Mechanical Modulation of ATP-binding Affinity of V1-ATPase*

Background: The ATP-binding reaction was hypothesized to be the main torque-generating step for V1-ATPase. Results: Upon mechanical manipulation, the ATP-binding reaction of V1-ATPase showed weaker angle dependence than that of F1-ATPase. Conclusion: The ATP-binding reaction is not the main torque-generating step in V1. Significance: This external manipulation technique should be applied to other reaction steps of ATP hydrolysis to get the whole view of the mechanochemical coupling mechanism in V1. V1-ATPase is a rotary motor protein that rotates the central shaft in a counterclockwise direction hydrolyzing ATP. Although the ATP-binding process is suggested to be the most critical reaction step for torque generation in F1-ATPase (the closest relative of V1-ATPase evolutionarily), the role of ATP binding for V1-ATPase in torque generation has remained unclear. In the present study, we performed single-molecule manipulation experiments on V1-ATPase from Thermus thermophilus to investigate how the ATP-binding process is modulated upon rotation of the rotary shaft. When V1-ATPase showed an ATP-waiting pause, it was stalled at a target angle and then released. Based on the response of the V1-ATPase released, the ATP-binding probability was determined at individual stall angles. It was observed that the rate constant of ATP binding (kon) was exponentially accelerated with forward rotation, whereas the rate constant of ATP release (koff) was exponentially reduced. The angle dependence of the koff of V1-ATPase was significantly smaller than that of F1-ATPase, suggesting that the ATP-binding process is not the major torque-generating step in V1-ATPase. When V1-ATPase was stalled at the mean binding angle to restrict rotary Brownian motion, kon was evidently slower than that determined from free rotation, showing the reaction rate enhancement by conformational fluctuation. It was also suggested that shaft of V1-ATPase should be rotated at least 277° in a clockwise direction for efficient release of ATP under ATP-synthesis conditions.


V 1 -ATPase is a rotary motor protein that rotates the central shaft in a counterclockwise direction hydrolyzing ATP.
Although the ATP-binding process is suggested to be the most critical reaction step for torque generation in F 1 -ATPase (the closest relative of V 1 -ATPase evolutionarily), the role of ATP binding for V 1 -ATPase in torque generation has remained unclear. In the present study, we performed single-molecule manipulation experiments on V 1 -ATPase from Thermus thermophilus to investigate how the ATP-binding process is modulated upon rotation of the rotary shaft. When V 1 -ATPase showed an ATP-waiting pause, it was stalled at a target angle and then released. Based on the response of the V 1 -ATPase released, the ATP-binding probability was determined at individual stall angles. It was observed that the rate constant of ATP binding (k on ) was exponentially accelerated with forward rotation, whereas the rate constant of ATP release (k off ) was exponentially reduced. The angle dependence of the k off of V 1 -ATPase was significantly smaller than that of F 1 -ATPase, suggesting that the ATP-binding process is not the major torque-generating step in V 1 -ATPase. When V 1 -ATPase was stalled at the mean binding angle to restrict rotary Brownian motion, k on was evidently slower than that determined from free rotation, showing the reaction rate enhancement by conformational fluctuation. It was also suggested that shaft of V 1 -ATPase should be rotated at least 277°in a clockwise direction for efficient release of ATP under ATP-synthesis conditions.
The vacuolar proton pumps, V-ATPases (V o V 1 type), are part of the ATPase/ATP synthase superfamily and share a common rotary catalytic mechanism with F 0 F 1 -ATPase (1-3). V o V 1 consists of two rotary motors: the membrane-embedded V o subunit and the water-soluble V 1 subunit, each driven by a proton flux to create a proton-motive force and by ATP hydrolysis, respectively. In cells, V o and V 1 bind to one another via the central rotor stalk and peripheral stalks and interconvert the energy liberated from ATP hydrolysis and proton translocation down the proton motive force into rotation of the central stalk (4,5). Although V o V 1 is primarily known as an ATP-driven proton pump that acidifies the inside of vacuoles in eukaryotic cells, V o V 1 also catalyzes ATP synthesis driven by the proton motive force in archaea and some eubacteria such as Thermus thermophilus.
