European Biophysics Journal

, Volume 42, Issue 2, pp 147–158

Estimating the rotation rate in the vacuolar proton-ATPase in native yeast vacuolar membranes


  • Csilla Ferencz
    • Institute of BiophysicsBiological Research Centre
  • Pál Petrovszki
    • Institute of BiophysicsBiological Research Centre
  • Zoltán Kóta
    • Institute of BiophysicsBiological Research Centre
  • Elfrieda Fodor-Ayaydin
    • Institute of BiophysicsBiological Research Centre
    • Institute of BiochemistryBiological Research Centre
  • Lajos Haracska
    • Institute of GeneticsBiological Research Centre
  • Attila Bóta
    • Department of Biological Nanochemistry, Institute of Molecular PharmacologyResearch Centre for Natural Sciences
  • Zoltán Varga
    • Department of Biological Nanochemistry, Institute of Molecular PharmacologyResearch Centre for Natural Sciences
  • András Dér
    • Institute of BiophysicsBiological Research Centre
  • Derek Marsh
    • Max Planck Institute for Biophysical Chemistry
    • Institute of BiophysicsBiological Research Centre
Original Paper

DOI: 10.1007/s00249-012-0871-z

Cite this article as:
Ferencz, C., Petrovszki, P., Kóta, Z. et al. Eur Biophys J (2013) 42: 147. doi:10.1007/s00249-012-0871-z


The rate of rotation of the rotor in the yeast vacuolar proton-ATPase (V-ATPase), relative to the stator or steady parts of the enzyme, is estimated in native vacuolar membrane vesicles from Saccharomyces cerevisiae under standardised conditions. Membrane vesicles are formed spontaneously after exposing purified yeast vacuoles to osmotic shock. The fraction of total ATPase activity originating from the V-ATPase is determined by using the potent and specific inhibitor of the enzyme, concanamycin A. Inorganic phosphate liberated from ATP in the vacuolar membrane vesicle system, during ten min of ATPase activity at 20 °C, is assayed spectrophotometrically for different concanamycin A concentrations. A fit of the quadratic binding equation, assuming a single concanamycin A binding site on a monomeric V-ATPase (our data are incompatible with models assuming multiple binding sites), to the inhibitor titration curve determines the concentration of the enzyme. Combining this with the known ATP/rotation stoichiometry of the V-ATPase and the assayed concentration of inorganic phosphate liberated by the V-ATPase, leads to an average rate of ~10 Hz for full 360° rotation (and a range of 6–32 Hz, considering the ± standard deviation of the enzyme concentration), which, from the time-dependence of the activity, extrapolates to ~14 Hz (8–48 Hz) at the beginning of the reaction. These are lower-limit estimates. To our knowledge, this is the first report of the rotation rate in a V-ATPase that is not subjected to genetic or chemical modification and is not fixed to a solid support; instead it is functioning in its native membrane environment.


ATPaseConcanamycinF-ATPaseNative membraneRotary enzymeV-ATPase


The proton-translocating adenosine-triphosphatase (first observed in vacuolar membranes, hence, called vacuolar proton-ATPase or V-ATPase) is nature’s most versatile and most universal proton pump, found in all eukaryotes (Finbow and Harrison 1997; Nishi and Forgac 2002; Cipriano et al. 2008; Jefferies et al. 2008). Similar to the more familiar and related F-ATP-synthase (F-ATPase) there are three catalytic sites—here for ATP hydrolysis—in the water-soluble V1 domain (F1 domain in F-ATPase), and transmembrane proton transport takes places in hydrophilic channels at the interface between the “c-ring” and subunit a of the membrane bound V0 (F0) domain (Wilkens et al. 1999; Grabe et al. 2000; Kawasaki-Nishi et al. 2001a, b; Wang et al. 2004; Beyenbach and Wieczorek 2006) (see Fig. 1). Crucial for proton transport are the unique glutamic acid residues, one on each c-subunit: binding, e.g., dicyclohexylcarbodiimide to this glutamic acid blocks both proton transport and ATP hydrolysis (Noumi et al. 1991; Hirata et al. 1997), proving that catalysis and transport are strongly coupled (Kawasaki-Nishi et al. 2001a). In both enzymes, this coupling involves a rotation of the rotor relative to the rest of the protein that can be considered as the stator (Yasuda et al. 1997; Fillingame et al. 2000; Futai et al. 2000; Yasuda et al. 2001; Rondelez et al. 2005; Ueno et al. 2005). The rotor versus stator subunits of the V-ATPase do not correspond exactly with those of the V0 versus V1 domains, because subunits a and d of V0 belong to the stator, and subunits D and F of V1 belong to the rotor (all other subunits of V0 and V1 belong to the rotor and stator, respectively) (Boekema et al. 1997; Ubbink-Kok et al. 2000).
Fig. 1

A membrane-bound molecular rotary engine, the vacuolar proton-ATPase (V-ATPase). The subunits of the water-soluble V1 domain are shown in green, whereas those of the membrane-bound V0 domain are shown in orangegrey colours. For better visibility, subunit a is transparent, not all subunits of the two domains are shown, and the c, c′ and c″ subunits are not differentiated. Missing are the d and e subunits of V0 and the C, E, G, and H subunits of V1. Binding and hydrolysis of one ATP molecule drives a 120° rotation of the rotor (consisting of the “c-ring”, a hexameric assembly of c-subunits (4 copies), c′ and c″ and the d subunit of the V0 domain plus D, F of the V1 domain) with respect to the stator. This results in the transport of two protons from the cytoplasmic side, either to intracellular compartments or to the lumen (depending on the cellular location of the V-ATPase), via the hydrophilic input and output channels formed between subunit a and the c-ring

