Quinone binding sites of cyt bc complexes analysed by X-ray crystallography and cryogenic electron microscopy

Cytochrome (cyt) bc1, bcc and b6f complexes, collectively referred to as cyt bc complexes, are homologous isoprenoid quinol oxidising enzymes present in diverse phylogenetic lineages. Cyt bc1 and bcc complexes are constituents of the electron transport chain (ETC) of cellular respiration, and cyt b6f complex is a component of the photosynthetic ETC. Cyt bc complexes share in general the same Mitchellian Q cycle mechanism, with which they accomplish proton translocation and thus contribute to the generation of proton motive force which drives ATP synthesis. They therefore require a quinol oxidation (Qo) and a quinone reduction (Qi) site. Yet, cyt bc complexes evolved to adapt to specific electrochemical properties of different quinone species and exhibit structural diversity. This review summarises structural information on native quinones and quinone-like inhibitors bound in cyt bc complexes resolved by X-ray crystallography and cryo-EM structures. Although the Qi site architecture of cyt bc1 complex and cyt bcc complex differs considerably, quinone molecules were resolved at the respective Qi sites in very similar distance to haem bH. In contrast, more diverse positions of native quinone molecules were resolved at Qo sites, suggesting multiple quinone binding positions or captured snapshots of trajectories toward the catalytic site. A wide spectrum of inhibitors resolved at Qo or Qi site covers fungicides, antimalarial and antituberculosis medications and drug candidates. The impact of these structures for characterising the Q cycle mechanism, as well as their relevance for the development of medications and agrochemicals are discussed.


Introduction
Isoprenoid quinones are a family of natural electron and proton carriers present in prokaryotic cellular membranes, in the mitochondrial inner membrane and in the chloroplast thylakoid membrane [1][2][3]. The various isoprenoid quinone species differ in their water-soluble ring system and the length of the hydrophobic isoprenoid tails [4][5][6] (Figure 1A-E). The electrochemically active part of this family of molecules is the quinone ring system, which accepts two electrons and two protons to become the fully reduced quinol ( Figure 1A), while the highly hydrophobic isoprenoid tail enhances its solubility in biological membranes. Isoprenoid quinone and quinol are substrates of respiratory chain and photosynthetic enzymes [7,8].
In the electron transport chain (ETC) of cellular respiration, NADH dehydrogenase (complex I) and succinate dehydrogenase (complex II) reduce quinone, harnessing the energy of redox equivalents obtained from metabolism while cytochrome bc 1 complex (cyt bc 1 complex, complex III) oxidises quinol and transfer the electrons to cytochrome c oxidase (cyt c oxidase, complex IV) via the electron carrier protein cytochrome (cyt) c. The cyt c oxidase catalyses the reduction in dioxygen to water. NADH dehydrogenase and cyt bc 1 complex couple quinone redox chemistry to proton translocation across the inner mitochondrial or bacterial cellular membrane to generate an electrochemical proton gradient and thereby power ATP synthesis [1,3]. In the ETC of photosynthesis, photosystem II utilises light energy to reduce quinone, and cyt b 6 f complex [9][10][11], a homologue of cyt bc 1 complex, oxidises quinol and passes electrons to photosystem I. Photosystem II and cyt b 6 fcomplex create a proton gradient across the chloroplast thylakoid membrane or the cyanobacterial plasma cellular membrane for ATP synthesis [12]. Therefore, cyt bc 1 and cyt b 6 f complex are substantial contributors to the driving forces of cellular energy conversion.