The V 1 domain from T. thermophilus, termed V 1 -ATPase, 2 has been extensively investigated as a model enzyme for the bacterial type V 1 due to its conformational stability and ease of biochemical handling (6 -8). V 1 -ATPase is composed of a hexameric stator ring of A 3 B 3 subunits and the central rotary shaft of DF subunits. The A 3 B 3 ring possesses three catalytic sites on each A-B interface, primarily on the A subunit. The D subunit is embedded inside the central cavity of the A 3 B 3 ring, whereas the F subunit binds to a protruding segment of subunit D (9).
Although V 1 -ATPase has several rotation features in common with F 1 -ATPase, such as rotation in the counterclockwise direction and a 120°stepping rotation, the rotational mechanism of V 1 -ATPase is distinct from that of F 1 -ATPase (10). One prominent difference between the two types of ATPase is that V 1 -ATPase does not show any rotational substep (7), whereas the elementary 120°step of F 1 -ATPase is composed of 80°and 40°substeps (11). Although the process by which the three catalytic A subunits participate in driving the unidirectional rotation remains unclear, it has been suggested that V 1 -ATPase executes all of the elementary reaction steps, that is, ATP binding, ATP hydrolysis, and product release, each at a pausing position. This hypothesis implies that each elementary reaction step is responsible for contributing to the 120°rotation, in contrast to the torque generation mechanism of F 1 -ATPase, in which individual reaction steps induce either of the 80°or 40°s ubsteps. Another difference between these two ATPases is that the torque of V 1 -ATPase is ϳ35 pN⅐nm, which is slightly * This work was supported in part by Grant-in-aid for Scientific Research 18074005 (to H. N.) and by a Special Education and Research Expenses grant from the Ministry of Education, Culture, Sports, Science, and Technology, Japan (to H. N.). 1 To whom correspondence should be addressed. smaller than that of F 1 -ATPase (40 pN⅐nm) (10,12). Comparative research on V 1 -ATPase and F 1 -ATPase could clarify the common working principles and unique mechanisms of these proteins.
We have previously conducted single-molecule stalling experiments to investigate how F 1 -ATPase modulates the chemical equilibrium and reaction rate of ATP-binding and ATP-hydrolysis steps via the rotation of the central rotary shaft (13). Although both reactions were exponentially enhanced during the forward rotation, the degree of reaction enhancement was distinctive; the ATP-binding rate was largely accelerated during the forward rotation, whereas the reaction enhancement of the ATP-hydrolysis step was only slight. The affinity of F 1 for ATP was also exponentially enhanced, suggesting that the F 1 -ATP complex is stabilized upon rotation. This finding suggested that F 1 generates a much greater torque during the binding change process than during the hydrolysis step. The torque generated upon ATP binding was quantitatively estimated from the angle dependence of k off .
Single-molecule manipulation with magnetic tweezers was also utilized in the experiment on V 1 -ATPase to attempt to activate V 1 -ATPase in the ADP-inhibited form by forcibly rotating the molecule with the magnetic tweezers (14). When rotated over ϩ110°, V 1 -ATPase resumed active rotation. The activation probability was notably dependent on the angular displacement from the inhibitory pausing position, as observed in the mechanical activation of F 1 -ATPase in the ADP-inhibited form (15). This observation suggests that V 1 -ATPase also possesses the ability to modulate the catalytic reaction upon rotation, similar to that of F 1 -ATPase. In the present study, we attempted to verify this hypothesis by examining how the ATPbinding process is modulated upon rotation of V 1 -ATPase. Interestingly, V 1 -ATPase displayed demonstrably weaker angle dependence of ATP binding than F 1 -ATPase, suggesting a smaller contribution of the ATP-binding process for torque generation than that observed for F 1 -ATPase. The results have been discussed in the light of the current understanding of the mechanochemical coupling mechanisms of V 1 -ATPase and F 1 -ATPase.

