The iron-sulfur cluster of the Rieske iron-sulfur protein functions as a proton-exiting gate in the cytochrome bc(1) complex.

The destruction of the Rieske iron-sulfur cluster ([2Fe-2S]) in the bc(1) complex by hematoporphyrin-promoted photoinactivation resulted in the complex becoming proton-permeable. To study further the role of this [2Fe-2S] cluster in proton translocation of the bc(1) complex, Rhodobacter sphaeroides mutants expressing His-tagged cytochrome bc(1) complexes with mutations at the histidine ligands of the [2Fe-2S] cluster were generated and characterized. These mutants lacked the [2Fe-2S] cluster and possessed no bc(1) activity. When the mutant complex was co-inlaid in phospholipid vesicles with intact bovine mitochondrial bc(1) complex or cytochrome c oxidase, the proton ejection, normally observed in intact reductase or oxidase vesicles during the oxidation of their corresponding substrates, disappeared. This indicated the creation of a proton-leaking channel in the mutant complex, whose [2Fe-2S] cluster was lacking. Insertion of the bc(1) complex lacking the head domain of the Rieske iron-sulfur protein, removed by thermolysin digestion, into PL vesicles together with mitochondrial bc(1) complex also rendered the vesicles proton-permeable. Addition of the excess purified head domain of the Rieske iron-sulfur protein partially restored the proton-pumping activity. These results indicated that elimination of the [2Fe-2S] cluster in mutant bc(1) complexes opened up an otherwise closed proton channel within the bc(1) complex. It was speculated that in the normal catalytic cycle of the bc(1) complex, the [2Fe-2S] cluster may function as a proton-exiting gate.

The cytochrome bc 1 complex, also known as ubiquinol-cytochrome c reductase or complex III, is the central segment of the energy-conserving, electron transfer chain of the mitochondria and many respiratory and photosynthetic bacteria (1). This enzyme complex catalyzes electron transfer from ubiquinol to cytochrome c (c 2 in bacteria) with concomitant translocation of protons across the membrane to generate a proton electrochemical gradient required for ATP synthesis by ATP synthase. The cytochrome bc 1 complexes from all species contain three core subunits, cytochrome b, cytochrome c 1 , and Rieske iron-sulfur protein (ISP), 1 that house two b-type cytochromes (b 566 and b 562 ), one c-type cytochrome (c 1 ), and a high potential Rieske [2Fe-2S] cluster, respectively. However, the number of nonredox group containing subunits, also called supernumerary subunits, in the complex varies from species to species.
Recently, the three-dimensional crystal structures of mitochondrial bc 1 complexes from bovine (2,3), chicken (4), and yeast (5), which contain seven to eight supernumerary subunits in addition to the three core subunits, have been obtained. The structures of the cytochrome b 6 f complex, a complex analogous to the cytochrome bc 1 complex that provides an electronic connection between photosystems I and II, have also been established for the thermophilic cyanobacterium Mastigocladus laminosus (6) and in the algae Chlamydomonas reinhardtii (7), respectively. The three-dimensional structural information for the mitochondrial bc 1 complex establishes the location of the redox centers, the number of transmembrane helices, quinone binding at the Qi site, and inhibitor binding at both the Qo and Qi sites (2-4, 8, 9). Moreover, it suggests mobility of the extrinsic head domain of ISP during bc 1 catalysis. Strong evidence in support of this movement has been reported (4, 10 -13, 15-19).
The proton-motive Q-cycle model (20,21) has been favored for describing electron and proton translocation in the cytochrome bc 1 complex. The key feature of this model is the presence of two separate ubiquinol/ubiquinone-binding sites as follows: a ubiquinol oxidation site (Qo) near the P side (intermembranes space) of the mitochondrial inner membrane, and a ubiquinone reduction site (Qi) near the N side (matrix). Because of a lack of information on the binding of ubiquinol at the putative Qo site from the three-dimensional structures, the detailed bifurcation of ubiquinol oxidation, the key step in the Q-cycle mechanism is difficult to establish. Several models of the bifurcated oxidation of ubiquinol at the Qo site have been proposed (22)(23)(24)(25)(26)(27)(28).
It has been established that for every electron transferred through the bc 1 complex, two protons are translocated across the membrane. This 2H ϩ /e Ϫ stoichiometry has been verified in a wide variety of species, in vitro, by studying bc 1 complexes in phospholipid (PL) vesicles. Proton pumping by the bc 1 complex inlaid PL vesicles requires an intact membrane that prevents nonspecific proton leakage. Any compromise of membrane integrity results in loss of the proton electrochemical gradient, as is observed when an uncoupler like carbonyl cyanide m-chlorophenylhydrazone (CCCP) is added to either intact mitochondria or PL vesicles harboring electron transfer complexes. The integrity of the membrane is maintained by its protein and lipid components.
