Logo of aacPermissionsJournals.ASM.orgJournalAAC ArticleJournal InfoAuthorsReviewers
Antimicrob Agents Chemother. Aug 2000; 44(8): 2100–2108.

Mutations in Plasmodium falciparum Cytochrome b That Are Associated with Atovaquone Resistance Are Located at a Putative Drug-Binding Site


Atovaquone is the major active component of the new antimalarial drug Malarone. Considerable evidence suggests that malaria parasites become resistant to atovaquone quickly if atovaquone is used as a sole agent. The mechanism by which the parasite develops resistance to atovaquone is not yet fully understood. Atovaquone has been shown to inhibit the cytochrome bc1 (CYT bc1) complex of the electron transport chain of malaria parasites. Here we report point mutations in Plasmodium falciparum CYT b that are associated with atovaquone resistance. Single or double amino acid mutations were detected from parasites that originated from a cloned line and survived various concentrations of atovaquone in vitro. A single amino acid mutation was detected in parasites isolated from a recrudescent patient following atovaquone treatment. These mutations are associated with a 25- to 9,354-fold range reduction in parasite susceptibility to atovaquone. Molecular modeling showed that amino acid mutations associated with atovaquone resistance are clustered around a putative atovaquone-binding site. Mutations in these positions are consistent with a reduced binding affinity of atovaquone for malaria parasite CYT b.

The widespread resistance of malaria parasites to standard antimalarial drugs is a serious global health problem. The urgent need for new antimalarial drugs has led to the development of atovaquone (566C80) which, combined with proguanil, has been licensed as Malarone. There is some concern that parasites may develop resistance to Malarone. In one study, 33% of patients treated with atovaquone alone experienced a recrudescence of parasitemia after treatment. These parasites tolerated up to 1,000-fold higher concentrations of atovaquone than did the pretreated parasites (16). Atovaquone-resistant parasites have been readily selected in vitro. Up to 1 in 105 parasites became resistant to the drug after having been cultured in the presence of 10−8 M atovaquone for 5 weeks (21, 23).

Atovaquone has potent blood schizonticidal activity and is also effective against the preerythrocytic (2, 4, 5) and sexual stages (8, 9) of the malaria parasite. It acts by inhibiting mitochondrial electron transport (10) and collapsing mitochondrial membrane potential (25). From these observations and on the basis of its structural similarity to ubiquinol, it has been postulated that atovaquone binds to parasite cytochrome b (CYT b) (31). The inhibitors stigmatellin and 5-n-undecyl-4,7-dioxobenzoxythiazol (UHDBT), which are structurally similar to atovaquone, have been shown to bind at the ubihydroquinone (Qo) site of CYT b and inhibit electron transport. Single point mutations within the Qo site confer resistance to these inhibitors in a variety of microorganisms (7). Two mutations in close proximity to the Qo site in Pneumocystis carinii are associated with atovaquone prophylaxis failure (33). Atovaquone-resistant Plasmodium yoelii lines have been derived from infected mice treated with suboptimal doses of atovaquone. All resistant lines have single or double amino acid mutations in their CYT b protein sequences, which are close to the predicted Qo site region. These mutations were associated with increased resistance to the collapsing of mitochondrial membrane potentials and the inhibition of respiration afforded by atovaquone (24). A potential drug-binding cavity has also been identified in the chicken CYT b. A similar study has been performed with Plasmodium berghei, and mutations in CYT b were again associated with atovaquone resistance (27). However, the three-dimensional structure of the Plasmodium falciparum CYT b and atovaquone-binding site has not been studied.

In this report, we describe mutations found in the CYT b genes of atovaquone-resistant P. falciparum and the resulting amino acid changes to the predicted atovaquone-binding site. The crystal structures of the chicken and bovine CYT bc1 complexes (13, 34) and recently found mutations in the parasite CYT b associated with atovaquone resistance (24, 27, 33) have made molecular modeling of the P. falciparum CYT b protein and prediction of the potential atovaquone binding site possible. Our modeling results demonstrate that the amino acids associated with atovaquone resistance in P. falciparum are clustered around the putative atovaquone binding site and that mutations in these positions are likely to reduce the binding affinity of atovaquone for P. falciparum CYT b.



P. falciparum TM93-C1088 was kindly provided by Dennis Kyle, Department of Immunology and Parasitology, U.S. Army Medical Component, Armed Forces Research Institute of Medical Sciences, Bangkok, Thailand. The parasite was isolated from a Thai patient who experienced a recrudescence of parasitemia after treatment with atovaquone (1,000 mg once a day for 3 days) and pyrimethamine (16). P. falciparum K1 is a laboratory line originally isolated from a Thai patient (28). K1-1D4 is a cloned line of K1 produced in the AMI laboratory. The origins and sources of HB3, 3D7, T9-94, MCK, D10, Palo Alto, Itg2, 7G8, W/L, NF36, MS2, AE7, FCQ50, and FCQ79 have been described elsewhere (15). Dd2 has been described by Oduola et al. (19).

In vitro cultivation of P. falciparum.

Parasites were maintained in vitro using conditions described by Trager and Jensen (30).

Cloning of the P. falciparum K1 line.

