Reconstructing the Qo Site of Plasmodium falciparum bc 1 Complex in the Yeast Enzyme

The bc 1 complex of the mitochondrial respiratory chain is essential for Plasmodium falciparum proliferation, the causative agent of human malaria. Therefore, this enzyme is an attractive target for antimalarials. However, biochemical investigations of the parasite enzyme needed for the study of new drugs are challenging. In order to facilitate the study of new compounds targeting the enzyme, we are modifying the inhibitor binding sites of the yeast Saccharomyces cerevisiae to generate a complex that mimics the P. falciparum enzyme. In this study we focused on its Qo pocket, the site of atovaquone binding which is a leading antimalarial drug used in treatment and causal prophylaxis. We constructed and studied a series of mutants with modified Qo sites where yeast residues have been replaced by P. falciparum equivalents, or, for comparison, by human equivalents. Mitochondria were prepared from the yeast Plasmodium-like and human-like Qo mutants. We measured the bc 1 complex sensitivity to atovaquone, azoxystrobin, a Qo site targeting fungicide active against P. falciparum and RCQ06, a quinolone-derivative inhibitor of P. falciparum bc 1 complex.The data obtained highlighted variations in the Qo site that could explain the differences in inhibitor sensitivity between yeast, plasmodial and human enzymes. We showed that the yeast Plasmodium-like Qo mutants could be useful and easy-to-use tools for the study of that class of antimalarials.


Introduction
Malaria is among the most serious health problems in the world leading to more than 655,000 deaths in 2010 mainly among children under five years old, according to the World Malaria report 2011. Although several antimalarial drugs have been developed, emerging resistance to many of these compounds and dissemination of drug-resistant P. falciparum are compromising the treatment of malaria patients. Thus there is an urgent need for new antimalarial drugs. The complex III or bc 1 complex is an attractive target for antimalarial drugs. This respiratory chain complex is essential for P. falciparum proliferation as its catalytic activity is critical for the maintenance of the mitochondrial membrane potential and for the reoxidation of ubiquinol, which is needed for the ubiquinone-dependent dihydroorotate dehydrogenase, and by consequence for the biosynthesis of pyrimidine (see for instance [1]). In addition, differences in the sequences of the active sites of the bc 1 complex between organisms facilitate the search for inhibitors with selective activity.
The hydroxynaphthoquinone atovaquone is currently the sole bc 1 complex inhibitor used (in combination with proguanil as MalaroneH) as antimalarial drug in treatment and causal prophylaxis. The compound is also used to treat Pneumocystis jirovecii pneumonia, Toxoplasma gondii toxoplasmosis, and other infections. In the USA, MalaroneH prescription accounted for more than half of all antimalarial prescriptions. However, the cost of atovaquone is so far prohibitive for more general use. The patent for MalaroneH expires this year, which might result in lower cost generics. With the possibility of more extensive use of atovaquone, the risk of resistance to atovaquone would likely increase.
P. falciparum atovaquone-resistant parasites have been reported to emerge during atovaquone-proguanil therapy, leading to typical treatment failure [2,3]. The resistance is caused by point mutations in the drug target. Therefore, new drugs that could circumvent the resistance would be required. Different compounds are currently being studied, such as 4(1H)-pyridones, acridones, acridinediones, and 4(1H)-quinolones (see for instance, [4] and also [5,6]). Known drugs used to control animal parasites [7] or plant pathogenic fungi [8] have been also revisited. It has been reported, for instance, that the fungicide azoxystrobin that targets the bc 1 complex was a potent inhibitor of P. falciparum [8].
The bc 1 complex contains two sites that bind its substrate quinol and inhibitors, the Q o and Q i sites. Atovaquone targets specifically the Q o site. The Q o site (and the Q i site) is provided by the main subunit of the bc 1 complex, cytochrome b. Cytochrome b is mitochondrially encoded in all eukaryotes while the other subunits of the complex are nuclearly encoded. The Q o site is a relatively large domain formed from components encompassing amino acid residues 120-150 and 260-280 of the cytochrome b. Hydrogen bonds form a 'loose stitching' that allows a degree of expansion of the site on occupant binding [9]. Some sidechains within the Q o site, such as components of the 'PEWY' motif (residues 271-274, yeast numbering), demonstrate considerable conformational flexibility on inhibitor binding [10]. The Q o site forms a bifurcated volume. On the basis of structural and mutational analysis, Q o inhibitors can be separated into two classes by their mode of binding [9,10]. For instance, azoxystrobin binds in the so-called 'haem b l proximal' domain of the Q o site whereas stigmatellin binds in the 'haem b l distal' domain, occupying the region of the Q o site close to the interface with another catalytic subunit of the complex, the iron-sulphur protein (ISP). Crystal structures for several inhibitor-liganded bc 1 complexes are available, such as azoxystrobin-and stigmatellin inhibited enzymes (see for instance, [10][11][12]). A crystal structure for the atovaquone-inhibited complex is not yet available. However, atovaquone is likely to be a 'distal' binding inhibitor due to its effect on the EPR spectrum of the ISP [13]. The quinolone compound RQC06 that showed a promising antimalarial activity might also bind in the 'distal' region of the Q o pocket, as suggested by the modelling analysis [14].
In order to facilitate the study of drugs targeting the P. falciparum bc 1 complex, we are developing the yeast (Saccharomyces cerevisiae) model. By modifying the inhibitor binding sites of the yeast native bc 1 complex, new versions can be generated that mimic the P. falciparum enzyme. In this study, we constructed a series of yeast mutants harbouring variants of the Q o site where yeast residues have been replaced by P. falciparum equivalents, or, for comparison, by the human equivalents. The yeast mutants with Plasmodium-like (and human-like) Q o site were used to investigate the structural determinant of the differential sensitivity to atovaquone and azoxystrobin and to explore the binding mode of the novel molecule RCQ06. We also constructed a mutant harbouring the atovaquone resistance mutation Y279S (Y268S in P. falciparum sequence) in the Plasmodium-like enzyme to study the effect of this mutation on new inhibitors such as RCQ06. The yeast mutants could be useful tools for the study of new drugs with selective activity and for the analysis of resistance mutations.