Rotation is needed to bring protons, bound at the protonated glutamic acid, from the input channel to the output channel, through the hydrophobic interface between the lipid matrix and the c-ring (Fig. 1). In the case of the V-ATPase, rotation is driven by ATP binding and hydrolysis. One ATP molecule drives a 120° rotation of the rotor and transport of two protons from the cytoplasmic side to the “other” side, which can be the lumen of intracellular organelles or the extracellular space, depending on the cellular location of the V-ATPase (Finbow and Harrison 1997; Nishi and Forgac 2002; Beyenbach and Wieczorek 2006; Cipriano et al. 2008; Jefferies et al. 2008). This stoichiometry is different from that of the F-ATPase, in which there are ~12 c-subunits, which have only two transmembrane helices, each with a unique proton-binding aspartic acid. So in F-ATPase, synthesis of one ATP is driven by a 120° rotation, which, however, requires movement of four c-subunits, hence, four protons (Van Walraven et al. 1996; Panke and Rumberg 1997; Stock et al. 1999; Ferguson 2000; Seelert et al. 2000; Stahlberg et al. 2001). In the case of the F-ATPase, rotation and ATP synthesis are driven by a transmembrane pH gradient, but it should be noted that both enzymes can work in either direction depending on the conditions (Yoshida et al. 2001; Itoh et al. 2004; Rondelez et al. 2005; Feniouk et al. 2007; Nakano et al. 2008). Another difference between these related rotary engines is that, whereas the c-ring of the F-ATPase is built from identical c-subunits, each having two transmembrane alpha helices (Dmitriev et al. 1999; Fillingame et al. 2000), the c-ring in V-ATPase consists of four c-subunits and one copy of c′ and c″ subunits each \( (c_4 c_{1}^{\prime} c_{1}^{\prime\prime}) \) (Hirata et al. 1997; Powell et al. 2000), although \( c_3 c_{1}^{\prime} c_{2}^{\prime\prime}\) assemblies are also reported (Gibson et al. 2002). The reason for this heterogeneity is not known but it might have to do with the regulation of the V-ATPase and the fact that, in different tissues, the c-ring appears in roles such as, e.g., gap junctional and neurotransmitter-release channels, or parts of the membrane fusion machinery, that are unrelated to the V-ATPase activity (Holzenburg et al. 1993; Baars et al. 2007; El Far and Seagar 2011; Strasser et al. 2011). In addition, subunit c″ has five transmembrane helices (Hirata et al. 1997; Gibson et al. 2002). Nevertheless, despite the difference in size between the c-subunits of F0 and V0, the first and second pairs of transmembrane helices for the V-ATPase are highly homologous to one another (although the unique glutamic acid is present only in helix 4) and to the two-helix c-subunit of the F-ATPase. Indeed, these and other highly homologous proteins belong to the same class, that were termed ductins (Finbow et al. 1994, 1995; Dunlop et al. 1995; Saito et al. 1998; Bohrmann and Bonafede 2001). In a number of studies on a 16-kDa gap junctional protein isolated from lobster (Nephrops norvegicus), which is not only a ductin protein but also can substitute functionally for subunit c of the yeast V-ATPase in a hybrid construct (Finbow et al. 1993; Finbow and Harrison 1997), we have shown that the c-ring is a hexameric assembly of 4-helix transmembrane bundles (Holzenburg et al. 1993; Pali et al. 1995), have determined the vertical membrane location of the unique glutamic acid and shown it to contact lipids (Pali et al. 1997, 1999), and have also discovered a divalent metal-ion binding site (Pali et al. 2006).

Specific inhibitors of the V-ATPase are important because the enzyme is a potential therapeutic target in certain diseases, viz., osteoporosis, deafness and cancer (Farina and Gagliardi 1999; Bowman and Bowman 2005; Lu et al. 2005; Morimura et al. 2008; Otero-Rey et al. 2008; Supino et al. 2008; Hinton et al. 2009; Perez-Sayans et al. 2009; McHenry et al. 2010; Nishisho et al. 2011). The macrolide antibiotics concanamycin A and bafilomycin are the most selective and most potent inhibitors of the V-ATPase, with IC50 values extending down to the nM region (Bowman et al. 1988; Farina and Gagliardi 1999; Gagliardi et al. 1999; Huss et al. 2002; Dixon et al. 2008). We have demonstrated that concanamycin A and its synthetic indole analogues (Dixon et al. 2003) incorporate readily into membranes (Dixon et al. 2004; Pali et al. 2004a), and interact with amino acid side chains of the 16-kDa lobster protein and also with subunit c of yeast V-ATPase (Pali et al. 2004b; Dixon et al. 2008). It was not the purpose of the latter studies to probe the interaction of the inhibitors with subunits c′, c″ and a of the yeast V-ATPase, but they do interact with polypeptides based on sequences of the putative interfacial transmembrane helices of the c-ring and subunit a (Kota et al. 2008). These inhibitors have also been shown to perturb the lipid-protein interface around the c-ring (Pali et al. 2004b).