K1 parasites were cultured in vitro and cloned once by the method of limiting dilution. Briefly, the number of infected red blood cells in the culture stock was determined by multiplying the number of red blood cells by the percentage of parasitemia. The culture stock was then diluted with medium containing 5% (vol/vol) uninfected red blood cells to concentrations at which the number of infected red blood cells per 100 μl was 3, 1, or 0.3. A 100-μl volume of the diluted cultures was transferred to each well of 96-well plates and cultured for 4 to 5 weeks. When the cultures had reached a 1% parasitemia, cells from the positive wells were transferred to 24-well plates and subsequently to 10- and 50-ml culture flasks. K1-1D4, the parent clone for this study, originated from one positive well on the one-per-well plate where approximately 7% of the wells yielded detectable parasites.

In vitro development of atovaquone resistance in P. falciparum K11D4.

Two protocols were used to select atovaquone resistant parasites from the K1-1D4 cloned line of P. falciparum, a step-wise selection and a single-step selection. The procedures are detailed as follows.

(i) Stepwise selection.

Stepwise selection (Fig. (Fig.1)1) was carried out as follows. Parasites were initially cultured in one flask and split into four cultures when parasitemia reached approximately 2%. These four cultures were maintained for one cycle until the parasitemia reached approximately 3%. Parasitized red blood cells from one of the four flasks were cryopreserved as the atovaquone-sensitive parents (parent). Parasites in the remaining three flasks were cultured continuously in medium containing 2 × 10−8 M atovaquone, a concentration equivalent to the 90% inhibitory concentration (IC90) (2.2 × 10−8 M) for this cloned line of parasites. Culture medium was changed every second day. Cultures were split in a 1:1 ratio approximately every 10 days until the parasitemia reached 0.1%. The parasites that survived were then cultured in two flasks (A and B) and sequentially subjected to 5 × 10−8, 1.5 × 10−7, 5 × 10−7, 1.5 × 10−6, and 1.5 × 10−5 M atovaquone. Aliquots of parasites were removed after surviving each drug concentration for cryopreservation and DNA preparation.

FIG. 1
Illustration of the procedures involved in the in vitro selection of atovaquone-resistant mutants.

(ii) One-step selection.

One-step selection was carried out as follows (Fig. (Fig.1).1). Parent parasites were split into nine flasks and were subjected to atovaquone treatment when parasitemias reached approximately 3%. Three sets of three flasks of parasites were cultured in the presence of 5 × 10−8, 1.5 × 10−7, and 5 × 10−7 M atovaquone. The three sets of cultures were maintained at these concentrations for 6 weeks. The surviving parasites were cryopreserved and extracted for DNA.

Cloning of atovaquone-resistant mutant K1-1D4 parasites.

Parasites that survived 2 × 10−8 M atovaquone initially and 5 × 10−8 M subsequently were cloned using methods described above.

In vitro drug susceptibility tests.

Two experiments were carried out using the following two protocols.

(i) The susceptibility of the K1 and TM93-C1088 lines of P. falciparum was determined for atovaquone (Jacobus Pharmaceutical Company, Inc., Princeton, N.J.), chloroquine (Sterling Pharmaceuticals, Sydney, Australia), cycloguanil (ICI Australia, Sydney, Australia) and WR99210 (Jacobus Pharmaceutical Company) using a modified radioisotopic method of Desjardins et al. (6). The K1 and TM93-C1088 lines were maintained in vitro as described earlier. A suspension of infected erythrocytes in RPMI 1640 (for chloroquine) or RPMI 1640 LPLF (for atovaquone, cycloguanil, and WR99210) (containing folic acid [0.01 mg/liter] and para-amicobenzoic acid [0.0005 mg/liter]) supplemented (10% serum) culture medium was added to 96-well microplates which were predosed with varying concentrations of drugs. The final culture suspension of the chloroquine plates had a hematocrit of 2% and contained 0.5 to 0.6% parasitized erythrocytes (>95% rings). [3H]hypoxanthine (10 μl containing 0.2 μCi) was added to each plate well at the start of the incubation, and the cultures were harvested 48 h later. The final culture suspension for the atovaquone, cycloguanil, and WR99210 plates had a hematocrit of 1.5% and contained 0.3 to 0.4% parasitized erythrocytes (>95% rings). These cultures were initially incubated for 48 h to allow merozoite reinvasion before [3H]hypoxanthine (10 μl containing 0.2 μCi) was added to each well. The cells were harvested after a further 24-h incubation. The level of incorporation of [3H]hypoxanthine was used as an index of inhibition of parasite growth. The IC50s and IC90s were determined by estimating the drug concentrations that inhibited parasite isotope incorporation by 50 and 90%, respectively, relative to the drug-free control cultures.

(ii) The parent clone and the selected mutant clones of K1-1D4 were tested for their susceptibilities to atovaquone, stigmatellin (Fluka Chemika-Biochemika, Buchs, Switzerland), and myxothiazol (ICN, Costa Mesa, Calif.). In this experiment, parasites were suspended in RPMI 1640 complete medium and plated in quadruplicate. [3H]hypoxanthine was added at the beginning of the test, and parasites were harvested 48 h later. Curves were best fitted to the quadruplicate readings using Sigmaplot. The IC50s and IC90s of the compounds for the cloned parasites and the standard errors were also calculated using Sigmaplot.

