The respiratory substrate rhodoquinol induces Q-cycle bypass reactions in the yeast cytochrome bc(1) complex: mechanistic and physiological implications.

The mitochondrial cytochrome bc(1) complex catalyzes the transfer of electrons from ubiquinol to cyt c while generating a proton motive force for ATP synthesis via the "Q-cycle" mechanism. Under certain conditions electron flow through the Q-cycle is blocked at the level of a reactive intermediate in the quinol oxidase site of the enzyme, resulting in "bypass reactions," some of which lead to superoxide production. Using analogs of the respiratory substrates ubiquinol-3 and rhodoquinol-3, we show that the relative rates of Q-cycle bypass reactions in the Saccharomyces cerevisiae cyt bc(1) complex are highly dependent by a factor of up to 100-fold on the properties of the substrate quinol. Our results suggest that the rate of Q-cycle bypass reactions is dependent on the steady state concentration of reactive intermediates produced at the quinol oxidase site of the enzyme. We conclude that normal operation of the Q-cycle requires a fairly narrow window of redox potentials with respect to the quinol substrate to allow normal turnover of the complex while preventing potentially damaging bypass reactions.

The mitochondrial cytochrome bc 1 complex catalyzes the transfer of electrons from ubiquinol to cyt c while generating a proton motive force for ATP synthesis via the "Q-cycle" mechanism. Under certain conditions electron flow through the Q-cycle is blocked at the level of a reactive intermediate in the quinol oxidase site of the enzyme, resulting in "bypass reactions," some of which lead to superoxide production. Using analogs of the respiratory substrates ubiquinol-3 and rhodoquinol-3, we show that the relative rates of Q-cycle bypass reactions in the Saccharomyces cerevisiae cyt bc 1 complex are highly dependent by a factor of up to 100-fold on the properties of the substrate quinol. Our results suggest that the rate of Q-cycle bypass reactions is dependent on the steady state concentration of reactive intermediates produced at the quinol oxidase site of the enzyme. We conclude that normal operation of the Q-cycle requires a fairly narrow window of redox potentials with respect to the quinol substrate to allow normal turnover of the complex while preventing potentially damaging bypass reactions.
The cytochrome (cyt) 2 bc 1 complex (EC 1.10.2.2) functions as the ubiquinol:cytochrome c oxidoreductase in the electron transport chains of mitochondria and bacteria, where it couples the energy released from the oxidation of ubiquinol (UQH 2 ) by high potential electron carriers (e.g. cyt c) to establish a proton motive force used to drive ATP synthesis (1,2). Although the cyt bc 1 complex usually conserves energy across the membrane, deleterious side reactions can occur that bypass normal electron flow in the enzyme, robbing the cell of energy and potentially producing reactive oxygen species (3)(4)(5)(6)(7). To deal with this problem, the cyt bc 1 complex has apparently developed strict control mechanisms to prevent side reactions of ubiquinone/ubiquinol (UQ (1)/UQH 2 ; see Fig.  1) reaction intermediates (3, 6, 8 -10). 3 The turnover mechanism of the cyt bc 1 complex is best described by the modified Q-cycle (11,12). The key feature of this mechanism is the bifurcated oxidation of UQH 2 at the quinol oxidase (Q o ) site of the enzyme that sends one UQH 2 electron to a high potential chain (composed of the Rieske 2Fe2S cluster and cyt c 1 ), whereas the other electron reduces a low potential chain (cyt b L and b H ) and is eventually recycled back into the Q pool through UQ reduction at a quinone reductase (Q i ) site on the opposite side of the membrane. In most models UQH 2 oxidation proceeds by the initial reduction of the Rieske 2Fe2S cluster (13)(14)(15), which produces an unstable semiquinone (SQ) intermediate localized to the Q o site (SQ o ), which then reduces the low potential chain and UQ at the Q i site, leaving a labile UQ in the Q o pocket.
