Metastable radical state, nonreactive with oxygen, isinherent to catalysis by respiratory and photosyntheticcytochromes bc1/b6fMarcin Sarewicza,1, Łukasz Bujnowicza,1, Satarupa Bhadurib, Sandeep K. Singhb, William A. Cramerb,and Artur Osyczkaa,2

aDepartment of Molecular Biophysics, Faculty of Biochemistry, Biophysics and Biotechnology, Jagiellonian University, 30-387 Krakow, Poland; andbDepartment of Biological Sciences, Purdue University, West Lafayette, IN 47907

Edited by Harry B. Gray, California Institute of Technology, Pasadena, CA, and approved December 23, 2016 (received for review November 15, 2016)

Oxygenic respiration and photosynthesis based on quinone redoxreactions face a danger of wasteful energy dissipation by diversionof the productive electron transfer pathway through the genera-tion of reactive oxygen species (ROS). Nevertheless, the widespreadquinone oxido-reductases from the cytochrome bc family limit theamounts of released ROS to a low, perhaps just signaling, levelthrough an as-yet-unknown mechanism. Here, we propose that ametastable radical state, nonreactive with oxygen, safely holdselectrons at a local energetic minimum during the oxidation ofplastohydroquinone catalyzed by the chloroplast cytochrome b6f.This intermediate state is formed by interaction of a radical with ametal cofactor of a catalytic site. Modulation of its energy level onthe energy landscape in photosynthetic vs. respiratory enzymesprovides a possible mechanism to adjust electron transfer ratesfor efficient catalysis under different oxygen tensions.

cytochrome b6f | reactive oxygen species | semiquinone | electronparamagnetic resonance | electron transport

Photosynthetic and respiratory cytochromes bc1/b6f (Fig. 1A)generate a proton-motive force (pmf) that powers cellularmetabolism by using the Gibbs free energy difference (ΔG) be-tween hydroquinone (QH2) derivatives (Fig. 1B) and oxidizedsoluble electron transfer proteins (e.g., cytochrome c or plasto-cyanin) (1, 2). To increase the efficiency of this process, which iscritical for the yield of the generated pmf, one part of the enzymerecirculates electrons to the quinone pool in the membrane(Q pool), whereas the second part steers the electrons to thecytochrome c pool, powering the electron recirculation (Fig. 1C).This mechanism, which is best established for the cytochrome bc1(cyt bc1) (3, 4), with supporting data for the cytochrome b6f (cytb6f) (5), discussed in ref. 2, is based on bifurcation of the routefor two electrons released upon oxidation of QH2 at one of thecatalytic sites—the Qp site, (Qp), (Fig. 1D) (3–5). A model forthe energetics of this reaction assumes that one electron derivedfrom the two-electron QH2 donor is transferred, through thehigh-potential cofactor chain (“steering part” in Fig. 1C) toplastocyanin or cytochrome c, whereas the second electron istransferred across the membrane through low-potential cofactors(“recirculation” part in Fig. 1C).The electronic bifurcation process requires formation of a

short-lived and reducing redox intermediate—ubisemiquinone(USQ) or plastosemiquinone (PSQ) (4, 6, 7). However, such anintermediate in an oxygenic environment would readily reduceoxygen to form superoxide radical, (O2

−), compromising theefficiency of energy conservation (8). Even in cyt b6f where thelevel of superoxide production through this pathway is at least anorder of magnitude greater than that from yeast cyt bc1, thebranching ratio for electron transfer to O2 forming O2

− is only 1–2% of the total flux (6). The low absolute level of O2

− productionin native proteins implies the existence of a mechanism that isnot understood. In fact, contemporary models are based on at-tempts to decrease stability of semiquinone (SQ) as a means to

decrease the stationary level of SQ to avoid reactive oxygenspecies (ROS) (9–12). This destabilization of SQ, in turn, in-evitably leads to an increase in the rate constant for the reactionof SQ with oxygen, which might have deleterious consequences,especially for enzymes exposed to the relatively high local oxygenconcentrations associated with oxygenic photosynthesis (8, 13).Here, it is shown that both cyt b6f and cyt bc1 generate a

metastable radical state, nonreactive with oxygen, under steady-state turnover. This result sheds light on thermodynamic prop-erties of intermediates of electronic bifurcation at the Qp, imply-ing a mechanism that explains how cyt b6f/bc1 maintain a balancebetween energy-conserving reactions and ROS production to se-cure the enzymatic reactions at physiologically competent rates.This mechanism provides a thermodynamic basis for the signifi-cant difference in the O2

− generation in the two enzymes, andadvances our understanding of the molecular mechanism of con-trol of electron flow through the photosynthetic and respiratorychains and its contribution to ROS generation, which are postulatedto function as signaling mediators released from bioenergeticorganelles.

