Integration of Structures, Spectroscopies and Mechanisms

Structural organization of the V-ATPase and its implications for regulatory assembly and disassembly

Meikel Diepholz, Michael Börsch, Bettina Böttcher


V-ATPases (vacuolar ATPases) are membrane-bound multiprotein complexes that are localized in the endomembrane systems of eukaryotic cells and in the plasma membranes of some specialized cells. They couple ATP hydrolysis with the transport of protons across membranes. On nutrient shortage, V-ATPases disassemble into a membrane-embedded part (V0), which contains the proton translocation machinery, and an extrinsic part (V1), which carries the nucleotide-binding sites. Disassembly decouples ATP hydrolysis and proton translocation. Furthermore, the disassembled parts are inactive, leading to an efficient shutdown of ATP consumption. On restoring the nutrient levels, V1 and V0 reassemble and restore ATP-hydrolysis activity coupled with proton translocation. This reversible assembly/disassembly process has certain conformational constraints, which are best fulfilled by adopting a unique conformation before disassembly.

  • ATP hydrolysis
  • plasma membrane
  • proton transport
  • regulatory assembly
  • structural organization
  • vacuolar ATPase (V-ATPase)


V-ATPases (vacuolar ATPases) are ATP-dependent proton pumps [1], which are located in the inner membranes of eukaryotic cells and in the plasma membranes of specialized cells. In the inner membranes, V-ATPases generate a potential difference of protons between the lumen and cytosol, which drives secondary transport processes [2], thus contributing to osmoregulation and pH and ion homoeostasis [3].

V-ATPases are multiprotein complexes that consist of 14 different polypeptides. The complex has a bipartite structure consisting of a membrane-integrated V0 (subunits a, c, c′, c″, d and e) and an extrinsic V1 (subunits A–H). Both parts are linked by a connecting region that is important for coupling proton translocation in V0 with ATP hydrolysis in V1. The connecting region consists of a central shaft and multiple peripheral elements to which V1 and V0 contribute (subunits C, D, E, F, G, H, d and a).

Similar to the related F-ATPases, V-ATPases function with a rotational mechanism [4,5]. Hydrolysis of ATP at alternate catalytic nucleotide-binding sites on subunits A in V1 induces rotational motion of the central shaft (subunits d, D and F). This motion is conveyed to a ring-like structure (c, c′ and c″) in V0. Part of this ring forms the rotating half of the proton channel which is completed by the C-terminal domain of subunit a, which remains static during proton translocation. The rotational movement of the ring against the static part of the channel drives protons across the membrane. In order to be productive, such a rotational mechanism requires an additional static connection between V1 and V0 that prevents co-rotation. This static connection consists of subunits a, E, G, C and H, with subunit a as the sole membrane anchor.

Overall organization

Most of our knowledge of the structural organization of V-ATPases originates from electron microscopic investigations ([6,7,914] and M. Diepholz, D. Venzke, S. Prinz, C. Batisse, B. Flörchinger, M. Rössle, D.I. Svergun, B. Böttcher and J. Fethiere, unpublished work). Projection maps show the bipartite organization of the enzyme where V1 and V0 are linked by a less dense connecting region [6,13], which is far more complex than in F-ATPases, where it consists of only a central stalk and a single peripheral stator [1519].

Several three-dimensional image reconstructions of V-ATPases from different species resolve the spatial arrangement of the connecting region. However, structural details disagree between different maps, making it difficult to derive a consensus model for the stator organization. For example the number of peripheral connecting elements ranges between one and three. In some reconstructions these stators are linked by arms parallel to the membrane, which form an arc of 120–180° [11,12], whereas other reconstructions do not resolve these arms [7] or show a complete circular structure [9].

A recent map of the V-ATPase of Saccharomyces cerevisiae (M. Diepholz, D. Venzke, S. Prinz, C. Batisse, B. Flörchinger, M. Rössle, D.I. Svergun, B. Böttcher and J. Fethiere, unpublished work) is the first one of the holo-enzyme that resolves the A- and B-subunits individually and reproduces their pseudo 3-fold arrangement faithfully. This lends support to the reliability of this particular map. The map shows three peripheral connecting elements that mediate binding of V1 to V0. The peripheral elements are interconnected by arms parallel to the plane of the membrane, which form an arc of 240°.

