Direct Demonstration of Half-of-the-sites Reactivity in the Dimeric Cytochrome bc1 Complex

We previously proposed that the dimeric cytochrome bc1 complex exhibits half-of-the-sites reactivity for ubiquinol oxidation and rapid electron transfer between bc1 monomers (Covian, R., Kleinschroth, T., Ludwig, B., and Trumpower, B. L. (2007) J. Biol. Chem. 282, 22289–22297). Here, we demonstrate the previously proposed half-of-the-sites reactivity and intermonomeric electron transfer by characterizing the kinetics of ubiquinol oxidation in the dimeric bc1 complex from Paracoccus denitrificans that contains an inactivating Y147S mutation in one or both cytochrome b subunits. The enzyme with a Y147S mutation in one cytochrome b subunit was catalytically fully active, whereas the activity of the enzyme with a Y147S mutation in both cytochrome b subunits was only 10–16% of that of the enzyme with fully wild-type or heterodimeric cytochrome b subunits. Enzyme with one inactive cytochrome b subunit was also indistinguishable from the dimer with two wild-type cytochrome b subunits in rate and extent of reduction of cytochromes b and c1 by ubiquinol under pre-steady-state conditions in the presence of antimycin. However, the enzyme with only one mutated cytochrome b subunit did not show the stimulation in the steady-state rate that was observed in the wild-type dimeric enzyme at low concentrations of antimycin, confirming that the half-of-the-sites reactivity for ubiquinol oxidation can be regulated in the wild-type dimer by binding of inhibitor to one ubiquinone reduction site.

The cytochrome bc 1 complex is a multisubunit enzyme that generates a transmembrane protonmotive force by transferring electrons from ubiquinol to cytochrome c. Energy conservation in this enzyme complex is ensured by the oxidation of ubiquinol and the reduction of ubiquinone at active sites located on opposite sides of the membrane, as described by the protonmotive Q-cycle (1). High resolution structures have shown that the bc 1 complex is a dimer composed of two copies of cytochrome b, the Rieske iron-sulfur protein, and cytochrome c 1 , which are the only polypeptides present in some bacterial enzymes (2), in addition to six to eight additional subunits present exclusively in each monomer of the eukaryotic complex (3)(4)(5).
The functional relevance of this dimeric arrangement has been supported by an extensive body of kinetic evidence (for review, see Ref. 6), which includes the key observations that only one ubiquinol oxidation site, or center P, is active when the two ubiquinone reduction, or center N, sites are occupied by inhibitors (7), that electrons rapidly equilibrate between the cytochrome b subunits (8,9), and that there is conformational communication between center P and center N sites (10). We have proposed that this half-of-the sites activity at center P also exists under normal steady-state conditions in the absence of inhibitors and that this mechanism minimizes the leakage of electrons to oxygen under conditions that would favor the accumulation of electrons in the cytochrome b hemes (8,11). Evidence consistent with this proposed dimeric mechanism has been obtained independently by other research groups (12)(13)(14). However, other interpretations of these data have been proposed to support a strictly monomeric mechanism of action in the bc 1 complex (15).
We have recently used the relatively simple three-subunit bc 1 complex from Paracoccus denitrificans to show that the inhibitor stigmatellin, which is considered to be a mimetic of ubiquinol bound at center P, binds with different rates and spectral effects to one of the center P sites in the dimer, thereby supporting the functional relevance of the dimeric structure of this enzyme (16). In the present work, we have generated P. denitrificans bc 1 heterodimers in which one center P site is largely inactivated by a mutation in cytochrome b, and we have determined its kinetics in comparison with wild-type and mutant homodimers. Our results directly demonstrate the half-of-the sites activity of the bc 1 complex dimer both in the presence and absence of center N inhibitors and confirm other postulates of the dimeric mechanism that we have proposed (6), such as the existence of intermonomeric electron transfer and the activa-tion of the second center P site upon binding of inhibitor to one center N site in the dimer.

EXPERIMENTAL PROCEDURES
Materials-Dodecylmaltoside was obtained from Anatrace. Stigmatellin, antimycin, decylubiquinone, and horse heart cytochrome c were purchased from Sigma. Decylubiquinol was prepared from decylubiquinone as described before (17). All inhibitors as well as decylubiquinol were quantified by UV spectroscopy (18) using reported extinction coefficients (19,20).