Although the proton translocation pathway in the cytochrome bc 1 complex is not fully understood, the involvement of the [2Fe-2S] cluster of ISP has been suggested. When bovine mitochondrial bc 1 complex is illuminated with a projector light in the presence of hematoporphyrin under aerobic conditions, the complex becomes inactivated as the [2Fe-2S] cluster in the treated complex is destroyed (29). When the photo-inactivated bc 1 complex is co-inlaid in PL vesicles with intact bovine bc 1 complex, no proton ejection is detected during ubiquinol oxidation. These findings suggest that a proton-permeable channel is created in the photo-inactivated complex. However, because the hematoporphyrin-promoted photo-inactivation site in the bc 1 complex is not specific, the formation of a proton-leaking channel in the photoinactivated complex may result from the oxidation of key amino acid residues in the bc 1 complex, rather than from destruction of the [2Fe-2S] cluster alone. Therefore, in order to establish unambiguously the involvement of the [2Fe-2S] cluster in the proton transfer pathway, a bc 1 complex either lacking the cluster or possessing a damaged one is necessary. We think that such a bc 1 complex can be obtained by site-directed mutagenesis of the histidine residues His-131 and His-152 of the ISP, which serve as ligands to the [2Fe-2S] cluster.
Here we report the generation and characterization of Rhodobacter sphaeroides strains expressing cytochrome bc 1 complexes with substitutions at one or both the histidine ligands of the [2Fe-2S] cluster. The proton-pumping activities of the protein-PL vesicles containing intact bovine mitochondrial bc 1 complex, or cytochrome c oxidase, and the mutant complex were determined and compared with those of protein-PL vesicles containing intact mitochondrial bc 1 complex or cytochrome c oxidase and wild-type bacterial bc 1 complex. Restoration of proton-pumping activity was explored by adding excess amounts of ISP head domain containing the [2Fe-2S] cluster to proton-leaking PL vesicles embedding mitochondrial bc 1 complex and thermolysin-digested bacterial bc 1 complex.
Generation of R. sphaeroides Strains Expressing the His 6 -tagged bc 1 Complexes with Mutation at the Histidine Ligand of the [2Fe-2S] Cluster of ISP-Mutations were constructed with the QuikChange TM sitedirected mutagenesis kit from Stratagene using a supercoiled doublestranded pGEM7Zf(ϩ)-fbcFB as template and a forward and reverse primer for PCR amplification. The pGEM7Zf(ϩ)-fbcFB plasmid (31) was constructed by ligating the EcoRI-XbaI fragment from pSELNB-3503 into the EcoRI and XbaI sites of the pGEM7Zf(ϩ) plasmid. The primers used were as follows: H131C(F), GCGTCTGCACCTGCCTCG-GCTGCGTGC, and H131C(R), GCACGCAGCCGAGGCAGGTGCAGA-CGC; H152C(F), GTTCTGCCCCTGCTGCGGATCGCACTACG, and H152C(R), CGTAGTGCGATCCGCAGCAGGGGCAGAAC; F and R in parentheses denote forward and reverse primers, respectively. The Eco-RI-XbaI fragment from the pGEM7Zf(ϩ)-fbcF m B plasmid was ligated into the pRKD418-fbcFB Km C H Q plasmid to generate the pRKD418-fb-cF m BC H Q plasmid. A plate-mating procedure was used to mobilize the pRKD418-fbcF m BC H Q plasmid from Escherichia coli S17-1 cells into R. sphaeroides BC-17 cells, as described previously (32). The presence of the engineered mutations was confirmed by DNA sequencing, before and after semi-aerobic growth of the cells. The expression plasmid, pRKDfbcF m BC H Q, was purified from an aliquot of semi-aerobically grown culture using the Qiagen Plasmid Mini Prep kit. Because R. sphaeroides cells contain four types of endogenous plasmids, the isolated plasmids lacked the purity and concentration needed for direct sequencing. Therefore, a 2-kb pair DNA segment containing the mutation sequence was amplified from the isolated plasmids by PCR and sequenced. DNA sequencing and oligonucleotide syntheses were carried out at the Recombinant DNA/Protein Core Facility at Oklahoma State University.
