Eukaryotic yeast V1-ATPase rotary mechanism insights revealed by high-resolution single-molecule studies

Vacuolar ATP-dependent proton pumps (V-ATPases) belong to a super-family of rotary ATPases and ATP synthases. The V1 complex consumes ATP to drive rotation of a central rotor that pumps protons across membranes via the Vo complex. Eukaryotic V-ATPases are regulated by reversible disassembly of subunit C, V1 without C, and VO. ATP hydrolysis is thought to generate an unknown rotary state that initiates regulated disassembly. Dissociated V1 is inhibited by subunit H that traps it in a specific rotational position. Here, we report the first single-molecule studies with high resolution of time and rotational position of Saccharomyces cerevisiae V1-ATPase lacking subunits H and C (V1ΔHC), which resolves previously elusive dwells and angular velocity changes. Rotation occurred in 120° power strokes separated by dwells comparable to catalytic dwells observed in other rotary ATPases. However, unique V1ΔHC rotational features included: 1) faltering power stroke rotation during the first 60°; 2) a dwell often occurring ∼45° after the catalytic dwell, which did not increase in duration at limiting MgATP; 3) a second dwell, ∼2-fold longer occurring 112° that increased in duration and occurrence at limiting MgATP; 4) limiting MgATP-dependent decreases in power stroke angular velocity where dwells were not observed. The results presented here are consistent with MgATP binding to the empty catalytic site at 112° and MgADP released at ∼45°, and provide important new insight concerning the molecular basis for the differences in rotary positions of substrate binding and product release between V-type and F-type ATPases.