Cloning Procedures-For the construction of the fbc operon with the deletion of the N-terminal acidic domain in cytochrome c 1 , a 1.5-kb SmaI-PstI cassette was subcloned into pTZ18R (Pharmacia). The resulting plasmid was then cut with XhoI and NotI in a double digestion, followed by the removal of the sticky ends using mung bean nuclease. This double digestion removed the sequence encoding the acidic domain (amino acids 39 -203 in cytochrome c 1 ) in-frame. After re-ligation of the vector and transformation, the newly obtained SmaI-PstI cassette was then subcloned into the cloning vectors described below.
The PCR products were SOE1 and SOE2 (580 and 1287 bp, respectively), obtained using the primer 5Ј-3ЈPCytbC-termHis-t_ext_ord and SOEing1 for the first fragment and 3Ј-5ЈPCytbC-termHis-t_ext_ord and SOEing4 for the second fragment. The two products, annealing via 30 bp at the 3Ј-end of SOE1 and the 5Ј terminus of SOE2, were then mixed together as templates for the third PCR, with the primers SOEing1 and SOEing4. This created a product of 1867 bp. The fragment was blunt-endligated in an empty pSL1180 vector and then cloned by Pfl23II and HindIII restriction into the expression vector pRI2 (21) containing the fbc operon. This plasmid (pMC1) was used to transform competent DH5␣ cells, and the resulting strain was used in a triparental mating procedure to transfer the expression vector into the MK6 P. denitrificans strain (22).
The Y147S mutation was obtained by one-primer site-directed mutagenesis of the wild-type fbc operon with the primer bY147SEco24I (5Ј-CCGCCTTCATGGGCTCGGTCCTGCC-CTGG-3Ј). Sequencing confirmed that mutagenesis was successful. The mutation was then subcloned via SacI-Pfl23II into a pUC18 vector containing a copy of the fbc operon with the deletion of the acidic domain in Cyt c 1 and the His tag. The clones obtained were screened by Eco24I digestion because the introduction of the Y147S mutation eliminates a restriction site for this enzyme in the operon. The clones showing the desired restriction pattern were then sequenced, and the newly obtained vector was digested with HindIII and SacI to subclone the fbc operon with mutation Y147S into the expression vector pEG400 (24). Clones were screened by digestion with the previously mentioned enzymes, and the resulting plasmid was named pMC12 (Fig. 1). The two plasmids, pMC12 and pTK56, were used to transform competent DH5␣ cells, and the resulting strains were used in a four-strain mating procedure to transfer the expression vectors simultaneously into the MK6 P. denitrificans strain (22) selected on rifampicin, kanamycin, gentamycin, and streptomycin, resulting in a strain stably containing both vectors.
Purification of Cytochrome bc 1 Complex-Cells from an overnight growth were harvested at an OD value of 3-5 and resuspended in a buffer containing 100 mM sodium phosphate, pH 8, and 1 mM EDTA, and frozen. Membranes were obtained as described previously (22). To solubilize the bc 1 complex, membranes were diluted to a protein concentration of 35 mg/ml. The suspension was then diluted 1:1 with solubilization buffer (50 mM sodium phosphate, pH 8, 300 mM sucrose, 1.2 M NaCl, with addition of trace amounts of avidine to titrate endogenous biotin compounds in the case of the Strep-tagged protein), and dodecylmaltoside was added to a final ratio 1.15:1 with respect to total protein. Solubilization of the complex was done on ice with stirring for 1 h. After centrifugation at 100,000 ϫ g for 1 h, the supernatant was filtered and loaded onto the appropriate column.