The most basic question with respect to the rotary mechanism, namely what is the rate of rotation of the rotor, is a difficult one to answer, because it can be measured directly only if the enzyme is modified to a great extent. All direct measurements (see (Nakanishi-Matsui et al. 2010) for a recent review), on both V-ATPase and F-ATPase, rely on gene-engineered modification of the enzyme in order to undertake single-molecule fluorescence resonance energy transfer experiments, or, more commonly, to attach a macroscopic probe, e.g., fluorescent filament or polystyrene or gold bead, to the rotor (or stator), and to fix the stator (or rotor) on a solid support, so as to allow visualisation of the rotation (Noji et al. 1997; Tsunoda et al. 2001; Nishio et al. 2002; Hirata et al. 2003; Nakanishi-Matsui et al. 2006; Xie 2009; Sekiya et al. 2010; Furuike et al. 2011; Kohori et al. 2011; Okuno et al. 2011). Such modifications may add or remove barriers to the rotation that are not present in the native system. Consequently, the reported rotation rates vary greatly in such studies, depending on the genetically engineered construct and on details of the artificial environment of the enzyme (Panke and Rumberg 1997; Masaike et al. 2000; Itoh et al. 2004; Imamura et al. 2005; Adachi et al. 2007; Takeda et al. 2009). Such direct studies are not possible on a native enzyme in a native membrane but, because the ATP/rotation stoichiometry is known, one could assay inorganic phosphate liberated from ATP hydrolysis by the V-ATPase in unit time and then relate this to the concentration of V-ATPase. However, reliable estimates of the rotation rate in the native V-ATPase in a native membrane by this method are still lacking because of several problems with this approach: (1) how to separate the activity of V-ATPase from other ATPases, (2) how to determine its concentration, (3) how to determine the fraction of inactive V-ATPases? [In some cases, as much as tenfold differences are reported by direct and indirect methods, based on activity measurements with the F-ATPase, even in the same reconstituted system, for which the enzyme concentration is known. This discrepancy was attributed to a fraction of as high as 90 % inactive F-ATPase (Ueno et al. 2005; Nakanishi-Matsui et al. 2006; Nakanishi-Matsui et al. 2010).]

In the present study, ATPase activity was measured in yeast vacuolar membrane vesicles. The activity of the V-ATPase was differentiated from that of other ATPases by using the specific V-ATPase inhibitor concanamycin A. Typically, about 60 % of the total activity comes from the V-ATPase in our system. Assay conditions were standardised, and the absorbance was calibrated for phosphate liberated in unit time. The concentration of the enzyme was determined from inhibitor titration experiments, which were fitted with a model of one inhibitor binding site per V-ATPase. Combination of the data yields a rate of 10 Hz for the full 360° rotation of the rotor, hence, the complete catalytic cycle. This estimate is a lower limit, when one considers the, probably less than 100 % activity of both inhibitor and V-ATPase (even in the absence of inhibitor).

Materials and methods


The lyticase enzyme (crude, from Arthrobacter luteus), concanamycin A, bafilomycin, ascorbic acid, sodium dodecyl sulphate (SDS), DL-Dithiothreitol (DTT), sodium orthovanadate (Na3VO4), 2-(N-Morpholino)ethanesulfonic acid hydrate (MES), ficoll-400 were purchased from Fluka (Sigma), Na2ATP (adenosine 5′-triphosphate disodium salt hydrate) from Serva. The D(−)-sorbitol, yeast extract, glucose and peptone were purchased from Molar, Tris(hydroxymethyl)aminomethane (TRIS), sodium azide (NaN3) and ammonium molybdate tetrahydrate from Reanal (Hungary). All chemicals were of analytical grade purity.

Yeast cell culture

Cells (Saccharomyces cerevisiae EMY 74.7) were grown in YPD medium (2 % glucose, 2 % peptone and 1 % yeast extract) at 30 °C in a water bath shaker (New Brunswick Scientific Co. Inc.) with continuous agitation. For isolation of vacuolar membrane vesicles, overnight cultures were diluted to OD600 = 0.1 in 1 litre YPD media, and cells were harvested when cultures reached OD600 = 0.8–1.

Isolation of vacuoles and preparation of vacuolar vesicles

Preparation of yeast vacuolar membrane vesicles was based on published methods (Ohsumi et al. 1983; Uchida et al. 1985) with some modifications. Briefly, exponentially growing cells were harvested by centrifugation at 5,400 rpm for 30 min at 4 °C in a GSA rotor (Sorvall rotors were used in a Sorvall RC-5C centrifuge unless stated otherwise). The pellet was resuspended in TRIS buffer (100 mM TRIS, 10 mM DTT, 10 mM NaN3, pH 9.4) and centrifuged in an SS-34 rotor at 6,500 rpm for 5 min at 4 °C. The pellet was resuspended in spheroplast buffer (1.5 M sorbitol, 50 mM TRIS, 2 mM MgCl2, 10 mM NaN3, pH 7.2), which was supplemented with lyticase to a final concentration of 1 unit/ml. The suspension was incubated at 30 °C for 90 min, during gentle shaking. Spheroplasts were cooled on ice, layered on top of 5 ml 1.9 M sorbitol and centrifuged in an HB-4 rotor at 2,100 rpm for 5 min at 4 °C. The spheroplasts so recovered were washed by centrifugation (3,800 rpm, 5 min at 4 °C) with 1 M sorbitol in the same rotor. The pellet was resuspended in buffer A (10 mM MES/TRIS, 0.1 mM MgCl2, 12 % Ficoll-400, pH 6.9) supplemented with protease inhibitor cocktail tablet (Roche). The resulting mixture was then homogenised by 30 strokes with a Dounce homogeniser and centrifuged at 5,200 rpm for 5 min at 4 °C in an HB-4 rotor. For the isolation of vacuoles, 5 ml of the supernatant was transferred into an ultracentrifuge tube, 5 ml of buffer A was gently layered on top, and centrifuged in a TH-641 rotor at 12,500 rpm for 30 min at 4 °C (Sorvall Discovery 90SE ultracentrifuge by Hitachi). Vacuoles, the white layer on the top, were recovered. Vacuolar membrane vesicles were formed following osmotic shock by diluting the vacuole suspension tenfold with buffer C (10 mM MES/TRIS, 5 mM MgCl2, 25 mM KCl, pH 6.9) during gentle shaking. The vesicle suspension was centrifuged in an AH-627 rotor at 15,000 rpm for 30 min at 4 °C, in the same centrifuge. The vesicles were collected as the pellet in the same buffer. The total protein content was determined (Lowry et al. 1951) and the vesicles were stored at −80 °C in buffer C and 10 % glycerol.