The V-ATPase belongs to the super family of rotary ATPases (Muench et al., 2011;Schep et al., 2016;Sobti et al., 2020;Vasanthakumar et al., 2019) that also include the F-type, A-type and V/A-type (Figure 1).The eukaryotic V-ATPase is composed of the integral membrane V O complex that provides the pathway for proton translocation, which is docked to the peripheral V 1 complex (V 1 V O ) (Benlekbir et al., 2012;Wang et al., 2020).The Saccharomyces cerevisiae V O complex consists of subunits a, d, e, f in addition to ten proteolipids (subunits c, c', c"), which form a ring structure (c-ring) (Mazhab-Jafari et al., 2016;Roh et al., 2018).The V 1 complex is comprised of eight different subunits A 3 B 3 CDE 3 FG 3 H (Zhang et al., 2003;Zhang et al., 2008;Benlekbir et al., 2012).Alternating subunits A and B form a catalytic hexameric ring consisting of three AB heterodimers, each with a catalytic site to hydrolyze ATP (Arai et al., 2013).The V 1 rotor (subunits D and F) is in the center of the (AB) 3 -ring and docks to the c-ring via V O subunit d for catalytic coupling.Three peripheral stalks (E/G heterodimers) connect the V 1 -ring of three AB heterodimers to the V O -proton transfer domain.The peripheral stalks provide the docking site for the V 1 regulatory H and C subunits in the V 1 V O holocomplex (Sagermann et al., 2001;Drory et al., 2004;Diepholz et al., 2008;Oot et al., 2012).Together, the, EG 1-3 heterodimers, subunit H, subunit C, and N-terminal domain of subunit a provide structural support between the ATPase and proton channel (Balakrishna et al., 2015).Subunit composition of eukaryotic V 1 V O ATPase from Saccharomyces cerevisiae (PDB-ID 6O7W), V/A 1 A O ATP synthase from Thermus thermophilus (PDB-ID 5GAR), and F 1 F O ATP synthase from E. coli (PDB-ID 6OQR).(A) The rotors of V 1 and V/A 1 are composed of subunit D and subunit F. The former has an α-helical coiled-coil that extends through the core of the (AB) 3 -ring, while the latter globular protein helps dock subunit D to the V O or V/A O rotor c-ring.Analogous to subunits D and F, subunit γ serves as the F 1 rotor, which contains a coiled-coil domain the extends through the F 1 (αβ) 3 -ring and a globular domain that docks to the c-ring of F O .The eukaryotic V 1 V O ATPase has unique regulatory subunits H and C and three peripheral stalks (EG) 3 connecting V 1 and V O , V/A 1 A O ATP synthase has two peripheral stalks (EG) 2 , and F 1 F O ATP synthase has one (b 2 ) peripheral stalk.(B) Cross section of eukaryotic V 1 from Saccharomyces cerevisiae (PDB-ID 7TMO), prokaryotic V 1 from E. hirae (PDB-ID 5KNB); and F 1 from E. coli (PDB-ID 3OAA).The helical coiled-coil of the rotors is comprised of the N-terminal helix (green) and C-terminal helix (pink) of V 1 subunit F and F 1 subunit γ and the globular foot of the rotor (cyan) is V 1 subunit D and the globular domain of F 1 subunit γ.Catalytic A subunits (V 1 ) and β subunits (F 1 ) have a helical domain (orange) and a Catch-Loop (red).The electrostatic interaction (EI) occurs between the rotor (blue) and the catch loop of the subunit-A about to release MgADP in V 1 , and between the rotor and the empty subunit β in F 1 .
The F 1 F O and A 1 A O complexes assemble into similar heterodimer rings that contain three catalytic sites surrounding a central rotor that is attached to a c-ring, which uses subunit a to facilitate proton translocation (Figure 1).However, the F-type and A-type motors have one and two peripheral stalks, respectively, and lack the eukaryotic V-type regulatory subunits H and C (Muench et al., 2011).The bacterial V/A-type ATP synthase, found in Thermus thermophilus, and bacterial V-type ATPase Na + pump found in Enterococcus hirae contain two peripheral stalks and also lack the subunits H and C that are essential for regulation of eukaryotic V-types (Muench et al., 2011;Schep et al., 2016;Vasanthakumar et al., 2019;Sobti et al., 2020).While F-, A-, and V/A-type ATP synthases can synthesize and hydrolyze ATP, eukaryotic V-type ATPases are ATP hydrolysis-dependent proton pumps in vivo (Kane, 2006;Forgac, 2007), even though Arabidopsis thaliana V 1 V O has been shown to synthesize ATP at exceptionally low ATP synthase rates in vitro (Hirata et al., 2000).
Eukaryotic V 1 catalyzes ATP hydrolysis at the nucleotidebinding domains within the interface of the subunits A and B dimers (Arai et al., 2013;Wang et al., 2020), and most residues that facilitate ATP hydrolysis reside on V 1 subunit A. Due to the different conformations of the three A subunits relative to subunit D of the rotor, these residues catalyze ATP hydrolysis via an alternating site mechanism (Kayalar et al., 1977;Boyer, 2002) where each catalytic site is in a different conformation (open/ empty state, tight/substrate bound state, and loose/product release state) at any given time.ATP hydrolysis drives V-ATPase rotation of subunits DFd and the c-ring (Hirata et al., 2003).Bacterial T. thermophilus V/A 1 was observed to rotate in an equivalent manner (Yokoyama et al., 2003).During rotational catalysis, carboxyl groups on the c-ring take up protons from the V 1 side of the membrane via the input channel in V O subunit a, which upon completing a full rotation, are deposited on the other side of the membrane via the subunit a output channel (Kawasaki-Nishi et al., 2001;Forgac, 2007;Oot et al., 2017).This enables V-ATPase-dependent proton pumping to maintain a nonequilibrium pH gradient across the cytoplasm and organellar membranes.