Cyt bc 1 and cyt b 6 fcomplexes form a large group of enzymes which all include a Rieske iron-sulfur protein (ISP), a b-type cytochrome (cyt b or cyt b 6 -SUIV, 'subunit four') and a c-type cytochrome (cyt c 1 , cyt f or di-haem cyt cc) as the core catalytic module ( Figure 1F,G) [2,13,14]. Cyt bc 1 and cyt b 6 f complexes are found in organisms from diverse phylogenetic clades [13], and they differ in composition in respect to number and types of peripheral subunits [10,[15][16][17]. In actinobacteria, the catalytic Rieske ISP, cyt b, cyt cc and the cyt aa 3 oxidase plus peripheral subunits comprise the cyt bcc-aa 3 supercomplex [18,19]. Therefore, they are collectively referred to as cyt bc complexes in this mini-review. which was co-crystallised with cyt bc 1 complex ( pdb 2ibz). The location of the quinone reduction Q i site is indicated by ubiquinone-6 (UQ-6) which was co-isolated with the enzyme (pdb 2ibz). Soluble cyt c is a substrate of cyt bc 1 complex. Its docking position is illustrated based on the X-ray structure of the electron-transfer complex ( pdb 3cx5). The three membrane-bound catalytic subunits of one protomer of the dimeric enzyme, namely cyt b, Rieske iron-sulfur protein (ISP) and cyt c 1 as well as the substrate cyt c, are separately illustrated in (G) The extrinsic domain of Rieske ISP undergoes diffusional movement and its position close to cyt b (b-position, pdb 2ibz) and close to cyt c 1 (c-position, pdb 1be3) are both indicated. The iron-sulfur cluster (FeS) is depicted in gray scale at the c-position and the extrinsic domain at the c-position is only outlined. P and N indicate the electropositive and -negative sides of the inner mitochondrial membrane. Iron atoms are depicted in brown, sulfur atoms are shown in yellow.
In respiratory and photosynthetic ETCs, the overall forward reaction of cyt bc complexes is to oxidise quinol molecules and to reduce cytochrome c or plastocyanin, which will further transfer the electron to cyt c oxidase or photosystem I, respectively. Cyt bc complexes do not directly pump protons across the membrane such as for instance cyt c oxidases, instead, proton translocation is achieved through the Mitchellian Q cycle mechanism ( Figure 2) [2,11,[20][21][22][23][24]. As the first step in a Q cycle, a quinol molecule is oxidised at the quinol oxidation (Q o ) site of cyt bc complex close to the positive side (P-side) of the membrane ( Figure 1F). Next, using the mitochondrial cyt bc 1 complex as an example, one electron of ubiquinol is transferred to the Rieske iron-sulfur cluster (FeS) and subsequently to haem c 1 . The extrinsic domain of Rieske ISP undergoes a substantial conformational change [16,[25][26][27] to bridge the 24 Å distance between the Q o site quinol and haem c 1 (Figure 2). Catalytic centres of dimeric cyt bc 1 complex from S. cerevisiae are illustrated in two ways. The structure of the right half shows the inhibitor stigmatellin and the natural substrate UQ-6 in the X-ray structure ( pdb 2ibz) as well as their hydrogen bonding partners. His181 RIP1 is also a ligand of the iron-sulfur cluster (FeS). Stigmatellin represents the position of a transition state of ubiquinol (QH 2 ) oxidation in the Q o site, and UQ-6 indicates the position of ubiquinone/semiubiquinone (Q/SQ • ) in the Q i site.
The structure of the left schematically shows the Q cycle mechanism. The four highly conserved residues of cyt b (COB): Pro271 COB (P), Glu272 COB (E), Trp273 COB (W) and Tyr274 COB (Y) in the Q o site form the Q o motif [14]. Electron transfer in cyt bc 1 complex can also cross the dimeric enzyme ( pink arrow) [111]. Owing to the large-scale movement of the extrinsic domain (ED) of Rieske iron-sulfur protein (RIP1), the FeS is shown at two positions, the b-position (based on pdb 2ibz) close to the Q o site quinol and the c-position close to haem c 1 (based on pdb 1be3). Whereas the Q o site ubiquinol releases two electrons and two protons upon oxidation, only one electron is transferred to Q i site, therefore the full reduction in the Q i site quinone requires oxidation of a second ubiquinol molecule at the Q o site and the uptake of two protons. The exact sequence of protonation steps at the Q i site is not differentiated in this simplified scheme. Iron atoms are depicted in brown, sulfur atoms are shown in yellow. The FeS and its ligand His181 RIP1 at the c-position are shown in gray scale. Hydrogen bonds are depicted as dashed lines. Electron transfer pathways are shown in red, and proton release and uptake routes are in green.