Amplification and sequencing of the P. falciparum CYT b gene.

Parasite DNA was isolated using methods described previously (1). A fragment of the CYT b gene (corresponding to nucleotides 12 to 950 of the gene) was amplified using primers cytb1 (5′ CTC TAT TAA TTT AGT TAA AGC ACA C 3′) and cytb2 (5′ ACA GAA TAA TCT CTA GCA CC 3′). PCR was performed using 2 mM Mg2+, a 0.2 mM concentration of each deoxynucleoside triphosphate (Promega, Madison, Wis.), 75 ng of each primer, and 1.25 U of AmpliTaqGold (PE Applied Biosystems). The reaction mixture was initially heated at 93°C for 10 min and then cycled at 93°C for 50 s, 45°C for 50 s, and 70°C for 1 min over 40 cycles. PCR products were purified using the QIAExII purification kit (QIAGEN, Hilden, Germany) and then sequenced using the ABI Prism Big-Dye Terminator kit. A second pair of primers, cytb3 (5′ AGC AGT AAT TTG GAT ATG TGG AGG 3′) and cytb4 (5′ ATT TTT AAT GCT GTA TCA TAC CCT 3′), were then used to sequence the amplified products from opposite orientations.

Amino acid comparison of the Qo site regions in CYT b from various species.

The sequences were retrieved from the GenBank or Swiss-Prot database and aligned using Pileup (Wisconsin package, version 8.1.0, March 1996 release). The origins of the sequences were as follows (accession numbers are shown in parentheses): horse (sp-P48665), rhinoceros (sp-Q96071), pig (sp-P24964), sheep (sp-P24959), bovine (sp-P00157), rabbit (sp-P34863), human (sp-P00156), chicken (sp-P18946), Drosophila yakuba (sp-P07704), Saccharomyces cerevisiae (gb-X84042), carrageen (sp-P48875), P. falciparum (gb-M99416), Plasmodium vivax (gb-AF055587), P. yoelii (gb-M29000), P. carinii (gb-AF074872), and Toxoplasma gondii (gb-AF015627).

Genetic typing for P. falciparum lines.

Two polymorphic genes, MSP1 and MSP2, were used as markers for genetic typing of P. falciparum isolates and cloned lines. A fragment from each gene was amplified from parasite lines using the protocols described by Ranford-Cartwright et al. (20). The fragment sizes of PCR products from each parasite line were analyzed.

Molecular modeling.

The P. falciparum CYT b model was constructed using the Homology software program of the Insight II molecular modeling system from Molecular Simulations Inc. (MSI). The amino acid sequences of chicken, bovine, and P. falciparum CYT b were aligned using Homology. This alignment was further refined using the predicted structurally conserved regions (SCRs) defined by PROMOTIF (12) of the chicken (1BCC [34]) and bovine (1BE3 [13]) CYT b crystal structures. Where the template and model did not have the same number of residues, suitable loops with appropriate conformations were assigned from protein structures found in the Brookhaven PDB database.

The model was refined so that splice points between SCRs and loops were fixed where omega bonds had a distance of 1.34 Å and an angle of 180 ± 1°. The MSI program Discover was then used to minimize the energy of the structure to an RMS deviation of less than 10−6 using steepest descents and conjugate gradients taking into account morse and cross terms and charges. Amino acid substitutions were made using Homology. The P. falciparum CYT b model was validated using Profiles 3D (within the Homology program) and Procheck (18).

Modeling of the antimalarial drug atovaquone into potential binding sites in the chicken, bovine, and P. falciparum CYT b structures was performed using the program for genetic optimization for ligand docking (GOLD) (14) from the Cambridge Crystallographic Data Centre. Potential atovaquone binding sites in the P. falciparum CYT b model were searched for on the cytoplasmic face of the protein. The binding predictions from GOLD were processed by the software package LUDI (Ligand Design) from MSI and given a score that was correlated to a theoretical dissociation constant. This constant was based on the number and quality of receptor-ligand hydrogen bonds, and the receptor-ligand hydrophobic contact area and was only used as a guide to assess potential binding sites. Minimization of the energy of the predicted P. falciparum- and chicken- and bovine atovaquone-docked complexes was performed using Discover. Contact residues were defined as those residues that resided within approximately 3.0 Å of atovaquone when bound in the putative site. Stigmatellin was positioned into the P. falciparum model based on a superimposition of the P. falciparum and chicken (stigmatellin-bound) CYT b structures.


The P. falciparum isolate TM93-C1088 has a marked reduction in susceptibility to atovaquone and a mutation in its CYT b gene.

The P. falciparum TM93-C1088 line was isolated from a Thai patient who experienced recrudescence following atovaquone treatment. The K1 line was also isolated from a Thai patient; however, this patient had not been exposed to atovaquone. When measured in vitro, both lines showed a reduced susceptibility to chloroquine, with IC50s of 244.23 ± 21.93 nM and 169.28 ± 8.74 nM, respectively, and to cycloguanil, with IC50s of 679.38 ± 16.86 nM and 229.11 ± 113.38 nM, respectively, but were sensitive to WR99210, with IC50s of 0.20 ± 0.02 nM and 0.13 ± 0.04 nM, respectively. Their susceptibilities to atovaquone differed markedly. The K1 line was susceptible to low concentrations of atovaquone, with the IC50s and IC90s being below 13.6 nM. The TM93-C1088 line was less susceptible to atovaquone, with mean IC50s and IC90s of 9,974 and 22,717 nM, respectively. These values were at least 730- and 1,600-fold higher, respectively, than those of the K1 line.