Growth of Bacteria-E. coli cells were grown at 37°C in LB medium. For semi-aerobic growth of the plasmid-bearing R. sphaeroides BC-17 cells, an enriched Sistrom's medium (33) supplemented with 20 amino acids and extra-rich vitamins were used. Cells harboring the mutated fbc genes on the pRKDfbcF m BC H Q plasmid were grown semi-aerobically for only one or two serial passages, in order to minimize any pressure for reversion. The semi-aerobic cultures were grown in 800 ml of enriched Sistrom's medium in 2-liter Bellco flasks with vigorous shaking (150 rpm) for about 26 h at 30°C in the dark. The inoculation volumes used for semi-aerobic growth were at least 5% of the total volume. Antibiotics were added to the following concentrations: ampicillin (125 g/ml), kanamycin sulfate (30 g/ml for E. coli and 20 g/ml for R. sphaeroides), tetracycline (10 g/ml for E. coli and 1 g/ml for R. sphaeroides), and trimethoprim (100 g/ml for E. coli and 30 g/ml for R. sphaeroides).
Enzyme Preparations and Activity Assay-Chromatophores and intra-cytoplasmic membranes (ICMs) were prepared as described previously (10) and stored at Ϫ80°C in the presence of 20% glycerol. The His 6 -tagged cytochrome bc 1 complexes were purified from frozen ICMs or chromatophores by the method of Tian et al. (10). Purified cytochrome bc 1 complexes were stored at Ϫ80°C in the presence of 10% glycerol. To assay ubiquinol-cytochrome c reductase activity, membrane preparations or purified cytochrome bc 1 complexes were diluted with 50 mM Tris-Cl, pH 8.0, containing 200 mM NaCl and 0.01% n-dodecyl ␤-D-maltopyranoside (DM) to a final cytochrome b concentration of 1 M. 5 l of the diluted samples were added to 1 ml of assay mixture containing 100 mM of sodium phosphate buffer, pH 7.4, 0.3 mM EDTA, 50 M ferricytochrome c, and 25 M Q 0 C 10 BrH 2 . Activities were determined by measuring the reduction of cytochrome c, by following the increase in the absorbance at 550 nm, in a Shimadzu UV-2101 PC spectrophotometer at 23°C using a millimolar extinction coefficient of 18.5 mM Ϫ1 cm Ϫ1 for calculations . The non-enzymatic oxidation of Q 0 C 10 BrH 2 , determined under similar conditions in the absence of the enzyme, was subtracted during calculations for the specific activity. Potassium cyanide was added to a final concentration of 30 M in the assay mixture to inhibit cytochrome c oxidase (CcO) when determining the bc 1 activity in ICMs or chromatophores.
Preparation of Electron Transfer Complex Inlaid PL Vesicles-Protein-PL vesicles were prepared by the cholate dialysis method of Kagawa and Racker (30). Bovine heart ubiquinol-cytochrome c reductase or CcO, either singly or in combination with R. sphaeroides wildtype or mutant bc 1 complex, was mixed with 1 ml of asolectin micellar solution to give an asolectin (mg)/protein (mg) ratio of 40. The asolectin micellar solution was prepared by sonicating 200 mg of acetone-washed asolectin in 4 ml of 50 mM sodium phosphate buffer, pH 7.4, containing 2% sodium cholate and 100 mM KCl (for use in preparation of bc 1 complex-PL vesicles), and in 4 ml of 100 mM Hepes-KOH, pH 7.3, containing 10 mM KCl and 2% sodium cholate (for use in preparation of CcO-PL vesicles), in an ice-water bath. Sonication was performed in an anaerobic environment by continually passing argon into the vessel.
The bc 1 complex-PL mixtures were incubated at 0°C for 30 min before overnight dialysis at 4°C against 100 volumes of 50 mM sodium phosphate buffer, pH 7.4, containing 100 mM KCl with three changes of buffer. The mixture was then dialyzed against 100 volumes of 150 mM KCl for 3-4 h. The CcO-PL mixture was dialyzed against 100 mM Hepes-KOH, pH 7.3, containing 10 mM KCl for 4 h followed by 10 mM Hepes-KOH, pH 7.3, containing 50 mM KCl and 50 mM sucrose with one change of buffer for 12 h each. Subsequently, the vesicles were dialyzed against two changes of 50 M Hepes-KOH buffer, pH 7.3, containing 55 mM KCl and 55 mM sucrose, for 12 and 4 h (34).
Determination of Proton Translocating Activity of bc 1 Complex-PL Vesicles-Proton translocation coupled to electron flow through the bc 1 complex-PL vesicles was measured at room temperature using an Accumet model 10 pH meter and a model 13-620-96 combination pH electrode. Twenty five nanomoles of Q 0 C 10 BrH 2 was added to the 1.6-ml reaction mixture containing 150 mM KCl, 4 M ferricytochrome c, 1 M valinomycin, and an appropriate amount of bc 1 -PL vesicles (30 -50 l). Electron flow was initiated by the addition of 5 nmol of ferricyanide, which oxidizes the cytochrome c, and thus provides an electron acceptor for the complex. Electron flow under conditions where no transmembrane ⌬pH formed was measured in an identical manner except that the protonophore, CCCP, was present at a concentration of 2 M to make the vesicles permeable to protons. Proton pumping (H ϩ /e Ϫ ) was calculated as the ratio of the decrease in pH upon ferricyanide addition to bc 1 -PL vesicles before and after treatment with CCCP.