Physiological electron transfer rates typically require a maximal distance of 14 Å between electron donor and acceptor [28,29]. The other electron is routed through the low potential haem b L , the high potential haem b H and reduces a quinone molecule in the Q i site to a semiquinone radical (SQ • ). In this process, the Q o site quinol releases two protons to the P-side of the membrane and the complete reduction and protonation of a quinone molecule in the Q i site needs oxidation of a second quinol at the Q o site and proton uptake from the N-side of the membrane. Consequently, bifurcated electron transfer must be achieved upon quinol oxidation to enable the Q cycle, i.e. the highly reactive SQ • at the Q o site must be controlled to avoid short circuits [2,[29][30][31][32][33][34] which lead to futile bypass reactions which would lower the efficiency of cellular respiration and can generate reactive oxygen species [29] that can cause oxidative damage to the cell [35].
Experimental structures of cyt bc complexes are essential to understand the molecular basis for efficient and safe electron and proton transfer mechanisms at Q o and Q i site. Position, geometry and distance of electron donors and acceptors, of substrate and analogous molecules as well as of prosthetic groups, are important to define electron transfer pathways [28]. Resolved positions of protonable amino acid side chains, hydronium ions (H 3 O + ) or water molecules enable to identify proton transfer pathways [36]. Owing to the central role of cyt bc complexes in cellular respiration and in photosynthesis, structural biology studies of these complexes based on X-ray crystallography and cryogenic electron microscopy (cryo-EM) have delivered, over the years, a great number of experimental structures of mitochondrial cyt bc 1 complexes [16,17,[37][38][39][40] as well as of respiratory supercomplexes [41][42][43][44][45][46][47][48], alpha-proteobacterial cyt bc 1 complexes [27,49,50], cyanobacterial [9] and chloroplast [10,51] cyt b 6 f complexes and actinobacterial cyt bcc-aa 3 supercomplexes [19,[52][53][54]. One should note that, the electrochemical properties of the redox-active centres of cyt bc complexes co-evolved with those of their native quinone substrates [13,[55][56][57]. Hence, comparison of structures of cyt bc complexes with bound substrates sampled from a wide spectrum of organisms sheds light on the conserved structural basis of the Q cycle's quinone catalysis as well as on adaptations reflecting its molecular evolution, and may support development of medications precisely targeting different pathogens.

Quinone binding positions at the Q o site
In cyt bc 1 and b 6 f complexes, the Q o site is embedded in subunit cyt b and at the interface with the mobile extrinsic domain of Rieske ISP (Figs. 1F, 2). The native substrate at the Q o site is quinol, the reduced form of quinone, and the oxidised reaction product quinone has to leave the catalytic position at the Q o site. So far, native quinone or quinol molecules were not resolved at the catalytic Q o site position in X-ray crystallography studies (Table 1), in particular because crystal formation requires a defined conformation of the complex, and the unrestrained motion of the extrinsic domain of Rieske ISP may hinder this process. Consequently, the characterisation of the binding mode of the substrate in the Q o site was supported by the use of inhibitors, and three binding positions at the Q o site were suggested [58]. The proximal position ( Figure 2) was assigned with myxothiazol, which is hydrogen-bonded solely to Glu272 of cyt b (Glu272 COB , yeast numbering) and shows no interaction to Rieske ISP [37]. The distal binding position ( Figure 2) is exemplified by HHDBT, which is hydrogen-bonded to the iron-sulfur-cluster (FeS) ligand (His181 