The CYT b gene from TM93-C1088 was sequenced and compared to the CYT b sequences of 17 laboratory P. falciparum lines, including K1. These 17 lines had not been exposed to atovaquone. Based on the PCR-amplified polymorphic markers, MSP1 and MSP2, each parasite line had a unique genotype (data not shown). Sequencing of a 939-bp cyt b fragment (85% of the gene) from these 17 laboratory lines revealed a high degree of conservation despite differences in the lines origins and other marker genes. The only synonymous nucleotide substitution found in the 17 lines of CYT b genes was in 7G8. Sequencing of the CYT b gene from TM93-C1088 revealed a unique single base change, A to C at nucleotide 803 (sequence numbers starting from the start codon ATG). This resulted in an amino acid change from Y(TAT)268 to S(TCT)268.

In vitro development of atovaquone resistance in the P. falciparum K1 cloned line is also associated with mutations in their CYT b genes.

The ATV-M1 parasites were selected after continuous exposure of the cloned parent parasites to 2 × 10−8 M atovaquone. A single base mutation was found in the CYT b gene of the ATV-M1 parasites from each of the three cultures. The mutation G to A at nucleotide 399 resulted in an amino acid change from M(ATG)133 to I(ATA)133.

The ATV-M1 parasites were then cultured in two flasks containing 5 × 10−8 M atovaquone. Sequencing of the CYT b gene from parasites that survived this concentration of atovaquone revealed an M133I mutation with mixed signals at several other positions, indicating multiple parasite populations. The parasites were then cloned using a limiting-dilution method to yield a total of 20 individual clones from flask A and 26 clones from flask B. CYT b genes were amplified and sequenced from these cloned parasites. In flask A, 3 clones had the single mutation of M133I (ATV-M1) and 17 had the double mutation M133I and P(CCA)275T(ACA)(ATV-M2). In flask B, 6 clones had the double mutation M133I and K(AAA)272R(AGA)(ATV-M3) and 20 had the double mutation M133I and G(GGT)280D(GAT)(ATV-M4).

ATV-M2 and ATV-M3 tolerated 1.5 × 10−7 M atovaquone but failed to grow when the atovaquone concentration was increased to 5 × 10−7 M. ATV-M4 parasites grew well in 1.5 × 10−6 M atovaquone but failed to grow in 1.5 × 10−5 M atovaquone. Figure Figure11 illustrates the procedures and results.

ATV-M5 was selected from one of the three cultures in which the parent parasites were subjected to 5 × 10−8 M atovaquone. The cyt b sequence of this line had three adjacent mutations that resulted in changes of two adjacent amino acids: L(ATT)283I(ATA) and V(GTA)284K(AAA). None of the three cultures that were subjected to 1.5 × 10−7 or 5 × 10−7 M atovaquone grew resistant mutants.

The susceptibilities to atovaquone of the mutant parasite clones ATV-M1 to ATV-M5 were measured and compared to those of the parent clone and the TM93-C1088 isolate (Table (Table1).1). All mutant parasites showed a significant increase in IC50s (Table (Table1)1) and IC90s (data not shown) of atovaquone. The TM93-C1088 isolate (Y268S) showed a 9,354-fold-higher IC50 than the K1 parent line. Among the K1 mutants, ATV-M4 (M133I-G280D), which survived the highest test concentration of atovaquone (1,500 nM) in culture, had an 897-fold increase in its IC50. This was followed by ATV-M3 (M133I-K272R), with a 537-fold increase; ATV-M2 (M133I-P275T), with a 220-fold increase; and ATV-M1 (M133I alone), with a 25-fold increase (Table (Table1).1). ATV-M5 (L283I-V284K) had a 76-fold increase in IC50 (Table (Table1).1). The IC50s of the mutants correlate well with the concentration of atovaquone at which they were originally selected. The mutations have been stable ever since the mutant parasites were cultured in drug-free medium for 3 months.

Susceptibilities of parasites to various drugs

By contrast, the susceptibilities of the K1 mutants to chloroquine were not markedly different enough from that of the parent line to assume any correlation with the mutations (Table (Table1).1). The mutants had the same MSP1 and MSP2 genotypes as the parent line, indicating that the mutants did not result from cross-contamination (data not shown).

Most of the mutations occur in amino acids that are highly conserved in eukaryotic CYT b.

Protein sequences of CYT b from 16 different eukaryotic species were aligned and compared (data not shown). Of the six mutations, four occurred in amino acids that are highly conserved in eukaryotic CYT b. Y268, M133, and G280 were conserved across all 16 species, and P275 was conserved among 15 species and replaced by an S in P. carinii, indicating the potential importance of these residues in maintaining CYT b structure and function. Two of the six mutations occurred in residues that are unique to primitive eukaryotes. K272 was seen in Plasmodium and Toxoplasma, whereas V284 was unique to the sequenced Plasmodium spp.