Certain conditions such as mutation (16 -18), imposition of a high proton motive force (17,19,20), or inhibitor treatment (3-5, 7) cause interruption of electron flow from SQ o to the low potential chain and can lead to a large fraction of reactions that bypass the normal Q-cycle. Most Q-cycle bypass reactions are thought to result from the buildup of a highly reactive SQ o species, although some are suggested to occur via reduced cyt b L (18); some of these reactions include (3, 4, 7) 1) double reduction of the high potential chain by SQ o , 2) re-oxidation of reduced cyt b L by SQ o , Q, or O 2 , 3) oxidation of SQ o by O 2 , and 4) direct reduction of cyt c by a labile SQ o that leaves the Q o site. Although each of these reactions is thermodynamically favorable in comparison to the Q-cycle, they are typically not detected during uninhibited turnover of the complex, leading to the hypothesis that the cyt bc 1 complex has evolved a mechanism to steer its reactive intermediate toward productive, rather than harmful, reactions.
Several variants of the Q-cycle have been proposed to explain the prevention of bypass reactions, most of which impart strict constraints on the chemical and thermodynamic properties of the substrate QH 2 and its proposed initial oxidation product, SQ o (6,8,(11)(12)(13)(14)(21)(22)(23)(24)(25)(26). Previous studies (27)(28)(29) have suggested a link between these thermodynamic properties and the overall turnover rate of the bc 1 complex, but it is unclear how the intrinsic redox properties of the QH 2 substrate and the influence of the protein environment on these properties work together to enforce normal electron flow in the Q-cycle while also preventing bypass reactions. We suggest that both these factors tailor QH 2 oxidation in the Q o site to minimize the concentration of reactive intermediates that can participate in these bypass reactions, which can be achieved through the use of a substrate whose oxidation results in a less reducing redox poise in the Q o site (i.e. by preventing the accumulation of either SQ o or reduced cyt b L ).
In this work we use synthetic analogues of the quinone substrates UQ and rhodoquinone (RQ (2), see Fig. 1) to test boundary conditions on how the intrinsic thermodynamic properties of the substrate and its associated SQ stability influence the rate of Q-cycle bypass reactions. RQ is used in certain eukaryotes (30 -36) and prokaryotes (37,38) to shuttle electrons from the NADH dehydrogenase to a fumarate reductase (31, 36, 39 -41). This alternative electron transport chain is used during anoxia to maintain cellular redox balance by using fumarate as a terminal electron acceptor instead of oxygen. Although RQ is chemically similar to UQ (a OCH 3 substituent on the ring is replaced by NH 2 ), the electron donating nature of its amino group gives it substantially more reducing power than its counterpart UQ (42)(43)(44), thus offering the opportunity to probe the efficiency of the Q-cycle under conditions where the relative stability of SQ species are modulated in both the Q o and Q i sites.
In certain organisms (nematodes (45), parasitic helminths (35, 46), mussels and oysters (35), and the photosynthetic bacterium Rhodospirillum rubrum (34)), RQ is co-localized in the mitochondrial or plasma membrane with UQ (32,34,47). This occurs even under aerobic conditions where biosynthesis of RQ is down-regulated but not entirely absent, providing the potential for the interaction of RQ or RQH 2 with the cyt bc 1 complex. The use of the synthetic substrates ubiquinone-3 (UQ 3 ) and rhodoquinone-3 (RQ 3 ), where n ϭ 3 isoprene units (instead of 9 or 10 in the natural substrates), allow us to probe the Q-cycle and bypass reactions as well as to measure the relevant redox potentials of the electrochemical couples involved in these reactions. From these results we predict the potential impact due to bypass of the Q-cycle in organisms that contain a mixed UQ/RQ pool (32,37,48) and test the relationship between the rates of bypass reactions and the thermodynamic properties of these two substrates.

