Structures of the bc1 complex reveal dynamic aspects of mechanism
Structures for the bc1 complex (1) show several features of interest to catalysis at the Qo-site. The iron sulfur protein (ISP) occurs in several different positions: at a catalytic interface on cytochrome (cyt) b close to the Qo-site, in H-bond contact with a heme propionate of cyt c1, or in several intermediate positions (1-3). The Qo-site is a bifurcated volume, in which myxothiazol and MOA-inhibitors bind in a proximal domain near heme bL, but stigmatellin and UHDBT bind in a distal domain, where they interact with the ISP docked on cyt b. The structures also allow for the first time insights into the functional significance of mutations that modify turn-over or inhibitor binding. These structural observations have led us to suggest several new mechanistic features for the operation of the Qo-site, and have suggested experiments to test these, some of which we report here.
Crystallographic procedures were as described in (1). Spectrophotometry and kinetic measurements of cytochrome changes were as described in (5). Redox potentiometry of the ISP was performed using CD-spectroscopy (6). Molecular engineering protocols, and introduction of specific cysteine residues will be described in detail elsewhere.
3.1 Movement of the ISP.
We have suggested that the displacement of the ISP seen in the crystal structures is essential for catalysis (1), and have investigated the movement under physiological conditions. In strains of Rb. sphaeroides with all but essential cysteines removed, we engineered unique reactive cysteines at suitable positions on cyt b and ISP, and labeled these with complementary donor (D) and acceptor (A) fluorophores. The expected distribution, assuming equal reactivities, was 25% each for DD, DA, AD, AA. Because the fluorescence spectra overlap at the red end of the spectrum, we measured excitation transfer between fluorophores by the quenching of the donor. The changes in quenching on changing redox state are shown in Fig. 1. Quenching was maximal at high Eh, or in the presence of stigmatellin. In the absence of inhibitor, quenching was lost in two phases; ~30% as ISP and cyt c1 became reduced, and ~60% as the quinone pool and cyt bH became reduced. Since 2/3 of the centers with D showed quenching, this suggests that the distribution of fluorophores was close to the ratio expected. The insert in Fig. 1 shows the residues at which cysteines were inserted. Ca-atoms were 12 and 30 Å apart in the stigmatellin and native structures, respectively. The data show that the ISP is close to cyt b in the presence of stigmatellin under all redox conditions, consistent with the structure, and the known binding. In the uninhibited state, the subunits are close when the complex is oxidized, and the ISP moves away as the complex becomes reduced. When ISP is reduced and the pool oxidized, a fraction of the ISP is retained close to cyt b, consistent with the complex giving rise to the gx=1.80 band in the ISP EPR-spectrum.
Fig. 1. Left: Unique cysteines were engineered in a his-tagged complex at V141C (light-gray, with stigmatellin (top); or dark-gray (bottom)) in ISP, and T386C (black) in cyt b. Structures are shown with stigmatellin (pale wireframe) (top), or without (bottom). Subunits are shown as back bone models: ISP (bottom, left), light gray (with stigmatellin) or dark gray (native); cyt b, black; cyt c1 (bottom, right), gray. Hemes are wire frame models, 2Fe2S-centers are gray spheres. Right: Movement of ISP was detected by the quenching of fluorescence from MTS-dansyl (donor) adduct by neighboring sulfo-rhodamine-MTS (acceptor) adduct. Ambient potential was adjusted electrochemically in an anaerobic cuvette.
3.2 Movement of semiquinone.
Although the bifurcated volume of the Qo-site provides support for the double-occupancy model of Ding and Dutton (7) and Brandt (8), the following features suggest that the changes of the gx=1.80 signal seen on reduction, extraction, or in mutant strains, have a simpler interpretation, in line with occupancy by a single quinone: i) None of the structures (1-3) shows occupancy of the Qo-site, although all show a quinone at the Qi-site. The double-occupancy data were interpreted as showing that Qos binds 20-fold tighter than Qi. ii) Mutations effect signals attributed to Qos and Qow the same; if one is lost or modified, so is the other. There is no indication of a set of protein ligands that differentially effect one of the proposed binding sites but not the other.
iii) Both signals are eliminated by either stigmatellin or myxothiazol at a stoichiometry of 1/Qo-site. iv) Some mutants show a strong gx=1.80 signal (diagnostic of Qow), but no electron transfer, some a weak gx=1.80 band, but substantial electron transfer, and some a correlation between gx=1.80 band and electron transfer rate. This is difficult to interpret in the double-occupancy model, but the structure allows a reasonable interpretation in terms of the distribution of residues (4).