Materials and MethodsReagents. Equine cytochrome c, decylubiquinone (DB), decylplastoquinone(PQ), antimycin A, dibromothymoquinone (DBMIB), sodium borohydride,sodium dithionite, potassium ferricyanide (PFC), and other reagents were

Significance

Photosynthesis and respiration are crucial energy-conservingprocesses of living organisms. These processes rely on redoxreactions that often involve unstable radical intermediates. Inan oxygenic atmosphere, such intermediates present a dangerof becoming a source of electrons for generation of reactiveoxygen species. Here, we discover that cytochrome b6f, a keycomponent of oxygenic photosynthesis, generates a meta-stable state nonreactive with oxygen during enzymatic turn-over. In this state, a radical intermediate of a catalytic cycleinteracts with a metal cofactor of a catalytic site via spin-spinexchange. We propose that this state is a candidate for regu-lation of cyclic vs. noncyclic photosynthesis and also allowsphotosynthetic and respiratory cytochrome bc complexes tofunction safely in the presence of oxygen.

Author contributions: M.S., Ł.B., and A.O. designed research; M.S., Ł.B., S.B., and S.K.S.performed research; M.S., Ł.B., W.A.C., and A.O. analyzed data; and M.S., W.A.C., andA.O. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1M.S. and Ł.B. contributed equally to this work.2To whom correspondence should be addressed. Email: artur.osyczka@uj.edu.pl.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1618840114 PNAS | February 7, 2017 | vol. 114 | no. 6 | 1323–1328

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

9, 2

021

http://crossmark.crossref.org/dialog/?doi=10.1073/pnas.1618840114&domain=pdfmailto:artur.osyczka@uj.edu.plhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1618840114

purchased from Sigma-Aldrich. Dodecyl-maltoside detergent was purchasedfrom Anatrace. PQ and DB were suspended in ethanol and DMSO, re-spectively, and reduced to hydroquinone form with H2 gas released fromacidic water solution of sodium borohydride in the presence of platinum.Ethanolic stock of reduced PQ was mixed with DMSO in 1:2 (vol/vol) ratio todecrease the rate of spontaneous oxidation of plastohydroquinone (PQH2).Both substrates were kept at −80 °C until used.

Enzymes. Cyt bc1 was isolated according to the procedure described in ref. 14from wild-type Rhodobacter capsulatus grown under semiaerobic condi-tions. Thylakoid membranes were isolated from spinach as described in ref.15. Cyt b6f was isolated according to the procedure described in SI Materialsand Methods.

Electron Paramagnetic Resonance Spectroscopy. Electron paramagnetic reso-nance (EPR) spectra were measured by the continuous wave method at 20 Kon a Bruker Elexsys E580 equipped with an Oxford Instruments liquid heliumtemperature controller. The X-band spectra were measured as described inref. 16. For Q-band measurements, a ER5106QT/W resonator inserted intoCF935O cryostat was used and calibrated at 20 K by using a trityl radicalsignal. Parameters for EPR measurements are described in SI Materials andMethods. Spectra were analyzed and processed by using the Eleana com-puter program (larida.pl/eleana).

Sample Preparation for EPR Spectroscopy. Samples for measurements shownin Fig. 2A were prepared by manual injection of PQH2 to the EPR tubecontaining a mixture of cyt b6f, plastocyanin (PC), and PFC. After the addi-tion of the substrate to the reaction mixture, the solution was rapidly frozenin an ethanol bath cooled to 200 K after 2 s or 5 min of incubation. Finalconcentrations of cyt b6f, PC, PFC, and PQH2 were 140 μM, 25 μM, 680 μM,and 910 μM, respectively. Samples for measurements shown in Fig. 2B (Xband) were prepared by the freeze-quench method as described in ref. 17with the exception that glycerol was not present in the buffer. Final con-centrations of cyt bc1, cytochrome c, antimycin, and DBH2 were 20, 190, 130,

and 190 μM, respectively. Samples with cyt bc1 for Q-band measurements(Fig. 2B) were obtained similarly as for cyt b6f, as shown in Fig. 2A. Finalconcentrations of cyt bc1, antimycin, PFC, and DBH2 were 300 μM, 500 μM,2.5 mM, and 4 mM, respectively. Spectra in Fig. 3 A and B were obtained forsamples containing 90 μM cyt b6f or 130 μM cyt bc1 supplemented with2 mM DBMIB.

Preparation of Samples Under Anoxic Conditions. Samples containing cyt b6fand cyt bc1 in aerobic and anaerobic conditions were prepared similarly asthose for X-band experiments presented in Fig. 2 A and B, respectively, with

Fig. 1. Structural and functional elements of cyt b6f/bc1 catalysis. (A) Cofactors and catalytic sites (red circles) overlaid on the protein surfaces (based oncrystal structures of cyt b6f and cyt bc1 (PDB ID codes: 1VF5 and 1ZRT, respectively). (B) Chemical structures of hydroxyquinones with estimated average Emvalues of Q/QH2 couples at pH 7. Red, redox-active groups. (C) General scheme showing the “steering” and “recirculation” parts. (D) Simplified sequentialscheme of enzymatic reaction. Yellow hexagons, quinones bound to the catalytic sites. Cofactors common for cyt b6f and cyt bc1: heme bp, bn (rhombuses),and FeS (circles) in oxidized (white) or reduced (red) forms. Gray rhombus, heme cn exclusively present in cyt b6f.