EXPERIMENTAL PROCEDURES
Rotation Assay-Sample preparation and experimental procedures were performed essentially as described in our previous study (14). Wild-type V 1 that had a minimal modification for the single-molecule rotation assay, A (His-10/C28S/C508S)3 B (C264S)3 D (E48C/Q55C) F, was prepared and examined in the rotation assay. Streptavidin-coated magnetic beads (Thermo Scientific) were used as rotational markers. The beads showed relatively great diversity in diameter. The small particles ( Ϸ 200 nm) were selectively observed due to the low frequency of physical interaction with the glass surface. Rotation of the bead was observed under a phase contrast microscope (IX70; Olympus, Tokyo, Japan) by using a 100ϫ objective lens. Images were captured with a charge-coupled device camera (FC300M; Takenaka System Co., Kyoto, Japan) and recorded at 30 frames/sec. A magnetic tweezers system was mounted on the specimen stage of the microscope and controlled with custom-designed software (Celery, Library, Tokyo, Japan). Analysis of rotation was also performed using the custom-designed software (Celery). All the experiments were carried out at a temperature range of 23-25°C.

RESULTS
V 1 -ATPase was immobilized on the glass surface through His tags introduced at the N termini of the A subunits of the stator A 3 B 3 ring. Rotation of V 1 -ATPase was observed by attaching a streptavidin-coated magnetic bead to the D subunit. The magnetic beads were used also as the handle for manipulation by the magnetic tweezers. The rotation assay was conducted under ATP-limiting conditions (1 or 1.5 M), well below the Michaelis-Menten constant (K m ) of the rotation assay with magnetic beads (8.1 M) (14). Under these conditions, V 1 -AT-Pase demonstrated a 120°stepping rotation (Fig. 1A). The mean times of the ATP-waiting pause were 0.57 and 0.32 s at 1 and 1.5 M, respectively (Fig. 1B). Note that the mean time for catalysis on V 1 -ATPase was 2.5 ms (7), which was much shorter than the ATP-waiting dwell or the mean time for the 120°rotation of the beads, so that the catalytic pause was obscured and undetectable in this condition. In a recent study, we reported that V 1 -ATPase spontaneously lapses into two types of inhibitory state, pausing rotation; V 1 -ATPase undergoes a second-scale inhibitory pause during continuous rotation that is terminated with a long and stable pause. Under ATP-limiting conditions, the occurrence frequency of the second-scale inhibitory pause is Ͻ0.4% of the total pause and is hence negligible (14). The observation of the rotation was terminated when the long pause appeared.
When V 1 -ATPase showed the ATP-waiting pause, the magnetic tweezers were turned on to stall molecules at the target angle. After a defined period elapsed, the magnetic tweezers were turned off to release the molecule. V 1 -ATPase essentially demonstrated two distinct responses. In the first response, V 1 -ATPase made a 120°step to the next ATP-waiting angle immediately after release ( Fig. 2A). Because V 1 -ATPase is unable to induce rotation unless it binds to ATP, this means that V 1 -ATPase had already bound to ATP when released from the magnetic tweezers. This response was termed the "on" event. In the second response, V 1 -ATPase simply rotated back to the original ATP-waiting angle (Fig. 2B), termed the "off" event. In the "off" event, the molecules demonstrated a spontaneous 120°step to the next ATP-waiting angle. The histogram of the waiting time until the next spontaneous 120°step displayed nearly the same distribution as that observed of the ATP-waiting time during free rotation (Fig. 3A). This suggests that V 1 -ATPase returned to the ATP-waiting state in the off event. We also analyzed the ATP-waiting time of the next step after an on event. The waiting time histogram also displayed the same distribution as that observed during the free rotation (Fig.  3B). Thus, the stall-and-release manipulation affected neither the catalytic nor the kinetic properties of V 1 -ATPase, suggestive of the high robustness of V 1 -ATPase. The probability of ATP binding was measured as the probability of an on event, P on . V 1 -ATPase also displayed unclassifiable behaviors, such as pausing for an unusually long period after returning to the original ATP-waiting angle (likely to be caused by the ADP-inhibited form). However, the occurrence rate of minor behaviors was very low (Ͻ5%); therefore, they were omitted from the analysis.
