Modification of quinone electrochemistry by the proteins in the biological electron transfer chains: examples from photosynthetic reaction centers

Abstract

Quinones such as ubiquinone are the lipid soluble electron and proton carriers in the membranes of mitochondria, chloroplasts and oxygenic bacteria. Quinones undergo controlled redox reactions bound to specific sites in integral membrane proteins such as the cytochrome bc₁ oxidoreductase. The quinone reactions in bacterial photosynthesis are amongst the best characterized, presenting a model to understand how proteins modulate cofactor chemistry. The free energy of ubiquinone redox reactions in aqueous solution and in the Q_A and Q_B sites of the bacterial photosynthetic reaction centers (RCs) are compared. In the primary Q_A site ubiquinone is reduced only to the anionic semiquinone (Q^•−) while in the secondary Q_B site the product is the doubly reduced, doubly protonated quinol (QH₂). The ways in which the protein modifies the relative energy of each reduced and protonated intermediate are described. For example, the protein stabilizes Q^•− while destabilizing Q⁼ relative to aqueous solution through electrostatic interactions. In addition, kinetic and thermodynamic mechanisms for stabilizing the intermediate semiquinones are compared. Evidence for the protein sequestering anionic compounds by slowing both on and off rates as well as by binding the anion more tightly is reviewed.

Quinones are the primary intramembrane, mobile, electron carriers in the energy-coupling electron transfer chains in mitochondria, chloroplasts and bacterial cell membranes (Cape et al. 2006). The relative concentration of the oxidized quinones, Q, and doubly reduced, quinol, QH₂, help define the redox poise in these membranes. Quinone oxidation/reduction reactions occur in proteins such as the NADH:quinone reductase (Complex I), the cytochrome bc₁ oxidoreductase (Complex II) and succinate–quinone reductase (Complex III) in mitochondria and oxygenic bacteria (Saraste 1999). In photosynthetic electron transfer, reactions involving quinones are found in the type II reaction centers (PSII or bacterial reaction centers (RCs)), b₆f (or in bacteria the bc₁) complexes and in the type I reaction centers (PSI). Quinone function is the characteristic that distinguishes between type I reaction centers where bound quinone is only transiently reduced to semiquinone and type II reaction centers where quinone is doubly reduced to QH₂ and released to the membrane Q_pool (Heathcote 2002). All proteins that use quinones must also guard against side reactions involving the free radical intermediates of quinone oxidation/reduction reactions, which can catalyze the formation of dangerous reactive oxygen species (Forquer et al. 2006).

Quinones, with their two electron, two proton electrochemistry have nine redox states (Fig. 1). All but Q and QH₂ are unstable in water or membrane under physiological conditions. However, the quinone reactions in biology occur in specific binding sites that can stabilize free radical, semiquinone intermediates. This permits a general mechanism using two reaction cycles to go between Q and QH₂. Thus, the bound quinones couple the obligate single electron carriers such as (bacterio)chlorophylls, hemes and iron sulfur complexes to the quinone pool where only Q and QH₂ are found. For example, in photosynthetic reaction centers, considered in this review, a primary quinone (Q_A) is first reduced to the anionic semiquinone state (Q_A^•−). This then reduces the oxidized secondary quinone (Q_B) to the semiquinone Q_B^•(H) (Q^•(H) here indicates either the anionic (Q^•−) or protonated (Q^•H) semiquinone). A second turnover of the protein re-reduces Q_A, which reduces Q_B^•(H). Thus, Q_B^•(H) must live long enough for the protein to undergo a second turnover to re-reduce Q_A. In addition, Q_A must be a good enough reductant to reduce both Q_B and Q^•(H). This is the same type of mechanism used for the Q_I site of the bc₁ complex. The Q_O site of QH₂ oxidation in the bc₁ complex uses a fundamentally different strategy (Osyczka et al. 2005; Cape et al. 2006). Here two different, simultaneously accessible reduction partners (cyt B_H and the Rieske iron sulfur complex) are found. Thus, oxidation of quinol at this site does not rely on a stable semiquinone waiting for a second turnover. Until recently no semiquinone had been seen in the Q_O site (Cape et al. 2007; Zhang et al. 2007).

Fig. 1.

The E_ms on the horizontal arrows are in millivolt. The pK_as, are in parenthesis on the vertical arrows. The quinone species in the dark circles are found in both the Q_A and Q_B sites. Thin circles indicate species found only in the Q_B site. The species in grey are not seen in RCs or in aqueous solution as they are expected to have pK_as below pH 0. All values are taken from (Zhu and Gunner 2005, see references therein)

To move between Q and QH₂ the protein binding sites can change the order of electron and proton transfers from that found in solution (Fig. 1) by modifying the relative energy of the intermediates. They can also kinetically trap the intermediates so they are not lost to the surroundings. This review discusses the electrochemistry of ubiquinone in aqueous solution, the general strategies that proteins use to modify the energies by changing electrostatic interactions and the evidence that the proteins modulate the on and off rates of reactants and products. The specific reaction sequence, thermodynamics and rates for reduction of quinones in the Q_A and Q_B sites in the photosynthetic reaction centers of the purple, non-sulfur bacteria Rhodobacter sphaeroides are described. This review focuses on the reactions of ubiquinone, however the chemistry of menaquinone (Rich and Bendall 1979) and plastoquinone are quite similar (Wraight 2004; Swallow 1982).