Eukaryotic V 1 V O is regulated by a unique mechanism among the rotary ATPase superfamily, which results in the reversible dissociation of the V 1 -ATPase complex from the V O protontransfer complex (Kane, 1995;Sumner et al., 1995;Parra and Kane, 1998;Parra et al., 2014;Hayek et al., 2019).Reversible dissociation helps to maintain cellular pH homeostasis in coordination with the metabolic state of the cell (Parra and Kane, 1998;Hayek et al., 2019).Nutrient stress conditions such as limiting glucose induce ATP hydrolysis dependent dissociation of the V 1 subunit C from V 1 V O , which in turn prompts dissociation of V 1 from V O and halts proton pumping until glucose is restored and the V 1 V O holocomplex reassembles.Upon disassembly, a conformational change of V 1 subunit H inhibits the futile hydrolysis of cytosolic ATP by the dissociated V 1 that is uncoupled from proton transport (Parra et al., 2000;Oot et al., 2016;Oot et al., 2017).In this conformation, subunit H bridges the V 1 rotor and a stator to halt rotation and trap an inhibitory Mg-ADP in one catalytic site of the autoinhibited V 1 .Glucose-dependent V-ATPase disassembly requires ATP hydrolysis (Parra and Kane, 1998), while reassembly involves the RAVE V-ATPase exclusive assembly factor (Jaskolka et al., 2021).The protein Oxr1p appears to aid in V-ATPase disassembly (Khan et al., 2022).
However, the mechanism of how rotational catalysis mediates disassembly and reassembly remains elusive.Structural studies have identified three states of S. cerevisiae V 1 V O that differ by 120 °rotational positions of the rotor relative to the asymmetric stator (Zhao et al., 2015).The rotary positions of these states are thought to correspond to the catalytic dwell positions of V 1 V O when ATP hydrolysis occurs at one of the three catalytic sites.Only one of these rotary states appears to be optimal to initiate disassembly (Mazhab-Jafari et al., 2016;Oot et al., 2016).These structures captured snapshots of individual protein conformations, but singlemolecule rotation studies are required to capture the nuances of the rotary mechanism.
Although progress has been made regarding rotational studies of bacterial V/A 1 -ATPase from T. thermophilus and V 1 -ATPase from E. hirae (Furuike et al., 2011;Iida et al., 2019), much less is known about the rotational mechanism of eukaryotic V 1 rotation.Single molecules of S. cerevisiae V 1 V O were observed to undergo ATPasedependent counterclockwise rotation as viewed from the membrane (Hirata et al., 2003), which was in the same direction as has been observed for F 1 , A 1 , and V/A 1 motors (Imamura et al., 2003;Spetzler et al., 2009;Furuike et al., 2011;Sielaff et al., 2016).In the S. cerevisiae V 1 V O experiments, the c-ring was attached to the slide and rotation was monitored by an actin filament attached to subunit G (Hirata et al., 2003).However, the drag imposed by the actin filament and the limitation of measurement time resolution obscured other rotational details of this motor including changes in its angular velocity or whether dwells interrupt rotation.To date, only the F 1 -ATPase, A 1 -ATPase and bacterial V 1 -ATPase rotary motors have been characterized by single-molecule rotation studies under conditions that resolve dwells and angular velocity changes (Spetzler et al., 2009;Ishmukhametov et al., 2010;Sielaff et al., 2016;Ragunathan et al., 2017;Iida et al., 2019).The rotary positions of ATP binding and product release have been found to differ among the F 1 , A 1 , and bacterial V 1 motors (Furuike et al., 2011;Martin et al., 2014;Suzuki et al., 2014;Sielaff et al., 2016;Martin et al., 2018;Iida et al., 2019;Kobayashi et al., 2020;Zarco-Zavala et al., 2020), although the mechanistic basis for these differences is a major unresolved question.
Årrhenius analysis of F 1 rotation indicated that the first 60 °of the 120 °power stroke resulted from release of elastic energy, which was postulated to result from interactions between the rotor coiledcoil domain and the surrounding catalytic sites (Martin et al., 2018).The electrostatic interaction between highly conserved residues of the C-terminal helix of the rotor and catch loop residues (Greene and Frasch, 2003) of the empty catalytic site is thought to contribute significantly to the elastic energy that powers the first 60 °of rotation.Notably, this electrostatic interaction in eukaryotic and prokaryotic V 1 -ATPases occurs between the N-terminus of the subunit F rotor and the catch loop of the catalytic site conformation that releases bound MgADP (Figure 1B).
Here, we characterized rotational dynamics of the eukaryotic S. cerevisiae V 1 -ATPase that lacks regulatory subunit H and subunit C using single-molecule studies with high resolution of time and rotational position.Rotation was observed in 120 °power strokes separated by dwells comparable to the power strokes and catalytic dwells observed in F 1 , A 1 , and bacterial V 1 and V/A 1 ATPases.However, the power strokes were interrupted by dwells at rotary positions that occurred most frequently 45 °and 112 °after the end of the catalytic dwell.The results support a mechanism in which the 45 °dwell results from dissociation of ADP while the 112 °dwell late in the power stroke is the result of ATP binding to the empty site.

