Cytochrome b Mutations That Modify the Ubiquinol-binding Pocket of the Cytochrome bc1 Complex and Confer Anti-malarial Drug Resistance in Saccharomyces cerevisiae*

Atovaquone is a new anti-malarial agent that specifically targets the cytochrome bc1 complex and inhibits parasite respiration. A growing number of failures of this drug in the treatment of malaria have been genetically linked to point mutations in the mitochondrial cytochrome b gene. To better understand the molecular basis of atovaquone resistance in malaria, we introduced five of these mutations, including the most prevalent variant found in Plasmodium falciparum (Y268S), into the cytochrome b gene of the budding yeast Saccharomyces cerevisiae and thus obtained cytochrome bc1 complexes resistant to inhibition by atovaquone. By modeling the variations in cytochrome b structure and atovaquone binding with the mutated bc1 complexes, we obtained the first quantitative explanation for the molecular basis of atovaquone resistance in malaria parasites.

Up to one-half billion people in the world suffer from malaria with various degrees of severity. The disease kills more than 2.7 million people a year, most of them children under 5 years old in sub-Saharan Africa (1). Over the past three decades, the parasite that causes malaria, Plasmodium falciparum, has developed resistance to almost every commonly available antimalarial drug, including chloroquine, pyrimethamine, cycloguanil, and sulfadoxine. Because some of these drugs are now almost useless in many parts of the world, the rapid spread of resistant parasites is a serious global health problem in endemic countries.
In the 1990s, the urgent need for new anti-malarial drugs for treatment and chemoprophylaxis led to the development of atovaquone (2-[trans-4-(4Ј-chlorophenyl)cyclohexyl]-3-hydroxy-1,4-hydroxynaphthoquinone). This recently introduced anti-malarial compound, which has broad spectrum activity against apicomplexan parasites (2)(3)(4), also prevents and clears Pneumocystis pneumonia (5). Atovaquone is a potent and specific inhibitor of the cytochrome bc 1 complex (EC 1.10.2.2) (6), an essential respiratory enzyme present in the inner mitochondrial membrane. This drug was approved by the FDA in 1995 and is now distributed as a synergic combination with proguanil hydrochloride under the trade name of Malarone®. Despite minimal side effects and high activity against malaria, the cost of Malarone® treatment has, until now, restricted its use to western non-immune adult and children (7) travelers and some European military personnel (8).
Unfortunately, there is growing evidence that malarial parasites may quickly develop drug resistance by mutation of amino acid residues located in or near the atovaquone-binding site on cytochrome b (9 -11). Within just a few years after its introduction, the mitochondrial DNA sequences of isolates of P. falciparum associated with Malarone® treatment failures exhibit the same Y268S (malaria numbering) mutation of the cytochrome b gene of the parasite. Because these isolates were obtained from infected patients returning from Thailand (12), Ivory Coast (13), Mali (14), Cameroon (15), and Kenya (16), growing concern about the spread and the high frequency of this mutation led to the design of PCR-based screening procedures in Europe for quick detection in case of Malarone® treatment failure (14,17,18).
Because the yeast bc 1 complex is also inhibited by atovaquone (6), we previously developed Saccharomyces cerevisiae as a model to study cytochrome b mutations conferring atovaquone resistance in Pneumocystis (19,20). In the present study we transferred several mutations associated with atovaquone resistance in Plasmodium yoelii (9) and P. falciparum (10,(12)(13)(14)(15)(16) into the yeast cytochrome b gene. This allowed us to biochemically confirm the linkage of atovaquone resistance to the cytochrome b mutations and to explain at the molecular level the mechanism by which malaria parasites counter the therapeutic cytotoxicity of this new drug.

EXPERIMENTAL PROCEDURES
Materials-Yeast extract and peptone were from Difco. Nitrogen base without amino acids but with ammonium sulfate was from U. S. Biological. Dodecylmaltoside was obtained from Roche Applied Science. DEAE-Biogel A was obtained from Bio-Rad. Decyl ubiquinone, diisopropylfluorophosphate, and dithionite were purchased from Sigma. Stigmatellin was purchased from Fluka Biochemica. Atovaquone was a gift from GlaxoSmithKline.