RIP1 ) of the Rieske protein, and to Glu272 COB with a water-mediated hydrogen bond [59]. The third binding position is characterised with stigmatellin, which is hydrogen bonded directly to both Glu272 COB and His181 RIP1 [17] (Figure 2). Stigmatellin also binds at the Q o site of the cyt bcc-aa 3 supercomplex of Corynebacterium glutamicum in a similar manner as in the mitochondrial cyt bc 1 complexes [19], therefore it exhibits a conserved binding pose in the Q o sites of cyt bc complexes which oxidise respectively ubiquinone or menaquinone. The Q o site pocket is unlikely to accommodate two isoprenoid quinol molecules simultaneously due to spatial constraints, thus these aforementioned three inhibitor binding positions may reflect the locations of reaction intermediates in different oxidation or protonation states, as well as their interactions with potential proton acceptors [22,60,61]. One of the proton acceptors is His181 RIP1 , which undergoes a pK a change dependent on the Rieske protein redox state [62,63]. The other hypothetical proton acceptor is Glu272 COB . Its substitution with other residues by mutagenesis partially compromises the turnover of the enzyme [14] but its exact function remains elusive. Glu272 COB is the second residue of the Q o motif of cyt b, a highly conserved motif of four consecutive amino acid residues ( Figure 2) present in all cyt bc complexes with systematic phylogenetic variations (PEWY in mitochondrial cyt b) [14]. The type of residue at the second position of the Q o motif is correlated with the redox midpoint potential of cyt bc complex cofactors as well as with the quinone species [14]. Substrate binding positions in experimental structures of cyt bc complexes from different organisms would be very important to derive the conserved structural basis of catalysis as well as species-specific adaptations.
Recently, native co-isolated quinone molecules at or in proximity to the Q o site were identified in several cryo-EM structures of respiratory chain supercomplexes (Table 1). In a mammalian respiratory I/III 2 supercomplex [48], which is composed of a NADH dehydrogenase (complex I) and a dimeric cyt bc 1 complex (complex III 2 ), an ubiquinone molecule was identified in the Q o site which is distal to complex I, whereas the Q o site proximal to the quinone reduction tunnel of complex I was unoccupied ( Figure 3A). The authors proposed that the Q o site close to complex I would accept ubiquinol reduced by complex I as they share the shortest diffusion distance [48]. The cryo-EM structure of cyt bc 1 complex from Candida albicans contains a ubiquinone molecule in the Q o site of both protomers [40] ( Figure 3B). By superimposition of the mammalian supercomplex I/III 2 with Candida albicans complex III, and yeast cyt bc 1 complex co-crystallised with stigmatellin, a trajectory of Q o site occupants can be deduced ( Figure 3C). In comparison, stigmatellin reached deepest into the Q o site pocket. The ubiquinone molecules resolved in the cryo-EM structures only partially overlap with the stigmatellin binding position. Concomitantly, the FeS cluster of the cryo-EM structures are further apart from the Q o site. The FeS of the yeast cyt bc 1 complex is located at the closest distance to the Q o site, as it is constrained by a hydrogen bond from its own ligand His181 RIP1 to stigmatellin ( Figure 2). In contrast, the FeS clusters of the mammalian supercomplex I/III 2 and the Candida albicans complex III are more distant from the Q o site. The distances between ubiquinone and the FeS histidine ligand in these two complexes are larger than 4.5 Å, which is too long for a hydrogen bond. These two positions in the cryo-EM structures likely represent the states of ubiquinone, the product of ubiquinol-oxidation, exiting the catalytic Q o site position.