Materials and methods
Construction of a His-tagged V 1 ΔHC and site-directed mutagenesis To generate S. cerevisiae that express the V 1 -ATPase for single molecule measurements, the regulatory subunits C and H were deleted, two cysteine substitution mutations (Y73C and T123C) were made on subunit D for biotinylation, and a 6xhis tag was added to c-terminus of subunit G for purification.
First, the genes encoding subunits C and H (VMA5 and VMA13, respectively) in the S. cerevisiae strain SF838-5Aα genomic DNA were replaced with NAT and KanMX selectable marker genes through homologous recombination, respectively.Additionally, the subunit G gene (VMA10) was replaced with the URA3 gene so the wild type subunit G would not compete for assembly in V 1 because only the mutant containing the 6xhis tag was expressed.Mutant colonies were selected by inoculating the cells on SC + nourseothricin + kanamycin-uracil plates.
To generate V 1 with a 6xhis tag on subunit G, yeast genomic DNA was isolated from SF838-5Aα cells.The subunit G gene (VMA10) was PCR amplified using primers that contained the His tag sequence, the PCR product was restriction digested, and the DNA fragment was subcloned the expression vector.Then, 5Aα VMA5Δ::NAT, VMA10Δ::URA3, VMA13Δ::KanMX cells were transformed with the recombinant plasmid.Mutant colonies were selected by inoculating the cells on SC-uracil-leucine plates.
To generate the cysteine substitution mutation for the biotinylation, the subunit D gene (VMA8) was PCR amplified from the genomic DNA, the PCR product was restriction digested, and the DNA fragment was subcloned into the pRS313 expression vector.The two cysteine substitution mutations (Y73C and T123C) were introduced through sitedirected mutagenesis.This plasmid was used as a template to PCR amplify the mutant VMA8 gene along with the HIS3 gene from the vector using primers containing the 5′ and 3' flanking regions of the VMA8 gene.Then, the double cysteine mutant VMA8-HIS3 PCR product was integrated into the genome of 5Aα VMA5Δ:: NAT, VMA10Δ::URA3, VMA13Δ::KanMX cells containing pRS315-VMA10-6His through homologous recombination.Mutant colonies were selected by inoculating the cells on SC-uracil-leucine-histidine plates.Mutations were confirmed after each step with agarose gel electrophoresis and DNA sequencing.

ATP hydrolysis assay
The rate of ATP hydrolysis by the purified V 1 ΔHC was measured with an ATP-regenerating NADH-coupled assay (Lotscher et al., 1984).The measurement was made with the final concentration of 25 mM Tris/HCl (pH 8.0), 60 mM KCl, 2.5 mM phosphoenolpyruvate, 0.3 mM NADH, 17.5 Units pyruvate kinase (rabbit muscle, Sigma Aldrich), 25 Units L-lactate dehydrogenase (rabbit muscle, Sigma Aldrich), at varying concentrations of ATP including 1, 0.5, 0.2, 0.1, 0.05, 0.02, 0.01 mM, twice the MgCl 2 concentration for each corresponding ATP concentration, and 3.22 × 10 −5 mM of purified V 1 ΔHC (0.0174 mg/mL) in a final volume of 2.5 mL.The rate was determined in three replicates as the change in absorbance at 340 nm using a Cary 100 spectrophotometer with Peltier temperature control at 25 °C.MgATP concentration was determined by the Maxchelator program MgATP calculator v1.3 using constants from NIST database #46 v8 (UC Davis Health).