Materials
Equine cytochrome c, decylubiquinone, azoxystrobin and atovaquone were obtained from Sigma. RCQ06 was kindly supplied by Prof. Paul O'Neill (Liverpool, UK).

Yeast Mutants
The mutants were generated by side-directed mutagenesis and mitochondrial transformation as described in [15]. They have identical nuclear and mitochondrial genomes with the exception of the mutations introduced in the cytochrome b gene.

Measurement of Quinol:Cytochrome c Reductase Activity
Yeast mitochondria were prepared as in [16]. Bovine mitochondrial samples were kindly given by Prof. Peter Rich (UCL, UK). Quinol:cytochrome c reductase activity measurements were performed in 10 mM potassium phosphate pH 7 and 20mM equine cytochrome c at room temperature. Mitochondria were diluted to 5-30 nM bc 1 complex. Concentrations of monomeric bc 1 complex were determined from the cytochrome b a-band in dithionite-reduced optical spectra, using e = 28.5 mM 21 .cm 21 at 562 nm minus 575 nm. Activity was initiated by the addition of 20mM decylubiquinol. Cytochrome c reduction was recorded at 550 nm versus 540 nm over a 3-min time-course in a Beckmann DU 640 spectrophotometer. Initial rates were measured. From these rates, turnover numbers (TN) were calculated as cytochrome c reduced per bc 1 complex per second.

Inhibitor Titration
Cytochrome c reduction activity was measured as described above in presence of increasing concentrations of inhibitors (six to ten different concentrations). Each measurement was repeated at least twice and averaged. The errors did not exceed 10%. The mid-point inhibition concentrations (IC 50 ) were determined from the titrations. As the titrations were performed using mitochondrial samples containing different concentrations of bc 1 complex (5-30 nM), the results were presented as ratio of IC 50 on the concentration of bc 1 complex.