Purification of Cytochrome bc 1 Complexes-The cytochrome b mutations were introduced into the mitochondrial cytochrome b gene by biolistic transformation as described elsewhere (19). Wild-type and mutated yeast strains were grown in yeast extract/peptone/dextrose medium and harvested by centrifugation. Cytochrome bc 1 complexes were then isolated from mitochondria as described previously (21,22).
Measurement of Oxygen Consumption and Ubiquinol-Cytochrome c Reductase Activity-Rates of oxygen uptake by intact yeast cells were measured at 25°C with a Gilson oxygraph and a polarographic oxygen electrode (Gilson Medical Electronics, Middleton, WI) as described previously (23). Cytochrome c reductase activities of the cytochrome bc 1 complexes were assayed in 50 mM potassium phosphate, pH 7.0, 250 mM sucrose, 0.2 mM EDTA, 1 mM NaN 3 , 2.5 mM KCN, 0.01% Tween 20, and 40 M cytochrome c at 23°C. The cytochrome bc 1 complex was diluted to 2.5 nM in the assay buffer, and the reaction was started by adding 2,3-dimethoxy-5-methyl 6-decyl-1,4-benzoquinol, an analogue of ubiquinol. The reduction of cytochrome c was monitored in an Aminco DW-2a spectrophotometer at 550 versus 539 nm in dual wavelength mode. Data were collected and analyzed using an Online Instrument Systems Inc. computer interface and software.
Molecular Modeling-Molecular modeling was carried out on a Silicon Graphics O2 work station using the commercially available Insight II® software package (Accelrys Inc., San Diego, CA). The starting structure was the energy-minimized atovaquone-bound yeast cytochrome bc 1 complex (6). Each cytochrome b mutation, Y279S, Y279C, I269M, L282V, and F278I, was built separately into the atovaquonebound center P site using the lowest energy manual rotamer selected by the Biopolymer® module.
For each structure, a flexible subset of adjacent residues within 2.0 Å of the mutation was selected. A surrounding 9.5-Å shell of residues, including atovaquone, was kept fixed. Residues beyond this shell were excluded from calculations in the interest of speed. The atom-based cut-off for non-bonding interactions was set at 9.5 Å, and the dielectric constant was set at 2.0. The CFF91 force field, which uses a 6 -9 Lennard-Jones potential to calculate van der Waals interactions, was used for all modeling.
A five-step simulated annealing run, from 800 to 298 K, was performed on each mutated structure using the Discover 3® module in Insight II®. At each of the five temperatures, the Nosé temperature control method was used for a simulation time of 10,000 fs/step, with a 0.5 fs/iteration time step. The final structure was minimized to a final convergence of 0.001 using Cauchy's steepest descent, conjugate gradient, and Newton methods consecutively. Five simulated annealing runs were carried out for each mutant structure to obtain an average final energy.
To estimate the energy cost of binding atovaquone, we calculated the energy of each mutant structure with the bound atovaquone and compared it with the energy of the wild-type enzyme with bound atovaquone. To compare non-bonding interaction energies (van der Waals and electrostatic) and internal conformation energies of mutant and wild-type structures, a common subset (20), including atovaquone, the residue His-181 on the iron-sulfur protein, cytochrome b residues within 4.0 Å of the inhibitor, and cytochrome b residues within 4.0 Å of cytochrome b residue Glu-272, was used for all calculations.