In prokaryotes, a co-isolated menaquinone at the Q o site was resolved in two cryo-EM structure of bcc-aa 3 supercomplex from the actinobacterium Corynebacterium glutamicum [19,54] (Figure 4A). This menaquinone molecule is positioned in ∼6 Å distance to the closest possible H-bonding partners His355 QcrA and Tyr153 QcrB , respectively (QcrA and QcrB are homologous to mitochondrial Rieske ISP and cyt b), and is 9.4 and 13.7 Å apart from FeS and haem b L , respectively [19]. In a cryo-EM structure of the actinobacterial cyt bcc-aa 3 supercomplex from Mycobacterium smegmatis, a menaquinone molecule was described in 14 Å and 16 Å distance from FeS and haem b L , respectively ( Figure 4B) [52]. This binding position agrees with a menaquinone molecule resolved in another M. smegmatis cryo-EM structure [64], as well as a menaquinone molecule identified   Figure 4C). By superimposition of the structures of the corynebacterial supercomplex with stigmatellin [19], with menaquinone [19,54], and the mycobacterial supercomplex structures with menaquinone [52,64,65], genus-specific consensus menaquinone binding positions can be deduced ( Figure 4D). The locations of FeS in these structures are static. The menaquinone molecules in the two structures of the corynebacterial supercomplex both partially overlap with the stigmatellin binding position, whereas the menaquinone molecules of the three structures of the mycobacterial supercomplex were consistently located closer to the entrance of the quinone exchange cavity ( Figure 4D). These experimentally resolved menaquinone molecules likely illustrate a migration path to the catalytic position of menaquinol, which is represented by the transition state analogue stigmatellin [66,67]. Interestingly, the Q o site menaquinone position assigned in a M. smegmatis supercomplex ( pdb 6hwh, Figure 3D) [53] does not agree with the Q o site menaquinone positions shown in other four actinobacterial supercomplex structures and its naphthoquinone ring was resolved in 21 Å and 19 Å to FeS and the haem b L iron [53], therefore this model is not included in Figure 4D. In addition to ubiquinone and menaquinone at the Q o site, a plastoquinone was described in the cryo-EM structure of cyt b 6 f complex from spinach chloroplasts [51], with its benzoquinone ring 26.4 Å apart from FeS and 16.2 Å from haem b L . It was described as in an approaching position to the Q o site ( Figure 5A). Moreover, the entrance of the Q o site in this structure is partially blocked by the phytyl tail of chlorophyll (Chl), which was suggested to gate the Q o site access [51].
Although quinone molecules were resolved in the Q o site of cyt bc complexes in several positions, structural information of the natural substrate in the catalytic relevant position in the Q o site with close distance to electron and proton acceptors is still lacking. So far, only the structures with inhibitors bound at the Q o site suggest the potential proton acceptors for quinol oxidation. Taken together, the cryo-EM structure of the ovine supercomplex I/III 2 provided a first hint of a co-isolated quinone in the Q o pocket in the context of substrate exchange between complexes I and III. The diverse binding positions of native co-purified ubiquinone, menaquinone and plastoquinone molecules resolved in structures of cyt bc complexes, most likely exemplify snapshots of their migration paths in and out of the active site and stand-by positions.

Quinone binding positions at the Q i site
In contrast with the Q o site characterisation, many X-ray and cryo-EM structures of cyt bc complexes described co-purified quinone molecules in the Q i site. A plausible explanation is that the Q i site substrate has to be stabilised within the cyt b pocket to ensure a full Q cycle turnover with the two-step reduction to semiquinone and quinol, which is strictly coupled to the oxidation of two quinol molecules in the Q o site (Figure 2). Binding poses of Q i site ubiquinone including ordered water molecules were obtained with high resolution X-ray structures of bovine [38], chicken [39,68], yeast cyt bc 1 complexes ( Figure 5D) [17] and that from Rhodobacter sphaeroides ( Figure 5B) [69]. In brief, the Q i site ubiquinone is consistently located within ca. 5 Å distance to the porphyrin ring of haem b H ( Figure 3D) in the different structures. Two proposed proton transfer pathways were assigned from the protein surface on the mitochondrial matrix side (the electro-negative side) to Asp229 COB and His202 COB (yeast numbering, Figs. 2, 3D). Each residue is connected via hydrogen bonds to a carbonyl group of the Q i site ubiquinone. The exact hydrogen bond pattern, whether it is a direct interaction or mediated by water molecules, varies in X-ray structures of the complex from different species [21]. That the binding of the Q i site inhibitor antimycin A replaced the natively occupied ubiquinone with Asp229 COB as its direct interaction partner (in the bovine structure, pdb 1ppj) [38].