Ligand Docking and Molecular Modelling
Atovaquone was docked into the Q o site of yeast cytochrome b (3CX5.PDB) as described in [17]. An atomic model of RCQ06 was created using PRODRG2 and the Dock Prep module of Chimera [18,19]. The energy-minimised RCQ06 model was docked into a 9Å radius sphere centred on the e2-oxygen atom of cytochrome b residue E272 in the Q o site of 3CX5.PDB using EADock DSS via SWISSDOCK [20]. Using an iterative search, ligand binding modes with favourable CHARMM energies were clustered taking account of the solvent effect with the FACTS implicit solvation model, and the resulting output files examined with Chimera and VMD.

1) Sensitivity to Q o Inhibitors: Comparison between Plasmodial, Mammalian and Yeast bc 1 Complexes
The sensitivity of the bc 1 complex (cytochrome c reductase) activity towards atovaquone, azoxystrobin and RCQ06 is presented in Table 1. As previously reported, yeast bc 1 complex, as P. falciparum enzyme, is highly sensitive to atovaquone (IC 50 of 4 (molar ratio)) while the bovine enzyme was less reactive (IC 50 of 75). Azoxystrobin has been shown to be a potent inhibitor of P. falciparum proliferation with an IC 50 in the nanomolar range (15 nM [8]). The drug inhibits yeast bc 1 complex with an IC 50 of 20 while it is less active against the bovine enzyme (IC 50 of 180). RCQ06 was also reported as potent inhibitor of the plasmodial bc 1 complex [14] with an IC 50 of around 1.3 nM, which is in the same range than the atovaquone IC 50 (around 3 nM) [21]. We found that the bovine enzyme was less sensitive to that compound (IC 50 of 40). The differential sensitivities to RCQ06 and atovaquone between the Plasmodium and mammalian enzymes are thus similar. By contrast, the yeast enzyme was highly resistance to the RCQ06 (IC 50 .500).
In order to study the structural determinants that could explain the differential sensitivity to these drugs, we monitored the impact of mutations introduced in the Q o site of the yeast bc 1 complex, single mutations and combined Plasmodiumor human-like changes.

2) Determinants of Atovaquone Sensitivity Studied in Plasmodium-like Yeast Q o Site Mutants
In previous works, the low atovaquone sensitivity of the bovine bc 1 complex (as compared to the sensitivity of the yeast enzyme) was investigated. A computed energy-minimised structure for atovaquone liganded to the yeast bc 1 complex suggested that residue 275 plays a key role in the differential sensitivity. The presence of a phenyalanine at position 275 in the bovine enzyme would hinder the drug binding. F275 being replaced by L in the yeast enzyme, the steric constraint is lessened and the sensitivity to atovaquone increases. When the mutation L275F was genetically introduced in the yeast bc 1 complex, its sensitivity was significantly decreased (over 20-fold) [13,15,22].
In the cytochrome b of P. falciparum (and of other Plasmodium and of Toxoplasma), the F275 variant is naturally present. However the parasite bc 1 complex is highly sensitive to atovaquone. Thus other structural variations should explain the differential sensitivity between the parasite and the mammalian enzymes.
Comparison of the cytochrome b sequences ( Figure 1A) shows that the Q o domain is well conserved between organisms. There are however variations that may affect atovaquone susceptibility.
In order to investigate their impact, we constructed a series of yeast strains harbouring multiple changes in the Q o domain. We replaced, in a step-wise manner, ten yeast residues by the P. falciparum equivalents (see Table 2): in positions 133-136 and 141 (mutant PF1), positions 133-136, 141 and 275 (mutant PF2), positions 275 to 299 (mutant PF8) and finally a combination of the ten changes from residue 133 to residue 299 (mutant PF11). The location of the mutated residues in the structure is shown in Figure 1B. The bc 1 complex level was not affected in the mutants produced in this study, as judged by the optical signal of cytochromes (not shown). None of the modifications introduced in the Q o site severely affected the overall assembly of the complex.
In addition to substitutions in the Q o site, the sequence comparison ( Figure 1A) showed a deletion of four amino acids, D158-161, in P. falciparum cytochrome b. The deletion, observed in all apicomplexan, is located in helix cd2 in the vicinity of the Q o site ( Figure 1B). The deletion, introduced in PF11, caused a significant decrease in bc 1 complex activity but had little effect on atovaquone sensitivity (not shown). The study of the deletion and of its possible compensations will be reported elsewhere.
The Q o pocket of PF11 best mimics P.falciparum Q o site as it harbours most of the variations that might affect the susceptibility to atovaquone or other Q o site inhibitors. The combined change was well tolerated: the bc 1 complex retained more than 40% activity (Table 1).
Interestingly, PF11 was highly sensitive to atovaquone despite the presence of F275 (IC 50 of 3) whereas F275 alone in the yeast Q o site (mutant L275F) caused a strong atovaquone resistance (IC 50 of 150). A similar behaviour was observed with PF2 and PF8 but the atovaquone sensitivity was less marked (IC 50 of 10 and 16, respectively). In Rhodabacter capsulatus cytochrome b, a phenylalanine is naturally present at this position. The IC 50 of atovaquone was estimated to be around 30 nM. Interestingly, the sensitivity of the bacterial complex to atovaquone was increased (IC 50 decreasing from 30 nM to 10 nM) by the introduction of two mutations replacing bacterial residues by the plasmodial equivalents, namely I281M and R283K (yeast numbering) [30], which is in agreement with our data as R283K is present in PF8. It seems therefore that when L275F is combined to other Plasmodium-like mutations, the impact of L275F is weakened. From these data, it appears that the atovaquone sensitivity of P. falciparum bc 1 complex could be explained by a combination of variations (variations from the yeast Q o site) that cancel the resistance effect of F275. We could hypothesise that the P. falciparum Q o site and the yeast Plasmodium-like Q o site have a larger volume that could accommodate atovaquone despite a phenylalanine in position 275. The modified yeast bc 1 complex harbouring a Plasmodium-like Q o and highly susceptible to atovaquone (PF11) could be useful model to study the effect of resistance mutations.