The P on obtained at limiting concentrations of ATP (1 M) was plotted against the stall angle (Fig. 4A). It is evident from the graph that P on significantly depended on the stall angle. Here, we defined 0°as the mean angle for ATP-waiting pause and the "plus" direction as the rotational one (i.e. counterclockwise). When the data points were replotted against the stall time, the time course of P on was obtained for individual stall angles (Fig. 4B). The values of P on at Ϫ100°were too low to provide a reliable time-dependent increment of P on ; therefore, these data points were omitted. The time courses displayed simple saturation curves, suggesting that the single event, i.e. ATP binding, triggers rotation. This supports the above expectation that the time constant of other reactions such as ATP-hydrolysis step is too short to affect P on . Note that all of the time courses displayed plateau levels below 100%, except for the ϩ100°stall. This result implies that the ATP-binding event is reversible; during a long period of stalling, V 1 -ATPase releases ATP into the medium. This process was nearly identical to the stall-and-release experiment of F 1 -ATPase (13). The time courses were fitted on the basis of a reversible reaction scheme to determine When the molecule entered into the ATP-waiting pause, we switched on the magnetic tweezers to generate a magnetic field and rotated the magnetic bead and therefore the shaft to the target angle. Here, the bead was rotated almost 60°from the original waiting angle, stalled there for 0.5 s, and then released. Upon release, the shaft proceeded directly to the next ATP-waiting angle. This behavior implies that V 1 was bound ATP at the time of release, which generated the torque necessary for advancing to the next step. B, off event. The same molecule from A was stalled again at 60°from another ATP-waiting angle. Here, shaft returned to the original (0°) angle, indicating that V 1 was not bound ATP at the time of release.

FIGURE 3. Dwell time analysis of spontaneous ATP binding immediately after an off event (A) or an on event (B).
To determine the effect of mechanical manipulation on kinetics of V 1 -ATPase, dwell time analysis was performed for both off and on events.
the rate constant of ATP binding (k on ) as well as the rate constant of ATP release (k off ). The dissociation constant of ATP (K d ) was also determined as the ratio of k off to k on . The kinetic parameters were plotted in semi-log plots (red data in Fig. 5). In Fig. 5A, k on from free rotation was also designated (red open circle). It is evident that k on exponentially increased upon V 1 -ATPase rotation, whereas k off was exponentially reduced (Fig. 5, A and B), resulting in an exponential reduction in K d (Fig. 5C). From a Ϫ60°to ϩ60°rotation, k on increased by ϳ22fold, whereas k off decreased by 8-fold, resulting in an decrease in Kd of ϳ173-fold. For comparison, kinetic parameters of F 1 -ATPase were also included in Fig. 5 (gray data) (13).