Location of the Cytochrome b Mutations Conferring Atova-
quone Resistance to Plasmodium-The molecular target of atovaquone is now known to be the ubiquinol oxidation pocket at the center P site of the cytochrome bc 1 complex (6). An energy-minimized structure of atovaquone in the ubiquinol oxidation pocket obtained by molecular modeling indicated that the hydroxyl group of the hydroxy-naphthoquinone bound via a hydrogen bond to the nitrogen of His-181 of the Rieske iron-sulfur protein. This hydrogen bonding accounts for the effects of atovaquone on the midpoint potential of the Rieske protein (6) and mimics the hydrogen bond between undecyl hydroxy benzoxythiazol and the Rieske protein in the crystal structure of the undecyl hydroxy benzoxythiazol-bound bc 1 complex (24). On the opposite side of the ring system, the carbonyl group at position 4 of the quinone ring of atovaquone formed a water-mediated hydrogen bond with Glu-272 of cytochrome b. The remainder of the molecular interactions between atovaquone and cytochrome b were essentially hydrophobic with a network of aromatic and aliphatic side chains surrounding the inhibitor. This binding pattern is similar to the one recently found in the crystal structure of the yeast bc 1 complex with undecyl hydroxy benzoxythiazol, a hydroxyquinone, bound to the ubiquinol oxidation pocket (24).
Because there is a high degree of sequence identity (about 68%) between S. cerevisiae and Plasmodium cytochrome b within the atovaquone-binding pocket ( Fig. 1), we chose the yeast for studying how mutations can affect the drug efficacy. All atovaquone-resistant mutations in the cytochrome b sequence found in Plasmodium were localized around this binding site (9 -14, 25). Five of these mutations associated with resistance in P. yoelii (9) and P. falciparum (10 -16) (Fig. 1) were transferred into the S. cerevisiae cytochrome b gene by the biolistic method (19).
The modeled structure of the yeast ubiquinol oxidation pocket at the center P site with bound atovaquone and the location of the amino acids residues affected in the atovaquoneresistant cytochrome b mutants are shown in Fig. 2. These five mutations are I258M, F267I, Y268C, Y268S, and L271V in Plasmodium numbering, equivalent to I269M, F278I, Y279C, Y279S, and L282V in the yeast numbering system (we will use S. cerevisiae numbering of cytochrome b hereafter). Except for I269M, which is located in the EF loop, all of the mutations affect residues of the EF helix of cytochrome b.
Loss of Fitness Associated with the Cytochrome b Mutations-Growth of the five mutated strains was monitored in a non-fermentable medium to study the impact of the mutations on the respiration of the yeast cells. The growth curves (Fig.  3A) reveal that, in contrast to a doubling time of 6 (Ϯ 0.5) h observed for the wild-type strain and also for the I269M and F278I mutants, the Y279C mutated strain had a doubling time The iron-sulfur cluster is at the top right, with iron and sulfur atoms colored purple and yellow, respectively. The carbon atoms in atovaquone (ATV) are green, oxygen atoms are red, and hydrogen atoms are white. The mutated residues are purple. of 9 (Ϯ 0.5) h. In addition, both the Y279S and L282V mutants show even more severe growth defects, with doubling times of 13 (Ϯ 0.5) h.
Measurements of oxygen consumption of the yeast cells in mid-log growth phase (Fig. 3B) confirmed a decrease of Ͼ60% of the respiratory function for the Y279C, Y279S, and L282V mutated strains. To further investigate this effect, the cytochrome bc 1 complex was purified from the mitochondria of each strain to assess the kinetic parameters of the enzymes (Table  I). Again, the bc 1 complexes from the three mutants Y279C, Y279S, and L282V that showed significant defects in growth rates and oxygen consumption also showed activity losses of Ͼ90% compared with the wild-type strain. These activities were too low to permit accurate determination of K m and V m  values. The activity of the bc 1 complex from the I269M mutant was also low, although the growth rates and rates of oxygen consumption by this yeast mutant were similar to and only slightly less than those of the wild-type strain. This suggested that the bc 1 complex from the I269M mutant suffered some loss of activity during isolation.