P. falciparum CYT b model.

A homology model of P. falciparum CYT b was constructed based on the deduced crystal structures 1BCC and 1BE3. The amino acid sequence alignment and α-helical elements of the three proteins are shown in Fig. Fig.2.2. The SCRs predicted in the P. falciparum CYT b model were comparable to those reported in the chicken and bovine crystal structures with only minor prediction discrepancies between three- and four-residue helices. A Ramachandran plot revealed that the model was of a high quality, with 87% of the residues being in the most-favored regions (data not shown). Profiles 3D also yielded a high result, with the model scoring 86 out of a possible 169. This was comparable to the chicken crystal structure, which scored 97 out of a possible 173. This score of 97 was used as a theoretical maximum due to the program's not being able to account for amino acids in membrane bound regions of proteins.

FIG. 2
Amino acid sequence alignment of chicken, bovine, and P. falciparum CYT b with their SCR elements represented by bars on the top, second top, and bottom, respectively. Putative residues involved in atovaquone binding are shown in boldface type. Residues ...

Putative binding site for atovaquone in P. falciparum.

Potential binding sites for atovaquone were located on the cytoplasmic face of the P. falciparum CYT b model using GOLD. Most sites predicted by GOLD were given LUDI scores below 650. There was only one site predicted in the model (Fig. (Fig.3A)3A) that scored well above this, obtaining a value of 950. This site coincided with a region where ubiquinol has been predicted to bind (13, 31, 34) and is the most likely site where atovaquone binds as judged from the LUDI scores of the predicted sites. Also, the equivalent region in the chicken CYT b structure has been found to support stigmatellin and myxothiazol binding (3). The putative contact residues at this atovaquone-binding site after energy minimization of the atovaquone-CYT b complex in P. falciparum were I119, F123, Y126, M133, V140, I141, L144, I258, P260, F264, F267, Y268, L271, V284, L285, and L288 (Fig. (Fig.22 and and3A).3A).

FIG. 3FIG. 3
(A) Putative P. falciparum CYT b contact residues involved in atovaquone binding and the residues that were mutated in the atovaquone-resistant parasites. Atovaquone is represented in green stick form, with its lone electron pairs shown in light blue. ...

Mutations associated with atovaquone resistance.

The mutations Y268S, Y268C, M133I, and V284K (Fig. (Fig.3B)3B) in Plasmodium CYT b that were associated with atovaquone resistance occurred at putative contact residues in the CYT b model (Fig. (Fig.3A3A and B). The mutations G280D, K272R, and P275T did not occur at any of the predicted contact residues. However, they were all found in tandem with the mutation at the putative contact site residue M133. Residues Y268, M133, and V284 are all within hydrophobic or H-bond interaction distance (approximately 2 to 3Å) of atovaquone when atovaquone was modeled into the putative binding site.

The mutations I258M, F267I, Y268C, and L271V (Fig. (Fig.3B)3B) associated with atovaquone resistance in P. yoelii CYT b also occurred at the putative contact residues in the P. falciparum CYT b model (Fig. (Fig.3A).3A). The mutation K272R did not occur at a predicted contact residue but was again found to coexist with the contact residue mutation L271V. The mutation L144F or L144S (Fig. (Fig.3B)3B) found in P. carinii and P. berghei CYT b, respectively (27, 33), also occurred at a contact residue. T142I was the only lone mutation not directly associated with a contact residue.

Equivalent putative binding site for atovaquone in the chicken and bovine CYT b structures.

The equivalent atovaquone binding site in bovine and chicken CYT b (Fig. (Fig.3C)3C) with atovaquone modelled into them after energy minimisation revealed a significant difference of contact residues from those predicted for P. falciparum. The residues I119, F267, V284, L285 and L288 that form one side of the binding pocket in the P. falciparum binding site are represented by A278 and V299 in the chicken and bovine equivalent binding site. These residue differences are bolded in Fig. Fig.22 and compared between Fig. Fig.3A3A and C.

The atovaquone-binding site may overlap with the stigmatellin and myxothiazol binding site.

Since stigmatellin and myxothiazol are well-characterized inhibitors in the CYT b Qo site, we were interested in whether the mutations in CYT b that are associated with atovaquone resistance also affect stigmatellin and myxothiazol binding. The parent parasites were less susceptible to stigmatellin and myxothiazol than to atovaquone by 530- and 66-fold, respectively, as judged by the IC50s (Table (Table1).1). In the five atovaquone-resistant mutant lines, four had significantly higher IC50s of stigmatellin compared with the parent line, with a 2.7-fold maximum difference. However, ATV-M5 showed a 664-fold decrease in IC50 compared to the parent. The situation was more complicated for myxothiazol. The ATV-M1, ATV-M3, and TM93-C1088 lines had significantly higher IC50s than the parent line by 2.56-, 1.38-, and 10.58-fold, respectively, while ATV-M2, ATV-M4, and ATV-M5 lines had significantly lower IC50s by 1.35-, 2.7-, and 59-fold, respectively. These results suggest that there may possibly be an overlap between the binding sites of atovaquone, stigmatellin, and myxothiazol in P. falciparum CYT b. This argument was supported by the P. falciparum CYT b model, in which the equivalent stigmatellin-binding site overlapped with part of the atovaquone-binding site (data not shown). The V284K mutation appeared to enhance both stigmatellin and myxothiazol binding to parasite CYT b, while the Y268S and G280D mutations appeared to only enhance myxothiazol binding to CYT b. The M133I mutation in contrast appeared to reduce the binding of both compounds to CYT b.