Electrostatic modification of the thermodynamics of quinone redox reactions

Proteins can greatly modify the in situ electrochemistry of bound cofactors. For example, the E_ms of protein bound hemes range from −400 to +450 mV, representing a change in the free energy of the oxidation reaction of over 18.4 kcal/mol (Churg and Warshell 1986; Gunner and Honig 1991; Voigt and Knapp 2003; Reedy and Gibney 2004; Zheng and Gunner 2008). It has been suggested that electrostatic interactions with the protein are the primary means by which proteins control electrochemistry (Warshel and Russell 1984; Honig and Nicholls 1995; Shurki et al. 2004; Gunner et al. 2006).

To understand how proteins change quinone electrochemistry, it is helpful to start with a description of their chemistry in solution (Rich and Bendall 1979; Swallow 1982; Prince et al. 1983, 1986; Wraight 2004; Zhu and Gunner 2005). The role of the protein then becomes clear when these E_ms and pK_as of different quinone redox states are compared to those in a particular binding site. While it is not easy to measure the high energy free, radical intermediates between ubiquinone and ubiquinol in solution, a consensus view of the free energy changes linking each species can be found in Fig. 1 (see Zhu and Gunner (2005) and references therein).

There are many methods for analyzing reaction chemistry in proteins, including classical, free energy perturbation techniques embedded in Molecular Dynamics Simulations (Marchi et al. 1993); semi-microscopic PDLD (PDLD/S) analysis (Churg et al. 1983; Parson et al. 1990; Sham et al. 1997, 1999); Density Function Theory (Mallik and Datta 2004; Cape et al. 2006) and QM/MM treatments (Hasegawa et al. 2003). For reactions such as electron or proton transfers where the primary difference between reactant and product is a change in charge, classical continuum electrostatics (CE) analysis has provided important insights into reaction mechanism (Beroza et al. 1995; Lancaster et al. 1996; Rabenstein et al. 1998; Alexov and Gunner 1999; Ishikita et al. 2003; Ishikita and Knapp 2004; Zhu and Gunner 2005; Klingen et al. 2007).

In a continuum electrostatics analysis the influence of the protein on the free energy of a reaction can be divided into three terms: the desolvation penalty where the stabilization of a charge by water is lessened when a charged group is moved into protein; the pairwise interactions between the reactants and the protein charges and dipoles; and dielectric relaxation, where the protein and surroundings change position to stabilize either the reactant or the product (Honig and Nicholls 1995; Gunner and Alexov 2000; Baker 2005; Gunner et al. 2006).

1. The desolvation penalty

Charged and polar solutes have favorable interactions with water that are diminished when they are bound to protein. This loss forms the basis of the substantial destabilization of charged groups in proteins (Kassner 1972; Warshel and Russell 1984). The desolvation energy for moving a sphere between solvents with different dielectric constants can be estimated with the Born equation (Bockris and Reddy 1973; Rashin and Honig 1985), which gives the energy of transfer from water to protein as:

$$\Delta \Delta G_{\text{desolv}}^{°}=-331\frac{q^2}{2r}\left(1/{\epsilon }_{\text{wat}}-1/{\epsilon }_{\text{prot}}\right)\text{kcal}/\text{mol}$$ (1)

With the constant 331, the distance in Å and the charge in multiples of the charge on an electron the outcome is provided in Kcal/mol. The desolvation energy increases with the charge (q) squared and becomes less important as the charge is delocalized over a molecule with a larger radius (r). With water as the reference solvent, the external dielectric constant is 80. The appropriate value of the interior dielectric constant is subject of much debate, with values from 2 to 80 being used (Gunner and Alexov 2000; Schutz Warshel 2001). This review will discuss calculations carried out with an ε_prot of 4 (Zhu and Gunner 2005). Roughly, the loss of solvation energy for a cofactor in a protein comes from the larger distance of the charge to water, i.e. the protein adds to the effective cofactor radius. Since proteins are of finite size, even deeply buried groups retain significant solvation energy (Kim et al. 2005).

Charged species always loose more solvation energy then neutral, polar species when moved into the protein (Kassner 1972). Thus the desolvation energy destabilizes the anionic Q^•− relative to Q or Q^•H, lowering the E_m. Likewise the neutral intermediate, Q^•H has less solvation energy to loose then Q^•−, raising the pK_a relative to the reaction in water.