To explain these anomalies, we suggest the following alternative model: a) The gx=1.80 signal is associated with a complex formed between quinone bound in the Qo-site at the distal end, and the reduced ISP docked firmly at the interface with cytochrome b. b) The signal is lost whenever this complex cannot form: - on reduction of Q, on binding of inhibitors, on extraction of quinone, in strains where mutation prevents binding of quinone at the distal end of the pocket, and in strains where mutation prevents docking of the ISP. c) Only one quinone species occupies the site. The spectroscopic effects (gx=1.783) attributed to Qos arise from the statistical distribution of quinone in the heterogeneous chromatophore population, with additional contributions from changes in environment of the ISP. This model accounts well for the experimental data, and provides a natural explanation for the effects which are anomalous in the double occupancy model.
Since the evidence favors a single occupancy model, it becomes necessary to account for the bifurcated volume of the Qo-site without invoking a double occupancy. We have propose that movement of the semiquinone (SQ) from the distal domain where it is formed to the proximal domain plays a critical role in the efficiency of the bifurcated reaction. We suggest that mutations around the proximal lobe which slow or eliminate electron transfer, but do not prevent formation of the gx=1.80 signal, prevent this movement, and thereby inhibit turn-over (4).
3.3 Activation barriers for the partial reactions of quinol oxidation as a function of pH.
Fig. 2A shows the kinetics of cyt b reduction at different pH values, with the pool
reduced to the same extent before flash-activation. The data are taken from a complete set in which we have measured the temperatures dependence at each pH (Fig. 2B). The traces show that cyt b reduction is strongly pH dependent, with a rate inversely proportional to [H+]. In contrast, the activation energy did not show a pH dependence, so the activation barrier could not be attributed to a reaction involving proton release in this range. We have made similar measurements for the kinetics of cyt c1 oxidation in the presence and absence of myxothiazol. For the partial reactions in the high potential chain, the rates were faster than cyt b reduction at all temperatures, the activation energies all had lower values, and the rates and barriers were not markedly affected by pH. The lag phase before onset of cyt b reduction, which includes all processes leading to delivery of an oxidizing equivalent to the Qo-site, showed an activation energy similar to that of the high potential chain (not shown), and a weaker pH dependence than the rate (Fig. 2A). It is noteworthy that the Arrhenius plots for cyt b reduction were non-linear at all pH values, suggesting that more than one process contributes to the barrier. As suggested by Crofts and Wang (5), the activation barrier is in processes after formation of the complex.
3.4 Dependence on pH of the redox potential of the ISP
The redox potential of the ISP has been investigated using CD- spectroscopy.
The characteristic CD features in the region between 450 and 550 nm were
found in the purified bc1 complex and membranes of Rb. sphaeroides.
The difference between reduced and oxidized CD-spectra showed a negative
band at about 500 nm with a half of width 30 nm that corresponds to the
specific dichroic absorption of the reduced Rieske protein (6). The redox
potential at pH 7.0 for the Rieske center in the isolated bc1
complex, and in chromatophore membranes from the R-26 strain of Rb.
sphaeroides, was 300 ± 5 mV. The
Em varied with pH in the range above pH 7, and the pH dependence
was well fit by two pK values, pK1 =7.2 and pK2 =9.8.
Similar titrations and pK values were found for the ISP in the isolated
bc1 complex and membranes from R-26 strain of Rb. sphaeroides.
These values are similar to those found in the isolated beef ISP (6).
