Three Molecules of Ubiquinone Bind Specifically to Mitochondrial Cytochrome bc 1Complex* 210

Bifurcated electron flow to high potential “Rieske” iron-sulfur cluster and low potential hemebL is crucial for respiratory energy conservation by the cytochrome bc1 complex. The chemistry of ubiquinol oxidation has to ensure the thermodynamically unfavorable electron transfer to heme bL . To resolve a central controversy about the number of ubiquinol molecules involved in this reaction, we used high resolution magic-angle-spinning nuclear magnetic resonance experiments to show that two out of threen-decyl-ubiquinones bind at the ubiquinol oxidation center of the complex. This substantiates a proposed mechanism in which a charge transfer between a ubiquinol/ubiquinone pair explains the bifurcation of electron flow.

A central question in membrane biochemistry is how cofactors interact with membrane proteins. Here we introduce a general NMR-based method to quantify the stoichiometry of binding of hydrophobic cofactors to membrane proteins to resolve a central controversy about the number of ubiquinone binding sites in mitochondrial cytochrome bc 1 complex. The cytochrome bc 1 complex plays a crucial role in oxidative phosphorylation, a universal process that converts most of the energy provided by foodstuffs into the general energy source adenosine 5Ј-triphosphate (ATP). Within this process, the cytochrome bc 1 complex connects hydrophobic ubiquinol and water-soluble cytochrome c, transferring electrons between these two freely diffusible intermediates and thereby linking the exergonic reaction to a vectorial proton translocation across the inner mitochondrial membrane. Molecular structures of the cytochrome bc 1 complex from different sources (1)(2)(3)(4) are fully consistent with the electron transfer scheme of the protonmotive ubiquinone cycle proposed earlier (5)(6)(7). The reaction most critical for energy conservation is an obligatory bifurcation of the electron path linked to the two-electron oxidation of ubiquinol ( Fig. 1). Molecular structures indicate that this unique reaction occurs in a rather spacious (Q o or Q P ) 1 pocket formed mostly by transmembrane cytochrome b and the tip of the mobile hydrophilic domain of the "Rieske" iron-sulfur protein (1)(2)(3)(4). As predicted by enzymological studies (8), methoxyacrylate-type inhibitors like myxothiazol and the chromone-type inhibitor stigmatellin were found to bind with very high affinity to different but overlapping sites within this pocket (9). In crystal structures, bound ubiquinone could only be seen in the ubiquinone reduction (Q i or Q N ) center facing the opposite side of the inner mitochondrial membrane (2,4). Presumably because of very weak binding of the substrate, no corresponding electron density could be identified in the ubiquinol oxidation pocket. It is still a controversial issue whether ubiquinol oxidation in the cytochrome bc 1 complex involves just a single quinone that may have to move to transfer the second electron (6,10) or whether two quinone molecules occupy this binding pocket simultaneously (11) and facilitate bifurcated electron flow (12) (Fig. 1B). Double occupancy of the ubiquinol oxidation pocket was proposed by Ding et al. (11) based on specific line shape changes in the EPR spectrum of the reduced Rieske iron-sulfur cluster of bacterial cytochrome bc 1 complex. However, Crofts et al. (10) proposed that the different line shapes may also reflect different states of the complex or different positions of the ubiquinone headgroup. Resolving this issue will be a prerequisite to understand the chemistry of this unique reaction.

EXPERIMENTAL PROCEDURES
Synthesis of 13 C-Labeled Ubiquinone-[ 13 C]Ubiquinone was synthesized anaerobically according to the method described for the synthesis of ethoxyubiquinone derivatives (13). The reaction was carried out in a Thunberg tube with a two-arm stopper. 1.5 ml of hexane solution containing 100 l of [ 13 C]methanol and 1 mg of sodium methoxide was placed in the bottom of the tube. 10 mg of Q 0 C 10 (2,3-dimethoxy-5methyl-6-n-decyl-1,4-benzoquinone) in 0.5 ml of hexane was placed in one arm of the stopper. 20 l of 10 N acetic acid was placed in the other arm of the stopper. The assembly was then subjected three times to evacuation and argon flushing. The Q 0 C 10 solution was then carefully tipped into [ 13 C]methanol/methoxide solution. This mixture was incubated at room temperature for 2 h in the dark with constant shaking. At * This work was supported by Deutsche Forschungsgemeinschaft Grant SFB 472 and Graduiertenkolleg Grant 145/12, National Institutes of Health Grant GM30721, the Max-Planck Society, the Bundesministerium fü r Bildung und Forschung, and the Fonds der Chemischen Industrie. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  the end of incubation the mixture was acidified by tipping in the acidic acid. The acidified mixture was concentrated under vacuum, redissolved in 0.3 ml of hexane, and subjected to thin-layer chromatography (TLC) separation. The TLC plate was developed with a mixture of hexane:ether (3.5:1.0). The yield of the synthesis was 75%. The pattern of 13 C labeling was analyzed by mass spectroscopy. 22.6% of the ubiquinone molecules were found to carry two, 57.9% one, and 19.5% no 13 C-methoxy group, corresponding to an average of 1.0 13 C-atom per molecule.