To avoid formation of a buried, anionic semiquinone the reaction sequence could ensure tightly coupled electron and proton transfer (Q + H⁺ + e⁻ → Q^•H). However the electron and proton transfers in proteins occur over very different distance scales. The electron-transfer reactions occur by electron tunneling between reaction partners that are generally more then 7 Å apart (Moser et al. 2003, 2006). In contrast, the proton must be transferred directly between two residues that are close enough that they could make a good hydrogen bond (Sham et al. 1999). Thus, mechanisms that require strictly coupled electron and proton transfer need a proton donor near the electron acceptor (and a separate proton acceptor near the distant electron donor). Instead, the anionic quinone state is often found as a stable intermediate in the various proteins in the bioenergetic electron transfer chains. Thus each protein is designed to provide favorable interactions to stabilize the bound anion despite the loss of solvation energy.

2. Pairwise interactions with protein side chains and backbone

The second contribution the protein makes to the change in the free energy of a redox reaction is via the interactions of the protein charges and dipoles with each redox and protonation state. While, the desolvation penalty always destabilizes both buried anions and cations, the pairwise interactions that stabilize an anion in a particular location will destabilize a cation. The shift in reaction free energy (ΔΔG°) is proportional to the charges on the quinone and the interacting group, the distance between them and the effective screening by the protein and surrounding water (Gunner et al. 1997, 2006).

3. The effects of dielectric relaxation

The third contribution of the protein to the reaction comes from changes in electronic polarization, atomic position and nearby residue protonation in response to the changes in charge distribution caused by the redox reaction. In standard Continuum Electrostatics simulation methods the atomic positions are kept fixed. The energy of electronic polarization and atomic reorganization are accounted for by the assigned protein dielectric constant. These methods do incorporate explicit changes in the protonation state of the residues in the protein (Bashford and Karplus 1990; Beroza et al. 1991; Yang et al. 1993). This relatively simple style of analysis has been able to provide detailed, molecular insight into how specific protein residues play a role in proton coupled electron transfer (Beroza et al. 1995; Lancaster et al. 1996; Rabenstein et al. 1998; Ullmann and Knapp 1999; Ishikita et al. 2003; Haas and Lancaster 2004). In hybrid methods the protein side chains, but not backbone, can change position (You and Bashford 1995; Beroza and Case 1996; Alexov and Gunner 1997). These methods identify more of the ‘dielectric response’ with specific conformational changes rather then with an averaged dielectric constant (Gunner and Alexov 2000; Simonson 2001; Baker 2005; Gunner et al. 2006). The work described below uses the hybrid MultiConformation Continuum Electrostatics (MCCE) method (Georgescu et al. 2002; Alexov and Gunner 1999).

Redox reactions can be stabilized by coupling electron and proton transfers. If one proton is bound per electron transferred then the reaction will not change the charge of the system. This will reduce the contributions of electrostatic interactions to the reaction free energy. However, one of key features of the chemiosmotic mechanism, elucidated by Michel, for storing energy in a transmembrane proton gradient is the alternation of cofactors that bind protons and those which do not (Mitchell 1975a, b). Thus, in principle, the electron transfer sequence has both reactions where the protein changes charge and those that are electroneutral.

The free energy of redox reactions where the reactants and products have different numbers of bound protons will be pH dependent. The proton can be bound to the cofactor itself (Rich and Bendall 1979; Fig. 1) or to the acidic and basic residues in the surrounding protein (Alexov and Gunner 1999; Haas and Lancaster 2004). The presence of nearby ionized groups and the cost of coupled changes in protonation affects the E_m (Fig. 2a). Changes in conformation coupled to the redox reaction can be analyzed in a similar manner, by considering the free energy of making the conformational change in the absence of the charge change (Mao et al. 2003). In standard continuum electrostatics methods the costs of these conformational changes are averaged using the protein dielectric constant. In MCCE a mixed response is used, with an averaged protein response of four, while side chain motions are included explicitly.

Fig. 2.