3) Determinants of Atovaquone Resistance Studied in Human-like Yeast Q o Site Mutants
The mammalian bc 1 complex is naturally resistant to atovaquone (as compared to the plasmodial and the yeast enzymes) ( Table 1).
To investigate further the observed resistance, we modified the Q o site of yeast bc 1 complex and introduced human variations, single or combined ( Table 3). The substitutions were well accommodated in yeast Q o site. The mutant enzyme combining eight human changes (HSc) showed nearly WT activity (89%).
The three human-like mutations, L275F, F278A and M295L, conferred some level of resistance towards atovaquone. The IC 50 increased 35-, 9-and 2.5-fold in L275F, F278A and M295L, respectively. The combination of these mutations (HSa and HSb) The mutated residues in the yeast strains are presented in Table 2 for the Plasmodium-like mutants (PF) and in Table 3 for the human-like mutants (HS). Their location in the sequence and the structure is shown in Fig. 1. For P. falciparum enzyme, the atovaquone IC 50 is taken from [21]. In the same study, the IC 50 s for bovine and human enzymes were approximately 70 nM. The RCQ06 IC 50 value is from [14]. Note that the bc 1 complex concentrations in the P. falciparum preparations used for inhibitor titrations were not available in the published studies. It was thus not possible to present the data as the molar ratio IC 50 /[bc 1 complex]. An IC 50 of azoxystrobin is not available. The inhibitor was reported to be highly active on the parasite growth [8]. doi:10.1371/journal.pone.0071726.t001 caused a further increase of resistance. When combined with amino-acid changes at positions 133 to 141 (HSc), the resistance to atovaquone was decreased, as observed with Plasmodium-like mutants. However, the human-like bc 1 complex remained 15-fold more resistant than the WT enzyme, with an IC 50 value of 60, similar to the IC 50 value of 75 obtained with the bovine enzyme. It can thus be suggested that the combination of F275, A278 and L295 in human bc 1 complex is responsible for the natural resistance towards atovaquone.