Light microscopy of yeast vacuoles

Transmitted light mode microscopy images of the vacuole suspensions were observed using differential interference contrast optics of an Olympus IX81 inverted microscope. Images were captured with an F View 12-bit monochrome CCD camera, using Olympus Cell-R software and a UPlanSApo 60× oil immersion objective with a numerical aperture of 1.35.

Freeze-fracture electron microscopy of vacuolar membrane vesicles

Freeze-fracture electron microscopy was used for direct visualisation of the membrane structures derived from the vacuoles. Glycerol was added to the vesicle dispersion as cryoprotectant at a final concentration of 20 % by volume. Addition of glycerol does not alter the bilayer structure, but inhibits the aggregation and ice crystal damage of the vesicles during the freezing process (Hope et al. 1986; Severs 2007). The gold sample holders used in freeze fracture were pre-incubated at 24 °C, the same temperatures as that of the samples. Droplets of 1–2 μl of the sample were pipetted onto a gold sample holder and frozen by plunging immediately into partially solidified Freon for 20 s, and then stored in liquid nitrogen. Fracturing was performed at −100 °C in a Balzers freeze-fracture device (Balzers BAF 400D, Balzers AG, Vaduz, Liechtenstein). Replicas of the fracture faces, etched at −100 °C for 30 s, were made by platinum–carbon shadowing, then cleaned with a water solution of surfactant and washed with distilled water. The replicas were placed on 200-mesh copper grids and examined in a Morgani 268D (FEI, Eindhoven, Netherlands) transmission electron microscope.

ATPase activity assay

Vesicle suspensions with 300 μg of total protein were used to assay ATPase activity in 250 μl of assay mixture. Our activity assay and the use of inhibitors was based on (Serrano 1978; Clelland and Saleuddin 2000; Lunde and Kubo 2000; Padilla-Lopez and Pearce 2006), with modifications, as follows. The assay mixture contained the activity buffer (50 mM MES/TRIS, 5 mM MgCl2, pH 7.0), 5 mM sodium azide (to inhibit mitochondrial ATPase), 0.2 mM ammonium molybdate (to inhibit acid phosphatases), 100 μM sodium orthovanadate (to inhibit plasma membrane proton-ATPase). In some of the substrate titration experiments (Fig. 4), the MgCl2 concentration was set to be twice that of Na2ATP. Except for the inhibitor titration experiment (Fig. 6), the final concentration of the specific V-ATPase inhibitor concanamycin A was 1 μM. Vacuolar vesicles were incubated in the activity buffer at room temperature for 30 min. ATP hydrolysis was started by adding 2 mM Na2ATP and proceeded at 30 °C for 20 min (Figs. 3, 4), or at 20 °C for 10 min for the inhibitor titration experiment (Fig. 6), or for variable periods in the time dependence (Fig. 5). The reaction was stopped with a solution containing 0.5 % SDS, 2 % H2SO4, 0.5 % ammonium molybdate tetrahydrate and 10 % ascorbic acid. The acid was added to start colour development in the molybdate reaction which follows inorganic phosphate production, a product of ATP hydrolysis. The vesicle suspension was incubated in this stopping buffer at room temperature for 15 min. Absorbance was measured at 750 nm in a Thermo Spectronic (USA) spectrophotometer. All data analysis and curve fitting used the IGOR scientific graphing and data analysis program (WaveMetrics, Lake Oswego).

Results and discussion

Figure 2a shows a typical bright field, transmitted light microscopy image of intact vacuoles as we purified them from yeast S. cerevisiae. We did not use vacuoles for activity measurements because: (1) the ATPase activity originating from other ATPases is rather high, in comparison to that of the V-ATPase (data not shown), (2) these vacuoles were rather unstable, and (3) the osmotic conditions needed to keep the vacuoles intact were not optimal for administering other compounds, in general for washing, and for the activity assays. Therefore, it was decided to wash the vacuoles and let the osmotic shock lyse them. Simply this step alone led to the spontaneous formation of vesicles from vacuolar membrane fragments. Figure 2b shows a typical freeze-fracture electron microscopic image of such vesicles. This vesicular system turned out to be rather stable; it tolerated storage up to 8 weeks at −80 °C (longer periods were not tested). In addition, and most importantly, this step allowed us to wash away most of the water-soluble ATPases from the suspension, maximising the relative contribution from the V-ATPase to the total ATPase activity of the sample. This is demonstrated in Fig. 3, which shows ATPase activities of yeast vacuolar vesicles in the presence and absence of different ATPase inhibitors. In these experiments, the concentration of inorganic phosphate is assayed after a 20 min incubation of the vacuolar vesicles at 30 °C in the presence of 2 mM Na2ATP and 5 mM MgCl2. Because the substrate of the reaction is MgATP, and MgCl2 is in excess relative to Na2ATP, the substrate concentration was close to 2 mM (at least at the beginning of the reaction). Several inhibitors that are known to inhibit other ATPases (Serrano 1978; Clelland and Saleuddin 2000; Lunde and Kubo 2000; Padilla-Lopez and Pearce 2006) were tested. These ‘other’ inhibitors remove only about 20 % of the total ATPase activity, even at relatively high inhibitor concentrations. Dicyclohexylcarbodiimide, which inhibits proton translocation also in the related F-ATPase (see, e.g., (Kopecky et al. 1981, 1982, 1983; Hermolin and Fillingame 1989; Wada et al. 2000)), is more potent, but the most potent inhibitors are, as expected, concanamycin A and bafilomycin, even when applied at the lowest concentrations (Bowman et al. 1988; Drose et al. 1993; Gagliardi et al. 1999). In this system, and under the present conditions, these inhibitors remove ~60 % of the total ATPase activity, which varied to some extent from preparation to preparation. Because these two inhibitors are known to be very potent and very selective inhibitors of the V-ATPase (Farina and Gagliardi 1999; Huss et al. 2002; Dixon et al. 2008), it is concluded that ~60 % of the total ATPase activity comes from the V-ATPase in these vesicular preparations. This value translates to ~81 nmole ATP/min/mg total protein at the beginning of the reaction (using the kinetics, reported in (Fig. 5), and the corresponding calibration for liberated phosphate (see inset in Fig. 6)). These two characteristic ATPase activities (i.e., the total hydrolysed ATP per min per total protein, and the concanamycin A sensitive fraction of it) compare very well with corresponding values obtained in vacuoles (see, e.g., (MacLeod et al. 1998; Owegi et al. 2005; Johnson et al. 2010)), taking into account the differences in the conditions, most importantly in the temperature. This observation strongly argues against the possibility of a significant fraction of V-ATPases being oriented inside out in these vesicles.
Fig. 2