Preparation of Cytochrome bc 1 Complex-Cytochrome bc 1 complex proteoliposomes were prepared by the cholate dialysis method essentially as described in Ref. 15. 1 g of a mixture of 75% phosphatidylcholine (99% Sigma Type III-E), 20% phosphatidylethanolamine (98% Sigma Type IV-S), and 5% cardiolipin (Ͼ80% bovine heart) was dissolved in 26 ml of 3% sodium cholate, 1% octyl glucoside, 100 mM KCl, 2 mM EDTA, 2 mM NaN 3 , 20 mM K ϩ /Mops, pH 7.2, by sonication. 40 -50 ml of a 20 M solution of cytochrome bc 1 complex in 10% glycerol was added to the dissolved lipids. After stirring for 15 min on ice, the mixture was placed into a dialysis tube and dialyzed overnight against a 100-fold volume of 100 mM KCl, 2 mM EDTA, 2 mM NaN 3 , 20 mM K ϩ /Mops, pH 7.2. The dialysis buffer was changed once after 4 h. The proteoliposomes were sedimented by centrifugation for 5-6 h at 50,000 ϫ g av ; the red pellet was resuspended in 50 ml of dialysis buffer made with D 2 O, and the liposomes were sedimented again by overnight centrifugation at 50,000 ϫ g av . The concentration of cytochrome bc 1 complex was determined spectroscopically using ⑀ 562-575 ϭ 28.5 cm Ϫ1 mM Ϫ1 for the sum of two heme b groups per monomer of cytochrome bc 1 complex.
Experimental Set-up for HR-MAS Measurements-As a novel approach to directly measure the binding of an extremely hydrophobic, but weakly bound ligand to a membrane protein complex, we have used high resolution magic-angle-spinning nuclear magnetic resonance (HR-MAS NMR) spectroscopy of the liquid heterogeneous system composed of water, liposomes containing cytochrome bc 1 complex, and ubiquinone to determine the binding stoichiometry of [ 13 C]6-n-decylubiquinone to membrane-bound cytochrome bc 1 complex. We used a mixture of unlabeled and labeled 6-n-decylubiquinone ([ 13 C]Q 0 C 10 , Fig. 2) carrying one 13 C-methoxy group at C-2 or C-3 or two 13 C-methoxy groups at C-2 and C-3 of the quinone ring (13). Since the ubiquinone diffused freely in the phospholipid bilayer that was not immobilized, e.g. by freezing or orientation between glass plates as used in solid state NMR investigations, the labeled methyl groups not only had identical 1 H and 13 C chemical shifts but also gave rise to sharp resonances (proton line width of 7 Hz) under HR-MAS conditions (Fig. 2). Solid state MAS on frozen liposomes was not chosen for this investigation, since double labeled ubiquinone would have been necessary for the suppression of the protein background, and the experiments would have suffered from 10 times lower sensitivity. In the HR-MAS measurements, the enzyme-bound ubiquinones did not contribute to the NMR signal, because immobilization by the large integral membrane protein complex caused the signal to broaden beyond detection, since cytochrome bc 1 -bound ubiquinones as-sume the correlation time of the membrane protein in the liposome, which is of the order of microseconds. Thus, by displacing the cytochrome bc 1 complex-bound ubiquinones with specific inhibitors, the number of bound ubiquinones can be inferred from the increase of the NMR signal.
