Reversible redox energy coupling in electron transfer chains Artur Osyczka1*, Christopher C. Moser1, Fevzi Daldal2 & P. Leslie Dutton1 1 The Johnson Research Foundation, Department of Biochemistry and Biophysics, and 2 Department of Biology, Plant Science Institute, University of Pennsylvania, Philadelphia, Pennsylvania, 19104, USA * Permanent address: Jagiellonian University, Faculty of Biotechnology, Krakow, �� Poland ........................................................................................................................................................................................................................... Reversibility is a common theme in respiratory and photosynthetic systems that couple electron transfer with a transmembrane proton gradient driving ATP production. This includes the intensely studied cytochrome bc 1 , which catalyses electron transfer between quinone and cytochrome c. To understand how efficient reversible energy coupling works, here we have progressively inactivated individual cofactors comprising cytochrome bc 1 . We have resolved millisecond reversibility in all electron-tunnelling steps and coupled proton exchanges, including charge-separating hydroquinone���quinone catalysis at the Qo site, which shows that redox equilibria are relevant on a catalytic timescale. Such rapid reversibility renders popular models based on a semiquinone in Qo site catalysis prone to short-circuit failure. Two mechanisms allow reversible function and safely relegate short-circuits to long-distance electron tunnelling on a timescale of seconds: conformational gating of semiquinone for both forward and reverse electron transfer, or concerted two-electron quinone redox chemistry that avoids the semiquinone intermediate altogether. The challenge of engineering efficient photosynthetic and respira- tory energy conversion is to favour productive electron and proton transfer reactions that generate or use membrane proton motive force (DmH��), while suppressing energy-wasting short-circuit reac- tions. Photosynthetic reaction centres favour productive light- induced charge separation and avoid wasteful charge-recombining short-circuits by expending much of the absorbed light energy to drive the forward charge-separating steps, and thereby slow the reverse uphill charge returns that make short-circuits more likely. For other crucial membrane energy-coupled oxidoreductases, such as the cytochrome bc1 and b6f family, the modest driving force provided by substrates makes this strategy impossible. Furthermore, as DmH�� builds up, the net reaction can be reversed, as shown in mitochondrial cytochrome bc1 and complex I by classic experiments that artificially added ATP to increase DmH�� and to stimulate ���reverse electron flow���1,2 on a timescale of minutes. Some litho- trophic organisms apparently rely on the reverse electron flow through cytochrome bc1 for growth3,4. Despite these observations, contemporary models for energy conversion in cytochrome bc1 (refs 5���16) neglect reverse reactions and the implications of rever- sibility on short-circuit vulnerability. These models fail if reversi- bility on a rapid catalytic timescale is fully proven. Here we have used photosynthetic membranes of the bacterium Rhodobacter capsulatus to investigate cytochrome bc1 reversibility. R. capsulatus provides the kinetic means for rapid, light-activated delivery of the substrate���s hydroquinone (QH2) and oxidizes cyto- chrome c to cytochrome bc1 (Fig. 1a), even as the cofactors are progressively knocked out genetically. This strategy avoids long- standing difficulties in resolving concurrent reactions in the ���b-chain��� (comprising haem b L , haem b H and the quinone of the Qi site) and the ���c-chain��� (comprising the iron-sulphur centre FeS, haem c1 and cytochrome c (mostly c 2 but also c y )). This strategy also establishes a rigorous single-turnover activation of cytochrome bc1 to oxidize and reduce only one quinone molecule at the Qo site, where the Q pool and the b- and c-chains meet and where energy conversion is catalysed. Drawing on substrate and cofactor redox midpoint potentials and their pH dependencies (Fig. 1b), we have exposed selected single-turnover cofactor