Determination of Proton-pumping Activity of CcO-PL Vesicles-Proton-pumping experiments involving CcO-PL vesicles were conducted at 23°C in a stopped-flow spectrophotometer from Applied Photophysics, SX.18MV. Reaction vessel A contains 2 M CcO-PL vesicles, 110 M phenol red, 2 M valinomycin in 50 M Hepes-KOH, pH 7.3, and 55 mM KCl. Reaction vessel B contains 30 M horse heart ferrocytochrome c in the same buffer (34). Electron flow was initiated by mixing equal volumes of the reaction components in vessels A and B. Proton pumping was measured by monitoring the absorbance changes of phenol red at 556.6 nm. At this wavelength, the absorption change due to cytochrome c is minimal as it is the isosbestic point for cytochrome c. Because the pH of the components of vessels A and B were the same, no nonspecific absorbance changes should be observed.
Preparation of the Head Domain of ISP (ISF) from R. sphaeroides bc 1 Complex-The method used for the preparation of ISF from the mitochondrial bc 1 complex (35) was adapted to prepare the ISF from wildtype R. sphaeroides bc 1 complex. Two M wild-type R. sphaeroides bc 1 complex was suspended in 50 mM Tris-Cl, pH 8, containing 5 mM CaCl 2 and 100 mM NaCl. 0.2 M thermolysin was added, and the reaction mixture was incubated at room temperature for 3 h. Thermolysin digestion of the bc 1 complex was stopped by the addition of 10 mM EDTA. The EDTA-containing reaction mixture was dialyzed overnight against 10 mM NaCl, 10 mM EDTA, and 10 mM phosphate buffer, pH 7.2, with one change of buffer. The dialyzed sample was centrifuged at 150,000 ϫ g for 3 h. The precipitate was discarded, and the supernatant was applied to a calcium phosphate-cellulose column (36) equilibrated with 150 mM NaCl and 10 mM phosphate buffer, pH 7.2. Calcium phosphate was prepared according to Jenner (37) and mixed at a 3:1 ratio with cellulose powder, prior to use in column chromatography, for enhancement of the flow rate. The ISF did not bind to the column under these conditions and was recovered in the column effluent, which was then concentrated to about 0.5 mg/ml using Centriprep YM-3 from Amicon. The concentrated ISF was applied to a Mono-Q column (from Amersham Biosciences) equilibrated with 20 mM MOPS, pH 6.2, containing 20 mM NaCl. Thermolysin was not absorbed by the column and was therefore removed in the effluent. The ISF-absorbed column was subjected to a stepwise NaCl gradient wash, and ISF was eluted at 100 mM NaCl. The complete removal of thermolysin in the ISF-containing fractions was ensured by the absence of protease activity, assayed using casein as substrate.
Other Biochemical and Biophysical Techniques-Protein concentration was determined by the method of Lowry et al. (38). Cytochrome b (39) and cytochrome c 1 (40) concentrations were determined according to published methods. SDS-PAGE was performed according to Laemelli (41) using a Bio-Rad Mini-protean dual slab vertical cell. The polypeptides separated in the SDS-polyacrylamide gel were transferred electrophoretically to a 22-m nitrocellulose membrane for Western blotting. Polyclonal antibodies generated against ISP of the R. sphaeroides bc 1 complex were used as the primary antibody to detect their respective antigens (10). Protein A conjugated to horseradish peroxidase, from Bio-Rad, was used as the second antibody. The [2Fe-2S] cluster of ISP was determined by EPR, using a Bruker EMX spectrometer equipped with an Air Products flow cryostat, and by circular dichroism (42,43), using a Jasco J-715 spectropolarimeter. Instrument settings are detailed in the legends of the relevant figures.  H131N, H152N, and H131C, and a double mutant, H131C/H152C. For mutants to be useful for this study, the resulting mutant complexes must lack the [2Fe-2S] cluster but have protein and redox components similar to those in the wild-type complex.

