Energetics of the exchangeable quinone, Q_B, in Photosystem II

Significance

The E_m values of the 2 Q_B redox couples, E_m(Q_B/Q_B^•−) and E_m(Q_B^•−/Q_BH₂), in Photosystem II are fundamental for understanding the enzyme function, but they remain vague. Using EPR signals arising from the semiquinone, Q_B^•−, we have now measured both redox couples. The semiquinone is thermodynamically stable and has a relatively high potential. This minimizes back-reactions and electrons leaking onto O₂, explaining the remarkable stability of Q_B^•−. The release of Q_BH₂ is thermodynamically favorable, and the binding of the substrate PQ is greatly favored over the product PQH₂: This optimizes PSII function in the presence of a largely reduced plastoquinone pool. This work corrects and explains a recent report of anomalous values for the Q_B redox couples.

Abstract

Photosystem II (PSII), the light-driven water/plastoquinone photooxidoreductase, is of central importance in the planetary energy cycle. The product of the reaction, plastohydroquinone (PQH₂), is released into the membrane from the Q_B site, where it is formed. A plastoquinone (PQ) from the membrane pool then binds into the Q_B site. Despite their functional importance, the thermodynamic properties of the PQ in the Q_B site, Q_B, in its different redox forms have received relatively little attention. Here we report the midpoint potentials (E_m) of Q_B in PSII from Thermosynechococcus elongatus using electron paramagnetic resonance (EPR) spectroscopy: E_m Q_B/Q_B^•− ≈ 90 mV, and E_m Q_B^•−/Q_BH₂ ≈ 40 mV. These data allow the following conclusions: 1) The semiquinone, Q_B^•−, is stabilized thermodynamically; 2) the resulting E_m Q_B/Q_BH₂ (∼65 mV) is lower than the E_m PQ/PQH₂ (∼117 mV), and the difference (ΔE ≈ 50 meV) represents the driving force for Q_BH₂ release into the pool; 3) PQ is ∼50× more tightly bound than PQH₂; and 4) the difference between the E_m Q_B/Q_B^•− measured here and the E_m Q_A/Q_A^•− from the literature is ∼234 meV, in principle corresponding to the driving force for electron transfer from Q_A^•− to Q_B. The pH dependence of the thermoluminescence associated with Q_B^•− provided a functional estimate for this energy gap and gave a similar value (≥180 meV). These estimates are larger than the generally accepted value (∼70 meV), and this is discussed. The energetics of Q_B in PSII are comparable to those in the homologous purple bacterial reaction center.

In photosynthesis, light is absorbed by chlorophyll, resulting in charge separation within a photosynthetic reaction center. In Photosystem II (PSII), the water/plastoquinone photooxidoreductase, the electron hole is transferred from the chlorophyll cation radical, P_D1^•+, via a redox-active tyrosine (Tyr_Z) to the Mn₄O₅Ca cluster. After 4 sequential photochemical turnovers and the resulting oxidations of the Mn₄O₅Ca cluster (S_0–4), 2 water molecules are oxidized (1). On the electron acceptor side, the electron is transferred from the pheophytin anion (Pheo_D1^•−) via a nonexchangeable plastoquinone (Q_A), which acts as a one-electron relay, to an exchangeable plastoquinone (Q_B), the terminal electron acceptor (2).

Q_B^•− decays in tens of seconds by charge recombination with S₂ or S₃ when present, but is stable for several hours if S₀ or S₁ are present (3). During the subsequent photochemical turnover, Q_B^•− accepts a second electron from the newly formed Q_A^•−. This is accompanied by the 2 protonation steps, thought to occur sequentially, one before and one after the arrival of the second electron (4), as occurs in the homologous site in purple bacteria (5). The Q_BH₂ formed is released from the site and enters the PQ/PQH₂ pool, from where it can deliver electrons to the cytochrome b₆f complex (6–8).

