INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

F_OF₁-ATP synthase uses the transmembrane electrochemical potential difference of the proton (_H+) to synthesize ATP from ADP and inorganic phosphate (for reviews, see Stock et al., 2000; Weber et al., 2000; Boyer, 1998; Junge et al., 1997). This bipartite enzyme is composed of the membrane-embedded ion-translocating F_O and the hydrophilic catalytic F₁ that protrudes for more than 100 Å from the plane of the membrane. In bacteria and chloroplasts the F_O part is formed from three different subunits (a, b, and c) in stoichiometry a₁:b₂:c_10-14 with c-subunits arranged as a ring (Stock et al., 1999; Seelert et al., 2000; Stahlberg et al., 2001). The crystal structure of F₁ shows a hexamer of ₃₃-subunits with the -subunit as a central shaft (Abrahams et al., 1994) that is connected to subunit  and the c-oligomer (Gibbons et al., 2000; Stock et al., 1999).

ATPase is a rotary machine: the ₁₁c_10-14 complex (the rotor) rotates relative to the other subunits (that form the stator) when driven by the hydrolysis of ATP (Duncan et al., 1995; Sabbert et al., 1996; Noji et al., 1997; Pänke et al., 2000). The mechanism for torque generation by proton flow and its coupling to ATP synthesis has been discussed (e.g., in Junge et al., 1997; Dimroth et al., 1999; Wang and Oster, 1998; Cherepanov et al., 1999). The common features of the proposed mechanisms are 1) a ring of c-subunits, each of them carrying one carboxy residue capable of proton binding; 2) the alternating accessibility of this proton binding residue from different sides of the membrane; 3) Brownian rotation of this ring relative to ab₂ subunit complex; and 4) electrostatic constraints enforcing the sequential deprotonation/reprotonation of the acidic residue on the c-subunit depending on its position relative to the subunit a and the lipid phase. In some organisms proton is substituted by sodium (Dimroth, 2000).

Preparations of chromatophores (vesicles derived from invaginations of the cytoplasmic membrane) of phototrophic bacterium Rhodobacter capsulatus proved to be convenient for investigation of the ion-conducting properties of the ATP synthase. Excitation of chromatophores with short flashes of light generates steps of protonmotive force. Its chemical component (pH) has been first monitored by absorption transients of pH-indicating dyes (Jackson and Crofts, 1969a) and calibrated by the amphiphillic dye neutral red (Mulkidjanian and Junge, 1994). The electrical component () has been monitored and calibrated by intrinsic electrochromic bandshifts of carotenoids at 520 nm (Jackson and Crofts, 1969b) Proton transfer across the coupling membrane, driving the synthesis of ATP, has been apparent from an accelerated decay of the electrochromic absorption transients (Jackson et al., 1975). It is sensitive to specific inhibitors of F₁ (efrapeptin) and F_O (venturicidin, DCCD, oligomycin). The charge transfer has been calibrated and correlated with the ATP yield, which has been measured in the same samples by the luciferin-luciferase system (Saphon et al., 1975; Feniouk et al., 1999, 2001).

In this work we addressed the question of the average number of the F_OF₁-ATP synthases per chromatophore vesicle. We obtained experimental evidence that routine preparations of chromatophores from Rb. capsulatus contained, on the average, approximately one molecule of ATP synthase per vesicle. This finding greatly reduced the statistical ambiguity over active/inactive vesicles or enzymes and paved the way for studying the operation of the ATP synthase in a single enzyme/single vesicle mode.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

Cell growth and chromatophore preparation

Cells of Rb. capsulatus (wild type, strain B10) were grown photoheterotrophically on malate as a carbon source at +30°C (Lascelles, 1959). The bottles were illuminated with four Osram L 36W/25 and three Osram L 18W/21-840 candle-shaped bulbs. The average light intensity was 18 W/m². Cells were harvested at the end of logarithmic growth phase, if not otherwise indicated; the cell mass of the culture was estimated by measuring optical density at 660 nm (Schumacher and Drews, 1979).