Arrangement of A- and B-subunits

The map of the V-ATPase of S. cerevisiae serves as a scaffold for building a ‘pseudo-atomic’ model from individual subunits and subcomplexes. The largest part of V1 is accounted for by the catalytic (A) and non-catalytic (B) nucleotide-binding subunits. Homologous structures of both subunits have been determined from Pyrococcus horikoshii and Methanosarcina mazei Gö1 respectively [20,21]. At low resolution, the main differences are two extensions to subunit A: one in the N-terminal part that has been designated the non-homologous region, and one in the C-terminal region that results in a longitudinal enlargement towards the connecting region. These differences are sufficient to unambiguously distinguish between A- and B-subunits in the three-dimensional map of the V-ATPase of S. cerevisiae (Figure 1). According to this assignment, the B-subunits are the stator-binding subunits. The AB-dimers deviate slightly from the ideal 3-fold symmetry, with one AB pair being less tightly packed than the other two (Figure 1). This resembles the arrangement of the nucleotide-binding subunits in the F1-ATPase [22], where all three catalytic subunits have a different nucleotide occupancy. Structural comparison suggests that the tight nucleotide-binding site is located closest to the membrane anchor of the stator (Figure 1), which is the same arrangement as observed for the F-ATPase from chloroplasts [16]. Interestingly, the A-subunit that forms the empty nucleotide-binding site is less well resolved, which is compatible with a lack of stabilization by bound substrate.

Figure 1 Organization of (AB)3 in V-ATPase

(A) Density of a section through V1 showing the catalytic (AB)3. The position of the section is indicated in (C). (B) Surface representation of the same section with the crystal structures of A and B [20,21] fitted into the individual subunits. A and B are in close contact in two of the AB pairs (highlighted by circles), while the third pair is more open. The tight pairs are most likely to be in nucleotide-bound states, whereas the open pair is probably empty. The AB pairs are labelled analogous to the F-ATPase [22] (T, ATP-bound; D, ADP-bound; and E, empty). (C) Surface representation of the V-ATPase from S. cerevisiae.

The organization of the stator

The map of the V-ATPase of S. cerevisiae shows three stator elements that attach to the B-subunits at the top and at the side. Each of these three stators is probably formed by subunits E and G. E and G interact tightly [2325] and exist in three copies in the complex [26]. Cross-linking experiments establish that subunit E binds to the top and side of the B-subunit [27], which is consistent with the observed position of the stator. The structure of the C-terminal domain of a homologue of subunit E has been solved by X-ray crystallography [28] and fills the extensions above the B-subunits. Consequently, subunit G and the N-terminal domain of subunit E form the link with the connecting region, which is consistent with the interaction of the two subunits in their N-terminal domains [29].

The EG-stator elements do not have a membrane anchor. Therefore they require direct or indirect binding to subunit a, which is the only membrane-integrated subunit of the stator. Arms parallel to the plane of the membrane, which are resolved in various electron microscopic investigations ([6,1113] and M. Diepholz, D. Venzke, S. Prinz, C. Batisse, B. Flörchinger, M. Rössle, D.I. Svergun, B. Böttcher and J. Fethiere, unpublished work), form an arc that interconnects the three EG-stator elements. The arms consist of the N-terminal domain of subunit a and subunits C and H. Immunolocalization experiments [12,30] and comparison with the A-ATPase of Thermus thermophilus [11,31], which is lacking subunits C and H, suggest that these subunits are located at opposite ends of the arc and that the N-terminal domain of subunit a produces the central part of the arc.

Subunit H is close to the membrane anchor and binds to one EG-stator element ([11] and M. Diepholz, D. Venzke, S. Prinz, C. Batisse, B. Flörchinger, M. Rössle, D.I. Svergun, B. Böttcher and J. Fethiere, unpublished work). The proximity between subunits H and a is confirmed by cross-link experiments [32]. In addition, the binding of subunit H to the EG subcomplex in vitro [23] is also compatible with its close proximity to a stator.