Single molecule gold nanorod rotation assay
The rotation of individual V 1 ΔHC molecules were observed with a single-molecule rotation assay using gold nanorods under a dark field microscope (Spetzler et al., 2009;Martin et al., 2014).Purified V 1 ΔHC molecules were immobilized on a microscope cover slip by the His-tag on the G-subunits, unbound molecules were washed off the slide with wash buffer (30 mM Tris/HCl pH 8.0, 10 mM KCl).The surface area of the cover slip that remained exposed around the bound V 1 ΔHC molecules was then coated with BSA-C, which prevented the gold nanorods from binding nonspecifically to the surface.The 80 × 40 nm gold nanorod (A12-50-600 purchased from Nanopartz) coated with Neutravidin was bound to the biotinylated subunit D, excess gold nanorods were washed off with the wash buffer, and rotation buffer (1 mM MgCl 2 , 2 mM ATP, 30 mM Tris/ HCl pH 8.0, 10 mM KCl) was added to the cover slip (Spetzler et al., 2009).The rotations of individual molecules were observed by measuring the fluctuation of polarized red light scattered off the AuNR using a single-photon detector.In each molecule observed, the orientation of the polarizing filter was adjusted to align with the minimum light intensity position that corresponded to one of the three catalytic dwells.The sinusoidal fluctuation of the polarized red-light intensity was measured as the gold nanorod rotated from 0 °to 90 °relative to the catalytic dwell position.Measurements were taken in the form of 5 s datasets at a frame rate of 100 kHz.The standard error measurements of histograms of the intensity of red light scattered from a single nonrotating nanorod fixed to a slide as a function of the rotational position of the polarizer varies between about 0.02 and 0.12 °as the scattered light intensity varied between minimum and maximum values (Ishmukhametov et al., 2010).

Results
A strain of S. cerevisiae lacking the genes for the V 1 regulatory subunit H and subunit C was used to express the was genetically modified to add a 6xhis tag to the C-terminus, and subunit D mutations Y73C and T123C were made to enable covalent modification by biotin maleimide.The V 1 ΔHC construct exhibited ensemble ATP hydrolysis commensurate with V-ATPases and single molecule rotational ATPase activity consistent with the general mechanism of the rotary ATPases.However, V 1 ΔHC rotation differed in notable ways.

Ensemble ATPase assays
The ATPase activity of purified V 1 ΔHC versus the MgATP concentration (Figure 2A) was measured using an ensemble coupled assay with pyruvate kinase and lactic dehydrogenase at 25 °C.The apparent V max was observed at 990 μM MgATP.The double-reciprocal plot of ATPase activity versus [MgATP] was not linear (Figure 2B).As such, V 1 ΔHC did not exhibit simple Michaelis-Menten kinetics, which would have a linear dependence in a double reciprocal plot defined by Eq. 1, where v is the observed velocity at a given MgATP concentration, V max is the apparent maximum velocity and K M is the Michaelis constant.The kinetic values were determined from Eq. 1 for each of three consecutive ATP concentrations in the double-reciprocal plot.
The V max increased with MgATP concentration to a maximum turnover number of 9.6 s −1 (Figure 2C, inset).The time required to hydrolyze an ATP at 990 μM MgATP (saturating) as well as at 490 μM and 5.7 μM MgATP, which were rate-limiting, is shown in Table 1.The V max /K M values decreased 30-fold versus MgATP in a non-linear manner (Figure 2C), which indicates that the affinity of the empty catalytic site to bind MgATP decreases as V 1 ΔHC approaches saturating MgATP concentrations.At saturating MgATP concentrations, the K M was 41.7 μM.Changes in both V max (~2-fold) and K M (100-fold) contributed to the decrease in V max /K M versus MgATP.