X-ray crystallographic analysis resolved highly ordered quinone molecules in the Q i site of crystallised cyt bc 1 complex. The power of cryo-EM to better cope with global or local protein dynamics brought forward a higher  Figure 5E) and the ovine supercomplex I/III 2 [48] ( Figure 3A) one ubiquinone molecule was resolved in each Q i site, in a position consistent to the known binding poses in X-ray structures of mitochondrial cyt bc 1 complexes (Figs. 3D, 5D). In contrast, in the cryo-EM structure of yeast supercomplex III 2 /IV 2 [46], an ubiquinone ring was modelled on the internal two-fold symmetry axis of the dimeric cyt bc 1 complex with two alternate conformations ( Figure 5F). The distance from the quinone ring to haem b H of each protomer is 15.3 Å.
In actinobacterial respiratory supercomplexes, a menaquinone molecule was identified in the Q i site of the cyt bcc-aa 3 supercomplex of C. glutamicum, M. smegmatis, and M. tuberculosis (Figs. 3E, 4A-C) [19,[52][53][54]65]. The interaction mode between haem b H and menaquinone is very similar to that of ubiquinone in cyt bc 1 complexes ( Figure 3D). In contrast with mitochondrial cyt bc 1 complexes, in which protons could be delivered to the Q i site ubiquinone via a histidine and an aspartate residue, of which the side chains have direct or watermediated hydrogen bonds to both carbonyl groups of the quinone, the menaquinone molecule resolved in the Q i site of the cyt bcc complex from C. glutamicum is single hydrogen-bonded directly to a glutamate side chain ( Figure 4E) [19]. Interestingly, a second menaquinone was identified near the Q i site of the bcc complex from M. smegmatis [52], with its naphthoquinone ring in 3.6 Å distance to that of the other menaquinone in the Q i site ( Figure 4B). This short distance between the two menaquinone molecules in and close to the Q i site would allow a consecutive reduction from one to the other. Menaquinone and ubiquinone are quinone species of low (−78 mV) and high (+90 mV) redox midpoint potential, respectively [13,14,[55][56][57]. The hyperthermophilic Aquifex aeolicus uses demethylmenaquinone (DMK) which has a potential of +36 mV [70], giving it a transitional position in the evolution of cyt bc complexes from low to high potentials [71]. The cryo-EM structure of the A. aeolicus cyt bc 1 complex with bound DMK molecules at the Q i site ( Figure 5C) revealed a 6.1 Å distance from the naphthoquinone ring to haem b H [72], which is in good agreement with the binding mode of the Q i site ubiquinone in yeast and Rhodobacter homologues as well as the Q i site menaquinone of the actinobacterial cyt bcc-aa 3 supercomplex.
The most unique Q i site architecture of cyt bc complexes is found in cyt b 6 f complexes. The position equivalent to the aforementioned ubiquinone and menaquinone ring plane in the Q i site is replaced by a high spin c-type haem (haem c i ), which is attached via a single thioether bond to cyt b 6 and which has no amino acid axial ligand [9,10]. A recent cryo-EM structure of spinach cyt b 6 f complex revealed the position of a plastoquinone molecule at the Q i site ( Figure 5A) [51]. The benzoquinone ring of this plastoquinone molecule is 4.4 Å apart from the haem c i porphyrin ring. In addition, one of its carbonyl groups is hydrogen-bonded to a propionate carboxylate of haem c i in 3.2 Å. Notably, the Q i site plastoquinone breaks the internal two-fold symmetry of cyt b 6 f complex ( Figure 5A). The isoprenoid tail of the Q i site plastoquinone extends into the entrance of the unoccupied Q i site of the other protomer while a second plastoquinone was modelled in a diagonal position with respect to the Q i site plastoquinone, in a position approaching the Q o site of the other protomer [51]. In addition, the Q i site occupancy of plastoquinone seems to be correlated to the orientation of the propionate group of haem c i , which may control access to a potential proton transfer pathway from the stromal side (the electronegative side) via Asp20 and Arg207 [51]. It was therefore hypothesised that both Q i sites are not simultaneously functional [51].
