Structure and dynamics of rotary V1 motor

Rotary ATPases are unique rotary molecular motors that function as energy conversion machines. Among all known rotary ATPases, F1-ATPase is the best characterized rotary molecular motor. There are many high-resolution crystal structures and the rotation dynamics have been investigated in detail by extensive single-molecule studies. In contrast, knowledge on the structure and rotation dynamics of V1-ATPase, another rotary ATPase, has been limited. However, recent high-resolution structural studies and single-molecule studies on V1-ATPase have provided new insights on how the catalytic sites in this molecular motor change its conformation during rotation driven by ATP hydrolysis. In this review, we summarize recent information on the structural features and rotary dynamics of V1-ATPase revealed from structural and single-molecule approaches and discuss the possible chemomechanical coupling scheme of V1-ATPase with a focus on differences between rotary molecular motors.


Introduction
All living cells use a chemical fuel called adenosine triphosphate (ATP) to maintain the functions required for life, such as protein biosynthesis, muscle contraction, and brain activity. Therefore, ATP is often referred to as the energy currency for the cell. Under aerobic conditions, the majority of ATP is produced by the F-type ATP synthase [1]. F-type ATP synthases (also known as F-ATPases) are ubiquitous rotary motor enzymes found in the inner membrane of mitochondria, the thylakoid membrane of chloroplasts, and the plasma membranes of bacteria. They catalyze ATP synthesis from ADP and inorganic phosphate using the energy of ion (proton or sodium) translocation caused by transmembrane electrochemical potential (proton or sodium ions) and, when operating in reverse, they also generate an electrochemical potential difference of ions using the energy released by ATP hydrolysis [2][3][4].
Eukaryotic vacuolar-type ATPases (V-ATPases) are also rotary motor enzymes; they are evolutionarily and functionally related to F-type ATP synthases [5,6]. They function in reverse of F-type ATP synthases, that is, they transport ions (protons or sodium ions) across the membrane using the energy derived from ATP hydrolysis. Acidification of vesicles by intracellular V-ATPases is important for various cellular processes, including receptor-mediated endocytosis, membrane trafficking, and protein processing and degradation. They also function on the plasma membrane of certain cells, such as tumor cells, renal intercalated cells, and osteoclasts. Aberrant function of V-ATPase is associated with a number of human diseases including tumor metastasis, distal renal tubular acidosis, and osteoporosis. Therefore, they are considered as potential drug targets [5,7]. V-ATPases are found in some bacteria, such as Thermus thermophilus (T. thermophilus) and Enterococcus hirae (E. hirae). V-ATPase from T. thermophilus functions as an ATP synthase under physiological conditions [8,9]. Therefore, this enzyme is sometimes called A-type ATPase found in Archeae (A-ATPase). A-ATPases function as ATP synthases similar to the F-type ATP synthase, although the structure and subunit composition of A-ATPases are more similar to 1 3 those of V-ATPases (Fig. 1a) [10][11][12]. V-ATPase from E. hirae functions as a sodium ion pump similar in nature to eukaryotic V-ATPase, and plays an important role in maintaining sodium homeostasis in cells under an alkaline environment [13][14][15].
All rotary ATPases are unique rotary molecular motors that function as energy conversion machines with a similar architecture and rotary catalytic mechanism [5,11,12,16]. They are large, multi-subunit complexes composed of a hydrophilic F 1 /V 1 /A 1 motor for ATP synthesis/hydrolysis and a membrane-embedded F o /V o /A o motor for ion transport. The bacterial F 1 /V 1 /A 1 and F o /V o /A o motors are connected by one central stalk and one or two peripheral stalks (Fig. 1). Interestingly, V-ATPases in eukaryotes have three peripheral stalks, although there is only one peripheral stalk in eukaryotic F-ATPases [16][17][18]. The peripheral stalks of eukaryotic V-ATPases play an important role in reversible dissociation of V 1 motor from V o motor with silencing of the ATP hydrolysis activity of the free V 1 motor [18], in contrast to F 1 motor which will rapidly hydrolyze ATP when isolated. This implies the fact that it was necessary for V-ATPases to evolve a regulatory mechanism for keeping the dissociated V 1 motors catalytically inactive to prevent wasteful energy consumption.