Harvested cells were washed twice with 30 mM HEPES-KOH (pH 7.4), 5 mM MgCl₂, 0.5 mM dithiothreitol, 50 mM KCl, 10% sucrose, and resuspended in the same pH buffer. A few flakes of DNase were added. The cells were disrupted by sonication on ice (Branson Sonifier B15, four or five exposures for 15 s with 1-min incubation in between), and centrifuged (20,000 × g, 20 min, 4°C) to remove large cell fragments. The pellet was suspended in the same buffer and resonicated as above; the resonicated suspension was centrifuged again (20,000 × g, 20 min, 4°C). Supernatants were collected and centrifuged (180,000 × g, 90 min, +4°C). The pellet was resuspended in 30 mM HEPES-KOH (pH 7.4), 5 mM MgCl₂, 0.5 mM dithiothreitol, 50 mM KCl, 20% sucrose. It contained chromatophores at a bacteriochlorophyll concentration of 0.3 to 0.7 mM. Chromatophores were stored at 80°C until use.

The concentration of bacteriochlorophyll in the samples was determined spectrophotometrically in the acetone-methanol extract at 772 nm according to Clayton (1963). The amount of functionally active reaction centers (RC)s was estimated from the extent of flash-induced absorption changes at 603 nm (as in Mulkidjanian et al., 1991).

Preparation of chromatophores stripped of F₁

Chromatophores were depleted of F₁ by EDTA-treatment (Melandri et al., 1970,1971; Baccarini-Melandri et al., 1970). Chromatophores from frozen stock were thawed and diluted 25-fold with 1 mM EDTA, pH 8, in daylight. The suspension was sonicated 5 times for 20 s with 1-min interval (on ice) and then centrifuged at 180,000 × g, 90 min, +4°C. The pellet was resuspended in a medium containing 20 mM HEPES KOH (pH 7.4), 5 mM MgCl₂, 0.5 mM dithiothreitol, 50 mM KCl to yield a bacteriochlorophyll concentration of 0.3 to 0.7 mM and used on the same day or after storage at +4°C overnight.

Spectrophotometric measurements

The kinetic flash-spectrophotometer was constructed according to Junge (1976). Monitoring light was provided by a halogen 200-W lamp, it was heat-filtered (KG 2 filter, SCHOTT, Mainz, Germany) and passed through a 11-nm wide interference filter peaking at 522 nm (SCHOTT). A shutter, placed in front of the cuvette, eliminated the actinic effect of the monitoring beam between the measurements. The dead time between the opening of the shutter and the actinic flash was 400 ms. Changes in transmitted light intensity (I) were monitored by a photomultiplier (9801B, Thorn EMI, Ruislip, UK) that was shielded from the actinic flash with two blue filters (BG 39/2, SCHOTT). The DC output from the photomultiplier was preset to 1 V (load resistor: 10 k) by varying the output current of the photomultiplier power supply. The photomultiplier output was connected to the positive input of a difference amplifier (AM502, Tektronix, Beaverton, OR). Before an actinic flash was fired, the signal was sampled and held by a homemade amplifier (N. Spreckelmeier) connected to the negative input of the difference amplifier. The difference signal was amplified 100-fold, digitized, and stored on an averaging oscilloscope (Pro 10, Gould Nicolet, Erlensee, Germany). The analogue bandwidth was 3 kHz and the digital time-per-address was 200 µs. The optical path was 1 cm, both for the exciting and for the measuring beam. The final concentration of bacteriochlorophyll in the cuvette was 8 to 12 µM.

Eight signals measured at a repetition rate of 0.08 Hz were averaged. The dark adaptation time of 12 s between flashes was chosen to allow for the complete decay of the transmembrane electrical potential difference (). This time interval was still short enough to prevent the deactivation of the F_OF₁-ATP synthase. (The expected life time of the -activated state of the F_OF₁ ATP-synthase under these conditions was 30-70 s (Turina et al., 1992).)