Subunit C is positioned furthest away from the membrane anchor and connects the second EG-stator element with the third EG-stator element. This role of subunit C as a linker is consistent with its in vitro interaction with the EG dimer [23,24,29] and with two similar binding motifs at opposite sides [33] that could support binding of two EG-stator elements.

Implications of a resting position in V-ATPases

The fact that the different conformations of the catalytic nucleotide-binding subunits are resolved in the three-dimensional map of the V-ATPase of S. cerevisiae (M. Diepholz, D. Venzke, S. Prinz, C. Batisse, B. Flörchinger, M. Rössle, D.I. Svergun, B. Böttcher and J. Fethiere, unpublished work) implies that the stalled V-ATPase stops in a defined resting conformation. Such a resting conformation must have the lowest energy. This raises the question of what distinguishes the resting conformation from the other possible conformations. In contrast with the situation in F-ATPases, where one of the nucleotide-binding sites is potentially different from the others by binding the stator element, in V-ATPases, all three non-catalytic B-subunits bind to a similar stator element. This makes the environment of the three catalytic nucleotide-binding subunits much more similar in V-ATPases than in F-ATPases. Therefore, for V-ATPases it appears less likely that the different property of one of the binding sites is caused by binding to the stator element as proposed for the chloroplast F-ATP synthase [16]. However, the energy of the conformational state is not only determined by the state of the nucleotide-binding sites but also by the conformation of the proton channel. Part of this proton channel is formed by the c-ring, which rotates during ATP hydrolysis. In V-ATPase of S. cerevisiae, the c-ring is formed by three different types of c-subunit [34] [c (VMA3), c′ (VMA11) and c″ (VMA16)], which have a specific relative arrangement [35,36]. Consequently, the c-ring is intrinsically asymmetric and one of the proton-binding sites could be energetically more favourable than the other sites, which would trap the stalled V-ATPase in a certain conformation. It is likely that this special subunit is c″, which has the acidic residue that is involved in proton translocation in transmembrane helix 2 instead of in transmembrane helix 4 in subunits c and c′ [37].

The significance of the resting position of V-ATPases is unclear. One possible involvement could be in the regulatory assembly/disassembly process that occurs depending on nutrient supplies [38,39]. Nutrient shortage leads to a rapid disassembly of the V-ATPase into inactive V1 and V0. In the disassembled V1, subunit H inhibits ATP hydrolysis probably by interacting with subunit F [40], which is part of the central rotating shaft. Such an interaction between subunit H in the stator and F in the rotor would stop rotation and therefore inhibit ATP hydrolysis. However, efficient interaction requires the central shaft to have a certain orientation with respect to subunit H, which could be fulfilled if the V-ATPase adopts a certain conformation before disassembly.

During reassembly, V1 reconnects to V0. This requires matching of the asymmetrically attached static elements. Without a unique resting conformation, V1 and V0 could be in any permissible conformation of the rotational mechanism (Figure 2). In this case, reassembly would either require a sorting mechanism that brings together V1 and V0 in matching conformations or a switching mechanism that resets the conformation of V1 and V0 appropriately. By adopting a unique resting conformation before disassembly, such a complicated mechanism would be obsolete and any free V1 would match any free V0 (Figure 2).

Figure 2 Possible conformations of the disassembled subcomplexes

(A) Cartoon of a possible conformation of V0 (seen from the cytosolic side). The relative orientations of subunits d and a (the soluble N-terminus of a is shown as a broken line) follow the three-dimensional map of the V-ATPase from S. cerevisiae. Positions of the unique subunits c′ and c″ in the ring are chosen arbitrarily. The footprint of the D/F binding site is indicated by a white broken line. (B1B3) Three possible conformations of V1 that occur as a consequence of the rotation of the central rotor consisting of subunits D/F relative to the static part of V1 [(ABEG)3H]. V1 shown in (B1) is the only conformation in which it could reassemble without further rearrangements with the V0 shown in (A).

How to follow conformational changes during regulatory assembly/disassembly in vivo?