V 1 -ATPase: structural studies
In the bacterial V-ATPase, the catalytic V 1 moiety is composed of A, B, D, and F subunits, in which three alternately arranged A and B subunits form a hexameric stator A 3 B 3 ring. The central rotary shaft of D and F subunits penetrates the central cavity of the A 3 B 3 ring and rotates using the energy of ATP hydrolysis [41][42][43][44]. Unlike the isolated eukaryotic V 1 moiety in which subunit H inhibits its activity [18], the isolated bacterial V 1 moiety can generally catalyze ATP hydrolysis and hence is called V 1 -ATPase. Structural studies have been conducted using a V 1 -ATPase from the thermophilic eubacterium T. thermophilus, which has high stability. The crystal structure of the A 3 B 3 subcomplex from T. thermophilus was determined at 2.8 Å resolution [41]. The diameter of the A 3 B 3 subcomplex is larger than that of the α 3 β 3 subcomplex in F 1 -ATPase, because it includes an outward protrusion domain in the A subunit (Fig. 2a, green circles), termed the "non-homologous region" that is absent from β subunit; the structure also provides molecular information about the B-A interface. The catalytic sites are located at the interfaces of the A and B subunits (Fig. 2a, red arrows), with the majority of the catalytic residues residing in the A subunits, similar to the catalytic β subunits and the non-catalytic α subunits in F 1 -ATPase [19,27]. The overall structure of V 1 -ATPase (A 3 B 3 DF complex) from T. thermophilus was first determined at 4.5-4.8 Å resolution (Fig. 2b) [42]. This structure provided the initial information about the position and orientation of the rotor DF subunits in the A 3 B 3 ring and revealed structural similarities and differences between V 1 -and F 1 -ATPase. However, the lack of high-resolution structural information for the overall V 1 -ATPase from T. thermophilus limits our understanding of its molecular architecture and operation. Meanwhile, high-resolution crystal structures of the A 3 B 3 ring and entire V 1 -ATPase from E. hirae have recently been solved with and without bound nucleotides [44,45].

Structure of A 3 B 3 ring from E. hirae: asymmetric structure
The high-resolution crystal structures of the A 3 B 3 ring from E. hirae were solved with and without a non-hydrolyzable ATP analog [adenosine 5′-(β,γ-imido)triphosphate or AMPPNP] at 3.4 and 2.8 Å resolution, respectively ( Fig. 3) [44]. The overall architecture of A 3 B 3 from E. hirae is similar to that of α 3 β 3 in F 1 -ATPase, but the structures show some differences. Each A and B subunit consists of an N-terminal β-barrel, central α/β domain, and C-terminal helical domain (Fig. 3a, right). Superimposition of the N-terminal β-barrel part of three A or B subunits shows that the conformations of each A and B subunit are not identical. If one of the A subunits is in the closed conformation (C),  [42]. Side view (left) and top view (right) from the membrane side. Only the C-terminal domains of the A 3 B 3 ring and the D subunit are shown in top view. The A, B, D, and F subunits are shown in blue, magenta, green, and red, respectively. A W , B W pair shows a wide-open conformation, as observed in a α E β E pair in F 1 -ATPase [19]. A N B N and A N′ B N′ pairs show a narrowly closed conformation, as do the α TP β TP and α DP β DP pairs in F 1 -ATPase [19].  (Fig. 3a), resulting in the asymmetry of the A 3 B 3 ring. Interestingly, even in the absence of nucleotides and central rotor DF subunits, each catalytic site shows the three distinct states. In contrast, F 1 -ATPase shows a threefold symmetric structure with three identical catalytic sites in the absence of bound nucleotides [50,51]; accordingly, the asymmetric structure of the stator ring is not observed in the case of F 1 -ATPase. In the presence of high concentration of AMPPNP (5 mM), the A 3 B 3 ring binds two AMPPNP molecules (Fig. 3b), resulting in changes in the conformation of A (O′ to C) and B (O to O′) subunits, but the A C B O′ pair shows little conformational change upon AMPPNP binding. Therefore, the three catalytic sites (A O B C , A O′ B O , and A C B O′ ) are termed 'empty', 'bindable', and 'bound' sites, respectively (Fig. 3a, b).