knockout systems to a range of driving forces from exergonic to endergonic (Fig. 1c) to define, step by step, the thermodynamic parameters and to reveal the timescale of reversibility of the operating cytochrome bc1. Electron transfer in cofactor knockouts Figure 2 presents the flash-activated haem b reduction kinetics in cytochrome bc1 with different combinations of cofactor knockouts. The system is initially poised at high redox potentials to oxidize the Q pool. The uninhibited wild-type system (Fig. 2a, black) monitors flash-generated QH2 arriving at the Qo site and the 8-ms concurrent reduction of both FeS and haem b L , followed by rapid haem bL to bH electron transfer across the membrane. Subsequent reduction of Qi to semiquinone by haem bH completes the electrical charging of the membrane (Fig. 1a). After reduction, FeS normally undergoes constrained diffusion to transfer the electron to haem c1 (refs 17���19). In turn, haem c1 reduces cytochrome c (data not shown). Antimycin eliminates Qi function (Fig. 2a, green), so that the electron advances only as far as haem b H to reveal its full reduced extent. There is no noticeable effect of inactivation of the Qi site on the rate of Qo site turnover and haem bH reduction. Further addition of myxothiazol (Fig. 2a, red) inhibits QoH2 oxidation in the first place and no haem b reduction is observed this rules out any other routes to haem b reduction in these experiments. In the first cofactor knockout (Fig. 2b), mutation of the meth- ionine ligand of haem c1 to leucine20 markedly drops the redox midpoint potential by hundreds of millivolts and disables electron transfer from either cytochrome c or FeS. But oxidation of QH2 at the Qo site proceeds to the same extent and with kinetics essentially identical to the wild type, and reduction of FeS by QoH2 is unimpeded. Once FeS is reduced, oxidation of a second QoH2 is impossible and, unlike the wild-type cytochrome bc 1, this knockout is a de facto Qo site single-turnover enzyme. In the second cofactor knockout (Fig. 2c), insertion of two alanines in the hinge region between the FeS head group and the transmembrane anchor severely interferes with the normal move- ment of FeS, thereby locking it in the Qo site position17. This also prevents communication between the Qo site and haem c1 and cytochrome c, and again provides rigorous single-turnover action. But like the c1 knockout, the FeS-motion knockout yields the same kinetics and extents as the wild type. Clearly, the diffusive, large- articles NATURE | VOL 427 | 12 FEBRUARY 2004 | www.nature.com/nature 607 �� 2004 Nature Publishing Group

scale domain motion of FeS is not needed for fully competent Qo site activity17,21. The third knockout (Fig. 2d) replaces the haem bH histidine ligand with an asparagine, and haem b H is lost22, providing an alternative way to restrict the Qo site to one turnover. But reduction of haem b L , together with reduction of FeS and haem c1, takes place with kinetics similar to that of haem bH reduction. The simultaneous knockout of two cofactors most severely strains cytochrome bc1 turnover and delineates the thermodynamic limits of Qo site action under high redox potential conditions. Pairing the bH knockout with either the c1 knockout (Fig. 2e) or the FeS-motion knockout (Fig. 2f) trims the multicofactor enzyme to just three components: Qo, FeS and haem bL. Remarkably, Qo site action has essentially the same rate of haem b reduction as the wild type. However, the extent of haem b reduction is only about a third under these conditions, indicating that the oxidation���reduction reaction of this simple three-component system may be energeti- cally balanced and fully reversible on this timescale. The reversibility of the system and the thermodynamically cooperative behaviour of the b- and c-chains become obvious when, following the guidelines of Fig. 1b and c, we manipulate the driving force provided to cytochrome bc1 by changing the pH and state of the Q pool reduction before the activation. Figure 3 shows the pH modulation of the extent of haem b reduction for wild type and the c1 knockout (Fig. 3a), and for the bH knockout and the bH c1 double knockout (Fig. 3b) at both the high redox potential condition