EXPERIMENTAL PROCEDURES
Chemicals for RQ 3 Synthesis-All reactions were performed under a positive pressure of Ar with minimal exposure to light. Diethyl ether, ethyl acetate, hexanes, methanol, and tetrahydrofuran were purchased from Fisher (ACS-certified grade). Tetrahydrofuran was distilled from sodium benzophenone ketyl, and methanol was dried over 3-Å molecular sieves. Mesyl chloride, sodium azide, sodium sulfate, triethylamine, and triphenylphosphine were purchased from Aldrich. Thin layer chromatography was performed on Whatman (F 254 ) 0.25-mm coated silica gel plates and visualized using 254-nm light or by heating samples with vanillin stain. Column chromatography was carried out using 230 -245 mesh silica gel from Bodman Scientific.
Steady State Turnover Measurements-Steady state turnover was measured after the reduction of cyt c using an extinction coefficient of 17 mM Ϫ1 cm Ϫ1 for the 550-nm Ϫ 542-nm difference absorption (3,4). Just before measurement, synthetic substrate, Q 3 H 2 (UQ 3 H 2 or RQ 3 H 2 ), was mixed in a cuvette containing 50 M cyt c and 1 mM NaN 3 in reaction buffer (50 mM MOPS, 50 mM Tricine, and 100 mM KCl at pH  OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41 JOURNAL OF BIOLOGICAL CHEMISTRY 34655 8.0). In previous work (3,4) we used only MOPS as a buffer but have now included Tricine to ensure a high buffering capacity in the pH range 8.0 and above; this additional component did not cause any measurable changes in the activity of the complex. Cyt bc 1 (1-3 nM) was added and mixed in the cuvette after collection of 20 s of background cyt c reduction. Turnover numbers were calculated by subtracting the background cyt c reduction rate from the initial rate observed after the addition of cyt bc 1 . Measurement of Q-cycle bypass reactions under partially inhibited conditions was performed by the addition of 30 M antimycin-A (AA) to the cuvette before Q 3 H 2 addition and mixing. Cyt bc 1 (20 -40 nM) was used for bypass reaction measurements involving UQ 3 H 2 due to the very low rate of this reaction (1-5 s Ϫ1 ). All experiments were repeated once with the addition of stigmatellin as a control, which completely abolished activity in all cases.

Rhodoquinol-induced Bypass Reactions in the Cyt bc 1 Complex
Measurement of Superoxide Production-Under partially inhibited conditions, i.e. conditions where AA blocks reduction of Q at the Q i site, the intermediate SQ o can reduce oxygen to superoxide, which then reduces cyt c directly in a diffusion-limited reaction (4). Thus, the rate of superoxide production from the Q o site can be determined by taking the difference in rates of AA-resistant cyt c reduction with and without added manganese superoxide dismutase (SOD). In most cases the addition of SOD inhibits the rate of cyt c reduction by 25-50%; the residual non-SOD-inhibitable cyt c reduction is caused by other Q-cycle bypass reactions such as the direct reduction of cyt c by SQ o or double reduction of the high potential chain (4). A typical superoxide production assay measuring cyt c reduction was performed by the addition of 300 units of manganese SOD to the steady state assay mixture described above.
The Extent of Cyt b Reduction in the Steady State-Measurement of the reduction state of both cyt b hemes during steady state turnover was performed by mixing 30 M Q 3 H 2 into a cuvette containing 265 nM cyt bc 1 and 5 M cyt c. The low concentration of cyt c was used to slow depletion of the substrate through normal turnover and background cyt c reduction, which allowed us to observe steady state turnover of the complex over a timescale of minutes. The reaction was followed using the 561-nm minus 578-nm wavelength pair using an extinction coefficient of 26 mM Ϫ1 cm Ϫ1 for total cyt b. Saturating concentrations of AA and stigmatellin were added as needed. Sodium azide was omitted from the assay medium to allow re-oxidation of cyt c by the cyt c oxidase, which is a trace impurity in our preparations of cyt bc 1 complex. The reduction state of the cyt c pool was measured optically to verify that the complex had actually reached the steady state. Re-oxidation of reduced cyt c after an initial spike in the concentration (up to 300 nM) brought the steady state reduced cyt c concentration to ϳ60 nM. Over this time, the extent of cyt b reduction was nearly constant after the initial increase in reduction, suggesting rapid establishment of steady state conditions in the low potential chain.
