX (NB: This url links to your confidential home page and associated information about manuscripts you may have submitted or be reviewing for us. If you wish to forward this email to co-authors, please delete the link to your homepage first.) We look forward to hearing from you soon. Best wishes, Joshua Joshua Finkelstein, Ph.D. Senior Editor Nature Reviewers' comments: Referee #1 (Remarks to the Author):

Lau and Rubinstein present a very informative structural description of a bacterial ATPase with homology to eukaryotic V-pumps. The work is technically a great accomplishment and allows for the pseudo-atomic modeling of all the components in the complex for which there are crystals structures, giving a detailed description of a number of interactions here described for the first time. Most importantly, the reconstruction allows a description, at the secondary structure level, of the membrane-bound components, the C-terminus of subunit I and the L-ring, and how these interact with each other and with the rest of the complex. From their structure the authors offer a model of how proton movement through the I subunit will result in rotary motion. I find the paper to be of exquisite technical quality. This kind of resolution, definitely subnanometer in nature in light of the very well defined alpha helices throughout the structure, is state of the art for complexes of this size and with no symmetry. Most importantly, the paper reports on a fundamental biological complex that has proven extremely challenging, giving us a description that represents a giant step forward and will make it straight into our undergraduate text books in molecular biology. Thus, I recommend publication in Nature. However, I believe the present manuscript could be significantly improved and would like to propose a number of changes and put forward some comments and questions to the authors. Right now important information is relegated to the supplementary materials. I believe both the "Supplementary results" paragraph and supplementary figure 3 should be part of the main text of the paper. Concerning the discussion on the linkage between proton pumping and rotary motion, I believe the authors could help make their proposed model clearer with schematics that are more detailed and informative that what they now show in figure 3, which a do not find very informative in its present form. Additionally, it will add to the biological value of the structure and the strength of the model they proposed, if the authors could discussed alternative models and how these are compatible or not with their structure. For example, those not closely following the field may not know whether an alternative model, where the two paths proposed are both capable of pumping protons bidirectionally, depending on the direction of the proton gradients, is a plausible alternative, and if so, whether the present structure can shed light on the likelihood of such alternative or not. The paper emphasizes the transmembrane segments in the structure, for which there is no previous direct structural information. I wonder if the authors could not push their analysis of these segments a little further. For example, using the predictions of alpha helical segments shown in supplementary figure 2, could the authors not attempt at a likely identification of these segments (based on proximity in the map and the length of connecting loops, as well as length of the segments, which would relate to the tilt of the helices, or map the position of known functionally important residues? In that respect, it would be useful to have if only a rough idea of the location of the residues described in their discussion within the context of the cryo-EM structure. It would be informative to add a supplementary figure that shows the detailed fitting of the A3B3DF into the cryo-EM reconstruction

illustrating how the three conformations of the AB modules crystal structure relate to those in the EM map, and thus the point made in the main text of how the ATP-bound AB locates with respect to the stalks and the IL components. Another comment/suggestion concerns the inner ring of density for the L-ring segment of the map. What do the authors think about the fact that density appears hard to interpret? Is this structural element expected to be alpha helical? If so, would the helices be packing tighter due to the presences of small side-chain amino acids? Alternatively, could this part of the structure be helical but suffered from disorder or conformational variability that is being averaged out, or not be helical in nature? The appreciation of the structure would benefit tremendously from having a Chimera movie of the structure with the docking and segmentation shown rotating around the main axis of the complex that could be included in the supplementary material. Will the author deposit their structure and docking coordinates in a database? Minor comments: The key for the symbols used in supplementary figure 2 is not given ("*", ":", "."). For supplementary figure 1, the Fourier correlation is not a measure that is often used in the field (rather than the FSC) and including it without a remark as to its relevance seems unnecessary. ------------------------------------------------------- Referee #2 (Remarks to the Author): This manuscript reports a 9.7-Angstrom map of the T. thermophilus ATP synthase derived from cryo-electron microscopy. It reveals a number of novel features of this enzyme, which bears a strong evolutionary resemblance to both proton-translocating V-ATPases and other ATP synthases. The quality of the map is strongly supported by fitting available crystal structures for individual subunits and sub-complexes into the map. Overall, this work is remarkable both technically, as a superb example of the information that can be obtained by cryo-EM for a very large, asymmetric membrane protein, and scientifically, as the first time the helical arrangement of the intact membrane domain has been visualized in either F-type or V-type ATPases/ATP synthases. Arrangement of these helices has important implications for the mechanism of proton translocation. In addition, a number of intriguing features in the peripheral ATPase domain of the enzyme are revealed in this structure, particularly the evidence of a unique low energy conformation that appears to have facilitated the alignment. This work is a major contribution to the field that will be of interest to a wide audience; there are only a few points of interpretation that I believe should be addressed before publication. 1. The most serious point concerns the statement on p. 3 that "the twelve-fold symmetry of the L12-ring is not broken, even where the L-subunits are in contact with the I-subunit, refuting models of proton

translocation that require major conformational changes in the rotor." I do not believe that the data shown in Fig. 2 are sufficient to support this statement. Although there is an approximate 12-fold symmetry for the L-ring in this figure, there appear to be subtle alterations in the helix conformations, particularly where the ring approaches the I subunit (including helices 11, 12, 1, and 2 in different sections). Although this small degree of asymmetry does not alter the proposal of two half-channels through the I subunit, it certainly does not refute "helical swiveling" models such as that proposed in ref. 15, which primarily propose a rotation, rather than a "major conformation change" in the helices. This point could be softened without altering the overall message of the paper. 2. The small contact area between the L and I subunits is a major, and very interesting, point in the paper. However, in Fig 2, the maximal distance between the ring and the I subunit looks like it is about the same as the distance between the outer helices of subunits in the L ring. Are the outer L ring helices separated as suggested in Fig. 2? This might fit with some of the current structures of "c-rings", but deserves some comment. 3. Some points in the abstract need clarification. First, virtually all ATP synthases can act as both synthases and pumps, whether they are F-type, A-type or the bacterial V-type. The distinction for the eukaryotic vacuolar-type ATPases is that they are dedicated pumps. Given existing confusion about nomenclature among these enzymes, this first sentence needs clarification. It is also unclear why in line 9 of the abstract, the authors refer to the T. thermophilus enzyme as a "minimal" ATP synthase, because it has a larger overall membrane domain (although fewer subunits) than the E. coli F1Fo. 4. p. 2, line 11: I believe the optimization and refinement are described in Materials and Methods, not Supplemental Results. ------------------------------------------------------- Referee #3 (Remarks to the Author): By a "low resolution method", cryo-EM, and taking advantage of known partial high resolution structures of F- and V-ATPases the authors ventured to "set the keystone" to the full structure of their particular enzyme. For the first time the contact between the ion-translocating ring and the channel-providing subunit (L in V, A in F-ATPases) is presented at such clarity in any such enzyme. Single particle cryoEM is risky and has produced unreliable results in the past. The present data however are convincing owing to the lucky circumstance that several single molecules seemed to be trapped in one and the same conformational state. This is an important and well-written article that could be published as it stands. The authors might consider to add some words on how the supposedly 5 TM helices in F-ATPase would alter their concept in Fig. 3B. **** End of referee comments ****

**** End of email **** This email has been sent through the NPG Manuscript Tracking System NY-610A-NPG&MTS