Characterizations of Mutants
When aerobically dark grown wild-type and mutant cells were inoculated into enriched Sistrom's medium at mid-log phase and subjected to anaerobic photosynthetic growth conditions, all four mutants did not grow. However, these mutants grew aerobically and semi-aerobically at rates comparable with that of wild-type cells. These results indicated that the four mutants have a defective cytochrome bc 1 complex, because this complex is absolutely required for photosynthetic growth of this bacterium.
When ICMs were prepared from semi-aerobically grown mutant cells and assayed for ubiquinol-cytochrome c reductase activity, none was detected in the four mutant membranes. Absorption spectral analysis revealed that the content and absorption spectral properties of cytochrome b and cytochromes (c 1 ϩ c 2 ) in all these mutant membranes were similar to those in the complement chromatophores or ICM (see Table I), indicating that the mutation did not affect the assembly of cytochrome b and c 1 into the membrane. EPR analysis revealed that these four mutant membranes contained no [2Fe-2S] cluster of ISP, indicating that the mutation resulted in an inability of the [2Fe-2S] cluster to be ligated to apo-ISP, thereby leading to loss of bc 1 activity. Western blot analysis using antibodies against R. sphaeroides ISP showed that H131N, H152N, H131C, and H131C/H151C mutant ICMs had 9, 58, 64, and 98% of the apo-ISP in the wild-type chromatophores, respectively (see Table I). Apparently, the stability of apo-ISP was affected by the type of amino acid substituting for the histidine ligand and whether the alteration was at His-131 or His-152. The finding that the H152N mutant ICM had more apo-ISP than did the H131N mutant membrane was similar to observations in yeast mutants of H161R and H181R (44). However, it differed from reports on mutation at histidine ligands of the [2Fe-2S] cluster in R. capsulatus, in which less than 3% of the apo-ISP was found in the mutant membranes (45).
When the His 6 -tagged bc 1 complexes were purified from the ICM of the four mutants, none of them had bc 1 activity or the [2Fe-2S] cluster of ISP. As shown in Fig. 1, EPR signals at g ϭ 2.03, 1.90, and 1.80, characteristics of the [2Fe-2S] cluster of ISP, were observed only in the purified wild-type complex. Similarly, a negative CD peak around 500 nm, a characteristic of the [2Fe-2S] cluster of ISP, was observed only in the purified wild-type complex.
The b/c 1 ratios in the complement and mutants H131N, H152N, H131C, and H131C/H152C, determined by absorption spectra (see Fig. 2   1.10 respectively, indicating that the binding affinity between cytochrome b and cytochrome c 1 was affected by the nature and position of the substituting amino acids. A decrease in the b/c 1 ratio indicates a decrease in binding affinity between the two cytochromes. Because the expressed R. sphaeroides bc 1 complex was His 6 -tagged at the C terminus of cytochrome c 1 , all cytochrome c 1 from the dodecyl maltoside-solubilized mem-brane, regardless of whether or not it was associated with cytochrome b, was absorbed to the Ni-NTA column used in the one-step purification of the complex. Cytochrome b, which was not tightly associated with cytochrome c 1 , will appear in the effluent. Consequently, the eluted complex would have a lower b/c 1 ratio as compared with the wild-type complex. The possibility that the lower cytochrome b/c 1 ratio observed in the mutant complexes resulted from a difference in the effectiveness of DM solubilization of the mutant complex from the membrane has been ruled out. DM solubilization of the bc 1 complexes from the mutant membranes was comparable with that of the complement (wild-type) chromatophores. It should be mentioned that EPR signals for both b L (g ϭ 3.76) and b H (g ϭ 3.49) were detected in the H131N mutant complex, indicating that this mutation did not affect the heme environments of cytochrome b, despite the loss of more than 65% of cytochrome b from the DM-solubilized membrane during purification by an Ni-NTA column. The EPR signals for b L and b H were observed in all of the [2Fe-2S] clusters lacking mutants (Fig. 3). Fig. 4 shows Western blot analysis of ISP in the complement and mutant complexes. Although the purified H131N mutant complex contained no detectable apo-ISP or subunit IV, purified mutant complexes of H152N, H131C, and H131C/H152C have 40, 55, and 82%, respectively, of the amount of apo-ISP or subunit IV found in the complement complex. When the latter three mutant complexes were subjected to SDS-PAGE analysis, four major protein bands corresponding to cytochrome b, cytochrome c 1 , ISP, and subunit IV were observed, as in the complement wild-type complex (see Fig. 4). Thus, a significant amount of ISP, 40 -82%, was still present in the [2Fe-2S] cluster-lacking mutant complexes but not in H131N. It has been reported that a yeast mutant strain whose [2Fe-2S] cluster is lacking has about 83% of the ISP, relative to cytochrome c 1 , in the purified complex, when compared with its wild-type counterpart (46).