Fig. 2. EPR spectra of FeS of cyt b6f (A) and cyt bc1 (B) at two resonancefrequencies (X and Q). Blue, spectra measured for samples obtained by rapidfreezing of the reaction solution 2 s after mixing with respective QH2. Black,spectra of the same samples measured after reaching the equilibrium (5 minafter mixing). Vertical dashed lines, rounded g values of major EPR transi-tions (see Fig. S1 for details).

1324 | www.pnas.org/cgi/doi/10.1073/pnas.1618840114 Sarewicz et al.

Dow

nloa

ded

by g

uest

on

June

9, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=STXThttp://larida.pl/eleanahttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=SF1www.pnas.org/cgi/doi/10.1073/pnas.1618840114

the exception that the final concentration of cyt b6f was 50 μM. To obtainanaerobic conditions, glucose oxidase (final activity 1 U/mL) and glucose(final concentration 15 mM) were added to the reaction mixture. Reagentsafter glucose addition were incubated for 30 min and mixed under atmo-sphere of sulfur hexafluoride gas.

ResultsDetection of PSQ Spin-Coupled to Reduced FeS in Noninhibited cyt b6fUnder Nonequilibrium Conditions. Fig. 2A shows that noninhibitedcyt b6f, exposed to its substrates, PQH2 and oxidized plastocya-nin, generates an intermediate of Qp detected by EPR at acharacteristic spectral line position defined by an approximate gvalue ∼1.95 (see Fig. S1 for exact g values). The g value of thistransition strongly depends on the resonance frequency. Theshift in the g value indicates that the transition must be a result ofmagnetic interactions between at least two paramagnetic centers(18, 19) and not an effect of structural changes that lead tomodifications of g values of [2Fe-2S] Rieske cluster (FeS). Fur-thermore, it closely resembles a transition (also frequency-dependent) found earlier in cyt bc1 (g ∼ 1.94 in Fig. 2B) andassigned as USQ magnetically coupled to reduced FeS via spin–spin exchange interaction (designated as USQ-FeS) (17). Byanalogy to cyt bc1, we propose that the signal in cyt b6f originatesfrom PSQ that undergoes electron spin–spin exchange interac-tion with reduced FeS (PSQ-FeS). The occupancy of SQ-FeScenter in cyt b6f and cyt bc1 (Fig. 2 A and B, blue) was estimatedas 13% and 42% of the total Qp sites, respectively (see details inSI Materials and Methods).To record SQ-FeS (this term corresponds to either USQ-FeS

or PSQ-FeS) in cyt b6f, as in cyt bc1, the steady-state enzymaticreaction must be interrupted, and the reaction mixture frozenbefore equilibrium between substrates (PQH2 and oxidizedplastocyanin) and products (PQ and reduced plastocyanin) arereached. When the enzymes used all substrates (at equilibrium),the signals were no longer present (Fig. 2 A and B, black).Nevertheless, the SQ-FeS in both enzymes, despite being farfrom the global energy minimum, is relatively long-lived incomparison with a putative unstable SQ that is commonly de-scribed (9, 10, 20, 21). A series of control experiments (Fig. S2)verified that the detected signal did not result from nonspecificand nonenzymatic reactions between substrates and/or bufferconstituents. These measurements confirmed that generation ofthe g ∼ 1.95 transition is possible only when cyt b6f catalyzes netelectron transfer from PQH2 to plastocyanin.Until now, all reports of detection of intermediates of Qp,

including USQ-FeS in cyt bc1, were obtained under conditionswhen the inhibitor antimycin blocked the recirculation pathway,i.e., blocked electron flow from the low-potential path to the Qpool (10, 17, 21). Such inhibition severely slows the turnover ofQp, because electrons entering the low-potential path must findan alternate route that restores oxidizing equivalents in Qpnecessary to support turnover. However, PSQ-FeS in cyt b6f wasdetected in the noninhibited enzyme (Fig. 2A), which shows thatthis state can be formed in the absence of inhibitors. This

observation implies that the probability of formation of SQ-FeSin noninhibited enzymes is greater in cyt b6f than in cyt bc1.

Detection of the Spin-Coupled State Between High-Potential QuinoneAnalog (DBMIB) and Reduced FeS in cyt b6f/bc1 Under Equilibrium. Asimilar paramagnetic state can be generated in cyt b6f and cyt bc1under equilibrium in the presence of the halogenated quinonederivative, DBMIB (Fig. 3 A and B, respectively), possessing arelatively high average redox mid-point potential (Em) (seecomparison of Ems of quinone/hydroquinone couples for ubi-quinone (UQ), PQ, and DBMIB in Fig. 1B). In fact, this signalwas observed in spectra of DBMIB-inhibited cyt b6f or cyt bc1(22–25). However, the nature of the g transitions in the presenceof DBMIB, as we now propose, was misinterpreted as an alter-ation of protein structure around FeS. If DBMIB changed the gtensor of FeS, the g values of the spectra for DBMIB-altered FeSshould have the same value regardless of the spectrometer fre-quency used for detection. The clear frequency dependence ofthe g values measured for samples in the presence of DBMIB(Fig. 3 A and B) indicates that it binds to reduced FeS as asemiquinone, and these two centers are subject to similar spin–spin exchange interaction as USQ-FeS (17) or PSQ-FeS coupledcenters in cyt bc1 and cyt b6f, respectively.