Cytochrome bc 1 complex isolated from bovine heart mitochondria (14) was reconstituted into unilamellar proteoliposomes at a molar ratio of 3000 -4000 lipids per cytochrome bc 1 complex dimer (15). Signals were calibrated using [ 13 C]acetate as an internal standard (cf. Fig. 2) that remains in the water phase and neither interacts with the lipid membrane nor with the cytochrome bc 1 complex. Calculated concentrations took into account corrections for differences in the NMR T 1 relaxation times of the nuclei giving rise to the signals from ubiquinone and acetate (see Fig. 2). Controls using liposomes without protein (not shown) confirmed that it was possible to calculate the concentration of the mobile, unbound species from the integral of the [ 13 C]Q 0 C 10 signal in a two-dimensional HR-MAS heteronuclear single-quantum correlation (HSQC) experiment (16 -22). To assess unspecific binding, we measured immobilization of ubiquinone by reconstituted bovine heart cytochrome c oxidase (23), a membrane-bound complex of similar size but not containing a ubiquinone binding site. Unspecific binding was significant in the millimolar range and increased with the concentration of ubiquinone added to the sample. However, the extent of unspecific binding was very similar for cytochrome c oxidase and cytochrome bc 1 complex (with the specific sites blocked, see below), suggesting that it was largely due to weak association of ubiquinone molecules with the membrane domain of the complexes. ]Q 0 C 10 was 2.1 mM, as evaluated by integration to determine the volume under its cross-peak relative to that under the cross-peak from sodium acetate present at known concentration. Corrected integrals I corrected were obtained from the measured integrals I measured according to I corrected ϭ I measured /(1 Ϫ e Ϫ⌬/T1 ) with T 1 values for ubiquinone of 0.75 s and acetate of 3.43 s (determined by using the inversion recovery method (29) and the repetition delay of the HSQC pulse sequence including the acquisition time ⌬).

RESULTS AND DISCUSSION
To determine the number of specific ubiquinone binding sites of cytochrome bc 1 complex, we used high affinity inhibitors (24) with precisely known binding sites determined by x-ray crystallography (9) as well defined competitors. Counting the number of ubiquinone molecules per cytochrome bc 1 complex displaced by these highly specific inhibitors in the presence of saturating concentrations of ubiquinone avoided interference by nonspecific immobilization of ubiquinone by the cytochrome bc 1 complex. Unspecific binding was always observed at the high concentrations of ubiquinone necessary to saturate the rather weak ubiquinone binding sites of the cytochrome bc 1 complex under conditions of the NMR experiment and made direct quantitative binding studies impossible (not shown).
For the ubiquinone reduction site, x-ray structures show that a single ubiquinone shares a common binding pocket with the inhibitor antimycin (2,4). Competition was used to validate our approach. We added [ 13 C]Q 0 C 10 (1.2-3.6 mM) to cytochrome bc 1 complex (0.15-0.29 mM) in proteoliposomes; antimycin (0.3-1.5 mM) displaced 1.04 Ϯ 0.15 mol of ubiquinone from each mole of cytochrome bc 1 complex as expected from the molecular structure (Fig. 3). It should be stressed that this ratio was calculated using independently determined concentrations in the sample for the cytochrome bc 1 complex via UV-visible spectroscopy and for ubiquinone via referencing of the NMR integral to the NMR integral of acetate. Stigmatellin, a chromone-type inhibitor of the ubiquinol oxidation site, was found to displace 1.87 Ϯ 0.07 mol of ubiquinone/mol of cytochrome bc 1 complex. If both antimycin and stigmatellin where added to the cytochrome bc 1 complex proteoliposomes, 2.94 Ϯ 0.11 mol/mol were released. Stigmatellin is known to bind with much higher affinity to the ubiquinol oxidation pocket when the Rieske iron-sulfur protein is reduced (24). However, reducing this redox center by sodium ascorbate prior to the competition experiment had no effect on the displacement stoichiometries. In the presence of the E-␤methoxyacrylate inhibitor myxothiazol, a somewhat higher displacement ratio of 2.28 Ϯ 0.16 mol/mol was measured, but antimycin plus myxothiazol again displaced only 3.07 Ϯ 0.20 mol/mol. A possible explanation for the slight difference between the stoichiometries with stigmatellin and myxothiazol alone is that in contrast to stigmatellin, myxothiazol may have a weak affinity for the ubiquinone reduction site and could therefore partially displace ubiquinone in competition with antimycin. This seems feasible as other E-␤-methoxyacrylate inhibitors have been shown to be inhibitors of the plastoquinone reduction site of plastidial cytochrome b 6 f complex (25). None of the found stoichiometries was affected upon variation of the ubiquinone