Mutant homodimer purification was performed using a Ni 2ϩ -NTA column, previously equilibrated with 6 column volumes of buffer A (50 mM sodium phosphate, pH 8, 300 mM NaCl, 0.02% dodecylmaltoside). The column was then washed with 1 column volume, and a 0 -200 mM histidine gradient in the same buffer (5 column volumes) was used to elute the complex. Wild-type homodimers were bound to a StrepTactin column (IBA), equilibrated with 5 column volumes of buffer A, and washed with 2 column volumes. A one-step elution was used with 1 column volume of buffer A containing 2.5 mM desthiobiotin. The heterodimer was purified using the 2 columns sequentially in tandem purification (first Ni 2ϩ -NTA, then StrepTactin), following the protocols described above. Fractions from the purifications were collected and analyzed by redox difference spectra, SDS-PAGE, and Western blotting, to define purity and protein composition. Protein concentrations were calculated using the extinction coefficients of 23.2 mM Ϫ1 cm Ϫ1 at 553-539 nm (25) and 25.6 mM Ϫ1 cm Ϫ1 for the average of the two hemes in cytochrome b (26). All enzyme samples yielded ratios very close to 2 b hemes:cytochrome c 1 .
Pre-steady-state Reduction of bc 1 Complexes-Pre-steadystate reduction of cytochrome c 1 and cytochrome b was followed at room temperature by stopped flow rapid scanning spectroscopy using the OLIS Rapid Scanning Monochromator as described before (7,27). Reactions were started by rapid mixing of 2 M enzyme with or without 2.5 M antimycin in assay buffer containing 100 mM Tris-HCl, pH 8.0, 2 mM sodium azide, 0.2 mM EDTA, and 0.05% dodecylmaltoside against an equal volume of the same buffer containing the indicated concentration of decylubiquinol. For each experiment, 8 data sets were averaged after subtracting the oxidized spectrum. The time course of absorbance change at the appropriate wavelengths was extracted using software from OLIS and exported to the Origin 5.0 program (OriginLab Corp.). The contribution of cytochrome b to c 1 absorbance was calculated and corrected for as previously described (7). Absorbance changes as a function of time were fitted to monophasic or biphasic exponential functions in Origin.
Steady-state Activities of bc 1 Complexes-Horse heart cytochrome c reduction was followed at room temperature by stopped flow rapid scanning spectroscopy using the OLIS Rapid Scanning Monochromator. Reactions were started by rapid mixing of 100 nM enzyme and 20 M cytochrome c in assay buffer against an equal volume of the same buffer containing the indicated concentration of decylubiquinol. For each experiment, 6 data sets were averaged after subtracting the oxidized spectrum. The time course of absorbance change between 550 and 539 nm was extracted using software from OLIS and exported to the Origin 5.0 program, and the initial increase in absorbance was fitted to a linear regression function and expressed as turnover number (moles of cytochrome c reduced/ mol of bc 1 monomer/s) using an extinction coefficient of 21.5 mM Ϫ1 cm Ϫ1 (28). Initial rates at different decylubiquinol concentrations were fitted to the Michaelis-Menten equation to obtain V m and K m values. For the antimycin titration experiments, the indicated concentration of inhibitor was added to a 10 M solution of enzyme on ice and incubated for at least 2 min before dilution in assay buffer at room temperature to a concentration of 100 nM enzyme in the presence of 20 M cytochrome c, and rapid mixing against an equal volume of buffer containing 80 M decylubiquinol. The very low K d for antimycin (ϳ30 pM) (19) allows for the stoichiometric binding of the inhibitor to be preserved FIGURE 1. fbc operon and mutagenesis of the operon to construct a heterodimeric cytochrome bc 1 complex with one active and one inactive monomer. Upper panel, P. denitrificans fbc operon and its three constituent genes, fbcF, encoding the Rieske iron-sulfur protein; fbcB, encoding cytochrome b; and fbcC, encoding cytochrome c 1 . WT is the operon with no mutations and includes the acidic N terminus that is normally present on cytochrome c 1 . MC1 is the construct in which the acidic N terminus has been deleted. TK56 is the construct in which the acidic N terminus has been deleted and the Strep tag has been encoded into the cytochrome b gene and with no mutations in fbcB. MC12 is the N-terminal deletion construct in which a His tag has been encoded into the fbcB gene and a Y147S mutation has been introduced into the cytochrome b gene. Lower panel, three redox subunits of the heterodimeric cytochrome bc 1 complex in which one cytochrome b subunit carries a His tag and contains a Y147S mutation that extensively inactivates ubiquinol oxidation at the center P (Q p ) site. The other cytochrome b subunit carries a Strep tag II and no cytochrome b mutation. even after diluting the enzyme to 100 nM. The time course of absorbance change at 550 versus 539 nm from 6 individual data sets/antimycin concentration was extracted using software from OLIS and exported to the Origin 5.0 program, where rates were calculated as described above. Initial rates at each antimycin concentration were fitted to kinetic models published previously (7) that assume that dimers with one antimycin bound contribute equally or twice as much as free dimers to the overall steady-state rate.