Modification of quinone E_ms by proton or protein binding. a Energetics of coupled electron and proton transfer. X is a base associated with a quinone binding site. When X is neutral the E_m for Q is E_m,0; when Q is neutral the pK_a for X is pK_a,0. Favorable interactions between X⁺ and Q^•− raise the E_m for Q by −ΔΔG°/nF. The same interaction stabilizes XH⁺, raising the pK_a by ΔΔG°/2.3RT. A ΔΔG° of −1.36 kcal/mol shifts the E_m up by 58 mV and the pK_a up by 1 pH unit. If the pK_a of X remains below the pH in the presence of Q^•− then X will never be protonated and E_m=E_m,0. If the pH is below the pK_a of X with Q oxidized, XH⁺ will be present throughout the reaction and the observed E_m will be E_m=E_m,0−ΔΔG°/F. In this case the reduction of Q feels the full stabilization by the adjacent base. In either case the reaction is pH independent. However, if the pH is at least ≈2 pH units below the pK_a with Q and ≈2 pH units above the pK_a in the presence of Q^•− (i.e. ΔΔG° >≈5.4 kcal/mol) then binding ≈1 proton will be coupled to electron transfer (when the pH is 2 pH units below the pK_a X will remain 1% protonated; when the pH is 2 pH units above the pK_a X will remain 1% XH⁺. Thus, on average 0.98 more protons will be bound). The resultant E_m will be E_m,0− [ΔΔG°+(pH−pK_a,0)/2.3RT]/F. The free energy needed to protonate X at this pH is (pH−pK_a,0)/2.3RT. It is this term which leads to the classic 60 meV/pH unit E_m shift with pH indicative of 1 proton bound/electron. The cost of rearranging the surroundings diminishes the E_m shift from that found if XH⁺ were present at the start of the reaction. If pK_a,0 for X is near the pH even a small ΔΔG° of interaction with a more distant Q^•− will lead to substoichiometric changes in protonation of X, leading to a pH dependence smaller then 60 mV/pH unit. b The relationship between the thermodynamics of quinone binding and quinone electrochemistry. If the E_m is more positive in the quinone binding site then the semiquinone is bound more tightly then the quinone. ΔΔG°_bind is 2.3RT(logK_dQ^•−− logK_dQ)

If the protein shifts the reaction free energy then its affinity for the reactant and product states must be different (Fig. 2b). A positive E_m shift indicates that the reduced cofactor is more tightly bound. A similar connection can be made between the pK_a and K_d, where a higher pK_a indicates that the protonated species binds more favorably then the unprotonated one (Zhu and Gunner 2005).

The reaction sequence in bacterial photosynthetic reaction centers

The photosynthetic reaction centers from purple bacteria (RCs) were the first membrane protein with a structure known to atomic resolution (Feher et al. 1989; Gunner 1991; Woodbury and Allen 1995). Since a flash of light initiates the reactions, it is possible to measure single turnovers, so the sequence and kinetics of the individual electron transfer reactions are known in detail. The electrochemistry of metastable intermediates have been determined by equilibrium redox titrations. The free energy differences between many reaction intermediates have also been established by measuring reaction rates that rely on a stable species remaining in equilibrium with a high energy intermediate (Woodbury et al. 1986; Xu and Gunner 2000). The overall reaction in RCs uses the energy of two photons to take electrons off two cytochromes c, reducing ubiquinone to the dihydroubiquinone. No protons are pumped across the protein. However, the protons bound to the reduced quinone are taken up from the cell interior, adding to the transmembrane ΔpH (Fig. 3).

Fig. 3.

The overview of the sequence of electron transfer reactions in the bacterial type II reaction centers. The reaction starts with a dimer of bacteriochlorophylls (P) excited by light. Electron transfer from the excited state (P*) leads to reduction of a nearby bacteriopheophytin (H). The reduced H^•− is used to reduce the primary quinone, Q_A, which in turn reduces the secondary quinone, Q_B, to the anionic semiquinone (Q_B^•−). Absorption of a second photon again leads to formation of Q_A^•− which now reduces Q_B^•− to the quinol, QH₂, which dissociates from the protein and quinone is rebound. The Q_B charge in parenthesis is found on the second turnover. The order in which electrons and protons are added to Q_B is described in Fig. 4 B. In the membrane of the purple non-sulfur photosynthetic bacteria the cytochrome c is rereduced and the QH₂ reoxidized by the cytochrome bc₁ oxidoreductase with the concomitant increase in the transmembrane proton gradient

The RCs bind UQ in two distinct quinone binding sites, which are then designated Q_A and Q_B. Only the oxidized Q_A and anionic semiquinone Q_A^•− are found. Q_A does not dissociate from the protein during the reaction cycle. In contrast, Q_B serves as the two-electron gate, where two single electron transfers from Q_A^•− form the doubly reduced Q_BH₂ for release into the membrane (Wraight 1979; Okamura et al. 2000). RCs are found with Q_B in three relatively stable redox states: unreduced quinone (Q), anionic semiquinone (Q⁻) and fully reduced and protonated dihydroquinone (QH₂). The anionic semiquinone is tightly bound to the protein, while the Q and QH₂ freely exchange with the quinone pool in the membrane (Diner et al. 1984; Madeo and Gunner 2005). The pathway for the second reduction indicates that of the two possible intermediates, Q_BH is easier to form than Q_B⁻², so proton binding occurs prior to electron transfer (Graige et al. 1996; Zhu and Gunner 2005). Thus, of the nine possible redox states for Q_B five are found on the reaction pathway (Figs. 1, 4 B). Computational analysis using variations of classical continuum electrostatics (either with a rigid protein (SCCE) or with conformational flexibility (MCCE)) have found fairly good agreement between experiment and calculation, providing insight into the reaction mechanism (Beroza et al. 1995; Lancaster et al. 1996; Rabenstein et al. 1998; Alexov and Gunner 1999; Rabenstein et al. 2000; Ishikita et al. 2003; Alexov et al. 2000). There is a second distal binding site for Q_B seen in the crystal structures (Lancaster 1998; Stowell et al. 1997). Kinetic measurements find no evidence for quinone reduction in this outer site (Remy and Gerwert 2003; Breton et al. 2002; Xu et al. 2002; Breton 2004; Pokkuluri et al. 2004) and simulation suggests the E_m of this quinone too low for it to be reduced by Q_A^•− (Zhu and Gunner 2005).