Due to the 2-electron chemistry of Q_B, the Q_B^•− state is the only state on the electron acceptor side that cannot be stabilized by forward electron transfer; i.e., there is no “kinetic control” on this step (9). “Kinetic control” of electron transfer occurs in photosynthetic reaction centers when the forward rate outruns both the backward rate and the charge pair recombination rate (9). As a stable intermediate, Q_B^•− is therefore available to back-react via Q_A^•− with the S₂ and S₃ states on the donor side (3). Two main competing back-reaction pathways occur within PSII: 1) the direct route via electron tunneling from Q_A^•− to P^•+ and 2) the indirect route via thermal repopulation of the P^•+Pheo_D1^•− state (10). Recombination from the P^•+Pheo_D1^•− state mainly forms the chlorophyll triplet state ³P₆₈₀ (11), which reacts with oxygen to form highly reactive and damaging singlet oxygen ¹O₂ (12).

The driving force (ΔG) for electron transfer between Q_A and Q_B determines the Q_A^•−Q_B ↔ Q_AQ_B^•− equilibrium and therefore the extent to which back-reactions from Q_B^•− (or Q_BH₂) to Q_A can occur. This equilibrium is determined by the difference between the midpoint potentials (E_m) of the donor and the acceptor, E_m(Q_A/Q_A^•−) and E_m(Q_B/Q_B^•−) or E_m(Q_B^•−/Q_BH₂), according to the following equation:

$$\mathit{\Delta }G=-\text{n}F\left(\mathrm{\Delta }E\right).$$ [1]

The redox state of Q_A can be monitored relatively easily using fluorescence measurements. Q_A^•− has an electrostatic effect on the potential of Pheo_D1/Pheo_D1^•−, decreasing the already small energy gap between P* (the primary donor) and the first radical pair, which results in a decrease in the yield of charge separation and an increase in fluorescence. A wide range of values have been reported (13). This scatter of reported values is, at least in part, due to the potential being modulated to regulate forward and particularly back electron transfer reactions (9). The Q_A redox potential is affected by the binding of the Mn₄O₅Ca cluster to its site (10, 13–15) and by the bicarbonate binding to the nonheme iron (16) (Fig. 1). The redox potential of the Q_A/Q_A^•− couple in the fully functional, bicarbonate-bound system is −144 mV (15, 16).

Fig. 1.

PSII acceptor side. Blue lines denote possible hydrogen bonds. Data from the 1.9-Å crystal structure (Protein Data Bank ID Code 3WU2) (53).

The measurement of the redox state of Q_B is more complicated than that of Q_A. Firstly, there is no easy experimental probe for the redox state of Q_B. Secondly, in contrast to Q_A, which undergoes a one-electron redox reaction forming Q_A^•− without the involvement of protons, the Q_B reduction involves 2 electrons and 2 protons (Eq. 2).

$${\text{Q}}_{\mathrm{B}}\hspace{1em}\stackrel{+e^{-}+{\text{H}}^{+}}{\leftrightarrow }\hspace{1em}{\text{Q}}_{\mathrm{B}}^{\cdot -}\left({\text{H}}^{+}\right)\hspace{1em}\stackrel{+e^{-}+{\text{H}}^{+}}{\leftrightarrow }\hspace{1em}{\text{Q}}_{\mathrm{B}}{\text{H}}_2.$$ [2]

Thirdly, Q_A is a tightly bound cofactor in both of its redox states, while Q_B has 2 of the 3 relevant redox states (Q_B and Q_BH₂) that are relatively weakly bound and exchangeable with PQ or PQH₂. As a consequence, kinetic data have been used to estimate the redox potential of Q_B by deriving the apparent equilibrium constant, K, for the electron transfer between Q_A^•− and Q_B (17–19). Taking K as 15 (K = 10 to 20 in refs. 17–19), the energy gap is 70 meV, so, with the E_m of Q_A/Q_A^•− of −144 mV (15, 16), the E_m of Q_B/Q_B^•− E_m would thus be −74 mV (17–19). Due to the complex nature of the experiments from which the kinetic parameters were extracted, uncertainties remain (17–20). A theoretical model based on thermoluminescence (TL) was used to obtain a similar value for the energy gap between the Q_A/Q_A^•− and Q_B/Q_B^•− couples (21, 22). Doubts remain, however, as experimental observables in TL do not always fit with thermodynamic parameters derived from the TL model (21, 23).