Atovaquone is potentially active against other protozoans.

P. carinii (33), T. gondii (29) and Eimeria tenella (11) have already been shown to be susceptible to atovaquone. CYT b amino acid sequences for P. carinii and T. gondii were retrieved from GenBank, and their residues share similar properties with the putative atovaquone contact binding site residues from P. falciparum (data not shown).


Mutations observed in atovaquone resistant parasite CYT b resulted from drug selection.

Atovaquone is a potent and selective mitochondrial inhibitor of P. falciparum and P. yoelii, and its primary site of action is the CYT bc1 complex (10). Resistance to atovaquone has been associated with sequence changes in CYT b in P. carinii, P. yoelii, and P. berghei. In P. carinii, mutations resulting in amino acid changes were found in one of the Qo site regions in the CYT b gene of the parasites that were isolated from two of four patients who had failed atovaquone prophylaxis (33). However, the direct correlation between mutations and atovaquone resistance could not be further analyzed due to difficulties in culturing P. carinii. The direct evidence came from studies of P. yoelii, a rodent malaria species, whereby mutations in CYT b conferred atovaquone resistance. In that study, atovaquone-resistant parasites were derived from suboptimal treatment of P. yoelii-infected mice with atovaquone (5 mg/kg of body weight). Single base changes were found in the mitochondrial genome of resistant parasites and were located in the cyt b. These changes resulted in single or double amino acid mutations in the parasite CYT b. Parasites that carried mutations in their CYT b were able to respirate and maintain mitochondrial membrane potential in the presence of atovaquone. In contrast, both these functions in the nonmutant parent parasites were inhibited in the same concentrations of atovaquone (24).

In this work, we derived atovaquone-resistant P. falciparum mutants from an atovaquone-sensitive cloned line, K1-1D4, by culturing parasites in medium containing subcurative levels of atovaquone with gradually increased concentrations. A single base mutation was detected in the CYT b gene of the parasite (ATV-M1) that survived the initial 20 nM concentration of atovaquone, resulting in an amino acid change of M133I (in the predicted QoI region) and a 25-fold increase in the IC50 of atovaquone compared to its nonmutant parent. When this parasite was again subjected to higher concentrations of atovaquone, extra mutations were selected in the parasite CYT b genes. The resultant amino acid changes K272R, P275T, and G280D were also located close to the predicted Qo region. These changes were associated with significant increases in IC50s of 220-, 537-, and 897-fold, respectively, higher than that of the parent. These values were also significantly higher than that of the single M133I mutant. The M133I mutation was also seen in an atovaquone-resistant line of P. berghei (27). The two P. falciparum mutations, Y268S and K272R, were also seen in the P. yoelii atovaquone-resistant mutants with Y268C in P. yoelii (24). When parasites were subjected directly to 5 × 10−8 M atovaquone, a double mutation in the parasite CYT b protein sequence was selected: L283I and V284K. The V284K change was probably directly responsible for the 76-fold increase in IC50, because I283 has been observed in atovaquone-sensitive parasites. The V284 mutation has also been reported in an atovaquone-resistant line of P. berghei (15). Our sequencing of 17 field isolates that had not been exposed to atovaquone revealed only one synonymous change consistent with the strong evidence that CYT b genes in Plasmodium are highly conserved (17). Therefore, it is highly likely that the amino acid changes we saw in the atovaquone-resistant parasites resulted directly from the presence of atovaquone.

Mutation rate.

The rate at which mutations occurred in the CYT b genes can be estimated from the starting number of parasites, the number of generations, and the number of mutants. As an example, the initial M133I mutation occurred in each of the three atovaquone treated cultures. This mutation could have occurred (i) in the original culture and was partitioned to each of the three cultures or (ii) simultaneously in the three individual cultures. If the mutation occurred before the culture was split, then the mutation must have occurred at least one generation prior to splitting in order for each of the three subcultures to contain at least one mutant parasite. At that time, the flask contained approximately 5 × 107 parasites, so to have at least a 10% chance of the mutation occurring, the overall mutation rate must have been at least 1 in 5 × 108 parasites per generation. If the mutation occurred after splitting, then the mutation must have occurred at least once in each of the three flasks at the time the drug was added. At this stage there were approximately 6 × 108 parasites per flask. Thus, to have at least a 10% chance of seeing at least one parasite per flask, the probability of one mutation occurring per flask was 0.54 (i.e., 1 − 0.13) which would give an overall mutation rate of 6 × 108/0.54 or approximately 1 in 108 parasites per generation. However, if the mutation rate was high enough to see it after splitting, then the rate would have certainly been high enough for the mutation to have occurred before splitting. Therefore, the simplest explanation to fit these observations is that the M133I mutation occurred prior to splitting the cultures.