Fig. 4.

Energy levels for sequential electron transfer from a primary semiquinone to a secondary quinone. A Q_AS:Q_BS indicate a complex of two quinones which have the same electrochemistry as two isolated ubiquinones in solution at pH 7 (Fig. 1). To start the cycle one quinone is reduced to the semiquinone forming Q_AS^•−:Q_BS. Given a solution E_h of 0 mV or a chemical electron donor with an E_m of 0 mV, the single reduction of quinone with an E_m of −145 mV to Q^•− is uphill by 145 meV. Since these quinones are identical, electron transfer from Q_AS^•− to Q_BS will be isoenergetic. Given the semiquinone pK_a of 4.9, protonating either semiquinone would be uphill by ≈120 meV. Thus, the lowest energy, singly reduced state will be a 50:50 mixture of Q_AS^•−:Q_BS and Q_AS:Q_BS^•−. The second turnover starts with the second reduction of Q_AS forming Q_AS^•−:Q_BS^•−. The thermodynamically preferred pathway has the electron transfer occurring before the first proton is bound. The formation of Q_AS:Q_BS⁼ requires only 55 meV because the second reduction of Q_BS is coupled to the favorable oxidation of Q_AS. The reaction path where Q_BS^•− is protonated to form Q_AS^•−:Q_BSH before it is reduced is 120 meV uphill. Q_AS:Q_BSH⁻ where one quinone has two electrons and one proton is 290 meV lower in energy then Q_AS^•−:Q_BS^•− and is essentially isoenergetic with the initial Q_AS:Q_BS state at pH 7. Once Q_AS:Q_BSH⁻ is formed the second protonation to form Q_AS:Q_BSH₂ is downhill by −220 meV. B The ubiquinone energy levels in R. sphaeroides RCs at pH 7 and E_h 0. In the initial reaction Q_A is reduced to the semiquinone forming Q_A^•−:Q_B. Calculations put this state near 0 mV, close to the measured values between −45 and −70 mV (all calculated values are from Zhu and Gunner (2005)). Q_B^•− is stabilized by 30 (calculation) to ≈70 meV more then Q_A^•−. The second turnover starts with formation of Q_A^•−:Q_B^•−. Calculations and the experiments of Graige and Okamura show that in the protein Q_A^•−:Q_B^•H is lower in energy then Q_A:Q_B⁼ (Graige et al. 1998). The calculations place the energy of Q_A^•−:Q_B^•H 260 meV above Q_A^•−:Q_B^•−, while the kinetics of forward electron transfer support a value of 160 meV (grey text; Graige et al. 1999). The second reduction of Q_BH⁻ is favorable. The anionic Q_AQ_BH⁻ is 110 meV more stable in the protein then in solution, while Q_AQ_BH₂ is 50 meV less stable favoring quinol dissociation

Quinone electrochemistry in solution

In many photosynthetic type II reaction centers the same chemical species of quinone is used for both the primary (Q_A) and secondary (Q_B) quinone acceptors. This is ubiquinone in R. Sphaeroides reaction centers (RCs) and the electrochemistry of this quinone will form the basis of the discussion. The protein shifts the quinone E_ms and pK_as in the two sites differently so each is changed from what would be found in solution. To understand how the protein perturbs the quinone electrochemistry it is useful to start with describing the intrinsic chemistry of the cofactor in water (Fig. 1). The uncertainty in quinone electrochemistry in aqueous solution is discussed in (Zhu and Gunner 2005).

The free energy of the different pathways for reduction of ubiquinone (Q) to ubiquinol (QH₂) in water using a semiquinone electron donor is shown in Fig. 4 A. The reaction starts with Q_AS:Q_BS. The final product Q_AS:Q_BSH₂ is formed via intermediates with reduced primary quinone, first with the secondary quinone oxidized (Q_AS^•−:Q_BS) and then with it reduced to the semiquinone (Q_AS^•−:Q_BS^•(H)). Q_AS and Q_BS have pK_as and E_ms unperturbed from their solution values (identified by the subscript s). The reaction sequence is described in detail in the legend to Fig. 4 A. While reduction of quinone to the dihydroquinone (quinol) is favorable at pH 7 and Eh 0, all protonated or reduced intermediates are at higher energy. The lowest energy pathway has the quinone donor, Q_AS^•−, reducing the secondary quinone twice forming Q_AS:Q_BS⁼ before any protons are bound.