RESULTS
Construction of a Fully Wild-type Dimeric Cytochrome bc 1 Complex, a Mutant/Wild-type Heterodimeric bc 1 Complex, and a Homodimeric Mutant bc 1 Complex by Mutagenesis of the P. denitrificans fbc Operon-The fbc operon of P. denitrificans is composed of three structural genes, encoding for the three subunits in the complex: the Rieske iron-sulfur protein, cytochrome b, and cytochrome c 1 (Fig. 1). Three different domains can be identified in the last protein: an N-terminal, acidic domain, a "core domain" coordinating the c type heme, and a C-terminal transmembrane domain. In this work, a cytochrome c 1 mutant has been used in which the acidic domain was deleted (24) to convert the enzyme from a dimer of dimers (29) to a dimer, as judged by the molecular mass obtained by gel filtration and mass spectrometry when this domain is deleted, without affecting the catalytic rate at center P, as will be described in detail in a future publication. 6 The construction of a heterodimeric bc 1 complex (Fig. 1) in P. denitrificans was achieved by transforming a bc 1 complexdeficient strain (MK6) with two plasmids: pMC12 and pTK56. The two plasmids were conjugated at the same time, using a four-strain mating procedure. Plasmid DNA obtained from the resulting colonies was screened with the restriction endonucleases BglII and Bpu1102I, cutting inside the antibiotic resistance genes of pTK56 (Gm) and pMC12 (Sm), respectively. Four positive clones were identified, followed by large scale expression and membrane preparation.
Two different affinity columns were used to achieve purification of the heterodimer, a Ni 2ϩ -NTA and a StrepTactin column ( Fig. 2A). From the clones transformed with both plasmids, three differently tagged dimers can be expressed (two His-tagged monomers, two Strep-tagged monomers, and the heterodimeric complex with one His and one Strep tag). The first column (Ni 2ϩ -NTA), retains the His-tagged dimer and the heterodimeric bc 1 complex. These could subsequently be separated using the StrepTactin column. In Fig. 2B, the Western blot analysis of the different fractions from the purification shows the presence of the expected tags in the fractions obtained from each column, with the heterodimer containing both tags.
Crystallographic structures (2)(3)(4)(5) show that the bc 1 complex dimer is stabilized by the transmembrane helices of the Rieske iron-sulfur protein, which traverse from one monomer to the other, and by the N terminus helical region of each cytochrome b subunit. The tilted arrangement of the iron-sulfur proteins is in turn stabilized by multiple interactions with cytochrome b and cytochrome c 1 . Therefore, re-equilibration of cytochrome b subunits between different dimers is not possible because it would require complete dissociation of the subunits of the two monomers of the complex and reassembly by in vivo factors that are not present in the purified enzyme solution.
Pre-steady-state Oxidation of Ubiquinol at Center P of the Fully Wild-type Dimeric, the Mutant/Wild-type Heterodimeric, and the Homodimeric Mutant bc 1 Complexes-In the absence of inhibitors, decylubiquinol reduced cytochrome c 1 under presteady-state conditions through center P with the rates shown in Fig. 3. Both the fully wild-type dimer and the mutant homodimers with the Y147S/Y147S cytochrome b mutations exhibited monophasic kinetics on the time scale shown but with a rate 6-fold lower in the case of the mutant dimer. The heterodimer, in contrast, exhibited biphasic kinetics with a fast and a slow rate identical to that obtained in the wild-type and mutant homodimers, respectively. All three dimer samples showed an initial lag in the reduction of cytochrome c 1 , which lasted longer in the case of the mutant homodimer. This lag phase could reflect a relatively slow binding or partitioning of the substrate toward center P immediately after rapid mixing.