These considerations on the assembly/disassembly mechanism lead to the following question. In which conformational state are the subunits and subcomplexes in vivo? In living cells, the conformational dynamics of the events before, during and after assembly and disassembly can be best followed by fluorescence microscopy using autofluorescent proteins fused to subunits of V-ATPase [41]. As shown in Figure 3, subunit C dissociates from the vacuolar membrane and is found in the cytosol during starvation. In contrast, after glucose addition, this subunit is transported back to the membrane. Staining the vacuolar membrane with a secondary fluorophore enables the co-localization of membrane and subunit C. In addition, confocal microscopy with single-molecule detection sensitivity allows for localized measurement of the diffusion coefficient by FCS (fluorescence correlation spectroscopy). The diffusion properties of proteins depend not only on the size [42], shape [43], conformation [44] and nucleotide binding [45], but also on association with other proteins [46,47]. FCS of subunit C during starvation reveals the mobility in the cytosol as well as the size of the diffusing particle (Figure 3C). In the reassembled V-ATPase, the slower diffusion times of subunit C on the membrane are found to be the same as for other membrane-bound proteins, i.e. approx. 100 ms [48]. However, a cytosolic diffusion time of 330 μs for C upon starvation indicates a significant fraction of freely moving protein not bound to immobile cellular structures, such as actin filaments, as suggested by the actin-binding motifs in the subunit C structure [33] and its capability for binding actin [49]. Furthermore, the cytosolic diffusion time corresponds to a protein size compatible with that of isolated subunit C [43], and is significantly smaller than expected for a complex of C with V1.

Figure 3 Fluorescence measurements of yeast cells where subunit C of the V-ATPase is C-terminally labelled with enhanced GFP (green fluorescent protein) (C_eGFP)

(A) Yeast cells in glucose-containing medium; C_eGFP (cyan) co-localizes with the vacuolar membrane marker FM 4-64 (red). (B) Yeast cells in glucose-free medium; the vacuolar membrane marker FM 4-64 (red) is not co-localized with C_eGFP (cyan). (C) Normalized autocorrelation functions of rhodamine 110 (green; diffusion time 60 μs), C_eGFP in the cytosol (black) of cells in glucose-free medium, and C_eGFP close to the fluctuating membrane (red) of cells in medium with added glucose. The latter autocorrelation function shows an additional longer diffusion time component of approx. 60 ms compared with C_eGFP in the cytosol (approx. 330 μs).

Future experiments with the V-ATPase doubly-labelled with two distinct fluorophores will aim at determining co-migration [50] to address the question of how yeast cells achieve the rapid transport of V1 and C through the cytosol and back for reassembly on the vacuolar membrane.


This work was supported by the EU (European Union) grant ‘3D repertoire’, contract number LSHG-CT-2005-512028, and by the Deutsche Forschungsgemeinschaft. M.D. was supported by the E-STAR (Early-Stage Training in Advanced Life Science Research) project, the EC (Enzyme Commission)'s FP6 (Framework Programme 6) Marie Curie Actions for Early-Stage Training, contract MEST-CT-2004-504640.


  • Integration of Structures, Spectroscopies and Mechanisms: Second Joint German/British Bioenergetics Conference, a Biochemical Society Focused Meeting held at University of Edinburgh, U.K., 2–4 April 2008. Organized by Ulrich Brandt (Frankfurt, Germany), Steve Chapman (Edinburgh, U.K.), Peter Heathcoate (Queen Mary, University of London, U.K.), John Ingledew (St Andrews, U.K.), Mike Jones (Bristol, U.K.), Bernd Ludwig (Frankfurt, Germany), Fraser MacMillan (University of East Anglia, Norwich, U.K.), Hartmut Michel (Max-Planck-Institute for Biophysics, Frankfurt am Main, Germany), Peter Rich (University College London, U.K.) and John Walker (MRC Dunn Human Nutrition Unit, Cambridge, U.K.). Edited by Ulrich Brandt and Peter Rich.

Abbreviations: FCS, fluorescence correlation spectroscopy; V-ATPase, vacuolar ATPase


View Abstract