Electrochromic absorption transients at 522 nm were used to monitor the transmembrane voltage (Clark and Jackson, 1981; Symons et al., 1977; Jackson and Crofts, 1971).

Saturating actinic flashes were provided by a xenon flash lamp (full width half maximum ~10 µs, red optical filter (RG 780, SCHOTT)). The energy density on the cuvette was 12 mJ/cm²). Excitation by a xenon flash gave 8 to 11% greater signals than by the laser pulse (duration < 100 ns). It was indicative of double turnover in ~10% of RCs, in agreement with the recent report on the presence of a 3-µs component in the reduction of Q_B in chromatophores of Rb. sphaeroides and Rb. capsulatus (Tiede et al., 1998).

Incubation medium

Chromatophores were suspended in a medium that contained 50 mM KCl, 0.3% bovine serum albumin, 5 mM MgCl₂, 200 µM succinate/2 mM fumarate (redox buffer), 5 or 10 µM 1,1'-dimethylferrocene (redox mediator), and 2 mM KCN (to block the terminal oxidase). Instead of succinate-fumarate redox system, 2 mM K₄Fe[CN]₆ was present in some experiments.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION REFERENCES

Fig. 1 A illustrates typical absorption changes at 522 nm in chromatophores of Rb. capsulatus in response to two saturating flashes of light given at 240-ms interval. These changes reflect mainly electrochromic bandshifts of carotenoids and can be described by the following expression:
in which A_i is the absorption change proportional to the transient transmembrane voltage change (_i) in the subset of chromatophore vesicles containing i molecules of F_OF₁, a_i is the weight of this subset, and b is a residual flash-induced absorption change that is unrelated to the delocalized transmembrane voltage (local electrochromic effects, absorption transients of the primary electron donor of the photochemical reaction centers, etc). Under the chosen conditions the relative extent of b was less than 10%. Its magnitude was checked by addition of K⁺ + valinomycin to rapidly quench the transmembrane voltage (data not shown). Therefore, to a first approximation the absorption changes at 522 nm represent the weighted sum of the transmembrane voltage transients in a heterogeneous ensemble of chromatophore vesicles containing different copy numbers of ATP synthase.

View larger version (23K): [in this window] [in a new window]   FIGURE 1   Flash-induced absorption changes at 522 nm in a suspension of Rb. capsulatus chromatophores. The supending medium contained 2 mM K2HPO4, 2 mM K4[Fe(CN)6], 5 µM 1,1'-dimethylferrocene, 2 mM KCN, 5 mM MgCl2, 50 mM KCl, 0.3% bovine serum albumin, pH 7.9. (Trace 1) Control, no additions; the following compounds were added consecutively to the sample: ADP, 500 µM (trace 2), efrapeptin; 200 nM (trace 3), venturicidin, 200 nM (trace 4). Difference traces reflect proton flow through FO(F1) (3-2) and FO (4-3), respectively. Actinic flashes are marked by arrows. See details in text. (A) Well-coupled chromatophores. (B) Chromatophores depleted of F1 by EDTA treatment.