These calculations only give a minimum mutation rate. The actual mutation could be much higher, for example similar to the one mutation in 105 parasites observed by Safwat et al. (23). If this is the case, then the mutation would almost certainly have occurred earlier during culture and the observed mutation in this experiment would reflect the first mutation leading to atovaquone resistance which chanced to occur following cloning. Three other mutations that occurred in tandem with M133I (K272R, P275T, and G280D) are most likely to have occurred after the initial M133I mutation because all mutants have the M133I mutation. This is also supported by the modeling results suggesting that K272R, P275T, and G280D are not predicted contact residues and that the combination of M133I and one of the other mutations is required to confer resistance.

In the one-step selection, higher concentrations of atovaquone were used and fewer mutants were obtained. The mutation rate is probably similar, but only the mutants that can survive the concentration were selected. Even at an overall mutation rate as low at one mutation per 5 × 108 parasites per generation, a patient with a parasitemia of 1,000/μl in 5 liters of blood, would have a minimum of 10 mutant parasites. Some of the mutants may confer a strong resistance to atovaquone that would result in a parasite recrudescence.

Subcurative dose of atovaquone can select mutants that would survive curative treatment.

The P. falciparum atovaquone-resistant mutants we report here were selected using subcurative levels of atovaquone. It has been reported that the maximum atovaquone levels in serum after the third dose of Malarone were 5.1 μg/ml or 13.9 μM in children (22) and 13.02 μg/ml or 35 μM in adults (32). The maximum concentration of atovaquone following prophylaxis with Malarone was 5.74 μg/ml or 15 μM (26). Since atovaquone has not been shown to concentrate in the cells, the serum level of atovaquone would be similar to that in the infected red blood cells. The IC90 for one of the mutants, ATV-M4, was 23 μM, which indicates that this parasite has the potential to break through Malarone prophylaxis and even survive the curative treatment regimen of Malarone in vivo, although it was selected at subcurative levels of atovaquone.

Mutations in CYT b affect drug binding.

In order to understand how these mutations affect atovaquone binding, a model for P. falciparum CYT b was constructed based on the deduced crystal structures of chicken and bovine CYT b. A potential binding site for atovaquone in P. falciparum CYT b was identified in the model. This putative binding site scored 300 points higher, with a Ki 3 orders of magnitude lower than any of the other potential atovaquone-binding sites. The equivalent site in bovine and chicken CYT b with bound atovaquone had a lower affinity compared with the site in P. falciparum. Residues I119, F267, V284, L285, and L288 in the P. falciparum binding site can therefore support tighter binding of atovaquone than A278 and V299 in the equivalent positions in the chicken and bovine binding sites. This is consistent with Fry and Pudney's result that atovaquone was 2,000-fold more active against P. falciparum mitochondria function than in rat liver (10).

In the P. falciparum model, three of the six P. falciparum mutations, four of the five P. yoelii mutations, one of the two P. carinii mutations, and all three of the P. berghei mutations were predicted to cause a decrease in binding affinity for atovaquone in the putative binding site. The P. falciparum Y268S and P. yoelli Y268C mutations resulted in a dramatically reduced hydrophobic contact with atovaquone as well as altering H-bond formation potential. This could explain the marked reduction of susceptibility to atovaquone by 9,354-fold and 1,667-fold, respectively. The V284K mutation added a steric and charge alteration into the hydrophobic end of the binding pocket. This was consistent with a 76-fold change in susceptibility to atovaquone.

In contrast, the M133I substitution resulted in a steric alteration at the binding site consistent with a 25-fold reduction in susceptibility to atovaquone. The mutations K272R, P275T, and G280D that occurred only in tandem with M133I most likely caused structural alterations coupled with the steric hindrance through the M133I change. The structural alterations may have enhanced the alteration by M133I but may have not caused enough change to confer resistance themselves.

The L144F mutation in P. carinii and P. berghei was predicted to cause a steric hindrance into the binding site due to the overlap of the phenyalanine ring with atovaquone. The T142I mutation in P. carinii can only be predicted to have caused an indirect structural disruption to the binding site through F123. Given that the amino acid sequence of P. falciparum was almost identical to P. carinii at this site, the two mutations seen in P. carinii may potentially occur in P. falciparum with a similar effect. The current data for atovaquone resistance supports the location of this putative binding site and the discovery of other mutations or the elucidation of an atovaquone-CYT bc1 crystal complex will provide a more detailed analysis.

Atovaquone is the main component of the new antimalarial drug, Malarone, which is now available in many countries for the treatment of malaria. Atovaquone is also being used for the treatment Pneumocystis, mainly in human immunodeficiency virus and AIDS patients. This information on the atovaquone-binding site may provide a guide for exploring the use of atovaquone in treating other opportunistic parasites. In view of its increasing use, it is important to understand the mechanism of resistance and to monitor the emergence of parasites resistant to this drug. The discovery of mutations associated with atovaquone resistance provides a molecular marker for monitoring the development and spread of drug-resistant parasites in the field.


We thank Alan Hudson and Michael Dooley for scientific discussions and Ross Brinkworth for his help in molecular modelling. We also thank Athony Kotecki for technical support.


Published with the permission of the Director General of Army Health Services.