of the oxidized Q pool described above, and at a lower redox potential at which the Q pool is half-reduced and the arrival of QH2 at the Qo site is not rate-limiting. All systems become progressively less competent as the driving force is lessened, either through raising the redox poise or through lowering the pH, which raises the midpoint potential of the Q pool (Fig. 1b, c). A simple equilibrium model The progressive failure of the extent of haem b reduction is neatly accommodated in an equilibrium model with four simple postu- lates (Fig. 3a, b, lines, and Fig. 3c���e, graphic illustrations). First, individual redox centres have the same midpoint potentials on the catalytic timescale as those measured in equilibrium redox titrations on a timescale of minutes (Fig. 1b). Second, the 2:1 ratio of reaction centres to cytochrome bc 1 monomer produces two oxidized cyto- chromes c 2 and one QH2 per flash (ref. 9 and Fig. 3d) this postulate may be less valid at the high pH limit of 10, where QH2 production by the QB site in reaction centres may begin to fail. Third, Qo site redox reactions are strictly coupled so that every electron exchanged between the Qo site quinone and FeS is accompanied by electron exchange between quinone and haem b L (refs 23, 24). Fourth, electron transfers between Qo and FeS and haem b L , as well as electron transfers between members of the b-chain and between members of the c-chain, continue until the redox potential of the Q pool equals the average of the redox potentials of the c- and b-chain that is, the net driving force for the Qo site reaction is zero (refs 23, 24, and Fig. 3e). The double knockout system (Fig. 3b, black) has the least driving force and is the first to fail as the pH is lowered (see Fig. 1c). When communication with haems c1 and c2 is restored in the single bH knockout (Fig. 3b), the thermodynamically cooperative involve- ment of these haems improves cytochrome bc1 robustness. At high redox potential (Fig. 3b, red), simple, rapid redox equilibrium contact with multiple oxidized redox centres in the c-chain favours more QH2 oxidation and haem b L reduction, which continues even Figure 2 Flash-activated haem b reduction. Kinetics are shown for wild type (a) and cofactor knockout combinations (b���f) at pH 9.0 and at an of E h of �� 250 mV to oxidize the Q pool, FeS and haem b before the flash. Haem kinetics, initiated by QH2 diffusing from the reaction centre, is uninhibited (black) or inhibited with antimycin (green), antimycin and myxothiazol (red, left), or myxothiazol alone (red, right). Reduction of haem b H (left) and haem b L (right) is monitored by absorption change presented in milli-units of optical density (mOD) at 560���570 nm and 566���573 nm, respectively the initial step reflects reaction centre spectral contributions at these wavelengths. Red crosses indicate the locations of knockouts and blue arrows delineate remaining electron transfer reactions. FeS e��� H+ QB QA H+ H+ Qi Qo c2 bL bH c1 Q pool Diffusion b a c 6 5 7 8 9 10 ���0.2 ���0.1 0.1 0 0.2 0.3 pH Em (V) c2 c1 bH bL FeS Q pool ���Go bc1 Reaction centre Cytochrome bc1 Light ���0.2 ���0.3 0.0 eV bL QoH2 FeSo Qo FeSr bL��� bH k.o. Ant bH��� SQi 6 7 9 10 pH 5 8 ���0.1 0.1 Figure 1 Cofactors and their energetics in R. capsulatus. a, Light-activated R. capsulatus initiates electron (blue) and proton (green) transfers through haems and chlorins (squares), quinones (hexagons) and the FeS cluster (double cross) of its reaction centre and cytochrome bc1 to generate DmH��. The reaction centre generates oxidized cytochrome c2 and hydroquinone (QH2), which diffuse to a dimeric cytochrome bc1. Oxidized cytochrome c2 oxidizes haem c 1. Haem c 1 oxidizes FeS, which, by limited diffusion, arrives at the Qo site to oxidize QH2 drawn from the pool. QoH2 oxidation reduces both FeS and haem b L. Red crosses indicate the positions of cofactor knockouts. b, c, pH dependence of cofactor equilibrium midpoint (Em) potentials (b) yields adjustable patterns of endergonic and exergonic steps beginning at the Qo site (c). Grey areas indicate steps inactivated by antimycin and the b H knockout. articles NATURE | VOL 427 | 12 FEBRUARY 2004 | www.nature.com/nature 608 �� 2004 Nature Publishing Group