concentrations from 1.6 up to 3.6 mM or upon variation of the inhibitor concentrations from 0.33 up to 1.5 mM (see Fig. 3 for details). When less than 1.6 mM ubiquinone was present, lower stoichiometries were observed for stigmatellin and myxothiazol, but not for antimycin (not shown), indicating incomplete saturation of the ubiquinol oxidation (Q o or Q P ) site  Fig. 1), glutamine 272 changes its conformation upon movement of the ubiquinone. The figure was prepared with a version of MolScript (30) modified to enable color ramping (31). PDB entry 2BCC (32) was used as the starting structure for modeling of the chicken enzyme. At the ubiquinol oxidation pocket, the inhibitor stigmatellin was replaced by a pair of ubiquinone-7 molecules, and the system was subjected to energy minimization and molecular dynamics simulations at 300 K using the program CNS (33).  (2). Variation in inhibitor concentrations within these ranges and addition of sodium ascorbate in the case of stigmatellin had no effect on the stoichiometry of ubiquinone release from the cytochrome bc 1 complex. The error bars give the S.D. of all measurements listed for a given inhibitor or combination of inhibitors. and somewhat tighter binding of ubiquinone to its reduction (Q i or Q N ) center. This is in agreement with ubquinone occupancy in molecular structures (2) and seems characteristic for mitochondrial cytochrome bc 1 complex. It should be noted however that in the bacterial enzyme Q oS (cf. Fig. 1) was reported to have the highest affinity (11).
Our results clearly indicate that a total of three ubiquinones bind specifically to mitochondrial cytochrome bc 1 complex: one binds at the ubiquinone reduction center and is displaced by antimycin, and two bind at the ubiquinol oxidation center and are displaced by stigmatellin and myxothiazol. To test whether this finding is in accordance with structural data obtained by x-ray crystallography, we modeled two ubiquinone molecules carrying a long isoprenoid side chain that for technical reasons could not be used in the experiments into the ubiquinol oxidation pocket of the cytochrome bc 1 complex (Fig. 4). Two ubiquinone molecules could be accommodated by changes of the order of 1.5 Å in the atomic positions of a few neighboring amino acid residues. Movements on this scale are only slightly larger than those that have been observed experimentally for the removal of ubiquinone from the reaction center from Rhodopseudomonas viridis (26). Fig. 4 shows one set of possible conformations of the two quinones in the ubiquinol oxidation site; other conformations are possible.
Our equilibrium binding approach, using high concentrations of ubiquinone and cytochrome bc 1 complex, inherently provides no information on the functional meaning of the binding of two ubiquinone molecules at the ubiquinol oxidation center. However, our finding is in perfect agreement with the "double occupancy Q o site model" by Ding et al. (11,27), suggesting a functional role for two ubiquinones. In this complementary study, specific line shape changes in the EPR spectrum of the reduced Rieske iron-sulfur cluster were interpreted as reflecting the presence of two functionally interacting ubiquinone species in the ubiquinol oxidation pocket of bacterial cytochrome bc 1 complex. However, the indirect way in which ubiquinone binding was monitored in this approach allowed alternative interpretations of the data. Crofts and colleagues (10) proposed that the different line shapes may also reflect different states of the complex or different positions of the ubiquinone headgroup. However, our compelling result that two ubiquinone molecules bind to the ubiquinol oxidation pocket and are specifically displaced by inhibitors of this site corroborates the interpretation of Ding and colleagues. Together, both approaches provide strong support for the functional implications that have been based on the "double occupancy model" (11,28). In particular, charge transfer between ubiquinone and ubiquinol molecules, as implemented in the "proton-gated charge-transfer" mechanism (28), appears as a chemically attractive paradigm for the role of a ubiquinone pair in bifurcated electron flow at the ubiquinol oxidation center of the cytochrome bc 1 complex.
The HR-MAS approach presented may be widely employed to study binding of hydrophobic cofactors to membrane proteins. It should be useful for the analysis of ubiquinone binding to other respiratory chain complexes such as complex I and ubiquinol oxidase, for which specific inhibitors are available but the binding stoichiometries are uncertain. The method may develop into a general procedure for analyzing the binding of hydrophobic ligands to membrane bound proteins. If tighter binding reduces the problem of unspecific binding, quantitative binding studies with no need to use inhibitors will also be possible.