DISCUSSION
In our previous study, the rate constants k on , k off , and K d were determined in the stall-and-release experiments of F 1 -ATPase (gray circles and lines; Fig. 5) (13). A comparison of V 1 -ATPase and F 1 -ATPase data suggests that the ATP-binding process does not contribute to torque generation in V 1 -ATPase as much as it does in F 1 -ATPase. One of the distinctive features of V 1 -ATPase data is that the ATP-binding site of V 1 -ATPase has significantly lower affinity to ATP than the ATP-binding site of F 1 -ATPase (6). In kinetic terms, the k on of V 1 -ATPase is lower than that of F 1 -ATPase over all stall angles, whereas k off and K d are higher. However, the individual data points do not provide any indication of the contribution of ATP-binding process to torque generation. More important in understanding the contributions to torque generation is the angle dependence of the kinetic parameters. In our previous report on the angle dependence of F 1 -ATPase (13), we estimated the contributions of ATP binding to torque generation from the angle dependence of k off , comparing it with that of the ATP-hydrolysis step. Torque generated by ATP binding corresponds to the slope of the rotary potential of the ATP-bound state (16). Because Ϫk B T lnk off () represents the relative energy difference between the ATP-bound state and the transition state of ATP binding/release, the differential function, Ϫk B T ((lnk()))/(), indicates the magnitude of torque generated by ATP binding. The assumption behind this estimation is that only the free energy of the ground state changes upon rotation, whereas the activa-  . Rate constants determined from the time course of ATP binding. k on (A), k off (B), and K d (C) were fitted with single exponentials. Data of F 1 -ATPase were also included for comparison (gray-colored) (13). In both V 1 and F 1 , angle dependences are apparent for all the kinetic parameters. Open circle in k on graph (A) designates the k on value in case of free rotation. From the slope of k off , the torque of ATP binding for V 1 -ATPase was calculated to yield 4 pN⅐nm per radian, which is smaller than 11 pN⅐nm per radian of F 1 -ATPase. tion energy level remains constant. The magnitude of Ϫk B T ((lnk()))/(), for V 1 -ATPase is only 38% that of F 1 -ATPase. The absolute values of estimated torque generation during the ATP-binding process were 4 pN⅐nm and 11 pN⅐nm for V 1 -AT-Pase and F 1 -ATPase, respectively. In practice, these values must be underestimated due to the elasticity of the rotary shaft (17)(18)(19). Despite this qualification, the estimates still suggest that the role of ATP binding in torque generation is not as important in V 1 -ATPase as in F 1 -ATPase, considering that the torsional rigidity of the rotor shaft estimated from the rotary fluctuation during the pausing state does not greatly differ between V 1 -ATPase and F 1 -ATPase (10,19). The conformational changes upon ATP binding of the A subunit are expected to be small compared with those of the ␤ subunit in F 1 -ATPase. The structural analysis of the A subunit with or without bound nucleotides is highly anticipated.
Interestingly, the ATP-binding rate determined from free rotation (open circle in Fig. 5A) was slightly (but noticeably) higher than the k on determined from the stalling experiment at 0°. The essentially same observation was reported for F 1 -ATPase. This observation is attributable to rate enhancement by thermal agitation; the rotary shaft of V 1 -ATPase always undergoes a thermally agitated rotary fluctuation, and an occasional large rotary fluctuation in the forward direction (counterclockwise) triggers ATP binding.
The present results also have implications regarding ATP synthesis. The K d determined at 0°was 0.7 M, which is too low to release ATP under physiological conditions where the ATP concentration is in the millimolar range. The angle dependence of K d predicts that when rotated more than Ϫ162°, the K d increases to the millimolar range, and V 1 -ATPase is able to release ATP into the medium. Moreover, the k off determined at 0°was 0.46 s Ϫ1 , which was also too slow to explain the maximum turnover rate of ATP synthesis (67-73 s Ϫ1 ) (6). The angle dependence of k off suggests that the rotary shaft would have to be rotated more than Ϫ277°to achieve the maximum turnover rate. Therefore, the present results suggest that the reaction scheme of ATP synthesis is not simply the reverse reaction of the hydrolysis scheme and that the angular dependence of kinetic and thermodynamic parameters must be taken into account (note that these values are also overestimated, considering the possible elasticity of the rotary shaft). The torsional rigidity of the rotary shaft remains to be clarified for a more precise estimation of the contribution of ATP binding to torque generation.