4) Sensitivity and Resistance to Azoxystrobin of Plasmodium-and Human-like Mutants
In the human-like mutants, F278A caused a 6-fold increase in azoxystrobin resistance, M295L and L275F had a weaker effect. The triple mutation (HSb) conferred 8-fold resistance. Changes in position 133 to 141, when combined with that triple mutation (HSc), did not affect the resistance (IC 50 of 140). Thus the 9-fold resistance of the mammalian enzyme to azoxystrobin (IC 50 of 180), compared to yeast enzyme (IC 50 of 20) could be explained by the combination of F275, A278 and L295 in the binding site. In the Plasmodium-like yeast mutants, modifications of residues 283 to 299 caused a 10-fold increase in resistance (IC 50 of 220 for PF8), but the changes in position 133 to 141 lowered the resistance by 5 fold (IC 50 of 40 for PF11). That combination of residues might explain the susceptibility of P. falciparum to azoxystrobin.
As a control, we compared the sensitivity to the Q i site inhibitor antimycin of the bc 1 complex of the human-and the Plasmodiumlike mutants, HSc and PF11, with that of the WT control. No or very little difference was observed (data not shown).

5) Effect of the Acquired Atovaquone Resistance Mutation Y279S (Y268S in P. falciparum)
In P. falciparum isolated from patients after treatment failure, three mutations have been reported, Y279S/C/N (Y268 in P. falciparum). Mutations Y279S and C are the most frequent mutations in the parasite. Their consequences on the catalytic activity and the stability of the complex have been characterized in the P. falciparum (Y268S) [17], the yeast [31,32] and the bacterial enzyme [30,33].
We introduced the mutation Y279S in PF11 to obtain a model mimicking P. falciparum resistant enzyme. As presented in Table 1, the resulting mutant enzyme (PF12) showed a more severe decrease in activity, retaining only 15% of the WT activity. This was expected as the mutation Y279S alone in the yeast Q o site caused a 70% decrease of the bc 1 complex activity. In the parasite bc 1 complex, the acquired resistance mutation resulted in a low activity and unstable ISP [17]. It has been suggested that residue Y279 is involved in the correct orientation of the quinol bound in the Q o site via tyrosyl-benzoquinone hydrophobic packing, facilitating quinol deprotonation and electron transfer to the ISP [34]. Substitution of the tyrosine by (the less bulky) serine is likely to abolish, or otherwise weaken, this predicted stabilising interaction interfering with optimal quinol binding, potentially slowing the reduction of the ISP [2Fe2S] cluster redox group.
When introduced in the Plasmodium-like Q o site (PF12) or in the yeast Q o site, Y279S caused a high level of resistance to atovaquone (IC 50 .850), similar to that observed with the Y268S plasmodial enzyme. By contrast, the sensitivity to azoxystrobin was little affected by the atovaquone resistance mutation (IC 50 of 40 and 30 for PF11 and PF12, respectively). This result was not unexpected since azoxystrobin has been shown to bind in the Q o site at a position proximal to the haem b l [10], while atovaquone binds at a position distal to the haem b l [13]. However it emphasizes the interest of developing molecules, such as azoxystrobin, that have both differential reactivity and can circumvent the atovaquone-resistance mutation.