Quinone electrochemistry in bacterial photosynthetic reaction centers

The energy levels of the quinone redox intermediates (Fig. 4 A) can be compared in solution and in RCs (Fig. 4 B). Both experimental and calculated E_ms and pK_as for reactions in the protein will be considered. In RCs there are only limits for the energies for most intermediates. In contrast, it is possible to calculate energy levels for all states, predicting the energies of unobserved states. The calculated shifts in state energies from solution can then be quantitatively analyzed in light of the protein structure to understand the physical forces that yield the changes in electrochemistry. The overall picture, ranking the energy levels is the same in both simulation and experiment (Rabenstein et al. 1998; Alexov Gunner 1999; Ishikita et al. 2003; Zhu and Gunner 2005). All calculated values quoted here are from (Zhu and Gunner 2005). In general calculated and experimental values differ by less then 60 mV or 1 pH unit, which will be viewed here as representing good agreement. Reactions where experiment and simulation do not match this well highlight interesting changes in energy levels accomplished by the protein.

The protein uses the low potential bacteriopheophytin with an E_m (≈−500 mV) as the electron donor (Gunner 1991), making Q_A reduction an essentially irreversible step in photosynthetic charge separation. Thus, reduction of Q_A by BPh^•− would be favorable even if the Q/Q^•− E_m was unchanged from the −145 mV found in solution. However, the E_m of both Q_A and Q_B are raised in the protein. One reason for this could be that it ensures the semiquinone is bound more tightly to the protein then the quinone (Fig. 2b). The calculated E_m for Q_A is near 0 mV, close to the measured values of −45 mV (Dutton et al. 1973) and −70 mV (Rutherford and Evans 1980).

Q_A^•−, in its deeply buried site, is calculated have almost 400 meV less favorable interaction with water then the free semiquinone (Zhu and Gunner 2005). If the quinone E_m was lowered by this amount (to −545 mV) even the very low potential BPh^•− would have difficulty reducing it. However, interactions of Q_A^•− with the backbone dipoles and with the protein side chains are each calculated to stabilize the anion by ≈−250 meV. Q_A can be reduced in samples frozen to 4 K (Kleinfeld et al. 1984b; Gunner and Dutton 1989; Xu Gunner 2000), and this is consistent with the simulation showing small conformational changes when Q_A is reduced (Alexov and Gunner 1999). The reduction of Q_A provides a clear example of a charged cofactor being stabilized in a deeply buried site within the protein (Gunner et al. 1996). In essence the protein replaces the stabilization of the charge by water's mobile dipoles with charge:charge interactions with permanent dipoles and charged residues within the protein.

The semiquinone is stabilized in the Q_B site by 30 meV (calculation; Zhu and Gunner 2005) to ≈70 meV (experiment; Mancino et al. 1984; Kleinfeld et al. 1984a) more then in the Q_A site. The simulation shows that the two quinones loose the same amount of solvation energy, the backbone dipoles stabilize Q_B^•− more and the side chains less then Q_A^•−. The various simulations all suggest that there is significant rearrangement of the charges and dipoles in the Q_B site required to stabilize the anion (Rabenstein et al. 1998; Alexov and Gunner 1999; Rabenstein et al. 2000; Ishikita et al. 2003; Ishikita and Knapp 2004; Alexov et al. 2000). Without these rearrangements electron transfer would be unfavorable. This is constant with the electron transfer from Q_A^•− to Q_B freezing out at 200 K (Kleinfeld et al. 1984b; Xu and Gunner 2001). Interestingly, the protein frozen when it is equilibrated around Q_B^•− will allow electron transfer to reform Q_B^•− at temperatures <20 K, indicating that the protein can be trapped in the higher energy, active conformation (Xu and Gunner 2001). The electron transfer rate is independent of the driving force (Graige et al. 1998; Li et al. 2000) which implies that conformational changes, rather then the electron transfer itself, represent the rate determining step in the reaction.

Once the state PHQ_AQ_B^•− is formed, the second turnover starts with BPh^•− reducing Q_A in the presence of Q_B^•− (Fig. 3). The intermediates Q_A:Q_B⁼ and Q_A^•−:Q_B^•H are unstable and are never seen. In solution, electron transfer to form Q_A:Q_B⁼ would precede proton uptake (Fig. 4 A). However, the calculations show that the protein switches the relative stability of the two intermediates (Fig. 4 B). This is in agreement with the elegant experiments of Graige and Okamura (Graige et al. 1996). Two changes contribute to the reordering of the relative energy of the intermediates. One is that Q_A^•− is more stable in the protein then in solution so that oxidation of the primary quinone donor contributes less driving force to the reaction. More significant is that the reduction to the unprotonated Q⁼, with its −2 charge, is very unfavorable in the protein.