Clearly, the assembly of the bc 1 complex depends critically on the structural integrity of the head domain of ISP, particularly at the tip of the ISP where interaction between ISP and cytochrome b takes place. A unique feature of the ISP head domain structure is the tip area where the [2Fe-2S] cluster is located; the surface topology of the tip is strikingly smooth with only main chain atoms facing the solvent environment. Substitutions of H131C and H152C by molecular modeling using the bovine ISP structure as a template (Fig. 5) demonstrated the formation of two possible disulfide bonds, between Cys-141 and

FIG. 2. Absorption spectra of purified bc 1 complexes of the complement wild-type (A) and mutants lacking the [2Fe-2S] cluster, H131C/H152C (B), H131C (C), H131N (D), and H152N (E).
Each optical spectrum is a calculated difference spectrum of the dithionite-reduced minus ferricyanide-oxidized cytochromes (A max ϭ 551 nm for cytochrome c 1 and A max ϭ 560 nm for cytochrome b). mutants H131N (B), H152N (C), H131C (D), and H131C/H152C (E). Sample preparations and instrument settings were the same as in Fig.  1, except that the microwave power used was 108.1 milliwatts, and the modulation amplitude was 20 G.

FIG. 3. EPR spectra of cytochromes b in the cytochrome bc 1 complexes of complement (A) and the [2Fe-2S] cluster-lacking
Cys-161 for one pair and between Cys-139 and Cys-158 for the other pair, in the absence of the [2Fe-2S] cluster. Together with the disulfide pair in the native structure between Cys-144 and Cys-160, the three pairs of disulfide bridges in the mutant would stabilize the structure of ISP at the tip region and, more importantly, perhaps maintain the smooth surface topology of the wild-type ISP, whereas any of the other single amino acid substitutions would inevitably destroy this important feature to a different extent. The H131N and H152N mutations were particularly unfavorable due to exposed large side chains.
Among the four [2Fe-2S] cluster-lacking mutant bc 1 complexes, H131C and H131C/H152C are suitable for probing the function of the [2Fe-2S] cluster in proton translocation, because they differ from the wild-type complex only in the lack of the [2Fe-2S] cluster. The contents of the subunits and spectral properties of cytochromes b and c 1 in these two mutant complexes were similar to those in the complement complex. Moreover, the amount of ISP in the purified complexes was over 50% compared with the complement (wild type) complex. Therefore, any changes in proton translocation activity detected in these two mutant complexes can be unambiguously attributed to the lack of a [2Fe-2S] cluster.
Proton Translocation Activity of Intact Mitochondrial bc 1 Complex Co-inlaid in PL Vesicles with Wild-Type and H131C/ H152C Mutant of R. sphaeroides Complexes-One important property of the cytochrome bc 1 complex is the ability to pump protons across the membrane during electron flow from ubiqui-nol to cytochrome c. This property has been well demonstrated in PL vesicles embedding the purified mitochondrial bc 1 complex (47)(48)(49)(50). Fig. 6A shows a typical proton translocation activity assay for PL vesicles embedding the bovine bc 1 complex. After pH equilibrium was reached for a mixture containing Q 0 C 10 BrH 2 , valinomycin, ferricytochrome c, and bovine bc 1 complex-PL vesicles, an aliquot of ferricyanide solution (addition 1) was added to initiate electron flow from Q 0 C 10 BrH 2 to cytochrome c. After pH equilibrium was attained, protonophore CCCP was added (addition 2) to render the liposome membrane freely permeable to protons, and then a second aliquot of ferricyanide was added (addition 3). The ratio of the pH changes produced by the addition of equal amounts of ferricyanide, before (x) and after (y) the addition of CCCP, was taken as a measure of the H ϩ /e Ϫ ratio for proton translocation activity of the bc 1 complex. The protons released after CCCP addition were the "scalar" protons only, as no contribution from the accumulation of "vectorially" translocated protons was possible. Ratios (i.e. x/y) between 1.6 and 2.0 are routinely obtained with liposomes containing the bovine mitochondrial bc 1 complex (48).
When wild-type R. sphaeroides bc 1 -PL vesicles, prepared in the same manner as that of mitochondrial-PL vesicles, were subjected to proton-pumping activity measurement, an H ϩ /e Ϫ ratio of 1.5-1.6 was obtained (see Table II). As expected, no proton translocation activity was found in PL vesicles embedding the H131C/H152C mutant complex because it lacked ubiquinol-cytochrome c reductase activity. When the wild-type bacterial bc 1 complex was co-embedded in PL vesicles with the intact mitochondrial bc 1 complex, an H ϩ /e Ϫ ratio of 1.7 was obtained. However, when the H131C/H152C mutant complex was co-inlaid in PL vesicles with the mitochondrial bc 1 complex, no proton pumping was observed, i.e. H ϩ /e Ϫ ratio equals 1.0 (see Fig. 6B and Table II). This indicates that a protonleaking channel was present in the H131C/H152C mutant complex whose [2Fe-2S] cluster was lacking. Incorporation of the mutant complex into actively proton-pumping liposomes had a similar effect as the addition of CCCP; they both allowed the free flow of protons across the vesicle, preventing the generation of a gradient.