Each flash caused a sharp, here unresolved rise of the absorption at 522 nm. It was attributable to the charge separation in the photochemical RC. It was followed by a slower rise due to charge transfer in the cytochrome bc₁-complex (for reviews on the RC and the cytochrome bc₁-complex, see Jackson, 1988; Crofts and Wraight, 1983). The biphasic rise was followed by a multiphasic decay that was markedly accelerated by the addition of ADP and inorganic phosphate (trace 1 and 2 in Fig. 1 A). Similar observations have been made in chromatophores of Rb. sphaeroides (Jackson et al., 1975; Petty and Jackson, 1979). The effect of ADP and phosphate was reversed by efrapeptin, a peptide inhibitor that binds between subunit  and ₃₃ hexamer in F₁ (Abrahams et al., 1996). The inhibitor decreased the decay rate almost to the level that was observed in the absence of ADP and phosphate (Fig. 2 A, trace 3). Venturicidin, oligomycin, and DCCD, all inhibitors of proton conduction through the F_O-portion of ATP synthase (von Brufani et al., 1968; Linnett and Beechey, 1979), decreased the rate of the decay further, even in the absence of ADP or phosphate or in the presence of efrapeptin (Fig. 1, A and B, trace 4). As discussed elsewhere (Feniouk et al., 1999, 2001), the greater efficiency of F_O-inhibitors as compared with the F₁-inhibitor efrapeptin is due to the fact that the former blocks the proton conductivity at F_O, independent of the presence/absence or functionality of F₁, whereas efrapeptin acts only on F_OF₁ via F₁ proper. When chromatophores were stripped of F₁ by EDTA-treatment as in Baccarini-Melandri and Melandri (1971), the decay of the electrochromic transient lost its sensitivity to ADP+P_i or efrapeptin, as expected, but it was still sensitive to venturicidin (Fig. 1 B). It was obvious from Fig. 1 B, that the faster decay of the electrochromic transient was limited in extent. A substantial slowly decaying level remained even 200 ms after the flash. This held true, both in well coupled and in F₁-stripped chromatophores. This residual level was markedly higher after the second flash. This behavior suggested the existence of a subset of chromatophores with very low conductivity because of the total lack of F_O(F₁). (Hereafter we denote by F_O(F₁) the proton-conducting, and venturicidin-sensitive entity, which is either bare F_O or whole F_OF₁ (coupled or slipping).) Excitation of this subset of chromatophores by repetitive light flashes led to a stepwise increase of the electrochromic transient. The presence of a F_O(F₁)-lacking fraction of chromatophores is in agreement with the previous observations that the removal of F₁ by EDTA treatment did not result in a fast and complete dissipation of and pH to the zero level after an actinic flash (Saphon et al., 1975; Melandri et al., 1970, 1972). The traces presented in Fig. 1 B allow to roughly estimate the relative fraction of such F_O(F₁)-lacking chromatophores as ~50%. A detailed analysis of kinetic traces, which has been presented elsewhere (Feniouk et al., 2001) yielded a somewhat lower estimate of 40%.

View larger version (27K): [in this window] [in a new window]   FIGURE 2   Schematic representation of the inhibitor titration of electrochromic absorption transients under two different assumptions on the ATP synthase-contents in chromatophores (see text for details).

We determined the average amount of F_O(F₁) per vesicle and asked for the homogeneity of the F_O(F₁) distribution over the ensemble of chromatophores. The observed rapid relaxation of the electrochromic absorption transients reflects the superimposition of events in many vesicles with different surface densities of F_O(F₁). We used an approach similar to one developed elsewhere (Schmid and Junge, 1975; Drachev et al., 1981; Lill et al., 1987) to investigate the distribution of F_O(F₁) in chromatophores. The scheme in Fig. 2 illustrates how the titration with an F_O inhibitor reveals the number of ion transporter molecules per vesicle. If several ATP synthases were present per vesicle, the inhibition of a few would only slow down the electric relaxation but not decrease its extent (unless the number of conducting F_O(F₁) was reduced to zero). Alternatively, if most of the vesicles contained none or just one single enzyme molecule, then the latter subset of vesicles caused an all-or-none response to the binding of the inhibitor. When the inhibitor concentration was increased, more and more of such units were switched off, and the extent of the rapid relaxation decreased. Fig. 3, A through C shows the results of a titration with efrapeptin of well-coupled chromatophores and Fig. 3, D through F the titration with venturicidin of F₁-depleted chromatophores. In both cases the electrochromic transients recorded in the completely inhibited sample were subtracted from those recorded in the presence of subsaturating, gradually increasing inhibitor concentrations. It resulted in a series of kinetic traces of proton transfer attributable to a decreasing fraction of active enzyme molecules (Fig. 3, B and E). In both cases, the extent of the rapid proton transfer dropped as a function of the inhibitor concentration, whereas the time constant of the proton transfer was only marginally changed (Fig. 3, C and F). According to the above rationale the results were qualitatively in line with the assumption that the average number of proton-conducting F_O(F₁) per vesicle was close to 1.0. 