Redox State of Hemes b Under Conditions Favoring Generation of SQ-FeS. Fig. 4 examines the redox state of the hemes b in cyt bc1complex under three different conditions relevant to experi-ments described above. Before mixing the enzyme with ubihy-droquinone (UQH2) b-type hemes located at n and p side ofmembrane (bn and bp, respectively) are oxidized (Fig. 4A, green).However, after 2 s required to generate a relatively large USQ-FeS signal (Fig. 2B) heme bn undergoes complete reduction,whereas heme bp is still fully oxidized (Fig. 4A, blue). This resultappears unexpected because it means that semiquinone that is inspin–spin exchange interaction with reduced FeS is unable toreduce heme bp. Even less reducing power is observed whensemiquinone form of DBMIB creates the spin-spin coupled statewith FeS. In this case, both hemes remain oxidized (Fig. 4B)because they are unable to take electrons from the high-potentialsynthetic semiquinone at the Qp.

Testing the Reactivity of PSQ-FeS and USQ-FeS with MolecularOxygen. Because the experiments revealing SQ-FeS (shown inFig. 2) were performed in the presence of oxygen, we infer thatSQ-FeS cannot be highly reactive with oxygen. It follows that theyield of generated SQ-FeS should not be significantly sensitive tothe presence or absence of O2. Indeed, in cyt b6f, the amount ofPSQ-FeS generated under oxygenic and anoxygenic conditions issimilar (Fig. 5A). However, in the case of cyt bc1, the amount ofUSQ-FeS is almost always higher when the reaction is carriedout in the presence of oxygen (see example in Fig. 5B). Thereasons why USQ-FeS is more efficiently formed in the presenceof oxygen is not understood. However, one may speculate that aportion of O2

− reduces UQ that stays hydrogen-bonded to

Fig. 3. EPR spectra of FeS measured for samples of cyt b6f (A) and cyt bc1(B) after 5 min incubation of the enzymes with DBMIB. Vertical dashed lines,rounded g values of major EPR transitions.

Fig. 4. EPR gz transitions of oxidized hemes bp and bn of cyt bc1. (A) Anti-mycin-inhibited enzyme before (green) and 2 s after initiation of reaction byaddition of UQH2 (blue). (B) Enzyme incubated with DBMIB in the absence ofantimycin.

Sarewicz et al. PNAS | February 7, 2017 | vol. 114 | no. 6 | 1325

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

9, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1618840114/-/DCSupplemental/pnas.201618840SI.pdf?targetid=nameddest=SF2

reduced FeS. As a result, O2− is converted to oxygen and UQ to

USQ that strongly interacts with FeS. In other words, the portionof USQ-FeS is a product of O2

− scavenging by UQ that is boundto FeS. Such the effect is clearly visible in cyt bc1, but not in cyt b6f,as only in cyt bc1 do the conditions, antimycin-inhibited cyt bc1 vs.2% in noninhibited cyt b6f (6, 26, 27), used for detection of USQ-FeS, represent the conditions in which the enzyme produces sig-nificant amount of O2

− per single oxidized QH2 (10–18%).

DiscussionOne of the important issues associated with the understanding offundamentals of oxidative metabolism concerns elucidation ofthe mechanism by which enzymes catalyzing reactions that pro-ceed through highly unstable radical intermediates have adaptedto safely function in the presence of molecular oxygen (8). Inaddressing this issue for the cyt bc1/b6f family, a concept emergesfrom the unexpected discovery of the PSQ-FeS state in Qp ofnoninhibited cyt b6f (Fig. 2A) and subsequent comparativeanalysis of the conditions favoring appearance of this state in cytb6f and cyt bc1 under oxygenic and anoxygenic environments(Figs. 2–5). Based on the results and analysis presented in thiswork, it is concluded that the generation of SQ-FeS in cyt bc1 orcyt b6f is an inherent part of enzymatic catalysis that can bedescribed as a metastable state nonreactive with oxygen. Theconcept of stabilization of SQ in the Qp site by interaction withreduced FeS of cyt bc1 was proposed by Link (28), who consid-ered antiferromagnetic coupling of high energy between thesetwo centers as a possible explanation for the failure to detect anSQ intermediate in this site. However, such a strong interactionshould result in disappearance of the EPR signal of FeS, whichwas not observed. Furthermore, this concept was in opposition toa more popular view explaining the lack of SQ detection by ahigh instability of SQ at Qp (9, 10, 12, 20). Our results indicatethat coupling between reduced FeS and SQ exists in both cyt bc1and cyt b6f, but its energy is small enough that, regardless of theantiferromagnetic or ferromagnetic character of coupling, it pro-duces detectable EPR transitions of SQ-FeS (with a characteristicg ∼ 1.94) at temperatures higher than a few degrees Kelvin.Quite importantly, SQ-FeS in cyt bc1 is observed along with