The destabilization of Q⁼ can be explained using a continuum electrostatics analysis (Zhu and Gunner 2005). The anionic semiquinone, Q_B^•−, is calculated to be stabilized by the fixed backbone dipoles, the proton side chains, and an adjacent non-heme iron by ≈500 meV. The calculated, favorable interactions are more then doubled for Q_B⁼ to ≈1,300 mV. The interactions of a rigid protein with Q_B⁼ would be twice that with Q_B^•−. However, there are significant rearrangements made to the protein side chains to stabilize the first reduction (Alexov and Gunner 1999; Ishikita and Knapp 2004). The reaction free energy reflects the energy needed to move from the protein equilibrated around Q_B to that equilibrated around Q_B^•− (or Q_B⁼; Fig. 2) (Mao et al. 2003). Thus, pairwise interactions with the protein favor the double reduction because the first reduction pre-organizes the protein into conformation and protonation states that favor an anion in the Q_B site. However, the solvation energy varies, to first order, with q² (Eq. 1). The stabilization of the dianion by water is seen in the comparison of the reaction in water where E_m for reduction of Q is −145 and for reduction of Q^•− to Q⁼ is −195 and in the aprotic solvent dimethyl formamide where the Q^•−/Q⁼ E_m is more then 600 mV more negative then that that of Q/Q^•− (Prince et al. 1983). The loss of solvation energy is ≈4 times larger for Q_B⁼ then for Q_B^•−, resulting in a calculated E_m for Q_B^•−/Q_B⁼ ≈500 mV lower then for Q_B/Q_B^•−. Thus, the separation of the two energy levels moving from water to the Q_B site reflects the desolvation penalty increasing with ≈q² while the favorable pairwise interactions only increases with ≈q.

In water Q^•− has a pK_a of 4.9, while Q_B^•− is calculated to have a pK_a of 2.7. At pH 7 the neutral radical would therefore be 260 meV higher in energy then the anion. The ΔG° between the two states can be estimated from the rate of electron transfer from Q_A^•− to Q_B^•−, which has been shown to proceed by rapid, uphill preprotonation of Q_B^•− followed by the electron transfer reaction (Graige et al. 1998; Graige et al. 1999). The rate is determined by the fraction of the RCs in the state Q_A^•−:Q_B^•H. The rate determining step is then the quantum tunneling of the electron from Q_A^•− to Q_B^•H. The energy of the protonated intermediate thus factors into the rate of electron transfer since the concentration of Q_B^•H is diminished by a factor of 10 for every 60 mV increase in the ΔG°. These kinetic measurements suggest Q_B^•H is between 80 and 240 meV higher in energy then Q_B^•−, with a best estimate of 160 meV (Graige et al. 1998). The higher calculated energy for Q_B^•H would slow the observed rate of the second electron transfer by ≈40 fold.

The apparent disagreement between experimental and calculated pK_a for Q_B^•− raises an interesting question about how the protein stabilizes the individual quinone redox states. The work of Okamura and Graige places the in situ pK_a for Q_B^•− at 4.5, a negligible shift from solution, indicating the proton stabilizes the charged Q_B^•− and neutral Q_B^•H by essentially the same amount. In contrast, in the simulations Q_B^•− is stabilized relative to both neutral Q and Q_B^•H by ≈140 meV, lowering the calculated semiquinone pK_a from 4.9 to 2.7. If the experimental assignment of the quinone pK_a is correct then the protein must stabilize Q_B^•H relative to Q by ≈140 meV more then found in the current simulations.

The final steps in formation of the quinol in the Q_B site involve the second reduction of Q_B followed by binding the second proton to form Q_BH₂. The doubly reduced Q_BH⁻ is calculated to be more stable in the protein then in solution (Fig. 4 B). However, the calculated pK_a for proton binding is ≈8 while the pK_a is 10.7 in solution indicating the doubly protonated QH₂ is destabilized in the binding site, with the quinol bound ≈50 fold less tightly then the quinone. This favors quinol dissociation, allowing a new quinone to be bound to restart the reaction cycle (Fig. 2 B).

Thermodynamics and kinetics of quinone binding

The reduction of ubiquinone in the Q_B site uses two turnovers of the single electron donor Q_A^•− to form the doubly reduced, product QH₂, which is released into the Q_pool (Fig. 3). The semiquinone is a potentially dangerous high-energy free radical, which is considered to be a primary source of damaging reactive oxygen species (Kramer et al. 2004). In addition, the efficiency of the conversion of light energy to chemical energy relies on Q_A^•− and Q_B^•− remaining bound to the protein. Thus, the semiquinone at the Q_A^•− site must be sufficiently stable that Q_B has time to bind and be reduced. Then Q_B^•− must live long enough that a second photon can rereduce Q_A even under low light conditions where photons are scarce.