View larger version (34K): [in this window] [in a new window]   FIGURE 3   Titration by efrapeptin and venturicidin of electrochromic absorption transients through FO(F1) in intact (A-C) and uncoupled (D-F) chromatophores of Rb. capsulatus, respectively. Actinic flashes are marked by arrows. (A and D) The increase in the inhibitor concentration slowed the decay rate. (B and E) Difference traces reflecting proton transfer through FO(F1) at subsaturating inhibitor concentrations (see text); (C and F) the relaxation time of proton transfer through FO(F1) as function of the inhibitor concentration.

The exact number was calculated using the statistical model that was developed for treating similar questions in isolated thylakoids from higher plants (Schmid and Junge, 1975; Lill et al., 1987). The model has been based on three assumptions: 1) the distribution of vesicles over their membrane area is narrow; 2) the number of active conducting entities (F_O(F₁) in the present case) per chromatophore obeys Poisson's distribution,

(1)

and 3) the voltage decay in a given vesicle with n channels is monoexponential with a rate constant proportional to n:
in which k₀ is the decay rate in a vesicle with a single molecule of ATP synthase and U₀ is the amplitude of the voltage change. The apparent decay of the transmembrane voltage in the entire ensemble of chromatophores is then described by the equation (Lill et al., 1987):

(2)

in which is the average number of channels per vesicle. The latter parameter can be experimentally controlled by the addition of an appropriate inhibitor. The interaction of the inhibitor with F_O(F₁) is described by a single equilibrium constant K_I so that

(3)

in which [E] is the concentration of the free (active) enzyme, [I] is the concentration of free inhibitor, and [EI] is the concentration of the inhibited (inactive) enzyme. The mass balance requires that

(4)

(5)

where [E]_total and [I]_total are the total concentrations of the enzyme and the inhibitor, respectively. Eqs. 3 to 5 yield the concentration of the active form of F_O(F₁):

(6)

The average number of uninhibited ATP synthase molecules per chromatophore, , as function of the total inhibitor concentration is thereby

(7)

in which is the average number of active F_O(F₁) molecules per chromatophore before the addition of the inhibitor.

Thus, the complete model includes five fit parameters: the initial extent of the voltage, U₀, the decay rate through a single channel, k₀, the total concentration of proton-conducting F_O(F₁) molecules, [E]_total, the average number of enzyme molecules per chromatophore, , and the inhibition constant, K_I.

Two sets of kinetic traces of charge transfer through the coupled and uncoupled ATP synthase obtained at different concentrations of efrapeptin and venturicidin are represented in Fig. 3, B and E, respectively. We used numerical optimization to describe these traces by the model. The found parameter values are listed in Table 1. The experimental and the calculated voltage decay curves as function of inhibitor concentration are shown in Fig. 4 A (titration by efrapeptin) and in Fig. 4 B (titration by venturicidin), respectively. Titrations by either inhibitor gave the same result: each chromatophore contained 1 molecule of F_O(F₁) on the average.

View larger version (24K): [in this window] [in a new window]   FIGURE 4   Fit of the difference traces presented in Fig. 3, B and E by the statistical model based on Poisson's distribution (see Eq. 2 and Lill et al., 1987). Fitting parameters are given in Table 1. See text for details.

The results of the inhibitor titration were independent on whether chromatophores were prepared by sonication or by a French-Press treatment (data not shown). Routinely chromatophores were prepared from cells that were harvested at the late exponential phase of their growth. We checked whether the average number of the ATP synthases per chromatophore depends on the growth stage of the cells. Chromatophores were isolated from cells that were harvested at an early growth stage (for the growth curve, see Fig. 5) and inhibitor titrations by efrapeptin and venturicidin (after EDTA treatment) were carried out as above. The results of the statistical analysis are presented in Fig. 5. The average number of ATP synthase molecules per vesicle was as small as at the end of the growth phase.