reduced heme bn and oxidized heme bp (Fig. 4A). On thermo-dynamic grounds, this observation implies that electron transferfrom SQ-FeS to heme bp is an uphill step. As a consequence, ifheme bp is unable to transfer electrons further to heme bn, theelectron moves back to quinone (Q) at Qp and SQ-FeS is re-formed. After reaching equilibrium, no net reactions that lead tosignificant occupation of the metastable state occur and, conse-quently, this state is no longer detected spectroscopically. Toobserve it at equilibrium, one must use a high potential quinoneanalog (DBMIB in our case). However, a large increase in Emdramatically stabilizes this state to the point that it becomes in-hibitory, at least for cyt b6f. The electron from DBMIB

semiquinone, because of its relatively oxidizing potential, cannever reduce hemes bp or bn, leaving them oxidized (Fig. 4B).Considering the results presented here, a mechanism can be

proposed for inclusion of the metastable SQ-FeS into the ther-modynamic diagram of electronic bifurcation (Fig. 6). This dia-gram follows the generally accepted scheme of enzymatic cyclebut adds a new state, state 4, which is a result of an energeticallydownhill electron transfer from heme bp to Q at Qp (transitionfrom state 3). This state protects the enzyme against ROS pro-duction by: (i) the fact that electron transfer from SQ-FeS tomolecular oxygen (to state 8) is energetically unfavorable and(ii) blocking Qp for the next QH2 oxidation until electrons areremoved from the low-potential chain. The metastable state existsuntil electrons from the low-potential chain are removed throughthe Qn site (Qn) back to the Q pool through state 3 of higher ΔG,followed by subsequent downhill reactions through states 5 and 6.Hence, any factor that decreases the rate of electron release fromthe low-potential chain to the Q pool, in relation to the rate of thereduction of this chain by Qp, creates conditions that favor ap-pearance of the SQ-FeS metastable state. Such conditions mayexist in some physiological states, e.g., in the presence of a hightransmembrane potential or a high concentration of QH2 in the Qpool. It is intriguing that a similar g = 1.94 transition of unknownorigin, which might reflect the USQ-FeS state, was reported toappear under ischemia and disappear under reperfusion of rathearts (29).The free energy (ΔG) diagram, shown in Fig. 6, which depends

not only on Em but also on other processes, e.g., reconfigurationof H bonds within Qp, provides a possible explanation for sig-nificant occupation of the energetic state representing PSQ-FeSin noninhibited cyt b6f. Although the difference in Em betweenhemes bn and bp in cyt b6f is not well defined (shown as the widthin the level of state 5 in Fig. 6A), the average is somewhat morenegative than in hemes b in cyt bc1 (30, 31). This difference,together with the fact that PQ possesses a more positive Em thanUQ (13) makes the energetic gap between state 4 (with SQ-FeS)and state 5 (with reduced heme bn) smaller in cyt b6f comparedwith cyt bc1. In other words, in cyt bc1, the electron transfer fromUSQ-FeS (state 4) to heme bn (state 5) is energetically muchmore favorable than the electron transfer from PSQ-FeS toheme bn in cyt b6f. In addition, the existence of high-spin heme cnis possibly a bottleneck for electron flow from the low-potentialchain to the Q pool, or reduction of PQ in Qn requires a co-operative two-electron transfer from hemes bn/cn (32, 33).With the proposed mechanism, it can be appreciated that

energetic states associated with oxidation of QH2 by cyt b6f/cyt bc1are positioned at levels that allow smooth catalysis while limitingreleased ROS to perhaps just signaling levels that carefully reportthe dynamically changing redox state of cofactors (6, 34). It isproposed that the SQ-FeS metastable state serves as a “buffer”for electrons that are unable to be relegated from Qp through thelow-potential chain. Availability of the buffer allows the enzymeto be held in the state that protects against energy-wasting reac-tions including the short circuits (two electrons from QH2 goingto the same cofactor chain) and the leaks of electrons to mo-lecular oxygen (1, 7, 8). At the molecular level, the stabilization ofSQ by its spin-coupling to reduced FeS is likely to occur throughthe creation of an H bond between SQ and histidine ligatingreduced FeS (transition from 2 to 4 in Fig. 6) (35).The stabilization of SQ-FeS can occasionally be broken (depic-

ted in the Fig. 6 as the transition from state 4 to 2), resulting in theformation of a highly unstable semiquinone (SQ not spin-coupledto FeS), which is able to reduce molecular oxygen. This semi-quinone remains the most likely state responsible for limited su-peroxide release. In the reversible transition between SQ-FeS andSQ that is not spin-coupled to FeS (states 4 and 2, respectively),stationary levels of these two states will be proportional. It followsthat the amount of ROS will correlate with the amount of the

Fig. 5. Comparison of the amplitudes of g ∼ 1.95/1.94 transition, in oxy-genic and anoxygenic environment for cyt b6f (A) and cyt bc1 (B), obtainedunder conditions similar to those of Fig. 2.

1326 | www.pnas.org/cgi/doi/10.1073/pnas.1618840114 Sarewicz et al.