Cyclic Voltammetry and Determination of Redox Potentials-Voltammetric measurements of the Q/QH 2 and Q/Q . couples for UQ 3 and RQ 3 were performed in aqueous solution under an Ar atmosphere at 23°C using a BAS C50-W potentiostat with an Ag/AgCl reference electrode, glassy carbon working electrode, and Pt wire auxiliary electrode. The working electrode was washed with dilute H 2 SO 4 , repeatedly rinsed with distilled water, and then polished with alumina before all measurements. Determination of E(Q/Q . ) was performed in unbuffered solution (100 mM KCl) as described in Shim and Park (50) at pH Ͼ 5.0. Aqueous E(Q/Q . ) values were determined at several pH values above pH 5.0 to ensure that the E1 ⁄ 2 and ⌬E values were independent of pH and, thus, reflect a pure electron transfer process. The two-electron Q/QH 2 potential was measured in 50 mM phosphate, 100 mM KCl buffer at pH 7.0. E1 ⁄ 2 values for each couple were taken as the average of the cathodic and anodic peaks at half height. All potentials given in the text are referenced to the standard hydrogen electrode.
We used experimentally determined values for the Q/Q . and Q/QH 2 couples of UQ 3 and RQ 3 in aqueous solution to estimate reasonable values of the aqueous stability constant, K S , and E m values for the various redox couples contributing to K S, according to Equation 1 (51-53), where pK a,SQ is the pK a for deprotonation of QH ⅐ . The pK a,SQ for UQ was previously determined to be 5.6 in aqueous solution using pulse radiolysis (54). We used a calibrated computational technique using a combination of density functional theory calculations and thermodynamic cycles to estimate the pK a of RQ 3 H ⅐ ; the details of these calculations are to be presented in a forthcoming manuscript. 4 This pK a was found to be similar to that of UQH ⅐ , 5.0 -5.6. 3 and UQ 3 -A novel and efficient synthesis of RQ 3 was developed. The farnesylated analogue of demethylubiquinone (3), a known biosynthetic precursor of UQ (55)(56)(57), was prepared from fumagatin using an improved method that utilized farnesyltrimethyl stannane and BF 3 -O(CH 2 CH 3 ) 2 followed by oxidation with Ag 2 O (Scheme 1) (58 -60). Compound 3 was converted to the mesylate 4 using mesyl chloride and triethylamine in a 77% yield (38). The mesylate was then treated with NaN 3 to generate the azide 5 via a nucleophilic additionelimination reaction (62). This reaction produced a 33% yield of azide; however, unreacted starting material could be recovered. Reduction of the azide with triphenylphosphine formed an iminophosphorane 6, which was not readily hydrolyzed during an aqueous work-up as previously reported (63,64). However, treatment of 6 with dilute acid or slow elution through a silica gel column provided pure RQ 3 (7) in a 64% yield (two steps). UQ 3 was prepared using published procedures (55).

Synthesis of RQ
Thermodynamic Characterization of UQ 3 and RQ 3 -Cyclic voltammetry of UQ 3 in unbuffered aqueous solution (50) at pH Ͼ 5 exhibited quasi-irreversible cathodic and anodic peaks (⌬E ϭ 190 mV). The lack of complete reversibility likely reflects adsorption of UQ 3 to the electrode surface. Rigorous cleaning of the electrode surface did not affect the reversibility of these peaks. The E1 ⁄ 2 value obtained from this measurement was Ϫ23 mV (see TABLE ONE), which is consistent with a previous report of E1 ⁄ 2 ϭ Ϫ30 mV for the UQ (Q/Q . ) couple in 80% aqueous ethanol (65). The (Q/Q . ) couple for RQ 3 also exhibited quasiirreversible voltammetric peaks (⌬E ϭ 109 mV), although to a much lesser extent than UQ 3 , with E1 ⁄ 2 ϭ Ϫ170 mV.