Whereas high-resolution X-ray structures revealed detailed binding modes of the Q i site ubiquinone in mitochondrial cyt bc 1 complexes, cryo-EM structures more recently provided additional information of ubiquinone positions in the context of supercomplexes, and previously unavailable structures of plastoquinone and Rhodobacter sphaeroides [69]; (C) cryo-EM structure of cyt bc 1 complex ( pdb 6kls) from Aquifex aeolicus [72]; (D) X-ray structure of cyt bc 1 complex ( pdb 2ibz) from baker's yeast [67]. The deposited structure contains only one protomer which belongs to the crystallographic asymmetric unit. Here the dimeric structure shown was generated by applying symmetry operation; (E) cryo-EM structure of a supercomplex containing a dimeric cyt bc 1 complex and a monomeric cyt c oxidase menaquinone-occupied Q i sites which show considerably different architecture as compared with mitochondria cyt bc 1 complexes. We anticipate that alternate reaction mechanisms will be required to accomplish quinone reduction and protonation at the Q i site in these complexes.

Inhibitors bound at Q o or Q i site of cyt bc complexes
The use of Q o and Q i site inhibitors was instrumental in studies of cyt bc 1 complexes in order to explore the molecular basis of the Q cycle mechanism and to elucidate electron transfer pathways [58]. Their binding positions in Q o and Q i site, in particular that of stigmatellin [16,17], myxothiazol, UHDBT, NQNO and antimycin A [37] were all analyzed as early as the first X-ray structures of cyt bc 1 complexes were determined ( Table 2). Stigmatellin is a semiquinone analogue, i.e. it mimics a transition state of quinol oxidation and reduction [66,67], which is difficult to be captured in protein crystals or cryo-EM specimens with natural substrates. Therefore, its binding poses in the Q o site of cyt bc 1 complex [17] and cyt bcc complex [19] provide insights in the catalytic position from which the protons and electrons are released to their respective acceptors. Parallel to fundamental research, cyt bc 1 complex inhibitors are also of great agricultural and medical importance: Azoxystrobin [37,73] belongs to the strobilurins [74], a group of chemically similar compounds [75] which accounted for 27% of the total fungicide worldwide sales in year 2015 [76]. The Q o site inhibitor Famoxadone is a fungicide for crops [77]. Atovaquone [78] is used in a fixed-dose combination with proguanil as antimalarial drug [79][80][81], and is also used for treating pneumocystis infection [82]. Note that both, atovaquone and strobilurin inhibitors target the Q o site, however, resistances were identified soon after these compounds were made commercially available [81,[83][84][85]. Consequently, development of cyt bc 1 complex inhibitors targeting the Q i site could provide a chance to bypass this issue [86,87]. Interestingly, in the past 5 years, almost all new antimalarial drug candidates resolved in structures of cyt bc 1 complexes published in the RCSB protein data bank (PDB, www.rcsb.org) are Q i site inhibitors (Table 2). This includes the X-ray structures of cyt bc 1 complex inhibited by the antimalarial 4(1H)-pyridones GSK 932121 and GW844520 [88], MJM170 [89], and a 2-pyrazolyl quinolone WDH2G7 [90]. Although X-ray structures can deliver information on protein-ligand interaction with atomic detail, structure-based drug discovery is often hindered by the amount of protein available, time required for crystallisation trials, and conformational heterogeneity or dynamic properties of proteins. The cryo-EM structures of cyt bc 1 complex with bound compounds SCR0911 and GSK 932121 [91] exemplified the scope of cryo-EM structures to characterise binding of drug candidates to target proteins with dynamic properties. Cryo-EM structures of the Mycobacterium cyt bcc-aa 3 complex with the tuberculosis drug candidate telacebec (Q203) [92] and with TB47 bound at the Q o site demonstrated this approach for bacterial cyt bc complexes and supercomplexes [64,65]. Cryo-EM has the advantage of lower sample consumption for single particle analysis as compared with X-ray crystallography. This is especially important for proteins isolated from scarce sources such as patient tissue [93], or from pathogens which are difficult or dangerous to cultivate [94]. In this respect, cryo-EM also opens new possibilities in obtaining structural information of cyt bc complexes to develop novel human medications as well as agrochemicals [95][96][97][98].