a Bright field transmitted light microscopy image of intact vacuoles purified from yeast Saccharomyces cerevisiae. The image was taken with differential interference contrast optics. Scale bar 2 μm. b Freeze-fracture electron microscopy image of yeast vacuolar vesicles formed from yeast vacuoles after osmotic lysis and washing. The vesicles were fixed in glycerol. Scale bar 1 μm
Fig. 3

Comparison of ATPase activities of yeast vacuolar vesicles in the presence and absence of different ATPase inhibitors. ATP hydrolysis was stopped after 20 min of incubation at 30 °C in the presence of 2 mM Na2ATP and 5 mM MgCl2, and the medium was assayed for inorganic phosphate, in all cases. Abbreviations: all ATP, no inhibitors; AMV, 5 mM Na-azide + 0.2 mM molybdate + 0.1 mM vanadate; AMV + NEM, as AMV plus 2 μM n-ethyl-maleimide; AMV + DCCD, as AMV plus 2 μM dicyclohexylcarbodiimide; AMV + ConcA, as AMV plus 1 μM concanamycin A; AMV + BAF, as AMV plus 1 μM bafilomycin. Results are normalised to the same “all ATP” activity. Bars indicate averages, errors indicate a range of ± standard deviation (n = 3). Corresponding activities of blank (no vesicles, no ATP, no inhibitor) and no ATP (no Na2ATP, no inhibitor) samples were 0.0147 ± 0.0006 and 0.0253 ± 0.0025, respectively (not shown)

In order to standardise conditions for the activity measurements, the optimal substrate concentration has to be determined. This was done by measuring the absorbance (at 750 nm) for inorganic phosphate liberated by ATP hydrolysis in the vacuolar vesicles, and assayed after 20 min incubation at 30 °C, as a function of the concentration of exogenously added Na2ATP, both at fixed (5 mM) MgCl2 concentration (circles) or when it was set to twofold that of the Na2ATP (squares). Figure 4 shows the concentration dependence for the whole system, i.e., no inhibitor (open symbols) and that of V-ATPase alone (solid symbols). The latter is obtained from the difference in absorbance with and without 1 μM concanamycin A. The data are normalised such that the maximum activity is the same, viz., unity, in the absence of the inhibitor. (It should be noted that both the total ATPase activity and the concanamycin A sensitive fraction of it varied from cell culture to cell culture. Therefore, only the shape of the two kinds of experiments, fixed versus varied MgCl2 concentration, and the location of the maxima can be meaningfully compared.) The whole system saturates at ~2 and ~10 mM Na2ATP, in the excess and fixed MgCl2 concentration cases, respectively, and the production of inorganic phosphate decreases at high substrate concentrations (only measured for the latter case). The reason for the downward shift of the saturation lies most probably in the different ATP/Mg stoichiometries in the two cases. This, however, does not influence the point of maximum activity of the V-ATPase (at 2–3 mM Na2ATP), because at that concentration even 5 mM MgCl2 can be considered as in excess. Note that 2 mM substrate is considered as the high ATP, or full-speed, condition in most studies (see, e.g., (Noji et al. 1997; Yasuda et al. 1997, 2001; Ueno et al. 2005; Furuike et al. 2011)). We do not know the exact reason for the drop in activity, also in the case of the V-ATPase, at high substrate concentrations. One can speculate that, due to spontaneous ATP ⟺ ADP + Pi dissociation and possibly even ADP impurities in the ATP stock, the ADP concentration might reach a level at which it significantly inhibits ATPases (De la Cruz et al. 2000; Nakano et al. 2008). Nevertheless, a choice of 2 mM Na2ATP and 5 mM MgCl2 ensures that inhibition of this kind is negligible, and that the V-ATPase is running at the highest speed with respect to substrate concentration, at least at the beginning of the incubation period (which also needs to be standardised in order to avoid consumption of all substrate during the reaction).
Fig. 4

Relative absorbance (at 750 nm) for inorganic phosphate (Pi) liberated on ATP hydrolysis by yeast vacuolar vesicles, and assayed after 20 min incubation at 30 °C, as a function of the concentration of exogenously added Na2ATP (open symbols, righty axis). The data are normalised to the same maximum absorbance. Delta absorbance is the difference between the above normalised absorbances in the absence and presence of 1 μM of the specific V-ATPase inhibitor concanamycin A (solid symbols, lefty axis). MgCl2 concentration was either kept constant at 5 mM (circles) or was set to 2-times the Na2ATP concentration (squares). Each dataset, at constant or varying concentration of MgCl2, was measured twice from independent cell cultures