Dow

nloa

ded

by g

uest

on

June

9, 2

021

www.pnas.org/cgi/doi/10.1073/pnas.1618840114

detected SQ-FeS, despite its nonreactivity with oxygen. Indeed,noninhibited cyt b6f produces larger amounts of superoxide thannoninhibited cyt bc1 (6), which stays in line with our observationthat in the case of noninhibited enzymes SQ-FeS can only bedetected in cyt b6f (Fig. 2A). Nevertheless, when semiquinone ispresent in Qp, the equilibrium is always shifted toward its sta-bilization by formation of the SQ-FeS spin-coupled centerwhereby ROS release is limited to the level of ∼2 molecules ofO2

− per 100 electrons transferred to FeS. Such residual ROSgeneration detected in cyanobacterial cyt b6f has been proposedto activate the p side of the chloroplast transmembrane Stt7 (36)and to carry out longer range signaling in the plant cell (37).The probability of creation of the SQ-FeS metastable state

may vary in different species depending on the relative Em valuesof quinones and low-potential cofactors. Perhaps it is adjusted tothe oxygen tension in the cellular environment (8). Indeed, cytb6f, which experiences more than an order of magnitude higherlevel of oxygen in chloroplasts than cyt bc1 in mitochondria, has agreater tendency to reside in this buffered state. Also, a possibleconsequence of the existence of PSQ-FeS in noninhibited cyt b6fis that it may be responsible for regulation of the electron transferpathways of oxygenic photosynthesis. As proposed, cyt b6f innative chloroplasts can catalyze PQH2 oxidation according to twoalternative mechanisms: (i) noncyclic, in which one molecule ofPQ in Qn undergoes a sequential reduction by two electronsderived from Qp and (ii) cyclic, in which one electron is deliveredto Qn from Qp, whereas the second electron comes from reducedferredoxin (2). We speculate that creation of the metastable statePSQ-FeS may serve as a factor that changes the efficiency ofcyclic vs. noncyclic electron transfer between photosystem I andII. This inference is explained by the fact that transient

stabilization of the PSQ-FeS blocks the oxidation of anotherPQH2 in Qp, and, thus, creates a condition in which the proba-bility of delivering the second electron needed to complete thereduction of PQ in Qn by ferredoxin is significantly increased.The phenomenon of spin–spin exchange interactions between

semiquinones and metal centers has been observed many timesin different biological systems, including a coupling between tightlybound semiquinone QA and Fe

2+ in photosynthetic reaction centers(38, 39), between flavin semiquinones and metal cofactors (18, 40),and between iron-sulfur cluster N2 and ubisemiquinone in mito-chondrial complex I. However, the role of metal cofactors in thevicinity of radicals is not always clear and remains a subject of de-bate, as exemplified by recent discussion on possible origin on theunusual properties of the SQ signals in complex I (41). In light of thepresent study, we envisage that FeS in cyt bc1/b6f has a dual role inelectron transfer. Its obvious role is to accept electrons from sub-strate but besides this function, it offers a mean of stabilization ofpotentially dangerous intermediates that are inherently associatedwith the stepwise quinone redox reactions. This metal center behavesas a Lewis acid with electrophilic properties. When it creates a bondwith SQ, its electron-withdrawing properties decrease the probabilityof reaction of SQ with oxygen. We propose that this feature is notrestricted to the iron-sulfur cluster of cyt bc1/b6f, but may be commonfor other metal cofactors that, besides having a role in electrontransfer, stabilize potentially reactive radicals.

ACKNOWLEDGMENTS. This work was supported by The Wellcome TrustGrant 095078/Z/10/Z (to A.O.). Faculty of Biochemistry, Biophysics andBiotechnology of Jagiellonian University is a partner of the Leading NationalResearch Center supported by Ministry of Science and Higher Education, partof which included Grant 35p/1/2015 (to M.S.) and a scholarship (to Ł.B.).Studies of W.A.C., S.B., and S.K.S. were supported by NIH Grant GMS-038323.

Fig. 6. Simplified diagram of relative energy levels at different stages of reaction catalyzed by cyt b6f (A) and cyt bc1 (B). Black dots represent electrons onthe respective cofactor or quinone molecule. Yellow squares, Qp empty or occupied by substrate; purple squares, Qp with bound DBMIB semiquinone. Greenand red arrows, downhill and uphill transitions between the states, respectively. Red frame indicates the states that are inaccessible in antimycin-inhibited cytbc1. State 1 represents the most populated initial state for QH2 oxidation under steady-state turnover (in antimycin-inhibited cyt bc1 heme bn is alreadyreduced after the first turnover that takes place within experimental dead time). This reaction results in reduction of FeS and heme bp (transition from 1 to 3involving unstable SQ in 2). From 3 downhill reactions to 4 or 5 are possible and 5 undergoes further downhill transition to 6 (oxidation of heme bn) allowingnext turnover. State 4 is metastable state containing SQ spin-coupled to reduced FeS. Population of 4 increases when transition to 5 is blocked (antimycin) ortransition from 5 to 6 is slow with respect to the transition from 1 to 3. The stationary level of 4 depends on energetic gap between 4 and 5 (gray doublearrow), which differs within bc family. State 4 lays below energetic level of 8 in which O2

− is generated. State 7 involves stable DBMIB semiquinone spin-coupled to FeS. The gray square with a gradient depicts uncertainty in Em values for hemes b in cyt b6f.