Western blots indicated no loss of iron-sulfur protein, and optical spectra indicated no assembly defects in cytochromes b or c 1 that could account for these decreases of respiratory function. Thus the results clearly suggest that these three cytochrome b mutations associated with atovaquone resistance in Plasmodium impair the ubiquinol-binding properties of the yeast bc 1 complex. Possible implications of this effect for the growth of the parasite are discussed below.

Effect of Cytochrome b Mutations on Inhibition of the bc 1
Cytochrome Complex by Atovaquone-Ubiquinol-cytochrome c reductase activities of the purified bc 1 complexes from the wild-type and mutated yeast strains were measured in the presence of increasing concentrations of atovaquone to test whether sensitivity of the enzymes to the drug was altered by the cytochrome b mutations. We have shown previously that the bc 1 complex of wild-type yeast is sensitive to atovaquone with an inhibitor concentration of ϳ25 nM required for 50% inhibition of the activity of the purified enzyme (6). As shown in Fig. 4, all of the cytochrome b mutations that confer atovaquone resistance to the Plasmodium also altered the IC 50 for inhibition of the yeast bc 1 complex and by various amounts. The I269M and F278I mutations had the smallest effect, but For clarity, only atovaquone (green), wild-type residues (white), mutated residues (yellow), important aromatic interacting residues (cyan), His-181 from the iron-sulfur protein (orange), and the affected residues Tyr-279 and Arg-283 are shown. D, views of atovaquone binding in the bc 1 complexes of wild-type yeast (D1) and the Y279S or Y279C mutants (D2). The figure was generated using GRASP software. The chlorine on atovaquone is yellow, oxygens are red, and hydrogens are white. The yellow surface corresponds to the Rieske iron-sulfur protein. The white surfaces were generated by hydrophobic and aromatic residues of the cytochrome b, and the red surfaces were generated by hydrophilic and nucleophilic side chains. even these conferred a 5-10-fold increase in resistance, with IC 50 values between 150 -250 nM. The Y279C, Y279S, and L282V were the three cytochrome b mutations showing the highest level of resistance to the atovaquone, with IC 50 values between 1 and 2 M. This corresponded to a 40 -80-fold increase over the concentration required to inhibit the wild-type enzyme.
Structural Changes in Cytochrome b Caused by Mutations Conferring Resistance to Atovaquone-The five cytochrome b mutations responsible for atovaquone resistance in Plasmodium were built into the previously modeled atovaquone-bound ubiquinol oxidation pocket at the center P site of the yeast bc 1 complex (6), and the structural changes resulting from the mutations were modeled in silico. The isoleucine at position 269, located in a hydrophobic region of the binding pocket, interacted directly with the naphthoquinone ring of atovaquone. Modeling of the I269M mutation (Fig. 5A) showed a decrease of the volume available in this hydrophobic pocket, which interfered with binding of the inhibitor.
The phenylalanine at position 278 interacted directly with the chlorophenyl substituent of atovaquone. Examination of this region revealed the presence of an aromatic cluster composed of Phe-121, Phe-278, Tyr-359 and the chlorophenyl group of the atovaquone (Fig. 5B). The F278I mutation removed the central aromatic element of this cluster, destabilized this area of the pocket, and thus interfered with binding of the drug.
Tyrosine 279 is also contained in a stable aromatic cluster composed of His-181 of the Rieske protein and the naphthoquinone group of the ligand. The modeling results in Fig. 5D clearly show how the Y279C or Y279S mutations, which replace an aromatic group by strong nucleophilic side chains, decreased the hydrophobic interactions between the binding pocket and the atovaquone.
In the L282V mutant, two changes taking place simultaneously explain the decrease of atovaquone binding as shown in Fig. 5C. First, Leu-282 interacted slightly with His-181 of the Rieske protein subunit, and the L282V mutation modified this interaction and influenced the degree of freedom of His-181. Secondly, the reduction in size of the aliphatic side chain created an empty space that the modeling compensated for by adjusting the backbone geometry of residues 280 -285. Examination of this region showed that this adjustment suppressed a hydrogen bond between Arg-283 and Tyr-279 (Fig. 5C), which caused a change in the orientation of Tyr-279 toward the binding pocket. Consequently, displacement of the Tyr-279 residue caused an atovaquone-binding defect similar to the ones observed for the Tyr-279 mutants and contributed to the effect of the His-181 positioning.