Two AMPPNP-bound structure of V 1 -ATPase: catalytic dwell state
In addition to the A 3 B 3 ring structures, the nucleotide-free entire V 1 -ATPase from E. hirae (eV 1 ) was also determined at 2.2 Å resolution [44]. Insertion of the rotor DF subunits into the stator A 3 B 3 ring induces conformational changes in the A and B subunits, even in the absence of bound nucleotides three different catalytic sites termed 'empty', 'bound', and 'tight' sites ( Fig. 4a). Furthermore, by soaking eV 1 in AMPPNP, the structure of nucleotide-bound V 1 -ATPase (bV 1 or 2 ATP V 1 ) was also determined at 2.7 Å resolution ( Fig. 4b) [44,45]. Two AMPPNP molecules were bound to the binding sites of the 'bound' and 'tight' sites of eV 1 , but the overall structure was very similar to that of eV 1 (Fig. 4a,  b). Even in the presence of high AMPPNP (2 mM), no electron density peak for AMPPNP was found in the 'empty' site, indicating that it has a very low affinity for AMPPNP. Comparison of the 'tight' and 'bound' sites in bV 1 revealed the movement of the Arg-finger (Arg-350) in the 'tight' site closer to the γ-phosphate relative to the 'bound' site ( Fig. 4c). This γ-phosphate moved closer to Glu-261 in the A subunit which is essential for ATPase activity in yeast V 1 -ATPase [52]. The corresponding Glu-188 of the β subunit in bovine mitochondrial F 1 -ATPase is an essential residue for ATP hydrolysis and interacts with the γ-phosphate of the nucleotide and lytic water molecules [19,28,53]. The closer proximity of the Arg-finger to the γ-phosphate may enhance the ATP hydrolysis reaction. Therefore, it is possible that the 'tight' site corresponds to the catalytic site waiting for ATP hydrolysis and this structure corresponds to the catalytic dwell state (Fig. 5a). Similar nucleotide-free and nucleotide-bound structures of yeast mitochondrial F 1 -ATPase have been reported [25]. These results suggest that interactions between the rotor and stator are as crucial as nucleotide binding for determining the structure of the catalytic sites of rotary ATPases.