Single molecule rotation assays
To measure ATP hydrolysis-dependent rotation of subunit D, purified V 1 ΔHC molecules were immobilized on a microscope cover slip by the His-tags on the three G subunits, and a 35 × 75 nm NeutrAvidin-coated gold nanorod (AuNR) was attached to the biotinylated Y73C and T123C mutations of subunit D (Figure 3A).Rotation of single V 1 ΔHC molecules was observed in the presence of MgATP by the changes in intensity of polarized red light that scatters from the AuNR.While concentrations of the MgATP complex used in ensemble ATPase measurements and single-molecule experiments were the same, the Mg:ATP ratio was 2:1 and 1:2, respectively.The 1:2 ratio used in the singlemolecule experiments was intentional to minimize MgADP inhibition.These measurements were also carried out in the absence of an ATP regenerating system because the small number of V 1 molecules on the cover slip did not consume significant amounts of ATP during the assay.After the scattered red light passed through a high wavelength band-pass filter to remove light of wavelengths shorter than 600 nm, and a polarizing filter that could be rotated to specific orientations, the light intensity changes were quantified by an avalanche photodiode with a 50 ns time resolution.In this manner the intensity of scattered  red light was at a minimum and maximum when the polarizer was oriented perpendicular and parallel to the long axis of the AuNR (Figure 3B).
Figure 4 shows the results of a polarizer rotation measurement when the AuNR is attached to subunit D of a single V 1 ΔHC molecule.In this experiment, the light intensity scattered from the polarizer was rotated 10 °in successive 5 s intervals for a total of 360 °.During this time, the intensity of light scattered from the AuNR was acquired by the single-photon counter at 1 kHz (equivalent to 1,000 fps).Since red light scatters specifically from the long axis of the AuNR, the scattered light intensity when subunit D was not rotating in this experiment (Figure 4A, control) varied in a sinusoidal manner relative to the rotational axis of the polarizer (Spetzler et al., 2006;Ishmukhametov et al., 2010;Frasch et al., 2022).
In the presence of 990 μM (saturating) MgATP (Figure 4B), three sinusoidal intensity curves were observed in the polarizer rotation measurement, which were off-set from each other by 120 °.The 1 kHz data acquisition rate provides a time resolution of 1 ms such that the light intensities reported the rotational position of the AuNR primarily when it was in the same position for more than 1 ms.This indicates that, in the presence of saturating MgATP, subunit D stopped at three rotary positions separated by 120 °that each lasted >1 ms.To do so, subunit D rotated between these dwells with a power stroke that occurred too fast for the 1 kHz data acquisition rate to record as much scattered light as was observed during the dwells.These dwells are hereafter referred to as catalytic dwells, since they are comparable to the duration and rotary positions of catalytic dwells observed between the 120 °ATPase-driven power strokes of F 1 , A 1 , V/A 1 and bacterial V 1 ATPases (Spetzler et al., 2006;Furuike et al., 2011;Minagawa et al., 2013;Sielaff et al., 2016;Zarco-Zavala et al., 2020).
To resolve the intermediate positions of the 120 °rotational events between the dwells observed in Figure 4, the intensity of light scattered from the AuNR was sampled at 200 kHz (5 μs per data point).Prior to the 5 s data acquisition of each V 1 ΔHC molecule, the rotational position of the polarizer was set so that the scattered light intensity was at a minimum at one of the three catalytic dwell positions.As a result, the light intensity of the subsequent power stroke increased from a minimum through a maximum as the AuNR rotated from 0 °to 90 °relative to that catalytic dwell, then decreased until it reached the next catalytic dwell upon rotating 120 °.An arcsine 1/2 function of light intensity (Sielaff et al., 2016;Martin et al., 2018) was used to calculate rotational position versus time.Changes in rotary position versus time (angular velocity) for each power stroke were then averaged and binned to every 3 °of rotation, where 0 °and 120 °refer to catalytic dwell positions.
Examples of V 1 ΔHC power strokes that were used to determine the angular velocity profiles versus rotary position are shown in Figure 5.Many V 1 ΔHC power strokes rotated continuously (Figure 5A), which are comparable to E. coli F 1 power strokes at saturating MgATP (Martin et al., 2018).However, many other V 1 ΔHC power strokes rotated in a faltering manner (Figure 5B) with frequent small oscillations.Some contained a clearly defined dwell during the first 60 °of the power strokes (Figure 5C) even in the presence of saturating MgATP, while other power strokes contained a dwell near the end of the power stroke (Figure 5D).Although the power strokes shown were observed in the presence of 5.7 μM MgATP, all four types were present at all three of the MgATP concentrations examined.
The distribution of rotary positions during the power stroke when V 1 ΔHC dwells occurred (Figure 6) shows that dwells were most commonly observed ~45 °and 112 °after the catalytic dwell.The distribution of the former dwell was significantly broader than the latter, and the peak of the distribution observed at 990, 490, and 5.7 μM MgATP appeared to occur at 45 °, 50 °, and 40 °, respectively (hereafter designated the 45 °dwell).This is the approximate rotary position at which F 1 -ATPases gives rise to an "ATP-binding" dwell when MgATP is rate-limiting (Yasuda et al., 2001;Bilyard et al., 2013;Martin et al., 2014;Sobti et al., 2021).The percentage of power strokes examined that contained a 45 °dwell increased from 24% at saturating MgATP by 6% and by a further 8% at the rate-limiting MgATP concentrations of 490 and 5.7 μM MgATP, respectively (Table 2).The 184 ± 2 μs average duration of these dwells did not change significantly at the ratelimiting MgATP concentrations.However, the occurrence of the 112 °dwell increased by ~3-fold at 5.7 μM MgATP with respect to that of the 8.5% occurrence observed at saturating MgATP.The duration of these dwells increased by 21% (391 μs) at 5.7 μM MgATP.
The rate returned to 300 °•ms -1 at ~100 °then decelerated to 100 °•m -1 (d4-deceleration) as it approached the next catalytic dwell at 120 °.The V 1 ΔHC angular velocity profile was closely similar to that of the E. coli F 1 -ATPase (Martin et al., 2018) with the exception that the angular velocity of the latter was significantly slower during the first 70 °, and was 21% slower during the spike in velocity at 90 °(Figure 7A).
The average time required for power strokes to rotate between catalytic dwells were calculated from the angular velocity profiles at each MgATP concentration (Table 1), since each profile reports the time required for each three degrees of rotation.At saturating MgATP (990 μM MgATP), the average power stroke duration was 625 μs while the average power stroke durations at the ratelimiting MgATP concentrations of 490 μM, and 5.7 μM MgATP were 745 and 910 μs, respectively.The power stroke durations at these limiting MgATP concentrations were 1.19-fold and 1.46-fold longer than that measured at saturating MgATP.These increases were comparable to the 1.16-fold and 1.53-fold longer times required to consume an ATP as determined from the ensemble ATPase measurements in the presence of 490 μM and 5.7 μM MgATP, respectively (Table 1).These results support the conclusion that the additional time required for MgATP to bind to the empty catalytic site when MgATP is rate-limiting is evident as a decrease in angular velocity during the power stroke.
The angular velocity profiles (Figure 8) were determined from the average of several thousand power strokes.Consequently, the decreases in average angular velocities observed at limiting MgATP may be due to slower rotation or may occur if a proportion of the power strokes briefly stop rotating, which would be observed as a dwell.The extent that limiting MgATP caused decreases in the angular velocity versus rotary position was determined by taking the difference angular velocity profiles at limiting MgATP from that at saturating MgATP (Figures 8A, B).When compared to the distribution of dwells (black squares) there is a clear correlation between the incidence of dwells and the decreases in angular velocity of the power stroke, but the frequency did not change as MgATP became increasingly limited.This suggests that the 45 °dwell does not result from MgATP binding.However, the largest decreases in power stroke angular velocity when MgATP is rate-limiting was observed between 80 °and 100 °when dwells did not occur.Consequently, these results indicate that the major contribution to the decrease in ATPase rate when MgATP is limiting occurs between 80 °and 120 °.
The consumption of each ATP requires a consecutive catalytic dwell and power stroke.The power strokes analyzed here are one of three required for a complete rotation of subunit D. Based on the number of power strokes analyzed for each V 1 ΔHC molecule during the 5 s data acquisition period, and the average power stroke durations obtained directly from the angular velocity profiles, the average duration of catalytic dwells was determined by subtraction of the time consumed by the power stroke from the total data acquisition time (Table 1).The average catalytic dwell durations in the presence of 990 μM, 490 μM, and 5.7 μM MgATP were calculated to be 7.2 ms, 12.3 ms, and 13.4 ms, respectively.The average durations of the power strokes and catalytic dwells were consistent with the polarizer rotation results that have a minimum time resolution of 1 ms (Figure 4).
The average time required to consume an ATP molecule was also calculated from the sum of the average durations of the power stroke and the catalytic dwells (Table 1).In the presence of 990, 490, and 5.7 μM MgATP, the average times to consume an ATP were  calculated from the single-molecule data to be 7.8, 13.0, and 14.3 ms, respectively.It is noteworthy that the times required to consume ATP as measured by the ensemble ATPase assay (Figure 2) at 990, 490, and 5.7 μM MgATP were 105.2, 118.9, and 200 ms (Table 1).These times were considerably longer than those determined by other single-molecule studies, where each molecule was known to be undergoing ATPase-dependent rotation.The ensemble assays reported the average of many V 1 ΔHC molecules without knowing how many molecules are actively consuming ATP.By comparing the ensemble and single-molecule results, we estimate that 7%-11% of the V 1 ΔHC molecules were actively consuming ATP at any moment in the ensemble assay (Table 1).