View larger version (22K): [in this window] [in a new window]   FIGURE 5   Growth curve of Rb. capsulatus cells as derived from the absorption of cell suspension at 660 nm. The average amount of ATP synthase molecules per chromatophore vesicle, as determined by the efrapeptin and venturicidin titrations, at different stages of Rb. capsulatus cell growth, is indicated.

Our own electron microscopy data on chromatophores from Rb. capsulatus (data not documented) were similar to published EM pictures of chromatophores from Rb. capsulatus and Rb. sphaeroides showing very few of mushroom-like structures, i.e., F_OF₁ on the chromatophore surface (Yen et al., 1982), in contrast with the inner mitochondrial membrane, that is functionally homologous to the chromatophore membrane, but is densely covered with "mushroom-like" ATP synthase complexes (see, e.g., Tsuprun et al., 1989). The finding that only one copy of F_OF₁ is present in most of the chromatophore vesicles might be attributed to the smallness of chromatophores (typical diameter 30 to 60 nm) and the necessity to house large amounts of the light-harvesting complexes in the membrane.

For  = 1, the fraction of chromatophores without F_O(F₁) can be calculated from Poisson's distribution as 37%. This value is pretty close to the estimate of 40% that has been previously obtained based on a different approach (see Discussion above and Feniouk et al., 2001). This similarity indicates that the F_O(F₁) distribution among chromatophore vesicles is nearly homogeneous. Another supporting evidence is the value of K_i for venturicidin, which we obtained from titrations of the F_O-related proton flux extent in of F₁-stripped chromatophores with this inhibitor. Assuming that one molecule of inhibitor per enzyme was sufficient for complete inhibition, we obtained a value of K_i = 0.41 ± 0.1 nM¹ (data not shown), which was quite close to the K_I value given in Table 1.

It is noteworthy that  = 1 implies that 37% of chromatophores carry zero, 37% one, 18% two, and 6% three proton-conducting enzyme molecules (Fig. 6, upper box). The electrochromic relaxation is then multiphasic. For conduction studies it is, of course, desirable to eliminate the statistical spread and to obtain a "digital" behavior, where almost all vesicles fall in either of two classes with zero or only one proton-conducting molecule. This can be achieved by blocking some of F_O(F₁), e.g., by venturicidin. On the one hand, it is disadvantageous that the proportion of nonconducting vesicles is thereby increased, but on the other hand, the ratio between vesicles with one single conducting molecule and those with more than one increases. With 40% of the enzymes blocked ( = 0.6), the ratio between proton-conducting chromatophores with n = 1 over those with n > 1 amounts to 2.7 with 80% blocked the ratio increases to 9.3. Under these conditions the subset of vesicles with more than one proton-conducting molecule is negligible. Despite the inhibition of a certain fraction of proton conducting molecules the signal-to-noise ratio is still high enough for single-molecule/single vesicle studies on ion conduction by F_O(F₁) (Figs. 3 and 4). The common crux in enzyme kinetics, namely the ambiguity between reduced turnover rate and inactivation, is here "automatically" solved. This ambiguity is particularly inconvenient considering site-specific mutations, which can affect the enzyme turnover rate and/or the enzyme assembly in the membrane.

View larger version (21K): [in this window] [in a new window]   FIGURE 6   Schematic illustration to Poisson's distribution of ATP synthase molecules in chromatophore vesicles. The upper box holds for chromatophore suspension in the absence of inhibitor of proton conduction by FO(F1), middle and bottom boxes illustrate partially inhibited samples.

We intend to exploit the technique described to investigate proton transfer through F_O and F_OF₁ in chromatophores of mutant strains of Rb. capsulatus. A proton conductor, like ATP synthase has a single-channel conductance of a few femto-Siemens, which is by orders of magnitude too low to be detected by patch-clamp technique. Whether the method described above can be extended to other ion-translocating enzymes with single-channel conductance in the range of femto-Siemens is an interesting possibility that we are currently investigating.