6) RCQO6 Binding, Comparison with Atovaquone
As shown in Table 1, the yeast bc 1 complex is highly resistant to RCQ06. Strikingly, the introduction of mutation L275F rendered the enzyme over 50-fold more sensitive to RCQ06 than WT. The sensitivity to RCQ06 was observed in all the mutants with L275F in the Q o site: PF2, PF8, and PF11. Thus RCQ06 binding is likely to be stabilised by an aromatic residue at position 275, whereas the binding of atovaquone is hindered by the bulkier residue. The chemical structures of RCQ06, atovaquone and azoxystrobin are presented in Figure 2.
These residues are shown in the sequence comparison ( Figure 1A). Their location in the Q o structure is presented in Figure 1B. -, unchanged residue. doi:10.1371/journal.pone.0071726.t002 Table 3. Human-like mutations in the yeast Q o site.
strains mutated residues and positions These residues are shown in the sequence comparison ( Figure 1A). Their location in the Q o structure is presented in Figure 1B The introduction of the atovaquone resistance mutation Y279S in Plasmodium-like Q o site (PF12) caused a high level of resistance to RCQ06, as it did for atovaquone. This suggests that the two inhibitors have similar binding mode. On the basis of this mutational analysis, a model of RCQ06 has been build. Figure 1C presents a molecular model of atovaquone and RCQ06 docked into the Q o site of the yeast bc 1 complex. As mentioned in the introduction, a crystal structure of the atovaquone-bound enzyme is not available but atovaquone is predicted to interact with the Q o site in a manner similar to that of stigmatellin via hydrogen bonding from its naphthoquinone moiety via a bridging water molecule to E272 (yeast numbering) of cytochrome b, with a second hydrogen bond formed to the imidazole group of the ISP [2Fe2S] cluster ligand H181 [13].
Our model suggests that RCQ06 binds in a manner reminiscent to that of stigmatellin with a 3.5Å hydrogen bond between its quinolone amine moiety and the glutamyl sidechain of cytochrome b residue E272. This association would require that the glutamyl is acting as the hydrogen bond acceptor, and is thus present in the deprotonated (carboxylate) form, which is thermodynamically accessible when haem b l is oxidised [35]. In the docking model, the quinolone carbonyl of RCQ06 is within hydrogen bonding distance of [2Fe2S] cluster ligand H181 (2.8Å ), although the formation of such a hydrogen bond would require the histidyl sidechain of this residue to be present as a protonated imidazole, favouring the reduced form of the [2Fe2S] cluster. Perhaps surprisingly, we predict no hydrogen bonding interactions between the trifluoromethoxy group of RCQ06 and the surrounding protein. A stabilising aliphatic/aromatic interaction (2.7Å ) is predicted to occur between the sidechain of L275 and the quinolone moiety of RCQ06. As shown in Table 1, the experimental data indicate that the determinant of the sensitivity towards RCQ06 is a phenylalanine in position 275, which is naturally present in plasmodial and mammalian cytochrome b. Inspection of the model presented in Figure 1C suggests that a favourable aromatic/aromatic interaction may be formed between F275 and the RCQ06 quinolone group, with a larger van der Waals contact area than that observed for the proposed L275/ RCQ06 interaction, effectively packing this sidechain against the inhibitor. A similar stabilising aromatic/aromatic interaction, with the aromatic groups oriented perpendicularly, is predicted to form between the sidechain of Y279 and the quinolone moiety of RCQ06. Replacement of Y279 by a serine would destabilise RCQ06, which would explain the observed resistance.

Conclusions
Acquired resistance to atovaquone developed readily, which is caused by the mutation of the cytochrome b residue Y279 (Y268 in P. falciparum). New drugs are needed that can circumvent the acquired atovaquone resistance: for example, drugs targeting the Q i site [36,37] and compounds binding at the Q o site in the haem b l -proximal region such as azoxystrobin or in the distal site but with different binding interactions than atovaquone [38].
Cytochrome b sequences from yeast, human and P. falciparum are highly conserved, making yeast an attractive model to investigate sensitivity and resistance to drugs targeting the bc 1 complex. Crystal structures of yeast bc 1 complex are available. Most importantly, genetic tools have been well developed to introduce designed mutations in yeast mitochondrially-encoded cytochrome b gene. Yeast is the only organism, with the green algae Chlamydomonas reinhardtii, amenable to mitochondrial transformation. By modifying the yeast enzyme, useful tools could be created for the discovery and the analysis of new drugs with potential antimalarial activity and for the study of the development of resistance mutations.
In previous reports, we studied acquired resistance mutations found in parasites and introduced in yeast Q o site [15,22,32]. In this study, we generated yeast mutants with modified Plasmodiumlike Q o sites that mimic the site of the plasmodial enzyme. We tested the sensitivity to atovaquone and azoxystrobin to validate the model. We showed that the differential sensitivity to atovaquone and azoxystrobin, and the high resistance caused by Y279S (Y268S) could be reproduced in the yeast models.
We then used the mutants to study the binding mode of a new molecule RCQ06. The experimental data, together with the model of RCQ06 binding in the Q o site, indicated that the drug shares the same binding pocket than atovaquone. Even though the binding mode of RCQ06 might not be identical to that of atovaquone, the efficiency of binding is affected by the Y279S mutation in the yeast enzyme.
More remodelling of the yeast Q o and Q i sites will be performed to obtain easy-to-use models that mimic more accurately the plasmodial enzyme. These mutants with a P. falciparum-like enzyme could be useful tools for the discovery and the analysis of new drugs with potential anti-malarial activity and for the prediction of sensitivity/resistance phenotype associated with