Two ADP-bound structure of V 1 -ATPase: ATP-binding dwell state
More recently, the crystal structures of two other nucleotide-bound states were reported [45], i.e., the two ADPbound structure (2 ADP V 1 ) and the three ADP-bound structure (3 ADP V 1 ). When soaking the eV 1 crystals in 20 µM ADP, it binds to the 'bound' and 'tight' sites of eV 1 , as in the case of bV 1, and the two ADP-bound structure was solved at 3.3 Å resolution (Fig. 5b). ADP binding to the 'tight' site in eV 1 induces changes in the A and B subunits to more open conformations (CR to C in A subunit; CR to C′ in B subunit), but the 'bound' site in eV 1 shows no conformational change upon ADP binding. The observed conformational changes result in the tilting of rotor DF subunits towards the ADP-bound site (A C B C′ ) (Fig. 5b). Interestingly, the 'empty' site shows a cooperative conformational change without ADP binding. The newly found catalytic sites (A C B C′ and A O′ B O′′ ) are termed 'ADPbound' and 'bindable-like' sites, respectively. The 'bindable-like' site is similar to the 'bindable' site in the A 3 B 3 structure (Fig. 3a) 5 Crystal structures of two AMPPNP-bound V 1 -ATPase (bV 1 ), two ADP-bound V 1 -ATPase (2 ADP V 1 ), and three ADP-bound V 1 -ATPase (3 ADP V 1 ) from E. hirae. a Crystal structure of two AMPPNP-bound V 1 -ATPase (bV 1 , 2.7 Å resolution, PDB ID: 3VR6) [44]. b Crystal structure of two ADP-bound V 1 -ATPase (2 ADP V 1 , 3.3 Å resolution, PDB ID: 5KNB) [45]. c Crystal structure of three ADP-bound V 1 -ATPase (3 ADP V 1 , 3.0 Å resolution, PDB ID: 5KNC) [45]. Top views of the C-terminal domain of A 3 B 3 rings and central rotor D subunit (green) viewed from the membrane side are shown on the left. Red, orange, and magenta arrowheads indicate the catalytic sites that bind to AMPPNP, ADP, and sulfate, respectively. Conformations of the individual A and B subunits superimposed at the N-terminal β-barrel domain (white) are shown on the right. O and O′ open conformation, HC half-closed conformation, C closed conformation, CR more closed 'closer' conformation than that of the 'empty' site. Therefore, the 'bindablelike' site is considered to be the catalytic site waiting for ATP binding, and the 2 ADP V 1 structure is regarded as the ATP-binding dwell state.

Three ADP-bound structure of V 1 -ATPase: ADP-release dwell state
The 3 ADP V 1 structure was solved at 3.0 Å resolution by soaking the eV 1 crystals in 2 mM ADP [45]. In this structure, all three catalytic sites are occupied by ADP and, in addition, a sulfate is bound to one catalytic site (Fig. 5c, magenta arrowhead). A comparison between 2 ADP V 1 and 3 ADP V 1 structures shows that ADP (and sulfate) binding to the 'bindable-like' site in 2 ADP V 1 induces a conformational change in the A subunit (O′) to the 'half-closed' conformation (HC), whereas the B subunit (O′′) shows no conformational change. This catalytic site (A HC B O′′ ) in 3 ADP V 1 is termed a 'half-closed' site. The conformational change in the catalytic site from the 'bindable-like' site to the 'half-closed' site induces a conformational change of the 'ADP-bound' site to a more tight-like conformation (A CR′ B CR′ ). This shifted 'ADP-bound' site in 3 ADP V 1 is termed a 'tight-like' site, because the nucleotide-binding site is more similar to that of the 'tight' site than to that of the 'ADP-bound' site (Fig. 6). The β-phosphate of ADP in the 'tight-like' site is more distant from the surrounding interacting residues compared to that in the 'ADP-bound' site, suggesting that an ADP will be easily released from the 'tight-like' site ( Fig. 6) [45]. Therefore, the "tightlike" site is considered to be a dwelling state before ADP release and the 3 ADP V 1 structure is regarded as the ADPrelease dwell state.

Structural difference between F 1 -and V 1 -ATPase: conformational features
A comparison of the two AMPPNP-bound structures of V 1 -ATPase from E. hirae [44,45] and F 1 -ATPase from bovine mitochondria [28] is shown in Fig. 7. Both nucleotide-binding sites show very similar arrangements of catalytically important residues and nucleotides (Fig. 7a). However, the overall structures of A and B subunits in V 1 -ATPase show some differences from those of the β and α subunits in F 1 -ATPase. The non-catalytic B subunit of this V 1 -ATPase does not bind to a nucleotide, whereas the non-catalytic α subunit of F 1 -ATPase binds to a nucleotide (Fig. 7b). Superimposition of the N-terminal β-barrel part of three A and B subunits in V 1 -ATPase and β and α subunits in F 1 -ATPase reveals conformational differences between A C and A CR , and B C and B CR of V 1 -ATPase, but the conformations of β TP and β DP as well as α TP and α DP of F 1 -ATPase are very similar (Fig. 7c). These differences are also evidenced by the positional displacement of residues between two of the three A and B subunits in V 1 -ATPase and β and α subunits in F 1 -ATPase (Fig. 8), which also shows that the structures of A C and A CR are largely different from that of A O (Fig. 8,  top left). In contrast, the central portions (residues 180-320) of the β TP and β DP structures show very similar conformations to that of the β E structure (Fig. 8, top right). These results suggest that the A subunit in V 1 -ATPase from E. hirae undergoes the whole conformational change upon nucleotide binding, whereas the β subunit in F 1 -ATPase undergoes the conformational change mainly in the P-loop and C-terminal domains.