Figure 5 shows the time-dependence of the difference in the absorbance assaying inorganic phosphate liberated by ATP hydrolysis by yeast vacuolar vesicles in the absence and presence of 1 μM of the specific V-ATPase inhibitor concanamycin A, at 20 °C in the presence of exogenously added 2 mM Na2ATP and 5 mM MgCl2. As shown above, the absorbance difference without and with 1 μM concanamycin A measures phosphate liberated exclusively by the V-ATPase. Because the substrate concentration decreases during the reaction, the rate of ATP hydrolysis by the V-ATPase decreases monotonically too. The semi-kinetic curve does not follow a single exponential, as demonstrated by the poorness of the single exponential fit over the whole incubation period. The reason is that the product, ADP, inhibits the enzyme (De la Cruz et al. 2000; Nakano et al. 2008). Obviously, the fit over the first 20 min, where the concentration of the product is much lower, is closer to a single exponential. It should be noted that the exponential fits are presented only for visualisation, and were not used in this study for any further analysis. Based on this experiment, the duration of the reaction is chosen to be 10 min, because this is short enough to avoid significant consumption of the substrate but long enough to obtain a conveniently high yield of Pi liberated by the V-ATPase. Comparing the ∆A/∆t slope of the fit to the first 5 points (onset slope) with the slope of the line connecting the points at 1 and 10 min (straight lines in Fig. 5), one can estimate the rate of ATP hydrolysis at the beginning of the incubation period if the mean (over the 1–10 min interval) is known. The ratio between the two slopes is 1.48.
Fig. 5

Delta absorbance obtained as the difference between the absorbances (at 750 nm) assaying inorganic phosphate (Pi) liberated on ATP hydrolysis by yeast vacuolar vesicles in the absence and presence of 1 μM of the specific V-ATPase inhibitor concanamycin A. Values are given as a function of time of incubation at 20 °C in the presence of exogenously added 2 mM Na2ATP and 5 mM MgCl2. The dotted and dashed lines are single exponential fits over the 1–20 min and 1–50 min regions, respectively, and are shown just as guides for the eye. The solid line spanning the 1–3 min region is a linear fit to the first 5 points. The solid line extending over the 1–10 region connects the means of the two data points at 1 and 10 min. Over the 1–20 min region, absorbances were measured twice from independent cell cultures

Under the above standardised conditions, including spectrophotometric calibration for inorganic phosphate (Fig. 6, see below), and using the specific inhibitor concanamycin A, one can determine the concentration of ATP hydrolysed exclusively by the V-ATPase in unit time. The enzyme concentration, which is needed to convert this data into revolutions of the rotor per second, can be determined from the experimental inhibitor titration curve (Fig. 6) fitted with the appropriate binding equation, because the inhibitor concentration is known and the shape of the curve depends (differently) on the enzyme concentration and the dissociation constant for enzyme-inhibitor binding. The general case is presented below with n independent and identical inhibitor binding sites per enzyme, assuming non-cooperative binding. Further, it is assumed that all inhibitor molecules (I) are active and that all V-ATPase molecules are active if no inhibitor is bound (the case when these assumptions do not hold is considered later), but are inactivated when at least one inhibitor is bound. The data in Fig. 6 should follow the equation:
$$ A\left( {I_{t} } \right) \, = A_{0} - \Updelta A*\left( {1 - P_{f} /P_{t} } \right), $$
where A(It) is the absorbance considered as a function of the total inhibitor concentration, It. A0 is the absorbance in the absence of inhibitor, a constant (three measurements gave an average of 1.248). ∆A is the difference in absorbance at zero and saturating concentration of the inhibitor. This could be determined experimentally, but it is a fitting parameter in the present study because the inhibitor concentration used is not sufficiently high. Pt is the (total) enzyme concentration and Pf is the concentration of the enzyme without any inhibitor bound (the indices t, f and b mean total, free and bound sites or molecules, respectively, and the symbols indexed are the corresponding concentrations). The relevant binding equations need to be solved to derive Pf/Pt as a function of It.
Fig. 6

Absorbance (at 750 nm) assaying inorganic phosphate (Pi) liberated on ATP hydrolysis by yeast vacuolar vesicles, after 10 min incubation at 20 °C in the presence of 2 mM Na2ATP (and 5 mM MgCl2), as a function of concentration of the specific V-ATPase inhibitor, concanamycin A. Three points at zero inhibitor are not visible because of the logarithmic axis. The solid curve is a released fit according to the quadratic binding equation assuming a single binding site per monomeric enzyme [Eq. (6)]. The dotted curve is a fit according to the model with five binding sites per monomer V-ATPase enzyme [Eq. (10)], but with an enzyme concentration fixed at 44.9 nM. The inset shows the calibration of absorbance versus amount of exogenously added inorganic phosphate [Pi], in the same preparations, same reaction volume and under the same conditions as titration with the inhibitor, except that no Na2ATP was added