As an additional assurance that the E1 ⁄ 2 values measured here are reasonable estimates of the actual equilibrium potentials, we compared the (Q/QH 2 ) couple of RQ 3 with the previously published polarographic value. Our result (TABLE ONE), E1 ⁄ 2 ϭ Ϫ45 mV, is in reasonable agreement with the previously published value of Ϫ65 mV (44). Measurement of the E1 ⁄ 2 for the UQ 3 (Q/QH 2 ) couple was not possible due to complete irreversibility of the cathodic peak, resulting in the absence of an anodic peak under these conditions.
Normal Turnover and Q-cycle Bypass Reactions Induced by UQ 3 H 2 and RQ 3 H 2 -TABLE TWO summarizes the results from steady state turnover of the cyt bc 1 complex driven by UQ 3 H 2 and RQ 3 H 2 under uninhibited ( Fig. 2 and TABLE TWO, row 1) and AA-inhibited (TABLE TWO, row 2) conditions. Cyt c reduction observed under AA-inhibited conditions results from partial turnover of the complex and one or more reactions that bypass the Q-cycle and directly reduce cyt c. Using UQ 3 H 2 as a substrate, the addition of 30 M AA decreased the rate of cyt bc 1 turnover from 143 to 3 s Ϫ1 , consistent with previous results (3, 4, 7). On the other hand, the use of RQ 3 H 2 exhibited an uninhibited turnover rate substantially lower than UQ 3 H 2 , 76 s Ϫ1 , and exhibited a relative insensitivity to AA treatment, 88 s Ϫ1 .
We verified that AA did indeed bind to the Q i site of the complex under these conditions by observing the red-shifted absorbance spectrum of cyt b H induced by AA binding in the presence of RQ 3 (66) (data not shown). Further titration of RQ 3 into the AA-treated complex did not reverse the red-shifted spectrum of cyt b H , showing that RQ 3 does not displace AA during the assay.
A large degree of curvature in the kinetic traces with RQ 3 H 2 was observed, suggesting that the true initial rate of cyt c reduction was obscured during the manual mixing time. We thus repeated the AA-induced bypass reaction experiments using stopped flow. Indeed, as shown in TABLE ONE, the rate of UQ 3 H 2 AA-resistant turnover was 2.3 s Ϫ1 , consistent with steady state measurements, whereas the rate of RQ 3 H 2 AA-resistant turnover was very high, at 207 s Ϫ1 , a difference in bypass rates of more than 100-fold. Even this rate might be limited by the 0.2-s time resolution of our spectrophotometer.
Both RQ 3 H 2 and UQ 3 H 2 exhibited similar apparent K m values with the bc 1 complex, 10 and 8 M for UQ 3 H 2 and RQ 3 H 2 , respectively. These K m values are similar to those measured previously for the isolated yeast bc 1 complex by various groups (7,(67)(68)(69).
Partial Inhibition of the Q-cycle by RQ 3 H 2 /RQ 3 -The curvature in the steady state kinetic traces described above suggested some form of product inhibition (Fig. 2). We first tested for inhibition of Q reduction at the Q i site induced by RQ 3 or UQ 3 . Typically, efficient oxidation of cyt b by Q at the Q i site results in only a small accumulation of reduced cyt b in the low potential chain under steady state conditions, whereas blockage of cyt b reoxidation at the Q i site with inhibitors results in a substantial accumulation (30 -50%) of reduced cyt b (10). The data shown in Fig. 3 confirm these observations in our system using UQ 3 H 2 as substrate. Approximately 10% of total cyt b was reduced under uninhibited conditions, whereas 52% went reduced in the presence of AA. In contrast, when RQ 3 H 2 was used (Fig. 3), 77% of cyt b went reduced under uninhibited conditions, suggesting that the re-oxidation of cyt b by RQ 3 was slowed with respect to its reduction. The addition of AA in the presence of RQ 3 H 2 resulted in eventual reduction of nearly all (92%)

Thermodynamic properties of UQ and RQ redox and protonation reactions
Values were determined as described under "Experimental Procedures." Electrochemical midpoint potentials are expressed against the standard hydrogen electrode.