The extent of reduction of each phase in the heterodimer was approximately half of the total absorbance change, indicating that the fast rate corresponds to the catalysis of the wild-type monomer and the slow phase to that of the monomer with the Y147S cytochrome b mutation. This implies that, with center N free to reoxidize cytochrome b, the slower catalysis of the mutant center P site was revealed after the more active wildtype site became unable to continue catalysis due to the accumulation of electrons in the Rieske iron-sulfur protein and in cytochrome c 1 . The total absorbance change was identical in all three enzymes and corresponded to the total concentration of cytochrome c 1 present in the assay, indicating that no significant fraction of enzyme in any of the three dimer preparations was inactivated during the purification procedure.
In the presence of antimycin, which blocks reoxidation of the b H hemes through center N, both cytochrome b and cytochrome c 1 undergo reduction exclusively through center P, although the observed rate of cytochrome c 1 reduction is slightly slower than that of cytochrome b because of the higher redox potential of the Rieske protein, which causes a delay in the appearance of the electron in the c 1 heme (as fully discussed in Ref. 30). Nevertheless, both c 1 and b reduction report the rate of quinol oxidation at center P with antimycin present. As shown in Fig. 4, the wild-type homodimer and the heterodimer showed very similar rates and extents of reduction for both cytochrome b and c 1 , whereas the mutant homodimer showed reduction rates ϳ7-fold lower, yielding a significantly decreased and incomplete extent of reduction on the time scale shown. The kinetic traces obtained showed that only one c 1 heme/dimer underwent reduction in the fully wild-type and heterodimeric enzymes, indicating that only one center P/dimer (the wild-type in the case of the heterodimer) was able to oxidize ubiquinol when antimycin was bound to both center N sites. However, the spectra showed reduction of close to 70% of the total cytochrome b present, without showing the shoulder at 566 nm that is indicative of b L reduction. These observations indicate that both b H hemes in the dimer were reduced   1 Complexes-In the presence of an excess of cytochrome c and without inhibitors, a steady state that allows multiple turnovers of the enzyme occurs as it operates catalytically.
As shown in Fig. 5, the rate of this steady-state activity in the heterodimer, as well as its affinity for substrate, was essentially identical to that of the wild-type homodimer. In contrast, the mutant homodimer showed a ϳ10-fold lower V m together with an increased apparent affinity for ubiquinol, possibly caused by the smaller relative contribution in the mutant of the catalytic rate constant (k cat ) to the K m , which does not necessarily rep- resents a true dissociation constant for substrate binding. The virtually identical activity between the fully wild-type and heterodimers demonstrates that only one of the two center P sites is active at a time in the homodimer, and in the heterodimer the one wild-type monomer oxidizes ubiquinol continuously and is able to catalyze the same steady-state rate as in the wild-type dimer. This identical activity also disproves the remote possibility that slow cytochrome b re-equilibration between dimers occurred in the heterodimeric preparation to form wild-type and mutant homodimers because the generation of the latter would have decreased the overall turnover rate.
Antimycin Titration of the Steady-state Activity of the Wildtype and Heterodimeric bc 1 Complex Dimers-As shown in Fig.  6A, the tightly bound center N inhibitor antimycin (K d ϳ30 pM) (Ref. 19) exhibits a paradoxical effect on the steady-state activity of the wild-type dimer, consisting of a stimulation of the observed rate of cytochrome c reduction that peaks at an inhibitor: monomer ratio of 25% followed by a decrease in the rate. This effect, which we have already observed in the yeast bc 1 complex with antimycin or ilicicolin H (8,9), could be fitted by assuming that, whereas free dimers have only one center P site active at a time, binding of antimycin to one center N activates the second center P site, allowing ubiquinol oxidation to proceed in both monomers simultaneously (Fig. 6A, inset). Consequently, a dimer with an inhibitor bound at one center N has twice as much activity as a free dimer due to activation of the otherwise silent center P site, being able to use a single center N site to catalyze quinone reduction due to rapid electron equilibration between cytochrome b subunits at the level of the b L hemes. This interpretation was supported by the results shown in Fig. 6B using the heterodimeric enzyme in which one center P site is inactivated. The stimulation of the activity by antimycin was virtually absent in the heterodimer, consistent with the assumption that in the free dimer activity is sustained only by the wild-type center P site and that binding of antimycin to one center N results in the activation of the mutant monomer that, because of its low intrinsic activity (ϳ10% of the wild-type, as shown in Fig. 5) conveyed by the Y147S cytochrome b mutation, contributes by a correspondingly small fraction to the total rate. Therefore, the activity of dimers with only one antimycin bound is equal to that of free dimers (Fig. 6B, inset), yielding a titration curve that lacks the stimulation observed with the wild-type homodimer. Because electrons coming from the wild-type center P site in heterodimers with one monomer occupied by antimycin are still able to reach the unbound center N, the titration curve exhibits a nonlinear hysteresis, given that a dimer would become fully inactive only when antimycin was bound to both center N sites. If electrons could not be transferred between cytochrome b subunits, wild-type center P sites with antimycin bound at the center N site in the same monomer would be inactive under steady-state conditions, and the titration curve would be linear (dotted line in Fig. 6B), which was clearly not the case.