It should be noted that the proton translocation activity (H ϩ /e Ϫ ) in protein-PL vesicles containing intact mitochondrial bc 1 and the H131C/H152C mutant complexes decreased as the relative amount of mutant complex in the vesicle increased. A complete loss of proton-pumping activity was observed when the mutant complex concentration was 3-fold that of the mitochondrial bc 1 complex. An excess of the [2Fe-2S] cluster-lacking mutant complex was necessary to ensure that every PL vesicle harbored at least one mutant complex in addition to a mitochondrial bc 1 complex.

Proton Translocation Activity of Intact Mitochondrial Cytochrome c Oxidase Co-inlaid in PL Vesicles with Wild-type and
Mutant Bacterial Complexes-The ability of the H131C/H152C mutant bc 1 complexes to form proton-leaking channels was further demonstrated by co-embedding with bovine mitochondrial CcO. Like the cytochrome bc 1 complex, electron transfer through CcO is coupled to proton translocation across the membrane in which the complex is housed. When purified bovine CcO was embedded in PL vesicles, the accumulation of vectorially translocated protons was detected during the oxidation of reduced cytochrome c as indicated by acidification of the external medium (Fig. 7). However, when CcO was co-inlaid in PL vesicles with a 5-fold or higher molar excess of the H131C/ H152C mutant bc 1 complex, the vesicles produced an instant alkalization phase during the oxidation of reduced cytochrome c. The pattern of pH increase was similar to that observed in CCCP-treated PL vesicles embedding CcO, thereby further indicating the presence of a proton-leaking channel in the bc 1 complex lacking the [2Fe-2S] cluster. It should be noted that the proton translocation activity of CcO remained unchanged when co-inlaid in PL vesicles with a 5-fold molar excess of wild-type bc 1 complex.
Restoration of Proton Translocation Activity to Protein-PL Vesicles Containing Intact Mitochondrial bc 1 and the Thermolysin-digested Bacterial Complex with the Head Domain of ISP-Because the cytochrome bc 1 complex lacking the [2Fe-2S] cluster had a proton-leaking channel, the cluster may function as a proton-exiting gate regulating the controlled, vectorial extrusion of protons across the bc 1 complex. If removal of the [2Fe-2S] cluster of ISP made the bc 1 complex proton-permeable, a bc 1 complex with the ISP head domain removed should be similar, because the [2Fe-2S] cluster is located in the head domain of ISP. To confirm this prediction, a bacterial bc 1 complex lacking the ISP head domain was prepared by thermolysin digestion and co-inlaid into PL-vesicles with intact mitochondrial bc 1 complex. The proton-pumping activity of the resulting protein-PL vesicles was measured. No proton pumping was observed in protein-PL vesicles containing thermolysin-digested bacterial bc 1 complex and mitochondrial bc 1 complex at an 8:1 molar ratio. The requirement of a large excess of thermolysin-digested complex to abolish completely the proton transfer activity may be due to incomplete removal of the head domain of ISP by thermolysin. To examine this possibility, the digested complex, which has no ubiquinol-cytochrome c reductase activity, was subjected to SDS-PAGE and Western blot analysis. About 20% of the ISP in the complex was resistant to thermolysin digestion (see Fig. 8). Also, the extent of Rieske ISP cleaved by thermolysin in the H131C/H152C mutant complex was similar to that observed for the wild-type complex (Fig. 8). This indicated that the thermolysin cleavage site in the Rieske ISP was equally accessible to the protease in both the H131C/H152C mutant and wild-type complexes. Thus, the loss of the [2Fe-2S] cluster did not result in a conformational change of the Rieske ISP. Similar observations of incomplete ISP digestion are made with R. capsulatus, where incubation of the bc 1 complex with thermolysin for prolonged periods, with fresh additions of protease at 4-h intervals, does not yield complete digestion (51).