Substrate-dependent properties of cyt bc 1 turnover
All kinetic measurements were performed at 23°C with the isolated cyt bc 1 complex as described under "Experimental Procedures." The velocity of superoxide production is calculated from the difference between V max (AAinhibited) and V max (AA-inhibited ϩ SOD).

Quinol donor Ubiquinol Rhodoquinol
a V bypass ϭ 0.5 (V max ϩ AA) due to the fact that the first electron reduces the high potential chain in all cases and, therefore, accounts for half of the overall rate. b V SO ϭ V(superoxide production) ϭ V max ϩ AA Ϫ V max ϩ AA ϩ SOD. c Fraction SO ϭ fraction of partially inhibited turnovers resulting in superoxide production ϭ V SO / V bypass . d Determined using decyl-UQH 2 as a substrate. of cyt b. The statistically significant overall lower accumulation of cyt b during RQ 3 H 2 oxidation in the absence of AA probably indicates that reduction of RQ 3 at the Q i site occurs, albeit at a lower rate than with UQ 3 .
In addition to functioning as a Q i site inhibitor, Fig. 4 demonstrates that RQ 3 also interferes with UQ 3 H 2 binding to the Q o site. The addition of up to 20 M RQ 3 both increased the apparent K m for UQ 3 H 2 , from 8.5 to 13.7 M, and lowered the apparent V max , from 98 to 84 s Ϫ1 , for UQ 3 H 2 oxidation. These results are consistent with RQ 3 acting as a weak mixed competitive inhibitor of the bc 1 complex.
Superoxide Production Resulting from Q-cycle Bypass Dramatically Increases Using RQ 3 H 2 as a Substrate-The SOD sensitivity of AA-resistant cyt c reduction was used to estimate the fraction of bypass reactions resulting in superoxide production. The initial rate of AA-resistant UQ 3 H 2 -induced turnover resulting in superoxide production was 1.0 s Ϫ1 , consistent with previous measurements (3,4,7), whereas the same measurements performed with RQ 3 H 2 resulted in an astonishing superoxide production rate of 29.5 s Ϫ1 . The ϳ30-fold increase in rates for RQ 3 H 2 -induced superoxide production relative to UQ 3 H 2 (see TABLE TWO, row 5) correlates well with the overall increased rate of bypass reactions when this substrate is used (see TABLE TWO, row 3), which also increases by a factor of ϳ30, suggesting that about 66% of all QH 2 oxidations result in the reduction of oxygen to superoxide (3,4,7) under partially inhibited conditions regardless of the substrate used (TABLE TWO, row 6).
We observed complex (or strongly multi-phasic) cyt c reduction kinetics, probably reflecting a progressive onset of product inhibition (see "Partial Inhibition of the Q-cycle by RQ 3 H 2 /RQ 3 "). Because the curvature was apparent even at short reaction times, we suggest that our measured rates of RQ 3 H 2 -induced superoxide production represent a lower limit. Unfortunately, even with presteady state kinetics using stopped-flow, it is not presently possible to distinguish between the first two non-inhibited turnovers (which result in cyt b H and b L reduction) and the subsequent true bypass reactions.

The Thermodynamic Properties of RQ and UQ and Predictions for
Reactivity-Our electrochemical data (TABLE ONE) show that, as expected, the redox potential of the E(QH ⅐ /QH 2 ) for RQH 2 is substantially more negative than the same couple for UQH 2 (by nearly 200 mV). We predict that the low potential of the RQH ⅐ /RQH 2 couple will result in increased concentrations of rhodosemiquinone (RSQ) in the Q o site due to an increase in the equilibrium constant for RSQ formation by over 3 orders of magnitude, thus increasing the rate of bypass reactions. Similarly, the aqueous values of E(RQ 3 /RQ 3 H 2 ) and E(RQ 3 / RQ 3 . ) are ϳ150 mV lower than the corresponding UQ species, likely decreasing the equilibrium constant for reduction of RQ at the Q i site by more than 2 orders of magnitude. We, thus, predict that the slow oxidation of the low potential chain during steady state turnover will lead to an AA-like inhibition of the complex, again having the effect of increasing the rate of bypass reactions. Consistent with these predictions, we observe that the rates of bypass reactions and superoxide production are drastically increased (more than 100-fold, see TABLE TWO) when RQ 3 H 2 is used as a substrate instead of UQ 3 H 2 .