In both the wild-type and heterodimeric enzymes, antimycin at concentrations at or above 100% per monomer did not inhibit the rate of cytochrome c reduction beyond ϳ70%. Presteady-state experiments with higher enzyme concentrations in the presence of cytochrome c and saturating antimycin showed that the ϳ30% residual activity corresponded to bypass reactions in which electrons reach cytochrome c without cytochrome b undergoing reduction. These bypass reactions were associated with superoxide production, being ϳ50% sensitive to superoxide dismutase (results not shown). This was expected considering that (as discussed in Ref. 31), after electrons accumulate in cytochrome b due to the antimycin blockage, center P transfers one electron from quinol to cytochrome c via the ironsulfur protein and cytochrome c 1 , whereas the remaining electron in the substrate is directly transferred to oxygen to form superoxide, which by itself is a good reductant of cytochrome c. Thus, superoxide dismutase only inhibits reduction of cytochrome c by superoxide but is not able to block reduction by the electron coming from the iron-sulfur protein and cytochrome c 1 , yielding a maximal 50% inhibition in the total antimycininsensitive reduction of cytochrome c.

DISCUSSION
The relative simplicity of the three-subunit bc 1 complex in P. denitrificans compared with its eukaryotic counterparts allowed us in the present work to tag differentially the C terminus of the cytochrome b subunits containing a largely inactivating center P mutation to purify heterodimers where only one monomer had the mutation. This would have been extremely difficult to achieve in a eukaryotic enzyme because the C terminus of the mitochondrially encoded cytochrome b is surrounded by a polypeptide of 14 kDa (4,5,12), which would have masked or even sterically clashed with the tag extension. Furthermore, we have already demonstrated that the half- of-the sites activity at center P observed in the yeast enzyme is also evident in the complex isolated from P. denitrificans (16).
It was necessary to delete an acidic N-terminal extension in the cytochrome c 1 subunit peculiar to this bacterium that results in the stable aggregation of pairs of dimers, as will be published elsewhere, to purify heterodimers with the Y147S mutation in one cytochrome b subunit. Nevertheless, pre-steady-state kinetic experiments indicated no difference in the rates of cytochrome b or c 1 reduction through center P between bc 1 complexes containing the wild-type cytochrome b with or without the c 1 acidic extension, nor were differences observed when comparing the effect of tagging the C terminus of wild-type or Y147S cytochrome b with His or Strep extensions (results not shown).
Our present results demonstrate directly that only one center P site is active in both the wild-type and heterodimeric enzymes when antimycin is bound at both center N sites (see Fig. 4) and under steady-state conditions (see Fig. 5) in which, as we have previously proposed (8), both center N sites are likely occupied with semiquinone most of the time. The identical turnover rates obtained in the heterodimer, in which one ubiquinol oxidation site was largely inactivated by mutation, relative to the wild-type enzyme (see Fig. 5), is clear evidence that one active center P site is sufficient to sustain full catalysis in the bc 1 complex dimer. The Y147S cytochrome b substitution does not preclude the binding of substrate, as evidenced by EPR studies that show normal occupancy at center P (32). This agrees with the even higher apparent affinity of the mutant homodimer for decylubiquinol (see Fig. 5) relative to the wild-type homodimer, and with its fast and tight binding of stigmatellin (results not shown), an inhibitor that occupies the same site as the substrate. Therefore, the observation that the wild-type monomer is the only one to manifest activity in the case of the WT/Y147S heterodimer suggests that ubiquinol oxidation at the faster center P site, and not merely its binding, contributes to the complete inactivation of the second center P site. This agrees with the proposal that electron transfer from b L to b H affects the mobility of the Rieske iron-sulfur protein, as evidenced by the much slower cytochrome c 1 reduction in enzyme that lacks the b H hemes (13).