In summary, the mechanism and energetics of the PSII acceptor side and especially Q_B are still relatively poorly understood. The redox potentials for the 2 couples, E_m(Q_B/Q_B^•−) and E_m(Q_B^•−/Q_BH₂), have received relatively little attention, despite being central to understanding the quinone reduction function of this enzyme. These redox potentials should provide thermodynamic information related to 1) the stability of the electron transfer intermediates and their tendency to back-react, or to react with O₂; 2) the driving force for forward reactions; and 3) the (de)binding properties of the substrate (PQ) and product (PQH₂).

An experimental estimate of the E_m values of the 2 Q_B redox couples, Q_B/Q_B^•− and Q_B^•−/Q_BH₂, based on equilibrium redox titrations was published recently (24). Fourier transform infrared (FTIR) was used to monitor Q_B^•− formation upon illumination by a single flash as a function of the applied potential. The data seemed to show that Q_B^•− was not thermodynamically stable, a surprising result in light of the mechanistic implications and given the conflicting results in the homologous purple bacterial reaction centers (20, 25–27).

Here we have used electron paramagnetic resonance (EPR) spectroscopy to estimate the redox potential of the 2 couples, Q_B/Q_B^•− and Q_B^•−/Q_BH₂. Two different EPR signals were measured: a Q_B^•−Fe²⁺ semiquinone signal (28) and a Q_A^•−Fe²⁺Q_B^•− biradical signal (29, 30). We also used TL to estimate empirically the energy gap between the Q_A/Q_A^•− and Q_B/Q_B^•− couples, without relying on the theoretical model previously employed. Our results differ from previous measurements and estimates and show that the semiquinone, Q_B^•−, is stabilized thermodynamically and that the plastoquinone is preferentially bound compared to the plastoquinol.

Results

EPR spectra of PSII were measured at a series of electrode potentials. The D2-Y160F mutant lacking tyrosine D (Tyr_D) was used to eliminate the Tyr_D^• signal, which would otherwise dominate the PSII EPR spectrum in the radical region (31). At each potential, dark spectra and spectra after illumination at 77 K were recorded. Fig. 2A shows a scan of the radical region around g = 2. The appearance and disappearance of the EPR signal as a function of potential can be observed. This signal has been assigned to the low-field edge of the ground-state doublet of the semiquinone, Q_B^•−Fe²⁺ (28).

Fig. 2.

EPR spectra of PSII poised at different potentials from an equilibrium redox titration done in the dark. (A) Radical region spectra showing the Q_B^•−Fe²⁺ at g = 2.0024 (microwave power, 205.1 mW; modulation amplitude, 10.53 G). (B) Wider scan of the same samples after 77-K illumination showing the Q_A^•−Fe²⁺Q_B^•− signal at g = 1.66 (microwave power, 20 mW; modulation amplitude, 25.35 G).

Fig. 2B shows a full spectrum scan of the same samples as used in Fig. 2A after illumination at 77 K. A peak at 4,000 G (g = 1.66) shows a potential dependence similar to that of the Q_B^•−Fe²⁺ signal. The g = 1.66 signal has been assigned to the Q_A^•−Fe²⁺Q_B^•− biradical state (29, 30). The low-temperature illumination generates Q_A^•− in nearly all of the centers. No electron transfer occurs from Q_A^•− to Q_B or to Q_B^•− at 77 K (30); therefore the biradical signal should only be observed if Q_B^•− were present before the 77-K illumination. Thus, the biradical signal can be used to monitor the presence of Q_B^•− independently of the Q_B^•−Fe²⁺ signal.

To assess the proportion of Q_B^•− formed during the titration, the signal size of the Q_B^•−Fe²⁺ signal if present in 100% of the centers was estimated. In a dark-adapted sample, Q_B^•− is present in ∼40% of the centers, while Q_B is present in the rest (3, 30). That proportionality can be inverted by illuminating at 77 K and subsequently thawing in darkness (3). Therefore, the sum of the amplitudes of the signals present before and after this treatment should yield the size of the signal when Q_B^•− is present in all of the centers. This experiment was done, and the estimated value for 100% Q_B^•− was used to calibrate the amplitudes of the EPR signals in the titrations (SI Appendix, Fig. S2).