Discussion
The single-molecule results of eukaryotic V 1 ΔHC ATPasedependent rotation presented here are consistent with a mechanism in which subunits D and F rotate in 120 °power strokes separated by catalytic dwells when ATP hydrolysis occurs.This is supported by the presence of 120 °power strokes that last for 0.63 ms-0.91 ms separated by longer 7.2-13.4ms duration dwells.These power strokes have the same velocity profile as those of the F 1 -ATPases and A 1 -ATPases examined except for the magnitudes of the velocities (Sielaff et al., 2016;Ragunathan et al., 2017;Martin et al., 2018).These longer dwells are also consistent with catalytic dwells observed in F 1 -ATPases.These mechanistic features are shared by all members of the super family of rotary ATPases examined to date (Figure 9) even though these rotary ATPases have been found to vary in the rotational positions where ATP binds, as well as where ADP and Pi are released.The results presented here support the mechanism of the V 1 ΔHC shown in Figure 9A, and provide important new insight concerning the molecular basis for the differences in rotary positions of substrate binding and product release among these rotary motors.
The V 1 ΔHC rotation differed in notable ways from the more thoroughly studied F 1-ATPase.First, rotation during the V 1 ΔHC power strokes often faltered with small back and forth oscillations during the first 60 °of rotation subsequent to the catalytic dwell.Second, the V 1 ΔHC power strokes often contained a dwell occurring 40 °-50 °and/or a dwell at 112 °after the catalytic dwell (8 °before the subsequent catalytic dwell).Third, as the MgATP concentration became increasingly limited, the duration of the power stroke lengthened not only as the result of an increase in the occurrence and/or duration of 45 °and 112 °dwells, but also because the angular velocity decreased between 80 °and 100 °when dwells did not occur (Figure 8).
Interactions between the rotor and the catalytic sites provide clues to the differences in V 1 ΔHC rotation (observed here) from that of F 1 ATPases.In both ATPases, the rotors are surrounded by contacts at the tip of the C-terminal helical domains (CHDs) of each subunit in the (AB) 3 -ring and by the (αβ) 3 -ring, where the catalytic sites are primarily located on the A-subunits and β-subunits.Both V 1 and F 1 have a strong electrostatic interaction between highly conserved catch-loop residues from one of the three catalytic sites with their respective rotors.These residues are A/D422, A/S424, and A/D425 and βD301, βT304, and βD305 in S. cerevisiae V 1 , and E. coli F 1 , respectively.However, the location of the rotor residues that form the electrostatic interaction differs significantly.In S. cerevisiae V 1 -ATPase, these residues (D/R12) are on the shorter N-terminal helix of the subunit D coiled-coil (Vasanthakumar et al., 2022) while these F 1 residues (γQ269 and γR268 in E. coli) are on the longer C-terminal helix of the coiled-coil.As a result, V 1 subunit D forms electrostatic interactions with the catch-loop of the subunit A catalytic site that is about to release ADP (loose conformation), while F 1 subunit γ interacts electrostatically with the catch-loop of the empty β subunit (open conformation).Mutations of any of the residues that comprise this electrostatic interaction in E. coli F 1 and S. cerevisiae V 1 -ATPase result in dramatic losses of catalytic activity (Greene and Frasch, 2003;Boltz and Frasch, 2006;Arsenieva et al., 2010).
Elastic coupling powered by ATP hydrolysis provides an explanation for faltering rotation (throughout first 60 °) during S. cerevisiae V 1 ΔHC power strokes, which result from small oscillations and the occurrence of 45 °dwells even at saturating MgATP.In F 1 , the restraints on the rotor imposed by the surrounding CTHs and by the catch loop interactions impose elastic strain by twisting the coiled-coil of the rotor.Unwinding this coiled-coil spring then powers the first 60 °of rotation (Martin et al., 2018;Frasch et al., 2022).If the first 60 °of V 1 rotation is also powered by unwinding the coiled-coil of the rotor, the spring constant is likely to be decreased due to the subunit D short helix location and its electrostatic interaction to the ADP release catalytic site.
The V 1 subunit D electrostatic link to the catalytic subunit conformation that releases ADP also suggests that the 45 °dwells result from the dissociation of ADP.E. hirae V-ATPase has equivalent subunit D residues forming the electrostatic interaction with the ADP-release conformation (Suzuki et al., 2016).Of the rotary ATPases studied to date (Figure 9B), the E. hirae V-ATPase appears most closely related to eukaryotic V-ATPases.E. hirae V 1 V O is an ATPase-dependent Na + pump that is incapable of ATP synthesis.Single-molecule studies of purified E. hirae V 1 show the presence of a dwell that occurs 40 °after the catalytic dwell, which results from ADP dissociation, while ATP-binding occurs at the catalytic dwell (Iida et al., 2019).It is noteworthy that ADP-dependent rotational backsteps are observed during E. hirae V 1 power strokes, which are much larger than the small oscillations of the V 1 ΔHC power strokes observed here.Recent structures of the T. thermophilus A/V 1 , which also have the subunit D electrostatic interaction at the catalytic site that releases ADP indicate that ADP and Pi dissociate during the 120 °power stroke, although at positions that are undefined to date (Figure 9B).
The V 1 ΔHC results presented here are consistent with MgATP binding to the empty catalytic site at 112 °.The duration and occurrence of the 112 °dwell increased when MgATP became rate-limiting, and large decreases in angular velocity were observed during the final 60 °of rotation.Supporting further this conclusion, the duration of the 45 °dwells did not increase with decreasing MgATP, even though the occurrence of these dwells increased to some extent.E. hirae V 1 single-molecule studies suggested that ATP binding occurs during catalytic dwells (Iida et al., 2019).It is noteworthy that the S. cerevisiae V 1 ΔHC catalytic dwell duration did not change significantly when the major decrease in ATP hydrolysis occurred (from 490 μM to 5.7 μM MgATP).However, dwell duration did increase by 1.86-fold relative to 990 μM MgATP (Table 1) suggesting that some V 1 ΔHC molecules may bind MgATP during the catalytic dwell 8 °later when MgATP is saturating.
The 112 °dwell is unique in several ways.Its frequency of occurrence is MgATP dependent.The dwell occurrence is significantly more frequent when MgATP concentration is limiting at 5.7 μM (23.6% occurrence) than saturating at 990 μM (8.5% occurrence) (Table 2).In addition, the dwell is distinctly preceded by a major angular velocity reduction (Figure 8) that is remarkably steep at limiting MgATP.A nucleotide binding dwell 112 °after the catalytic dwell has not been observed in other rotary ATPases to our knowledge.The eukaryotic V-ATPase is uniquely regulated by reversible disassembly of V 1 and V O , which is an important regulatory mechanism that requires V 1 V O ATP hydrolysis (Parra and Kane, 1998) that traps the dissociated V 1 complex in a specific rotational state (Oot et al., 2016;Vasanthakumar et al., 2022).More work is required to determine whether this distinct 112 °dwell is a functional adaptation of the eukaryotic rotary V-ATPases, and/or has been observed here as the result of increased resolution of our single-molecule assay.
Developing therapies that target V 1 V O -ATPase with precision requires understanding dynamics of V-ATPase rotation at high resolution at a molecular level.Here we report single-molecule rotation studies of the yeast V 1 -ATPase complex with high resolution of time and rotational position.Single molecules of V 1 -dependent rotation occurred in 120 °power strokes similar to those of other rotary ATPases.However, these 120 °rotational steps were interrupted by dwells at 45 °and 112 °when the product (ADP) was released and a new substrate (ATP) bound, respectively.This nucleotide binding sub-step at 112 °, which may be unique to eukaryotic V 1 -ATPases, was distinctly preceded by a major reduction in angular velocity.This is important because current V-ATPase inhibitors immobilize the V 1 V O assembled state, although V-ATPases are regulated by reversibly disassembling V 1 and V O in vivo.These results will help to design drugs that target a specific rotational sub-step to prevent reassembly and trap a disassembled and naturally inhibited state, leading the way in the development of a new generation of treatments that reversibly control V-ATPase function.