Calculation of Changes in Atovaquone-binding Energy Resulting from Mutations in Cytochrome b-The energy required for binding atovaquone was calculated with each of the modeled structures. The calculated changes in binding energy were then compared with the experimentally measured changes in IC 50 values. The changes in atovaquone-binding energy with the bc 1 complexes from the mutants relative to the binding energy obtained for the wild-type enzyme and the changes in IC 50 values for the mutant enzymes compared with the wildtype enzyme are shown in Fig. 6. The relative increase in calculated binding energy correlated well with the relative increase in IC 50 values for the bc 1 complexes for each mutation, with the exception of L282V. This mutation, which displayed the largest change among the IC 50 values, did not yield a comparably large increase in calculated binding energy. Possible reasons for this difference are discussed below. DISCUSSION Only a decade after the introduction of the anti-malarial drug atovaquone, new strains of P. falciparum that are resistant to the therapy have emerged worldwide. In this work, we used the budding yeast S. cerevisiae as a surrogate to recapitulate and understand the molecular basis of the atovaquone resistance found in the malaria parasites. It is well known that atovaquone suppresses the respiratory function of the parasite by inhibiting the cytochrome bc 1 complex (26). The drug is also a potent inhibitor of the yeast bc 1 complex (6), and the binding sites of the enzyme in the two microorganisms show a high degree of identity.
We have previously elucidated the mechanism of atovaquone inhibition of the bc 1 complex and developed a molecular model of atovaquone interaction with the ubiquinol oxidation pocket of the enzyme that accounts for the effects of the drug on the isolated enzyme (6). The strong and highly specific binding between the bc 1 complex and atovaquone results from a combination of aromatic interactions and several hydrogen bonds, some of which mimic the binding determinants of the natural substrate, ubiquinol. Atovaquone is thus a competitive inhibitor of the enzyme.
Most of the atovaquone-resistant malaria strains contain point mutations in the mitochondria-encoded cytochrome b gene. We were able to transfer these mutations into the yeast mitochondrial genome using a biolistic transformation procedure. Some of the resulting mutants were unable to grow on non-fermentable carbon sources, and thus it was not possible to demonstrate atovaquone resistance in vivo. However, the bc 1 complexes purified from the mutated strains exhibited large decreases in sensitivity toward atovaquone. We were thus able to biochemically confer the genetic evidence that linked these mutations to atovaquone resistance.
A previous study done on Plasmodium (9) demonstrates that these mutations can be ranked by the effects of atovaquone on the collapse of the mitochondrial membrane potential of the parasite. This ranking resulted in moderate atovaquone resistance for I269M and F278I and greater resistance for Y279C and L282V. Notably, these mutations transferred into S. cerevisiae produced the same ranking order, demonstrating that the yeast bc 1 complex is a valuable model for atovaquone resistance.
In a recent study of mutations conferring atovaquone resistance in pathogenic Pneumocystis fungi, we show that one-half of the mutations are distal from the drug-binding pocket, and these are able to remotely modify the drug-binding properties of the bc 1 complex (20). Interestingly, all of the Plasmodium mutations that we duplicated in yeast in the present study affected residues located in the binding pocket and interacting directly with the atovaquone. The molecular modeling simulations were able to describe the effects of each point mutation on the geometry of the atovaquone-binding pocket at the molecular level.
The molecular modeling revealed that the I269M mutation, which includes the insertion of a bulkier residue, is a typical example of resistance acquired by increasing steric constraints on the inhibitor. Several atovaquone-resistant mutations (I147V, L150F, and L275F) found in Pneumocystis have been shown to have analogous steric effects (20).