Fig. 3 shows a plot of the normalized Q_B^•−Fe²⁺ and Q_A^•−Fe²⁺Q_B^•− signals versus the measured potential, combining data from 3 individual titrations (including those in Fig. 2). Titrations were carried out in oxidizing and reducing directions. Due to poor mediation between ∼0 mV and 50 mV, fewer reliable data points could be collected in that region, and the data are more scattered. The maximum amplitude, which occurs at 67 mV, represented about 55% of the centers in the stable Q_B^•− state. Data were fitted using the model first established by Michaelis (32). The resulting potentials for the 2 couples were E_m(Q_B/Q_B^•−) = 92 ± 23 mV and E_m(Q_B^•−/Q_BH₂) = 43 ± 23 mV.

Fig. 3.

Equilibrium redox titration of the Q_B^•− semiquinone in the dark done using 2 different EPR signals. Open squares, oxidizing titration of the Q_B^•−Fe²⁺ signal; closed squares, reducing titration of the Q_B^•−Fe²⁺ signal; closed diamonds, reducing titration of the Q_A^•−Fe²⁺Q_B^•− signal; open diamonds, oxidizing titration of the Q_A^•−Fe²⁺Q_B^•− signal.

The retention of the Mn₄O₅Ca cluster during the course of the titration was verified in 2 ways: firstly, by the presence of free “hexaquo” Mn²⁺ signals in the spectra, representing a loss of the Mn₄O₅Ca cluster, and, secondly, by the ability to form the S₂ multiline signal (33). The Mn₄O₅Ca cluster was retained in the majority of centers even after exposure to low potentials (SI Appendix, Fig. S5 and associated text).

In addition to the redox titrations of the EPR signals associated with Q_B^•−, pH-dependent TL measurements were used to estimate the energy gap between the Q_A/Q_A^•− and Q_B/Q_B^•− couples in PSII cores containing TyrD. TL measures the emission of luminescence associated with the heating-induced back-reaction of a stable charge-separated state. The peak temperature of TL is indicative of the energy stored in the charge-separated state and is determined by redox potentials of both the recombination partners, in this case, S₂/S₁ and Q_B/Q_B^•− (3, 21). Below pH of ∼7, the S₂/S₁ couple involves very little protonation and is almost independent of pH (34). The Q_B/Q_B^•− couple involves proton release when Q_B^•− is reoxidized and is expected to follow Nernst behavior (19, 35, 36), with the redox potential changing by −59 mV per pH unit. Thus, the pH dependence of the S₂Q_B^•− recombination peak position should reflect this process and can be used as an empirical calibration of the change in emission temperature in terms of the change in the redox potential of Q_B.

Fig. 4 shows the TL curves of long dark-adapted PSII samples after one saturating flash at different pH values. A clear shift of the peak positions to lower temperatures with increasing pH can be seen. Fig. 4, Inset shows a plot of peak temperature versus the pH, from which a linear dependence of −11.6 °C per pH unit can be observed. Using the ΔE relationship given by the Nernst equation, this translates to −5.1 meV °C⁻¹. A similar slope was observed by Vass and Inoue (37) in their study of the pH dependence of TL in plant PSII (see also SI Appendix, Figs. S8–S13).

Fig. 4.

Thermoluminescence of long-dark-adapted PSII cores from T. elongatus. Blue, after one saturating flash at different values of pH; red, after the addition of DCMU. Inset shows a plot of the S₂Q_B^•− peak position versus the pH with a slope of −11.6 °C/pH unit and a plot of the S₂Q_B^•− peak over the restricted zone of pH in which it is pH independent. At pH 6.5, a TL band of ∼70 °C becomes the dominant emission but is not attributed to the S₂Q_B^•−; it is more likely to be from the so-called C band (54).