If the binding sites (B) are equivalent and independent, the binding reaction at a given site is:
$$ B + I \Leftrightarrow BI, $$
where BI means occupied binding site, i.e., bound inhibitor. The binding equation with the corresponding concentrations is:
$$ B_{f} *I_{f} = I_{b} *K_{d} , $$
where Kd is the intrinsic dissociation constant for inhibitor binding to a single, independent binding site. Because Bf = n * Pt − Ib and If = It − Ib, it follows that:
$$ (n*P_{t} - I_{b} ) \, * \, (I_{t} - I_{b} ) \, = I_{b} *K_{d} . $$
This quadratic equation can be solved for the concentration of bound inhibitor:
$$ I_{b} = \, (I_{t} + n*P_{t} + K_{d} - ((I_{t} + n*P_{t} + K_{d} )^2 - 4 \, *n*P_{t} *I_{t} )^{1/2} )/2 \, . $$
If there is only a single binding site per enzyme (i.e., n = 1), then Bf = Pf = Pt − Ib and, hence:
$$ P_{f} /P_{t} = \, (P_{t} - I_{t} - K_{d} + ((I_{t} + P_{t} + K_{d} )^{2} - 4 \, *P_{t} *I_{t} )^{1/2} )/(2 \, *P_{t} ). $$
Substituting from Eq. (5) in Eq. (1) thus yields:
$$ A(I_{t} ) \, = \, (A_{0} - \Updelta A) \, + \Updelta A* \, (P_{t} - I_{t} - K_{d} + \, ((I_{t} + P_{t} + K_{d} )^2 - 4 \, *P_{t} *I_{t} )^{1/2} )/(2 \, *P_{t} ), $$
which is the expression needed for fitting the single binding site model to the data in Fig. 6. In these fits, It is the independent variable and Pt, Kd and ∆A are fitting parameters.
Let us now consider the case of multiple inhibitor binding sites. For the enzyme with n binding sites, the progressive binding equilibria are:
$$ P_{0} + I \Leftrightarrow PI_{1} ,PI_{1} + I \Leftrightarrow PI_{2} , \ldots ,PI_{n - 1} + I \Leftrightarrow PI_{n} , $$
where P0 is the enzyme without any inhibitor bound. The PIi represents the enzyme with i inhibitors bound (i = 1, 2, …,n) and Pi is its concentration, so that Pf ≡ P0. Considering the single-site binding equilibrium for the PIi−1 species, one can write the binding equations for the corresponding concentrations in a generalised recursive form [see, e.g., (Tanford 1961)]:
$$ P_{i - 1} *\left( {n + \, 1 - i} \right)*I_{f} = P_{i} *i*K_{d} ,\quad {\text{with}}\quad i = \, 1, \, 2, \ldots ,n, $$
where Kd again is the intrinsic dissociation constant [cf. Eq. (2)]. The factor (n + 1 − i) is the number of free binding sites on PIi1, and i is the number of the occupied (bound) sites on PIi. With the definition Ik ≡ If/Kd = (It − Ib)/Kd, repeated application of the recursion relation in Eq. (7) leads to:
$$ P_{i} = n!/\left[ {\left( {n - i} \right)!*i!} \right]*(I_{k} )^{i} *P_{0} ,\quad {\text{with}}\quad i = 1,2, \ldots ,n. $$
Because \( P_{t} = P_{0} + P_{1} + P_{2} + \cdots + P_{n} \), this leads directly to the binomial expansion and the closed form of the result becomes:
$$ P_{f} /P_{t} = P_{0} /P_{t} = 1/\left( {1 \, + I_{k} } \right)^{n} . $$
Note that if n = 1, Eq. (9) reduces to the single binding site situation given above. Combining Eq. (9) with Eq. (1) gives:
$$ A\left( {I_{t} } \right) = \left( {A_{0} - \Updelta A} \right) + \Updelta A*\left[ {1 + \left( {I_{t} - I_{b} } \right)/K_{d} } \right]^{ - n} , $$
which, after substituting the solution for Ib from Eq. (4), is the expression required for fitting the multiple binding site model to the experimental data in Fig. 6. Here, n can also be a fitting parameter. If n is fixed, the fitting parameters are again Pt, Kd and ∆A, as in the n = 1 case.
A fit of Eq. (10), where all parameters are released (not shown), including the number of binding sites n, yields n = 0.75 ± 0.58. Because a non-integer number of binding sites on a monomeric enzyme is impossible, and there is no reason to assume binding of three inhibitors to four enzymes, one can fix n = 1, i.e., settle on the single binding site model. Indeed, as a test, neither the n = 2 nor the n = 5 model gives a better fit (n > 5 would not make sense because no more than five c-ring subunits are available for binding). In fact, the fits with these models yield physically impossible parameters, e.g., negative enzyme concentration (Pt), even if some of the parameters (∆A or Kd) are fixed at their values from the above fit with n = 1. To illustrate this, Fig. 6 also shows the best fit with n = 5 (dotted line) when, however, Pt is fixed at the value from the fit to the n = 1 model. Obviously, the resulting value of ∆A is incorrect and the fit is much worse than that with n = 1. It should be noted that for n > 1 it is theoretically possible that enzymes are inhibited only if all of their n binding sites are occupied. Since in this case Pf = Pt − Pn(here meaning the concentration of active enzyme), the above equations lead to a corresponding fitting function
$$ A(I_{t} ) \, = A_{0} - \Updelta A* \, (I_{k} / \, (1 \, + I_{k} ))^{n} . $$
Fitting this function to the same data (this fit is not shown) yields again negative enzyme concentration for n = 2, 3, 4, or 5 fixed, whereas the fully released fit yields n = 1.2 ± 0.7. One can conclude that the present data are compatible only with the single binding site (n = 1) model. The fit with this model (solid curve in Fig. 6) yields the following results when all parameters are released (except that n = 1 is fixed): Pt = 44.9 ± 31.6 nM, Kd = 49.3 ± 18.2 nM, ∆A = 0.871 ± 0.023 (errors are ± standard errors from the fit). (If ∆A, the most accurate fitting parameter, is then fixed at 0.871, the other parameters do not change but their standard deviation becomes smaller: Pt = 44.9 ± 24.6 nM and Kd = 49.3 ± 11.5 nM.) The absolute amount of inorganic phosphate corresponding to ∆A is liberated exclusively by 44.9 ± 31.6 nM V-ATPase. The inset in Fig. 6 shows the phosphate calibration in the very same system. The slope of the linear regression is 0.00452 ∆OD/nmole Pi, from which the concentration of Pi liberated by V-ATPase is 770 μM. Taking the ratios of the mean concentrations one concludes that in 10 min ca. 17,000 Pi molecules, i.e., approximately 30 per second, were liberated by each V-ATPase molecule. Current understanding of the V-ATPase catalysis assumes hydrolysis of three ATP molecules in a complete 360° rotation cycle (see., e.g., Beyenbach and Wieczorek 2006), which means that the rotation rate in V-ATPase in our vacuolar membrane vesicle system is 10 Hz at 20 °C and with excess ATP. If we consider the ± standard deviation of the fitting parameter Pt, i.e., the one with the largest relative fitting error, the range of rotation rates is 6–32 Hz. Comparing the slopes in the time dependence (Fig. 5) at the onset of the reaction and that connecting the 1–10 min points, the mean rotation rate extrapolates to ~14 Hz and to a range of rotation rates of 8–48 Hz, for the beginning of the reaction.