Sarewicz et al. PNAS | February 7, 2017 | vol. 114 | no. 6 | 1327

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

9, 2

021

1. Sarewicz M, Osyczka A (2015) Electronic connection between the quinone and cyto-chrome c redox pools and its role in regulation of mitochondrial electron transportand redox signaling. Physiol Rev 95(1):219–243.

2. Cramer WA, Hasan SS (2016) Structure-function of the cytochrome b6f lipoproteincomplex. Cytochrome Complexes: Evolution, Structures, Energy Transduction, andSignaling, eds Cramer WA, Kallas T (Springer, Dordrecht), pp 177–207.

3. Mitchell P (1975) The protonmotive Q cycle: A general formulation. FEBS Lett 59(2):137–139.

4. Crofts AR, Meinhardt SW, Jones KR, Snozzi M (1983) The role of the quinone pool inthe cyclic electron-transfer chain of Rhodopseudomonas sphaeroides: A modifiedQ-cycle mechanism. Biochim Biophys Acta 723(2):202–218.

5. Joliot P, Joliot A (1988) The low-potential electron-transfer chain in the cytochrome bfcomplex. Biochim Biophys Acta 933(2):319–333.

6. Baniulis D, Hasan SS, Stofleth JT, Cramer WA (2013) Mechanism of enhanced super-oxide production in the cytochrome b6f complex of oxygenic photosynthesis.Biochemistry 52(50):8975–8983.

7. Osyczka A, Moser CC, Dutton PL (2005) Fixing the Q cycle. Trends Biochem Sci 30(4):176–182.

8. Rutherford AW, Osyczka A, Rappaport F (2012) Back-reactions, short-circuits, leaksand other energy wasteful reactions in biological electron transfer: Redox tuning tosurvive life in O2. FEBS Lett 586(5):603–616.

9. Cape JL, Bowman MK, Kramer DM (2006) Understanding the cytochrome bc com-plexes by what they don’t do. The Q-cycle at 30. Trends Plant Sci 11(1):46–55.

10. Cape JL, Bowman MK, Kramer DM (2007) A semiquinone intermediate generated atthe Qo site of the cytochrome bc1 complex: Importance for the Q-cycle and superoxideproduction. Proc Natl Acad Sci USA 104(19):7887–7892.

11. Quinlan CL, Gerencser AA, Treberg JR, Brand MD (2011) The mechanism of superoxideproduction by the antimycin-inhibited mitochondrial Q-cycle. J Biol Chem 286(36):31361–31372.

12. Bergdoll L, ten Brink F, Nitschke W, Picot D, Baymann F (2016) From low- to high-potential bioenergetic chains: Thermodynamic constraints of Q-cycle function.Biochim Biophys Acta 1857(9):1569–1579.

13. Song Y, Buettner GR (2010) Thermodynamic and kinetic considerations for the re-action of semiquinone radicals to form superoxide and hydrogen peroxide. Free RadicBiol Med 49(6):919–962.

14. Valkova-Valchanova MB, Saribas AS, Gibney BR, Dutton PL, Daldal F (1998) Isolationand characterization of a two-subunit cytochrome b-c1 subcomplex from Rhodo-bacter capsulatus and reconstitution of its ubihydroquinone oxidation (Qo) site withpurified Fe-S protein subunit. Biochemistry 37(46):16242–16251.

15. Hurt E, Hauska G (1981) A cytochrome f/b6 complex of five polypeptides withplastoquinol-plastocyanin-oxidoreductase activity from spinach chloroplasts. Eur JBiochem 117(3):591–599.

16. Sarewicz M, Dutka M, Pietras R, Borek A, Osyczka A (2015) Effect of H bond removaland changes in the position of the iron-sulphur head domain on the spin-lattice re-laxation properties of the [2Fe-2S]2+ Rieske cluster in cytochrome bc1. Phys ChemChem Phys 17(38):25297–25308.

17. Sarewicz M, Dutka M, Pintscher S, Osyczka A (2013) Triplet state of the semiquinone-Rieske cluster as an intermediate of electronic bifurcation catalyzed by cytochromebc1. Biochemistry 52(37):6388–6395.

18. Fournel A, et al. (1998) Magnetic interactions between a [4Fe–4S]1+ cluster and aflavin mononucleotide radical in the enzyme trimethylamine dehydrogenase: A high-field electron paramagnetic resonance study. J Chem Phys 109(24):10905–10913.

19. Calvo R (2007) EPR measurements of weak exchange interactions coupling unpairedspins in model compounds. Appl Magn Reson 31(1–2):271–299.

20. Zhu J, Egawa T, Yeh S-R, Yu L, Yu C-A (2007) Simultaneous reduction of iron-sulfurprotein and cytochrome bL during ubiquinol oxidation in cytochrome bc1 complex.Proc Natl Acad Sci USA 104(12):4864–4869.