This calibration was then used to estimate the gap between the Q_A and Q_B couples. Fig. 4 (red circles) shows the S₂Q_A^•− recombination band in the presence of the Q_B-site inhibitor, DCMU (3-(3,4-dichlorophenyl)-1,1-dimethylurea) occurring at ∼14 °C. The difference in peak positions of 36 °C between the S₂Q_A^•− and S₂Q_B^•− peaks corresponds to an energy gap of 183 meV. This may be extended by taking into account the upshift of S₂Q_A^•− caused by the binding of DCMU of ∼52 mV in plant PSII (38). However, this effect of DCMU has yet to be demonstrated in Thermosynechococcus elongatus PSII.

Discussion

In the present work, the midpoint potential of the terminal electron acceptor of PSII, Q_B, was measured with EPR. The results show that the semiquinone, Q_B^•−, is stabilized thermodynamically at pH 7. The redox potentials derived from the data for the 2 redox couples are E_m(Q_B/Q_B^•−) ≈ 90 mV and E_m(Q_B^•−/Q_BH₂) ≈ 40 mV. In addition, we have estimated the difference in redox potentials between the Q_A/Q_A^•− and Q_B/Q_B^•− couples using the pH dependence of S₂Q_A^•− and S₂Q_B^•− recombination measured by TL. The energy difference obtained from this approach is ∼180 meV (230 meV if there is a DCMU-induced shift on the Q_A potential, as occurs in plant PSII). This value is similar to the ΔE of 234 meV between the E_m = −144 mV for Q_A/Q_A^•− (15, 16) and +90 mV for Q_B/Q_B^•− presented here.

Stabilization of the Semiquinone State within the Q_B Site.

A fit of the Nernst model to our data shows the difference between the Q_B/Q_B^•− and Q_B^•−/Q_BH₂ couples, ΔE_m, to be ∼+50 mV, a value that agrees with the measured maximum of ∼55% of stable Q_B^•− in the EPR signal generated in the titrations. This ΔE_m is indicative of the degree of stabilization of the semiquinone radical, Q_B^•−, in the site, with larger values equating to a more stable semiquinone. When titrating a free quinone in the membrane, the semiquinone is not stabilized, the redox transition occurs as a steeper n = 2 curve (i.e., a 2-electron transition), no semiquinone can be observed, and the ΔE_m is negative. Although the membrane is aprotic, the quinone head group can access the aqueous solvent, and thus the semiquinone is destabilized [ΔE_m ≈ −500 meV for ubiquinone (39)]. Our data indicate that Q_B^•− is strongly bound and stabilized by the Q_B site in comparison to a quinone in the membrane. The structure of the Q_B site (Fig. 1) shows several features that likely contribute to the electrochemical properties of Q_B^•−, including 1) the proximity to the nonheme ferrous iron, 2) hydrogen bonds from D₁His215 and D₁Ser264 to both carbonyls of the quinone, and 3) the likely protonation of the distal H-bonding D₁Ser264/His252 pair (4).

This stabilization can be rationalized in part as a damage prevention mechanism. In PSII, back-reactions from P^•+Q_A^•− result in damage to the complex (9, 16). A large gap between the Q_A/Q_A^•− and Q_B/Q_B^•− couples would favor Q_B reduction and lower the equilibrium concentration of Q_A^•−, thereby diminishing the likelihood of a damaging back-reaction from Q_A^•−_. This stabilization of the semiquinone would come at a cost, because a more positive E_m(Q_B/Q_B^•−) must yield a more negative E_m(Q_B^•−/Q_BH₂) in order to maintain the chemical requirement that the average E_m for the 2 couples remains unchanged (see below). Back-reactions from the fully reduced quinol, Q_BH₂, to Q_A would therefore become more likely. These back-reactions, however, would only occur when the plastoquinone pool is reduced, because quinol/quinone exchange occurs orders of magnitude faster (∼10 ms; ref. 40) than the S₂Q_BH₂ back-reaction, which we predict to decay with kinetics between the recombination rates of Q_A^•− and Q_B^•− with S₂ [i.e., between ∼1 s and ∼30 s (41)].