Detergent, lipids and the native membrane
Owing to the nature that membrane proteins are located in the lipidic compartments of the cell [99], structural biology studies of membrane proteins have greatly benefited from the use of detergents to solubilise them from their native environment into aqueous solution. Detergent molecules bind to hydrophobic surfaces of membrane proteins and increase their solubility in aqueous environment. Detergents differ in their chemical and physical properties and the selection of the type of detergent is key to prepare well-diffracting membrane protein crystals [100] as well as cryo-EM grids with good contrast and particle distributions [101]. However, detergents compete with the binding of lipids and lipidic compounds such as quinone thus delipidation is unavoidable. Severe delipidation compromises the stability and eventually the integrity of isolated membrane proteins, which may cause artificial structural disorder and may account for poor resolution of X-ray and cryo-EM structures. Reintroducing the detergent solubilised membrane protein back into lipidic cubic phase (supercomplex III 2 /IV, pdb 6giq) from baker's yeast [47]; (F) cryo-EM structure of a supercomplex containing a dimeric cyt bc 1 complex and a dimeric cyt c oxidase (supercomplex III 2 /IV 2 , pdb 6hu9) from baker's yeast [46]. (L.C.P.) for crystallisation [102,103] and the application of lipidic nanodiscs in solubilisation or reconstitution of isolated membrane protein complexes for cryo-EM specimen preparation [104,105] have shown superior stabilisation effect so as to improve resolution. This can be exemplified by the X-ray structure of Thermus thermophillus cyt caa 3 oxidase (2.36 Å resolution, L.C.P. [106]), cryo-EM structures of Escherichia coli cyt bd oxidase (2.68 Å resolution, nanodiscs [107]) and the cryo-EM structure of Paracoccus denitrificans cyt c oxidase (2.37 Å resolution, nanodiscs [108]; all resolution of cryo-EM data refer to the FSC = 0.143 criteria for the same basis of comparison). Respiratory chain complexes and supercomplexes in nanodiscs may provide additional information about partitioning of co-purified quinone molecules and their trajectories to fully reflect the native electron transport chain in the hydrophobic environment. Finally, structural studies using in situ cryogenic electron tomography (cryo-ET) permits the determination of higher order assemblies of protein complexes as well as structural dynamics directly in cells [109]. Although many technical limitations, such as to resolve small molecules with sufficient resolution still need to be overcome, the rapid and intensive development of cryo-ET [110] will eventually allow to visualise the respiratory chain and photosynthesis complexes in cellular context and maybe in action.

Perspectives
• Structural biology research of cyt bc complexes will contribute to an in-depth understanding of redox-driven proton translocation via the Q cycle and its regulation as well as support the design of fungicides, anti-malarial and anti-tuberculosis drugs.
• Structural characterisation of cyt bc complexes from a wide spectrum of species as well as in different types of supercomplexes is important to expand the knowledge on conserved and species-specific binding modes of native substrates, drugs, and inhibitors at the quinone binding sites.
• Structural information on the enzyme-substrate complex and defined catalytic states of cyt bc complexes is still lacking. We encourage that the cryo-EM specimens or crystals should be prepared in lipid environment.