The SQ stability constant, K S , is often used alone as a qualitative predictor of SQ reactivity and steady state concentration for many Q o site models (6,8,11,70). Despite the clear differences between these substrates in our estimates for the one-electron couples, E(QH ⅐ /QH 2 ) and E(Q/Q . ), the estimated semiquinone stability constants (TABLE  ONE) for RSQ and USQ differ by only a factor of 2, with RSQ (K S ϭ 1 ϫ 10 Ϫ7 ) slightly more stable than USQ (K S ϭ 5.3 ϫ 10 Ϫ8 ). The semiquinone stability constant by itself is, thus, an incomplete predictor of reactivity and SQ concentration and must be used in conjunction with the overall two-electron Q/QH 2 couple to derive the one electron couples, so that meaningful predictions can be made.
It is important to note that the in situ potentials of the Q/QH 2 species in the Q o and Q i sites are likely to differ substantially from the aqueous potentials measured and estimated here due to the unique electrostatic environment of the protein and dielectric effects (71)(72)(73)(74). Interestingly, the two Q binding sites in the complex likely exhibit disparate effects on the stability of the SQ species; the Q i site has been shown to stabilize the SQ (74, 75), whereas some models suggest the Q o site greatly destabilizes the SQ (11-13, 70). Nevertheless, it is reasonable to assume that the  changes in the redox potentials of the RQ species in the Q o pocket will parallel those of the UQ species so that the differences measured in solution will persist in the Q o pocket.
RQ/RQH 2 Alters the Redox Poise of the bc 1 Complex to Favor Bypass Reactions-Turnover of the cyt bc 1 complex using RQ 3 H 2 as a substrate exhibited three features distinguishing it from that with UQ 3 H 2 (Fig. 2); 1) upon manual mixing of substrate into reaction buffer (with a mixing time of ϳ2 s) the initial rate of uninhibited turnover (76 s Ϫ1 ) was about half that with UQ 3 H 2 (143 s Ϫ1 ), 2) the kinetic trace exhibited a high degree of curvature as the reaction proceeded, lowering the steady state rate over time, and 3) unlike UQ 3 H 2 oxidation, the addition of AA had little effect on the initial rate of turnover.
We suggest that all of these observations can be explained by product inhibition at both the Q o and Q i sites. This interpretation is supported by the "backup" of reducing equivalents in the low potential chain (Fig.  3) when RQ 3 H 2 is used as a substrate and the mixed inhibition observed from RQ 3 using UQ 3 H 2 as a substrate (see Fig. 4).
The Q i site inhibition by RQ 3 /RQ 3 H 2 can be explained by two possible, non-exclusive models, both of which are consistent with the predictions based on solution thermodynamics given in the previous section. First, using a pure thermodynamic argument, RQH 2 is a considerably stronger reductant (ϳ150 mV) than UQH 2 , allowing for the establishment of a more reducing cyt b redox poise. Likewise, the lower redox potential of the RQ/RQ . couple should make it a poor substrate for the Q i site, effectively "trapping" electrons in the cyt b chain. Second, using a kinetics argument, RQH 2 , RQ, or even RSQ could act as inhibitors of or slow substrates for the Q i site. The simplest interpretation is that both the kinetic and thermodynamic properties of RQ/RQH 2 result in a more reduced cyt b chain, leading to a buildup of reactive intermediates in the Q o site, possibly an SQ, which increases the rate of superoxide production and other bypass reactions (TABLE  TWO). These data offer an explanation for previous findings that RQH 2 oxidation by the cyt bc 1 complex in R. rubrum chromatophores is insensitive to AA (34) (i.e. it is likely that the bc 1 complex is already partially inhibited by RQ).