FIGURE 2 V 1
FIGURE 2 V 1 ΔHC ATPase Ensemble Assays.(A) Rate of ATP hydrolysis vs. MgATP concentration at 25 °.(B) Double reciprocal plot of ATP hydrolysis rates.Values of K M and V max were determined from the intersects on the x and y-axes (−1/K M and 1/v), respectively, of a straight line of each three consecutive MgATP concentrations.(C) Values of V max /K M vs. [MgATP].Inset: V max values vs. [MgATP].

FIGURE 3
FIGURE 3Use of red light scattered from a 75 × 35 nm gold nanorod (AuNR) for single-molecule rotation measurements.(A) Molecules of V 1 ΔHC were attached to a cover slip via His-tags on subunit E, and an AuNR was attached to each V 1 ΔHC via biotinylated cysteines on subunit D created by mutagenesis (Y73C and T123C).Red scattered light was recorded by an avalanche photo diode after passing through a pinhole to allow light from a single AuNR, a polarizer, and a band-pass filter to remove shorter wavelengths of light.(B) Intensity of red light from a single AuNR when the short and long axes of the AuNR are parallel to the direction of the polarizer.

FIGURE 4
FIGURE 4Eukaryotic V 1 ΔHC catalyzes ATPase-dependent 120 °power strokes that occur on a sub-ms time scale that are separated by ms-duration dwells.(A) Consecutive histograms of light intensity scattered from a AuNR attached to a molecule of V 1 ΔHC that was not rotating show a single sinusoidal dependence.(B) Consecutive histograms of light scattered from a AuNR attached to a molecule of V 1 ΔHC catalyzing ATP-dependent rotation at saturating MgATP concentration show three overlapping sinusoidal intensities off-set by 120 °.Light intensity was collected from the AuNR at 1 kHz (data collected at 1 ms intervals) after each 10 °rotation of the polarizer.

FIGURE 5
FIGURE 5Examples of V 1 ΔHC power strokes plotted as rotary position vs. time.Scattered red light intensity from the AuNR was collected at 200 kHz from power strokes that began at the minimum light intensity at the end of a catalytic dwell (0 °) and passed through the maximum light intensity (90°).Red arrows indicate the position at which power stroke rotation was clearly interrupted by a dwell.(A) Power strokes that rotated continually for 120 °. (B) Power strokes that rotated in a faltering manner with small oscillations.(C) Power strokes interrupted midway by a dwell.(D) Power strokes interrupted by a dwell near the end.(E) Power strokes with dwells midway and near the end.

FIGURE 6
FIGURE 6Observation of dwells occurring maximally at ~45 °and 112 °during a power stroke versus MgATP concentration.Distribution of rotational positions of dwells observed at 990 μM ( ), 490 μM ( ), and 5.7 μM ( ) MgATP.The frequency that the dwell occurred at a rotational position is shown for the subset of power strokes that contained a dwell that interrupted the power stroke.

FIGURE 7
FIGURE 7Average angular velocity vs. rotational position of V 1 ΔHC (black) versus E. coli F 1 (blue) power strokes at 990 μM Mg-ATP (A), 490 μM MgATP (B), and 5.7 μM MgATP (C).Rotational positions of decelerations (d) and accelerations (a) are numbered in the order that they occur during the power stroke.

FIGURE 8
FIGURE 8Rotary positions where decreases in average angular velocity occur at limiting versus saturating MgATP.Extent of rate decreases between 490 and 990 μM MgATP (A) and between 5.7 μM versus vs. 990 μM MgATP (B).Distribution of the dwell occurrence that interrupts the power stroke (black squares) at 490 μM MgATP (A) and at 5.7 μM MgATP (B) where the scale of dwell frequency was normalized to the decreases in angular velocity of the 45 °dwell.

TABLE 1
Durations of V 1 ΔHC power strokes, dwells and ATP consumption derived from single-molecule measurements compared to those derived from the ensemble ATPase assay versus MgATP concentration.
a Number of V 1 ΔHC, molecules examined.b Average ATP, consumption time (sum of power stroke and catalytic dwell durations).c Average ensemble ATP, consumption time (1/apparent V max ).d Percent of V 1 ΔHC, molecules actively consuming ATP, in ensemble assay (single-molecule ms/ATP/ensemble ms/ATP).

TABLE 2
Occurrence and duration of dwells that interrupt the V 1 ΔHC power stroke versus MgATP concentration.
a Percent occurrence of dwell versus total power strokes analyzed at each MgATP, concentration.