The high potential E_m (∼90 mV) for the Q_B/Q_B^•− also means that Q_B^•− is a poor reductant for O₂ forming superoxide (E_m(O₂/O₂^•−) ≈ −160 mV). It has been suggested that Q_B^•− may be a source of reactive oxygen species (42); however, its very long lifetime, with half-times of hours in the presence of S₁ and S₀ (3, 30), argues against this. The present work showing the thermodynamic stabilization of Q_B^•− provides a good explanation for its lack of reactivity with oxygen and for its very long lifetime in the dark.

The presence of a stabilized Q_B^•− directly contradicts a recent report (24) (discussed in detail below) but is supported by other reports. A reductive titration of the g = 1.66 biradical signal with PSII particles from Phormidium laminosum was reported previously as part of the work done to identify the origin of the biradical signal (29). Although no reversibility was demonstrated and the signal size was not quantified, a thermodynamically stable Q_B^•−, similar to that reported in this work, was clearly present in the reductive titration.

A thermodynamically stable Q_B^•− was also present in the homologous purple bacterial reaction centers from Rhodobacter sphaeroides (25), Chromatium vinosum (26), and Blastochloris viridis (27). These titrations show ∼30 to 50% of the total quinone form as a stable semiquinone (see SI Appendix, Fig. S7 for a corrected fitting of these titrations). The exact ΔE_m and E_m^avrg all differ from each other to some extent, perhaps reflecting different mechanistic requirements in the different species, but, in all reaction centers, the semiquinone, Q_B^•−, is clearly thermodynamically stabilized. Together, these studies in the literature (25–27, 29) provide strong support for the current results.

Preferential Binding of the Quinone vs. the Quinol.

Although the redox potential of the plastoquinone in the pool from T. elongatus has not been measured, it is expected to be very similar to that in other organisms, i.e., a 2-electron transition at 117 mV (43, 44). The average for the 2 Q_B redox couples measured here (E_m^avrg ≈ 65 mV) is therefore about 50 mV lower than that of the plastoquinone pool. This represents a significant driving force for the release of the PQH₂ to the pool. This should be considered as an additional energy loss in the ongoing effort to understand energy use in PSII. A smaller but qualitatively similar energy loss was reported earlier for UQH₂ release from the Q_B site in the purple bacterial reaction center to the pool, based on a computational chemistry approach (39).

From the difference in redox potential, the ratio of the binding constants for the quinone and the quinol forms can be calculated (16) (see SI Appendix, Fig. S6 and associated text for details). It is found that the quinone is bound to the Q_B site ∼50 times more tightly than the quinol. Again, this fits qualitatively with the situation reported in purple bacterial reaction centers (39, 45). For PSII to function optimally, it makes sense that the binding of the substrate (the quinone) is favored over that of the product (the quinol). Preferential quinone binding will allow PSII to function when the pool is significantly reduced.

In the literature, it has often been assumed that the binding constants of the quinone and quinol in the Q_B site are equal both in PSII and in purple bacterial reaction centers (e.g., refs. 18 and 20). Nevertheless, the Q_B^•− redox titrations in purple bacterial reaction centers indicated the preferential binding of the quinone over the quinol (25), and other reports favor this binding regime (6, 46), partly because of the experimental indications (25–27, 39) and partly because it seemed mechanistically more likely [see SI Appendix for further discussion and a possible exception in C. vinosum (26)]. The present work is good experimental evidence for preferential binding of the quinone over the quinol in PSII.

The Difference in Redox Potentials Between Q_A and Q_B.