Given the importance of trapping the bound semiquinones at the Q_A and Q_B site it might seem as though the quinone binding sites should be designed to bind the semiquinone extremely tightly. However, stabilization of any individual redox intermediate will have consequences for the overall electron and proton transfers (Fig. 4 B). An over-stabilized Q_A^•− would not be able to transfer electrons to Q_B. If Q_B^•− were too stable, the succeeding intermediates would not be formed. Alternatively the protein could bind all quinone redox states very tightly to ensure the semiquinone were not lost while not changing the reaction free energy landscape. However, this would slow down turnover since, the product QH₂ must be released from the Q_B site at the end of the reaction sequence.

A study of the binding of neutral and anionic quinones to the Q_A site of RCs may provide insight into how the protein can sequester an anionic semiquinone while not making it too stable (Madeo and Gunner 2005). Quinone on and off rates were measured for a series of tailless neutral, methyl and methoxy substituted benzoquinones and naphthoquinones. The semiquinone is not stable in aqueous solution so its affinity and binding rates cannot be determined directly. Rather hydroxyl quinones were used and measurements carried out above the hydroxyl pK_a. While these anionic quinones are not reduced in the Q_A site they are found to be good, reversible competitive inhibitors for the binding site.

If the reactant and product are bound with different affinities the electrochemistry of the reaction will be shifted in the protein (Fig. 2 B). The E_m for the reduction to the anionic semiquinone is 70–100 mV more positive in the Q_A site then in water. This indicates the semiquinone binds 15–45 times more tightly then the quinone. The k_off for the tailless ubiquinone0 (2,3-dimethoxy, 5 methyl-benzoquinone), is 6/s (Madeo and Gunner 2005). If the change in affinity were translated into a change in the off rate (K_eq = k_off / k_on) the t_1/2 for semiquinone release would be 2.5–7.5 s. However, the series of neutral, alkyl quinones tested show a negligible change in k_off with affinity. Rather for the neutral quinones the changes in affinity strongly correlate with changes in k_on. If the semiquinone behaved similarly it would dissociate from the protein in 100 s of milliseconds and so could be easily lost.

The neutral, alkyl substituted and anionic, hydroxy substituted quinones studied have similar structures and similar K_ds for the binding site, ranging from 0.1 to 10 μM (Madeo and Gunner 2005). However, the anionic, hydroxyl quinones have a binding mechanism that is different and remarkably slower then the neutral quinones. Thus, the rate-determining step for binding the neutral quinones is bimolecular while it is first order for the anionic quinone, indicating the latter process is gated in some way. The binding rates k_on cannot be directly compared for the two types of compounds since the bimolecular reaction has units of per molar per second while it is per second for the unimolecular reaction. However the off rates both have units of per second. The anionic compounds are found to dissociate a remarkable 10⁴ times more slowly then do the neutral compounds. The anionic, hydroxyl quinones bind and dissociate in tens of minutes while the neutral compounds do this in milliseconds. Thus, the protein is seen to decouple the kinetic and thermodynamic stability to firmly bind anionic quinones. If the anionic, hydroxyl quinones are a good analog for the anionic semiquinone, the protein could shift the K_d for Q_A^•− relative to Q_A by less then 100 fold, constant with the modest E_m shift, while slowing the rate of Q_A^•− dissociation by 10⁴ fold.

Conclusion

The photosynthetic reaction centers are a well-studied system where we have remarkable knowledge of the protein structures, reaction pathways, and cofactor electrochemistry and binding constants. This information makes it possible to understand how quinone properties are altered when bound to specific binding sites in the transmembrane electron transfer proteins. A continuum electrostatics analysis shows how the protein can stabilize an anionic semiquinone (Q_B^•−) while destabilizing the dianion (Q_B⁼). Comparison of experimental and calculated results show that the protonated semiquinone is bound especially tightly, enabling the rapid preprotonation of Q_B^•−. Further analysis of anionic inhibitor binding suggests that the protein uses kinetic barriers to sequester the valuable, high-energy semiquinone intermediates.

The lessons learned here may provide insight into how the other transmembrane electron transfer proteins such as the cytochrome bc₁ oxidoreductase control their electron transfer reactions. As in RCs the bc₁ complex binds two ubiquinones whose roles are defined by their binding sites. The Q_i site carries out a clear two-step reaction sequence that is similar to that found in the Q_B site, with two turnovers leading to quinone reduction. At the Q_O site there is only transient semiquinone production as the bound QH₂ reduces two different reaction partners (Cape et al. 2007; Zhang et al. 2007). The protein controls the relative energy levels of the different redox intermediates for function and stability. For example, increasing the thermodynamic stability of the semiquinone in the Q_O site leads to the production of harmful reactive oxygen species (Cape et al. 2005; Forquer et al. 2006). It has been shown here that proteins can decouple thermodynamics and kinetics of reactions, slowing the binding and dissociation of anionic quinones without greatly increasing their affinity for the proteins. Mechanisms of this kind might play a role in inhibiting thermodynamically favorable undesirable reactions such as the reduction of the high potential cytochrome c₁ by the lower potential b-type cytochromes in the bc₁ complex (Rich 2004; Osyczka et al. 2005)