Addition of excess purified ISF to proton-leaking PL vesicles co-embedding mitochondrial bc 1 and thermolysin-digested bacterial bc 1 complex partially restored the proton-pumping capability with H ϩ /e Ϫ ratios ranging from 1. 35    introduction of the [2Fe-2S] cluster containing the head domain seals the proton channel across the bc 1 complex, preventing the uncontrolled and unimpeded proton flow across the protein-PL vesicles. As expected, the addition of excess ISF to PL vesicles embedding the mitochondrial bc 1 complex and the mutant complex, H131C/H152C, did not restore the proton-pumping activity because the mutant complex lacked only the [2Fe-2S] cluster, not the entire ISF, providing no room for added ISF to dock. Similarly, the addition of excess ISF to PL vesicles embedding the mitochondrial bc 1 complex and the 2-band bc 1 complex (a mutant bc 1 complex containing only the cytochrome b and cytochrome c 1 subunits) did not restore the proton-pumping capability of such vesicles. The tail domain of ISP, which is present in the thermolysin-digested complex and absent in the 2-subunit bc 1 complex, may play a role in maintaining the conformation of the bc 1 complex required for docking of ISF.
The Existence of Proton Pathway in the Cytochrome bc 1 Complex-The observation of proton leakage in the bc 1 complex with damaged ISP suggested a passage in the complex, channeling proton backflow from the periplasmic to the cytoplasmic side (from the inter-membrane space to matrix in mitochondria). It is unlikely, although it cannot be ruled out, that such a channel exists in the trans-membrane (TM) region at the interfaces of different subunits, such as that between cytochromes b and c 1 , because the only damage needed to produce the proton leakage is the removal of the [2Fe-2S] cluster. It is conceivable that the culprit lies in the cytochrome b subunit as it contributes most to the TM region of the complex. Based on the crystallographically refined structures (5,8,9), a solvent distribution around the subunit cytochrome b was generated (Fig. 9), from which three principal solvent-accessible regions in cytochrome b could be defined where water molecules penetrate deep into the subunit; they are the Qi site, the Qo site, and the dimer interface at the matrix side (Fig. 9). There is, however, a gap of roughly 17 Å in the mid-section of the TM domain where no ordered water molecules were found crystallographically.
Proton transfer pathways involving internal water molecules that provide hydrogen bonds and facilitate proton diffusion have been identified in other membrane proteins like bacteriorhodopsin (52) and CcO (53). Although the mechanisms for proton translocations are different for different proteins, the underlying principle for bringing water molecules or protons to the active sites is similar. Several pathways for proton leakage are possible. As indicated in Fig. 9, the proton entry for the leakage must be located at the Qo site where protons are ejected under normal circumstances. All other venues near the intermembrane space side are sealed. In the native cytochrome bc 1 complex, the [2Fe-2S] cluster undergoes redox state changes during the catalytic cycle. The ejection of protons must be controlled by the protonation and deprotonation of the histidine ligands of the [2Fe-2S] cluster. The histidine ligands uptake a proton from the substrate, ubiquinol, upon reduction of the [2Fe-2S] cluster and release them to the inter-membrane space when oxidized by cytochrome c 1 , as was suggested recently (14). It is less clear how the second proton is pumped out, although Glu-271 of cytochrome b is speculated to be the acceptor of the second proton from ubiquinol (14). The absence of the histidine ligands, and thus the loss of the [2Fe-2S] cluster, explain well the loss of the proton-pumping capability in PL vesicles incorporating the mutant bc 1 complexes. The exact mechanism of how protons leak into the Qo site in the absence of the [2Fe-2S] cluster remains to be seen.
There are two likely exit pathways for the proton leakage; one is via the Qi site, and the other is through a hole at the dimer interface (Fig. 9). We have suggested earlier (9) that two residues in the Qi pocket may be involved in fetching protons from the matrix side; Lys-227 and His-201 undergo conformational changes that are coupled to ubiquinone reduction at the Qi site. The same residues could serve as an exit gate in the case of proton leakage, except that the gate is decoupled to ubiquinone reduction and is spontaneously open. Alternatively, the proton could exit from the hole at the dimer interface near the matrix side (Fig. 9). The hole is just outside His-201 and is filled with ordered water molecules.
The possibility of a direct proton translocation path going through the interior of cytochrome b is slim because the overall structure is rather rigid as analyzed by the binding of various inhibitors (8). A likely pathway linking the entrance at the Qo site to the exit gate at the Qi site for spontaneous proton leakage is the large cavity formed between the two symmetryrelated cytochrome b subunits. Although this cavity is fairly hydrophobic, one could argue that the existence of the membrane potential (pH gradient) could overcome the thermodynamic barrier and facilitate proton movement from one side to the other. Heme moieties are given as the ball-and-stick models and are as labeled. Highlighted in a purple circle is the Qo site; in green circle is the Qi site; and in red rectangle is the hole at the dimer interface near the matrix. The central cavity is also indicated.