The difference in equilibrium redox potentials between the Q_A/Q_A^•− and Q_B/Q_B^•− couples reported here, ∼234 meV, is larger than previously estimated (∼70 meV) from the kinetics of the forward and backward electron transfer reactions between Q_A and Q_B (21, 41). The following factors could contribute to the mismatch: 1) The previous dynamic measurements are based on assumptions that may be complicated by the so-called “gating effect,” which determines the rate-limiting step of electron transfer between Q_A and Q_B rather than the driving force (47). This dynamic effect is not accounted for in the previous estimates and may at least contribute to the discrepancy. 2) While the Q_B redox couples themselves do not suffer from the problem of being titrated adjacent to an already reduced cofactor, the E_m of Q_A/Q_A^•− may be systematically underestimated (a lower potential) in redox titrations due to the binding of Q_BH₂ which would result in a smaller energy gap. 3) The Q_A/Q_A^•− couple may be more specifically sensitive to modulation from the electron donor side, which would also affect the energy gap estimate (see SI Appendix for a more detailed discussion).

Given the uncertainty about the energy gap between the E_m Q_A/Q_A^•− from the literature and the E_mQ_B/Q_B^•− measured here, we made an independent estimate of this energy gap by estimating empirically the energy gap between S₂Q_A^•− and S₂Q_B^•− as a function of pH. It was assumed that the energy gap determining the peak position of the TL follows Nernst behavior and shifts by 59 mV per pH unit. This assumption seems justified given the experimentally determined proton stoichiometries in the literature (34, 36). The resulting value, ∼180 meV or ∼230 meV if we assume a DCMU binding effect (40), is similar to the energy gap obtained from the equilibrium redox titrations (∼234 meV). A similar value (∼212 meV) is obtained by using the data from Vass and Inoue (37). This estimate is done in the presence of S₂ (as were the previous kinetic estimates) rather than S₁ (as were the equilibrium redox titrations), thus indicating the absence of a significant influence of the S state on the energy gap. We remain somewhat skeptical about the similarity of the 2 values because in the TL method: 1) The determination of the S₂Q_A^•− and S₂Q_B^•− peak temperatures are not very accurate, and 2) the added DCMU shift of 50 mV has not yet been determined in PSII cores from T. elongatus. Despite these shortcomings, the TL measurements seem to provide independent support for the energy gap being significantly bigger than 70 meV estimated previously.

A large energy gap would make sense in functional terms. As Q_B^•− is the terminal electron acceptor, it cannot be prevented from back-reacting by “kinetic control,” i.e., by ensuring the forward electron transfer is faster than the back-reactions (9). Thus wasteful and damaging back-reactions can be minimized by increasing the energy gap between the Q_A/Q_A^•− and Q_B/Q_B^•− couples.

Rationalizing the Conflicting Report in the Literature.

The findings in the present work differ significantly from a recently published study on the redox potentials of Q_B (24). In that work, redox titrations were performed in which the ability to form Q_B^•− by a single saturating flash was monitored by FTIR measurements in a spectroelectrochemical thin cell (24). This is not a direct measurement of Q_B^•−; rather, it is a measurement of the ability to form Q_B^•− upon flash illumination (or to form Q_BH₂ if Q_B^•− were already present). This method should, in principle, provide a valid Q_B titration. The results in ref. 24, however, showed no evidence for stable Q_B^•− formation. Instead a redox curve was reported that was essentially indistinguishable from an n = 2 curve, i.e., a 2-electron transition, with an E_m = 155 mV at pH 6.5; this is equivalent to 125 mV at pH 7. The E_m and the n = 2 curve are both characteristic of a titration of free plastoquinone (43, 44). While the PSII cores have no membrane and thus no membrane-localized quinone pool, they do contain 1 or 2 free quinones in addition to Q_B, and these quinones act as a limited plastoquinone pool (30, 48, 49).

Here we show that the E_m values of the quinone couples in the Q_B site are more negative than that of free quinone; therefore, in a reductive titration, the free quinone will be reduced before the Q_B quinone. Since quinone exchange happens on the millisecond timescale, this free quinol will exchange with a quinone in the Q_B site, allowing the free quinone to be reduced. If the mediation with the Q_B site is insufficient, electrons will not be removed from Q_BH₂, and therefore the redox state of the quinone in the Q_B site will reflect the redox state of the pool, irrespective of the true quinone potential in the Q_B site. Under these conditions, an n = 2 Nernst dependency at the E_m of the pool would be observed even when probing exclusively the Q_B site. Given that 1) only 3 mediators were used in the titration, only one of which is in the appropriate range, and 2) the flash-induced state monitored had to be stable for several minutes in order to allow the data collection, it seems likely that the redox mediation was insufficient and therefore the E_m of PQ was measured (24). See SI Appendix for a detailed analysis of ref. 24.