The Physiological Impact of a Mixed Q Pool in RQ-producing Organisms-The results discussed above beg the question, How do organisms that use both RQ/RQH 2 and UQ/UQH 2 in their quinone pools prevent deleterious side reactions at the cyt bc 1 complex? For example, Euglena gracilis possesses both RQ/RQH 2 and UQ/UQH 2 simultaneously (although RQ synthesis is down-regulated under aerobic conditions, residual RQ persists for extended periods, with a ratio of UQ:RQ approaching 3:1 under fully aerobic conditions) (32). Simultaneous accumulation of RQ and UQ has also been observed in certain nematodes (45), parasitic helminths (35, 46), mussels and oysters (35), and the photosynthetic bacterium R. rubrum (34). The presence of the two quinones is apparently physiologically necessary for rapid transitions between aerobic and anaerobic respiration in these organisms. Our model system, S. cerevisiae, does not synthesize RQ, but we suggest that if the cyt bc 1 complex in these RQ-containing organisms is similar to that in yeast, then co-mingling of RQ and UQ in cyt bc 1 -accessible compartments should result in production of large amounts of superoxide and other unwanted bypass reactions.
We consider three possible mechanisms by which this superoxide production could be prevented in these organisms. First, the cyt bc 1 complexes from RQ-containing organisms might have different substrate specificities. Sequence alignments of cyt b proteins between organisms known to contain mixed RQ/UQ pools (Mytilus edilus, R. rubrum) and UQ-only species (chicken, yeast) did not reveal any discernable trends in conserved residues known to be involved in Q o and Q i site binding (not shown), making this possibility seem unlikely. Nevertheless, we are currently testing this possibility directly by characterizing the cyt bc 1 complexes from several RQ-containing organisms. Another possibility is that the mitochondrial inner membrane or bacterial plasma membranes are compartmentalized either into separate membrane systems or domains (76,77) that contain separate electron transfer chains. A third possibility is that the cyt bc 1 complex and/or fumarate reductase are under regulatory control to prevent the co-existence of RQH 2 and an active cyt bc 1 complex in the same bioenergetic membrane.
Implications for the Function (and Dysfunction) of the Cyt bc 1 Complexes-This work shows that the propensity of the bc 1 complex to participate in Q-cycle bypass reactions can be greatly modified by changing the thermodynamic properties of the substrates Q and QH 2 , particularly those that affect the steady state concentration of reactive intermediates in the Q o site. In principle either SQ o or cyt b L could be responsible for the reduction of oxygen during bypass reactions, although it has been argued that cyt b L is not sufficiently reducing to act as an effective O 2 reductant (4). Our data show that the total amount of reduced cyt b does accumulate to a larger extent with RQ 3 H 2 than with UQ 3 H 2 , but in most Q-cycle models the accumulation of reduced low potential chain components would also imply an increase in the concentration of SQ o . Alternatively, RSQ might be very labile relative to USQ in the Q o site, thus allowing it to more readily escape and collide with O 2 .
In principle, both thermodynamic and kinetic factors will influence the amount of reactive intermediates formed in the steady state, i.e. how much of the intermediate is formed, how fast it is formed, and how fast it is consumed. All of these factors will be influenced by the redox and binding properties of SQ o /QH 2 and Q/SQ couples, which in turn are determined by the structural properties of the substrate and the binding site.
We propose that the Q-cycle requires a fairly narrow window of substrate thermodynamic (redox and binding) properties to ensure a low steady state concentration of reactive intermediates while allowing for rapid Q-cycle turnover. UQ/UQH 2 seems to be particularly well suited to this task in comparison to RQ/RQH 2 . Intriguingly, this "matching" of the quinone/quinol redox properties with those of the cyt bc 1 complex appears to be well conserved; menaquinol-oxidizing cyt bc-type complexes have electron carriers with redox potentials about 150 mV lower than those in UQH 2 -oxidizing complexes, perfectly matching the 150-mV lower potential of the substrate Q/QH 2 couple (61, 78 -80).