21. Zhang H, Osyczka A, Dutton PL, Moser CC (2007) Exposing the complex III Qo semi-quinone radical. Biochim Biophys Acta 1767(7):883–887.

22. Malkin R (1981) Redox properties of the DBMIB-Rieske iron-sulfur complex in spinachchloroplast membranes. FEBS Lett 131(1):169–172.

23. Schoepp B, Brugna M, Riedel A, Nitschke W, Kramer DM (1999) The Qo-site inhibitorDBMIB favours the proximal position of the chloroplast Rieske protein and induces apK-shift of the redox-linked proton. FEBS Lett 450(3):245–250.

24. Roberts AG, Bowman MK, Kramer DM (2002) Certain metal ions are inhibitors ofcytochrome b6f complex ‘Rieske’ iron-sulfur protein domain movements. Biochemistry41(12):4070–4079.

25. Roberts AG, Bowman MK, Kramer DM (2004) The inhibitor DBMIB provides insightinto the functional architecture of the Qo site in the cytochrome b6f complex.Biochemistry 43(24):7707–7716.

26. Borek A, Sarewicz M, Osyczka A (2008) Movement of the iron-sulfur head domain ofcytochrome bc1 transiently opens the catalytic Qo site for reaction with oxygen.Biochemistry 47(47):12365–12370.

27. Sarewicz M, Borek A, Cieluch E, �Swierczek M, Osyczka A (2010) Discrimination be-tween two possible reaction sequences that create potential risk of generation ofdeleterious radicals by cytochrome bc1. Implications for the mechanism of superoxideproduction. Biochim Biophys Acta 1797(11):1820–1827.

28. Link TA (1997) The role of the ‘Rieske’ iron sulfur protein in the hydroquinone oxi-dation (Qp) site of the cytochrome bc1 complex. The ‘proton-gated affinity change’mechanism. FEBS Lett 412(2):257–264.

29. Baker JE, Felix CC, Olinger GN, Kalyanaraman B (1988) Myocardial ischemia and re-perfusion: Direct evidence for free radical generation by electron spin resonancespectroscopy. Proc Natl Acad Sci USA 85(8):2786–2789.

30. Cramer WA, Hasan SS, Yamashita E (2011) The Q cycle of cytochrome bc complexes: Astructure perspective. Biochim Biophys Acta 1807(7):788–802.

31. Zhang H, et al. (2008) Quinone and non-quinone redox couples in Complex III.J Bioenerg Biomembr 40(5):493–499.

32. Zhang H, et al. (2004) Characterization of the high-spin heme x in the cytochrome b6fcomplex of oxygenic photosynthesis. Biochemistry 43(51):16329–16336.

33. Baymann F, Giusti F, Picot D, Nitschke W (2007) The ci/bH moiety in the b6f complexstudied by EPR: A pair of strongly interacting hemes. Proc Natl Acad Sci USA 104(2):519–524.

34. Bleier L, Dröse S (2013) Superoxide generation by complex III: From mechanistic ra-tionales to functional consequences. Biochim Biophys Acta 1827(11-12):1320–1331.

35. Pietras R, Sarewicz M, Osyczka A (2016) Distinct properties of semiquinone speciesdetected at the ubiquinol oxidation Qo site of cytochrome bc1 and their mechanisticimplications. J R Soc Interface 13(118):20160133.

36. Singh SK, et al. (2016) Trans-membrane signaling in photosynthetic state transitions:Redox- and structure-dependent interaction in vitro between Stt7 kinase and thecytochrome b6f complex. J Biol Chem 291(41):21740–21750.

37. Baniulis D, Hasan SS, Miliute I, Cramer WA (2016) Mechanisms of superoxide gener-ation and signaling in cytochrome bc complexes. Cytochrome Complexes: Evolution,Structures, Energy Transduction, and Signaling, eds Cramer W, Kallas T (Springer,Dordrecht), pp 397–417.

38. Calvo R, Passeggi MC, Isaacson RA, Okamura MY, Feher G (1990) Electron para-magnetic resonance investigation of photosynthetic reaction centers from Rhodo-bacter sphaeroides R-26 in which Fe2+ was replaced by Cu2+. Determination ofhyperfine interactions and exchange and dipole-dipole interactions between Cu2+

and QA-. Biophys J 58(1):149–165.

39. Sedoud A, et al. (2011) Effects of formate binding on the quinone-iron electron ac-ceptor complex of photosystem II. Biochim Biophys Acta 1807(2):216–226.

40. Ohnishi T (1998) Iron-sulfur clusters/semiquinones in complex I. Biochim Biophys Acta1364(2):186–206.

41. Hirst J, Roessler MM (2016) Energy conversion, redox catalysis and generation of re-active oxygen species by respiratory complex I. Biochim Biophys Acta 1857(7):872–883.

42. Liu J, et al. (2014) Metalloproteins containing cytochrome, iron-sulfur, or copper re-dox centers. Chem Rev 114(8):4366–4469.

1328 | www.pnas.org/cgi/doi/10.1073/pnas.1618840114 Sarewicz et al.

Dow

nloa

ded

by g

uest

on

June

9, 2

021

www.pnas.org/cgi/doi/10.1073/pnas.1618840114