Brandt and Okun (7) have shown in the beef and yeast bc1 complexes that the rate of quinol oxidation is markedly pH dependent, and is characterized by pK values at 6.7 and 9.2 for pKA and pKB respectively, and an activity that depends on dissociation of the group with pKA, and association of the group with pKB. They also showed that, in the alkaline range, the activation energy for steady-state electron transfer changed by -5.7 kJ mol-1 pH-1, indicating an involvement of proton release in the rate determining step, and suggested that the activated step is the release of the first proton on binding QH2 to the catalytic site. However, their data show a weaker pH dependence of the activation energy over the acidic range, consistent with the results above.
On the basis of kinetic studies (5) and the structural information now available (1), binding of quinol involves multiple steps. In order to explain the pH dependence of the rate over the acidic range, we suggest that the probability of forming the reaction complex depends on reaction with the dissociated form of the ISP, and that the rate reflects the concentration of this species, determined by pK1.
1. Oxidation of ISP: ISPr cyt c1+ Û ISPo cyt c1
2. Movement of ISP to cyt b: ISPo (cyt c1) Û ISPo (cyt b)
3. Dissociation of His-161: +HN'-ISPo-NH+ Û +HN'-ISPo-N + H+
4. Formation of the reaction complex with bound quinol:
Here the reaction complex is indicated by square brackets. It seems likely that above pK1, the concentration of the dissociated form is not rate limiting, but becomes increasingly limiting as the pH is lowered below pK1. From the structure, His-153 (His-161 in beef) forms the bridging H-bond, and we can therefore attribute pK1 to this residue.
3.5 Conclusions: Mechanism of quinol oxidation
(a) The reaction proceeds from a complex between the oxidized ISP and quinol, with QH2 bound at the distal end of the pocket, and the oxidized ISP docked tightly at the interface on cyt b. Binding of QH2 and formation of the complex requires changes in protein conformation to accommodate the substrate, and to allow access of the ISP to the quinol. (b) Binding likely also leads to release of 1 H+, probably through dissociation of the His-161 2Fe.2S ligand before formation of the H-bond with quinol. The H-bond formed in the stigmatellin complex provides a model (1). Below pK1 at 7.2, the probability of forming the reaction complex will be determined by i) the concentration of quinol, ii) the concentration of the dissociated form of the ISP, accounting for the pH dependence of electron transfer in this range. (c) Electron transfer from QH2 to the ISP leads to formation of semiquinone. Formation of the SQ is improbable, and represents the step with a high activation barrier (5). The SQ is a transient species, because, once formed, it is removed rapidly. We suggest that the reaction complex between quinol and ISPox is relatively stable, and as a consequence, formation of SQ, release of the ISPred and the electron transfer to cyt bL are concerted. Because the subsequent reactions of both high and low potential chains are not rate determining, electrons appear at the acceptor termini (cyt c for the high potential chain, and cyt bH for the low potential chain blocked by antimycin) with the same kinetics, determined by the rate-limiting formation of the semiquinone. (d) As a necessary step in its formation, the SQ moves in the pocket to the proximal end, near heme bL, which facilitates its rapid oxidation. This movement also requires changes in the protein conformation, and may contribute to the activation barrier. Although the SQ is normally rapidly oxidized, under static head conditions, since cyt bL is poised half reduced, a significant population of SQ might be expected at the site. The movement of SQ to the proximal domain insulates it from further reaction with the re-oxidized ISP, and thus prevents the decoupling of electron transfer from proton pumping. This location also minimizes reaction with O2, and formation of O2-. (e) The changes in protein conformation facilitate the separation of the reduced ISP, and also insulate the SQ from further reaction with the ISPox. (f) The SQ passes its electron to the cyt bL heme, and thence to heme bH, and the quinone exits the Qo-site. In the absence of antimycin, the electron is transferred to the Qi-site occupant. (g) At some point between c) and f), the second proton is released. This might be required for formation of the semiquinone, or its movement in the pocket, or the transfer of the second electron to the low potential chain. Possibly the pH dependence of the activation barrier in the alkaline range measured under steady-state conditions (9) reflects this second deprotonation. (h) Meanwhile, the reduced ISP can deliver an electron to cyt c1 by the tethered diffusion mechanism (1, 4), and return in the oxidized form to initiate another turnover of the Qo-site.