It should be noted that the dissociation constant, Kd, of concanamycin A binding to V-ATPase is much larger than in some early reports (Bowman et al. 1988; Drose et al. 1993) but similar to a more recent report (Whyteside et al. 2005; Dixon et al. 2008). In addition to the difference in the host systems, this discrepancy might be caused by several factors as explained in (Whyteside et al. 2005), of which, the most likely might be potential removal of a structural component (e.g., polypeptide or lipid) required for higher affinity binding. The state of the membrane might play a role here, because concanamycin A has been demonstrated to penetrate into the membrane and interact directly with both lipids and side chains of the c-subunits (Pali et al. 2004a, b; Dixon et al. 2008; Kota et al. 2008). There is no reason to assume that the composition and concentration of the lipids differed significantly from preparation to preparation (the samples were taken to have the same mass of total protein) but the membrane environment of the V-ATPase might be very different in different studies.


The present approach, to use the inhibitor titration curve in order to determine the concentration of the enzyme and the dissociation constant of the inhibitor, depends on the mechanism of the V-ATPase/inhibitor interaction. The picture on concanamycin A binding to V-ATPase is not clear in the literature, and the present data give further restrictions to localise the binding site of concanamycin A. Our data are compatible only with a single binding site per monomeric enzyme. Because concanamycin A most likely binds to intramembranous subunits (Huss et al. 2002; Pali et al. 2004a, b; Whyteside et al. 2005; Bowman et al. 2006; Dixon et al. 2008), the present result suggests that an interaction with a single c-subunit, of which three or four copies are present in the c-ring, is not sufficient for high-affinity concanamycin A binding: either the binding site consists of more than one c-subunit or it includes either subunit a or c′, of which there is only one copy each in V0. Interaction of concanamycin A with the lobster c-ring and other recent data (Pali et al. 2004b; Bowman et al. 2006; Dixon et al. 2008) indicate that the single-copy subunits a and c′ are not the sole components of the concanamycin A binding sites. According to (Gibson et al. 2002) there are two c″ subunits in the yeast c-ring. In view of the present data, concanamycin A binding to c″ would assume an inhibitor/\(c_{2}^{\prime\prime}\) stoichiometry. Nevertheless, it is important to note that the concentration of the enzyme (from the curve fit), needed to estimate the rotation rate, does not depend strongly on the details of the mechanism of the inhibitor binding because, based on our data, there is one binding site per enzyme.

Reports on the rate of rotation in F-ATPase and V-ATPase vary widely. One reason is that the enzyme is in a very different state and environment in activity measurements with native membranes versus direct observations of the rotation of attached fluorescent or gold particles when it is fixed on a solid support. In some cases there are as high as tenfold differences between activity-based estimates and direct observations even for the very same system (Nakanishi-Matsui et al. 2010). Such discrepancies are explained by the differences in the ratio of inactive/active ATPase, an unknown factor, in different systems. In view of the limited variations from preparation to preparation, and in comparison with other ATPases in the system (Fig. 3), we believe that such large variations in the inactive/active ratio of the V-ATPase are unlikely.

The accuracy of our approach could be further improved by adding more points to the inhibitor titration curve. However, it would not change the fact that the estimated rotation rate is a lower limit, because the activity of the inhibitor, which cannot be determined easily, is most probably less than 100 %. In addition, it is also very likely that not all enzymes are active in the vesicle preparations, and we do not know whether the inhibitor binds to otherwise already inactive V-ATPases. Because we assumed 100 % activity, for both inhibitor and enzyme, smaller activities would scale up the estimated rotation rate. For instance, assuming that both inhibitor and enzyme were only 80 % active (and the inhibitor only binds to the active enzyme) would result in an ~15 Hz average rotation rate, instead of ~10 Hz, i.e., a factor of 0.8−2 larger, for the very same data. Obviously, these uncertainties cause larger errors in the estimation of the rotation rate than those in the titration experiment. Although our approach allows us to report an estimate for the rotation rate in a native V-ATPase in its native membrane environment, it must be concluded that there is still a great need for a more accurate measurement in native systems that is independent of the chemistry of any inhibitor binding to the enzyme.


We thank Ferhan Ayaydin (Cellular Imaging Laboratory, Biological Research Centre Szeged, Hungary) for his help in taking the transmitted light microscopy images of yeast vacuoles. We also thank E. Kónya for excellent technical assistance. Financial support was received from the Hungarian National Science Fund (OTKA K68804 and K101633). D.M. thanks Christian Griesinger and the Department for NMR-based structural biology for financial assistance.

Ethical standards

It is declared that the experiments presented in this manuscript comply with the current laws of the country in which they were performed.

Copyright information

© European Biophysical Societies' Association 2012