Thus, the conformational features of the A and B subunits in V 1 -ATPase from E. hirae are apparently different from those of the β and α subunits in bovine F 1 -ATPase, despite the highly similar nucleotide-binding sites of these ATPases. These structural differences and similarities may be related Fig. 6 Nucleotide-binding site of the three ADP-bound V 1 -ATPase (3 ADP V 1 ). Nucleotide-binding site of the "tightlike" site in 3 ADP V 1 (colored) is superimposed at the adenosine onto those (transparent grey) of the "ADP-bound" site in 2 ADP V 1 (left) and the "tight" site in bV 1 (right) to the 'unique' and 'common' mechanisms of rotary catalysis between these rotary ATPases, such as the chemomechanical coupling scheme of the rotation [40,46].

Dynamics of rotary V 1 -ATPase: single-molecule studies
After the establishment of the single-molecule rotation assay of F 1 -ATPase in 1997 [29], the dynamics of rotary ATPases from various species have been studied at the single-molecule level [38-40, 46, 54-57]. The first rotation of V 1 -ATPase has been directly visualized under an optical microscope by the attachment of large beads to the rotor DF subunits in V 1 -ATPase from T. thermophilus [58]. V 1 -ATPase from T. thermophilus rotates stepwise in a counterclockwise direction, consuming one ATP molecule at each step when viewed from the membrane side. The basic step size is 120°, which is similar to that of F 1 -ATPase, and no substeps have been resolved in the rotation, even when using a slowly hydrolyzable ATP analog (adenosine 5′-O-(3thio)triphosphate or ATPγS) and high-speed imaging of gold nanoparticles [48,59]. These results indicate that ATP binding and ATP cleavage (and/or phosphate release) occur at the same angle in this V 1 -ATPase. In the case of F 1 -ATPases, some or all of these elementary reaction steps occur at different angles and the basic 120° step is further divided into two or three substeps [31,32,39,40], i.e., 80° and 40° substeps in thermophilic Bacillus PS3 F 1 -ATPase [31,32], 85° and 35° substeps in Escherichia coli F 1 -ATPase [39], and Fig. 7 Structural differences between F 1 -and V 1 -ATPase. a Comparison of the nucleotidebinding site of the "tight" site in E. hirae V 1 -ATPase (bV 1 , PDB ID: 3VR6) with that of the "ADP-bound" site in bovine F 1 -ATPase (PDB ID: 2JDI) [28]. b Top views from the membrane side of bV 1 (left) and bovine F 1 -ATPase (right). The catalytic and non-catalytic sites that bind to AMPPNP molecules are indicated by red arrowheads. c Superimposed structures at the N-terminal β-barrel (white) of three structures of A and B subunits in bV 1 compared with the β and α subunits in F 1 -ATPase 65°, 25°, and 30° substeps in human F 1 -ATPase [40]. The 80°, 85°, and 65° substeps are triggered by ATP binding and ADP release, while the 40° and 35° substeps are triggered by ATP cleavage and phosphate release. In human F 1 -ATPase, it is proposed that ATP cleavage and phosphate release trigger different substeps of 30° and 25°, respectively.