The half-of-the-sites dimeric mechanism we have proposed for the functioning of the bc 1 complex (6) also postulates that an asymmetry in binding of ligands to the center N sites allows simultaneous activity at the center P sites, as evidenced by the stimulation exerted by low concentration of antimycin on the steady-state activity of the dimer. Our present data showing the absence of this stimulation in the heterodimer (see Fig. 6) are a clear indication that activation of the second center P site is indeed responsible for this paradoxical effect of center N inhibitors. Our pre-steady-state results in the absence of inhibitors and cytochrome c (see Fig. 3) are also consistent with this aspect of the dimeric mechanism. Given that center N was allowed to oscillate freely between states with different semiquinone occupancies, the second center P site in the dimer was allowed to manifest its intrinsic activity once the first site was prevented from further ubiquinol oxidation events by the accumulation of electrons in cytochrome c 1 and the Rieske protein. This effect was evidenced by the biphasic c 1 reduction kinetics obtained with the heterodimeric enzyme. However, under normal turnover conditions in which cytochrome c continuously oxidizes c 1 , only one monomer would be active at a time, as observed under steady-state conditions. This also supports our previous proposal that dimeric states in which semiquinone is present only at one center N site are probably transient under normal turnover conditions (6), having the effect of The fitted V m and K m values, respectively, were 294.6 Ϯ 14 s Ϫ1 and 77.5 Ϯ 6.6 M for the wild-type dimer, 279.3 Ϯ 7.4 s Ϫ1 and 78.8 Ϯ 3.7 M for the heterodimeric enzymes, and 28.2 Ϯ 1.6 s Ϫ1 and 31.7 Ϯ 2.9 M for the homodimeric mutant enzyme. The reciprocal plots of the rates obtained at each substrate concentration (B) illustrate more clearly the large difference in V m and K m values obtained with the homodimeric mutant dimer relative to the wild-type and heterodimeric bc 1 complexes, which are virtually identical to each other. FIGURE 6. Antimycin titration of the steady-state activity of P. denitrificans bc 1 complex dimers. The catalytic rates of reduction of 10 M horse heart cytochrome c by 50 nM wild-type (A) or heterodimeric (B) bc 1 complex were obtained by mixing with a final concentration of 40 M decylubiquinol after incubating in the presence of the indicated ratio of antimycin/monomer. Solid curves correspond to the simulated activity expected assuming that activation of the previously silent center P site in dimers with only one antimycin molecule bound contribute to the activity twice as much as free dimers in the case of the wild-type enzyme (A, inset) or to the same extent as free dimers in the heterodimeric enzyme (B, inset), where the activated second center P site is too slow to contribute significantly to the rate. In both cases, intermonomeric electron transfer is assumed to allow either center N site in the dimer to recycle electrons from center P in the opposite monomer. Without this assumption, a linear titration curve (dotted line in B) would be expected. Antimycin at concentrations higher than 100%/monomer did not inhibit the activity beyond a rate of ϳ30 s Ϫ1 , indicating stoichiometric binding as expected from the very low K d (ϳ30 pM) of this inhibitor. randomly switching which monomer will be active in the next cycle of ubiquinol oxidation.
Our present kinetic results also convincingly demonstrate the existence of non-rate-limiting electron equilibration between cytochrome b subunits, which allows both b H hemes to undergo reduction through only one active center P site (see Fig. 4) and permits the recycling of electrons coming from either or both center P sites through only one center N (see Fig.  6). We have already provided direct evidence for rapid electron transfer (8,9) at the level of the b L hemes, given the close distance between them as shown in crystallographic structures, which is optimal to sustain fast electron tunneling rates (8,33). Therefore, our present work clearly demonstrates that the bc 1 complex does not function as independent monomeric units, as has been argued recently (15), and supports our previous conclusion (10) that a strong selective pressure has preserved the dimeric structure and function of this energy-transducing enzyme to regulate its activity in response to the redox poise of the substrate pool and minimize formation of reactive oxygen species.