The three mutations F278I, Y279S, and Y279C removed an aromatic residue localized near the atovaquone and suppressed important interactions between the drug and the binding pocket. Likewise, molecular modeling of the L282V mutation showed that the mutation caused the rotation of Tyr-279 and, consequently, a variation of aromatic interaction with the atovaquone. Tight binding to the same locus using a very similar aromatic network has been described for the fungicide famoxadone, another inhibitor of the bc 1 complex (27). Interestingly, Plasmodium, Saccharomyces, and Pneumocystis, which have atovaquone-sensitive bc 1 complexes, all have a phenylalanine in position 278 of cytochrome b, whereas the bovine and human cytochrome bc 1 complexes, which are relatively insensitive to atovaquone, have an alanine at the equivalent cytochrome b position. The 278 locus thus appears to be important for the differential efficacy of the drug, similar to the position 275 (6).
The importance of Tyr-279 in cytochrome b for bc 1 activity is supported by its conservation among species. It has been postulated that the aromatic side chain of this tyrosine participates in the positioning of ubiquinol, the natural substrate of the enzyme (24). In Rhodobacter sphaeroides (28), mutation of the homologous tyrosine has no effect on the activity when replaced by phenylalanine, a small effect (3-fold decrease) when replaced by leucine, and a dramatic effect (20-fold decrease) when replaced with a glycine. In yeast (29), mutation of the same tyrosine into alanine causes an ϳ5-fold decrease in bc 1 complex activity. Therefore, the 279 position appears to require an aromatic side chain, or at least a large hydrophobic residue, to maintain a functional bc 1 complex. Furthermore, when this position is occupied by a nucleophilic residue such as cysteine (Y279C) or serine (Y279S), the activity is severely compromised. The sulfur atom of the cysteine being less electronegative than the oxygen atom of the serine would account for the differences between the two mutations. The Y279C mutation has also been observed clinically in a human bc 1 complex, in which case the mutation was associated with severe exercise intolerance, deafness, and mental retardation (29,30).
The comparison of experimentally measured changes in IC 50 values for inhibition by atovaquone and theoretically calculated changes in binding energies provides a useful check on the validity of the molecular modeling. With one exception, there was generally good agreement between these two parameters. The one outlier, the L282V mutation, exhibited a relatively greater change in IC 50 than would be predicted from the calculated change in binding energy. This may reflect the fact that IC 50 is a free energy (⌬G) term, which includes an entropic parameter, whereas the calculated binding energy is solely enthalpic (⌬H). Alternatively, the divergence for this mutation may be because of a failure of the molecular mechanics algorithms to accurately evaluate the energy variation associated with the rotation of an element of a stable aromatic cluster. It is extremely challenging and costly to establish malaria cultures from patient blood (and probably impossible to do so in third world country facilities), so it is difficult to prove that a specific novel mutation of the cytochrome b gene is responsible for a specific patient failing treatment with Malarone®. The yeast model makes it possible to demonstrate and quantify which mutations confer atovaquone resistance and which are just coincidental allele variations. Two of the mutations, Y279S and L282V, severely impaired the ability of the yeast strains to grow on a non-fermentable carbon source. It has been observed that some atovaquoneresistant mutations cause changes in cytochrome b, leading to a loss of fitness in the malaria parasite (31). This is frequently observed in mutations that confer drug resistance to microorganisms by alteration of structurally important regions of the target molecules (32,33). In the case of atovaquone, this loss of fitness results from mutations close to the active site that reduce the binding of the ubiquinol substrate by the enzyme. There is no report that the widespread Y279S mutation causes a loss of fitness in P. falciparum, but our yeast model strongly predicts it. The implication of this prediction is important for the future development and use of the drug Malarone®, because it suggests that the geographic spreading of a single locus change such as the Y279S mutation will be stopped by the removal of the drug. As the resistant strain has to compete against a more fit, wild-type parasite, the reversal of atovaquone resistance will happen spontaneously.