Recently, rotary dynamics of V 1 -ATPase from E. hirae have been characterized using single-molecule analyses at a submillisecond temporal resolution employing gold nanoparticles and an objective-type total internal reflection dark field microscope (Fig. 9a) [46][47][48]. V 1 -ATPase from E. hirae rotates in basically the same manner as that from T. thermophilus which functions as ATP synthase [55,59]. This V 1 -ATPase also shows only three pauses separated by 120° at all concentrations ranging from below to above the Michaelis constant (K m ), where distinct elementary reaction steps of ATP hydrolysis (ATP binding, ATP cleavage, or product release) become the rate-limiting step (Fig. 9b), suggesting that 120° stepping rotation without substeps is a common property of V 1 -ATPase [46], despite the difference in physiological function between V-ATPase from E. hirae (ion pump) and that from T. thermophilus (ATP synthesis). These results imply that the basic properties of rotary dynamics of F-ATPases and V-ATPases are determined by their overall structures and that the difference in physiological function derives from regulatory mechanisms such as Mg 2+ ADP inhibition and inhibitor proteins [9,18,23,26]. Figure 10a shows a proposed chemomechanical coupling scheme of V 1 -ATPase from E. hirae based on recently determined structural features and rotary dynamics [44][45][46][47]. As mentioned above, three distinct structures of E. hirae V 1 -ATPase have been solved (bV 1 = 2 ATP V 1 , 2 ADP V 1 , and 3 ADP V 1 ) [44,45]. These three structures are regarded as the three different dwelling states in the rotation waiting for the elementary reaction steps of ATP hydrolysis corresponding, respectively, to the catalytic (ATP cleavage) dwell (bV 1 = 2 ATP V 1 ), ATP-binding dwell (2 ADP V 1 ), and ADPrelease dwell (3 ADP V 1 ) states.

Chemomechanical coupling scheme of V 1 -ATPase
In this model, ATP binds to the 'bindable-like' site in 2 ADP V 1 , which induces conformational changes of 'bindablelike' and 'ADP-bound' sites to 'half-closed' and 'tight-like' sites, respectively. This results in a slight shift of the rotor DF subunits toward the 'tight-like' site, but does not induce the obvious rotation of the rotor DF subunits. Then, ADP release occurs at the 'tight-like' site in 3 ADP V 1 , and all three catalytic sites change the conformations from 'half-closed' to 'bound', from 'bound' to 'tight', and from 'tight-like' to 'empty'. Coupled with these conformational changes, the DF subunits rotate 120°. After this rotation, ATP bound at the Fig. 9 Single-molecule rotation of E. hirae V 1 -ATPase. a Schematic illustration of the single-molecule rotation assay of E. hirae V 1 -ATPase. The A 3 B 3 ring is immobilized on a glass surface via a His-tag on the A subunit, and an optical probe (gold nanoparticle, 40 nm in diameter) is attached to the D subunit to visualize the rotary motion of rotor DF subunits using an optical microscope [46,68]. b Rotations of E. hirae V 1 -ATPase for various concentrations of ATP. Left: 40 mM ATP, considerably higher than the Michaelis constant (K m , 154 µM). Center: 100 µM ATP, near the K m . Right: 10 mM ATP, considerably lower than the K m [46] Ni-NTA glass  [45]. The structure models are based on the crystal structures of 2 ADP V 1 (ATP-binding dwell), 3 ADP V 1 (ADP-release dwell), and bV 1 (catalytic dwell). The catalytic sites that bind to nucleotides are indicated by arrowheads (top). Each blue circle represents the chemical state of each catalytic site, and the central red arrow represents the orientation of the rotor DF subunits. ATP* represents the pre-or post-hydrolysis state of ATP (middle). Correspondence table for all catalytic sites observed in the crystal structures of the A 3 B 3 and V 1 -ATPase (bottom). b Possible chemomechanical coupling scheme of E. hirae V 1 -ATPase [44-46, 48, 49] (left) and human F 1 -ATPase [40] (right) for 360° rotation. 0° is set as the ATP-binding angle for the catalytic site at the 12 O'clock position (green). In the model of E. hirae V 1 -ATPase, ATP bound at 0° is cleaved into ADP and Pi at 240°. Among these, phosphate first dissociates at 240°, and then, ADP release occurs at 240°. Other catalytic sites also obey the same reaction scheme offset by 120° and 240°. In the model of human F 1 -ATPase, ATP bound at 0° is cleaved into ADP and Pi at 210°, ADP dissociates at 240°, and then, phosphate release occurs at 305° 'tight' site in bV 1 (= 2 ATP V 1 ) is cleaved into ADP and phosphate. The release of phosphate, which has a lower affinity than ADP, is coupled with a conformational change from a 'tight' site to 'ADP-bound' site and from an 'empty' site to 'bindable-like' site. Consequently, the rotor DF subunits tilt toward the 'ADP-bound' site, without showing obvious rotation. Finally, it returns to the initial rotational state with 120° rotation (Fig. 10a).