Conclusion

Fig. 5 summarizes the results of the present work and provides a consistent energetic description of PSII, now including the 2 redox potentials of the Q_B couples. This provides insights into the redox tuning of Q_B with respect to the redox potentials of its neighboring redox partners, Q_A and free plastoquinone. The energy gap between the Q_A/Q_A^•− and the 2 Q_B redox couples reported here is significantly larger than previously assumed. The redox potentials need to be high enough compared to that of Q_A to provide for sufficient driving force and to minimize back-reactions. The redox potential of the plastoquinone pool limits the average value of the 2 Q_B couples. The measured value shows that, in PSII, ∼50 meV of driving force is expended to ensure rapid debinding of the quinol and the preferential binding of the quinone. These data indicate that the protein tunes the thermodynamics of the Q_B redox chemistry to optimize function over a wide range of plastoquinone pool reduction states, while minimizing back-reactions and side reactions with O₂.

Fig. 5.

Redox scheme of PSII. Redox potentials values were taken from the present work for the Q_B couples and from ref. 16 for Q_A, from ref. 55 for Pheo_D1, from ref. 2 for P680, and from refs. 43 and 44 for the quinone pool. The yellow arrow represents excitation of the P680 chlorophyll. The straight black arrows are electron transfer events. The curved arrows show excitation transfer.

Materials and Methods

Isolation of PSII from T. elongatus.

PSII cores were isolated from D2-Y160F; CP43-His strain (31) using a method based on that of Sugiura and Inoue (50) with specific modifications described in SI Appendix.

EPR-Detected Potentiometric Titrations.

Multiple PSII preparations were pooled to yield 7 mL to 10 mL of purified PSII at 0.7 mg (Chl)⋅mL⁻¹ in titration buffer. Redox titrations were carried out essentially as described by Dutton (51) at 15 °C under a bicarbonate-enriched argon atmosphere in absolute darkness and in the presence of the following redox mediators all at 50 μM: N,N,N′,N′-tetramethyl-p-phenylenediamine (300 mV), 2,6-Dichlorophenolindophenol (217 mV), Phenazine methosulfate (80 mV), Thionine (64 mV), Phenazine ethosulfate (55 mV), Methylene blue (11 mV), Pyocyanin (−34 mV), Indigotetrasulfonate (−46 mV), and Resorufin (−51 mV). Reductive titrations were carried out using sodium dithionite; oxidative titrations were carried out using potassium ferricyanide. EPR spectra were recorded on a Bruker ElexSys X-band spectrometer fitted with an Oxford Instruments liquid helium cryostat and temperature control system. Illumination at 77 K (20 min) was carried out in an unsilvered dewar with a halogen lamp (LQ 2600; Fiberoptic-Heim AG). See SI Appendix, Fig. S1 for further information on data treatment and fitting.

TL was measured using a laboratory-built apparatus (52). PSII cores (20 μg [Chl]⋅mL⁻¹) in buffer 1 (MOPS was used instead of MES at pH > 7 and HEPES at pH > 8, 20% glycerol instead of 10%) were dark-adapted for >1 h at 4 °C, and 200-μL samples were loaded in darkness. If required, DCMU (in ethanol) was added to the sample on the sample plate (50 μM final concentration). The pH was measured in darkness directly prior to the experiment. Samples were cooled (<30 s) to −20 °C with liquid N₂ directly after a ∼5-ns laser flash at 523 nm, the second harmonic of a Nd-YAG laser (Minilite II; Continuum). The frozen samples were then heated at a constant rate of 20 °C⋅min⁻¹, and TL was detected with a photomultiplier (H7422-50; Hamamatsu). The signal was amplified using a transimpedance amplifier (C7319; Hamamatsu) and digitized using a microcontroller board based on the Atmel SAM3X8E ARM Cortex-M3 CPU (Arduino Due).