Currently, there are no single-molecule studies on V 1 -ATPase from E. hirae that directly demonstrate the dwell angles for ATP cleavage, ADP release, and phosphate release at a single catalytic site over one revolution. However, considering the results of single-molecule studies and structural studies, the model for 360° rotation cycle is conceivable [48].
If the ATP-binding angle is defined as 0° in the 360° rotation cycle (Fig. 10b, left), ATP is cleaved at 240°. Then, phosphate is released first at 240°, and finally, ADP is released at 240°. In comparison, the proposed chemomechanical coupling scheme of F 1 -ATPase is more complicated. In the case of thermophilic Bacillus PS3 F 1 -ATPase, ATP cleavage, ADP release, and phosphate release occur at 200°, 240°, and 320° [37], respectively, although the timing of phosphate release is controversial. Furthermore, in the case of human F 1 -ATPase, ATP cleavage, ADP release, and phosphate release occur at 210°, 240°, and 305°, respectively (Fig. 10b, right) [40]. The coupling scheme of human F 1 -ATPase can be considered a variation of that of thermophilic Bacillus PS3 F 1 -ATPase, in which the ATP cleavage and the phosphate-release dwells are split into different angles. Of course, it is possible that V 1 -ATPase also shows the substeps, because the angles waiting for ADP release and phosphate release have not been directly demonstrated by advanced single-molecule experiments, as has been performed in the case of F 1 -ATPase [34,35,37]. Furthermore, multiscale molecular dynamics simulations predict the formation of rotational intermediate states of this V 1 -ATPase that have not yet been resolved [60]. Interestingly, the recent information-based soft clustering method revealed that thermophilic Bacillus PS3 F 1 -ATPase makes a small rotational movement during the catalytic dwell triggered by the ATP hydrolysis reaction; this had not been previously resolved using the conventional analysis methods [61]. Such advanced single-molecule techniques and analysis methods may reveal the unresolved reaction scheme and movements of V 1 -ATPase.

Future prospects
The recent high-resolution structural studies and singlemolecule studies reviewed have begun to clarify the rotation mechanism of V 1 -ATPase. By comparing the differences and similarities in the rotation mechanism between V 1 -ATPase and F 1 -ATPase, we can determine the 'unique' and 'common' mechanisms by which these rotary ATPases function and thereby establish the working principle of rotary ATPases. However, to fully understand the rotation mechanism of rotary ATPases, it is necessary to further improve our 'knowledge and understanding' by designing rotary ATPases with improved, modified, or novel functions based on our current 'knowledge and understanding,' and by experimentally verifying designed proteins. Such a synthetic approach has become a trend in biology and nanobiotechnology [62][63][64][65][66][67], and this approach will be extremely helpful to understand the mechanisms by which rotary ATPases operate.