1. Cheng Q, Lawrence G, Reed C, Stowers A, Ranford-Cartwright L, Creasey A, Saul A. Measurement of Plasmodium falciparum growth rate in vivo: a test of malaria vaccines. Am J Trop Med Hyg. 1997;57:495–500. [PubMed]
2. Chulay J D. Challenges in the development of antimalarial drugs with causal prophylactic activity. Trans R Soc Trop Med Hyg. 1998;92:577–579. [PubMed]
3. Crofts A R, Berry E A. Structure and function of the cytochrome bc1 complex of mitochondria and photosynthetic bacteria. Curr Opin Struct Biol. 1998;8:501–509. [PubMed]
4. Davis C S, Pudney M, Matthews P J, Sinden R E. The causal prophylactic activity of the novel hydroxynaphthoquinone 566C80 against Plasmodium berghei infections in rats. Acta Leidensia. 1989;58:115–128. [PubMed]
5. Davis C S, Pudney M, Nicholas J C, Sinden R E. The novel hydroxynaphthoquinone 566C80 inhibits the development of liver stages of Plasmodium berghei cultured in vitro. Parasitology. 1993;106:1–6. [PubMed]
6. Desjardins R E, Canfield C J, Haynes D, Chulay J D. Quantitative assessment of antimalarial activity in vitro by a semiautomated microdiluting technique. Antimicrob Agents Chemother. 1979;16:710–718. [PMC free article] [PubMed]
7. Di Rago J P, Coppee J Y, Colson A M. Molecular basis for resistance to myxothiazol, mucidin (strobilurin A), and stigmatellin. J Biol Chem. 1989;264:14543–14548. [PubMed]
8. Fleck S L, Pudney M, Sinden R E. The effect of atovaquone (566C80) on the maturation and viability of Plasmodium falciparum gametocytes in vitro. Trans R Soc Trop Med Hyg. 1996;90:309–312. [PubMed]
9. Fowler R E, Billingsley P F, Pudney M, Sinden R E. Inhibitory action of the antimalarial compound atovaquone (566C80) against Plasmodium berghei ANKA in the mosquito, Anopheles stephensi. Parasitology. 1994;108:383–388. [PubMed]
10. Fry M, Pudney M. Site of action of the antimalarial hydroxynaphthoquinone, 2-[trans-4-(4"chlorophenyl)cyclohexyl]-3-hydroxy-1,4,-naphthoquinone (566c80) Biochem Pharmacol. 1992;40:914–919. [PubMed]
11. Fry M, Hudson A T, Randall A W, Williams R B. Potent and selective hydroxynaphthroquinone inhibitors of mitochondrial electron transport in Eimeria tenella (Apicocomplexa:Coccidia) Biochem Pharmacol. 1984;33:2115–2122. [PubMed]
12. Hutchinson E G, Thornton J M. PROMOTIF-a program to identify and analyse structural motifs in proteins. Protein Sci. 1996;5:212–220. [PMC free article] [PubMed]
13. Iwata S, Lee J W, Okada K, Lee J K, Iwata M, Rasmussen B, Link T A, Ramaswamy S, Jap B K. Complete structure of the 11-subunit bovine mitochondrial cytochrome bc1 complex. Science. 1998;281:64–71. [PubMed]
14. Jones G, Willett P, Glen R C, Leach A R, Taylor R. Development and validation of a genetic algorithm for flexible docking. J Mol Biol. 1997;267:727–748. [PubMed]
15. Limpaiboon T, Shirley M W, Kemp D J, Saul A. 7H8/6, a multicopy DNA probe for distinguishing isolates of Plasmodium falciparum. Mol Biochem Parasitol. 1991;47:197–206. [PubMed]
16. Looareesuwan S, Viravan C, Webster H K, Kyle D E, Hutchinson D B, Canfield C J. Clinical studies of atovaquone, alone or in combination with other antimalarial drugs, for treatment of acute uncomplicated malaria in Thailand. Am J Trop Med Hyg. 1996;54:62–66. [PubMed]
17. McIntosh M T, Srivastava R, Vaidya A B. Divergent evolutionary constraints on mitochondrial and nuclear genomes of malaria parasites. Mol Biochem Parasitol. 1998;95:69–80. [PubMed]
18. Morris A L, MacArthur M W, Hutchinson E G, Thornton J M. Stereochemical quality of protein structure coordinates. Proteins. 1992;12:345–364. [PubMed]
19. Oduola A M, Milhous W K, Weatherly N F, Bowdre J H, Desjardins R E. Plasmodium falciparum: induction of resistance to mefloquine in cloned strains by continuous drug exposure in vitro. Exp Parasitol. 1988;67:354–360. [PubMed]
20. Ranford-Cartwright L C, Taylor J, Umasunthar T, Taylor L H, Babiker H A, Lell B, Schmidt-Ott J R, Lehman L G, Walliker D, Kremsner P G. Molecular analysis of recrudescent parasites in a Plasmodium falciparum drug efficacy trial in Gabon. Trans R Soc Trop Med Hyg. 1997;91:719–724. [PubMed]
21. Rathod P K, McErlean T, Lee P C. Variations in frequencies of drug resistance in Plasmodium falciparum. Proc Natl Acad Sci USA. 1997;94:9389–9393. [PMC free article] [PubMed]
22. Sabchareon A, Attanath P, Phanuaksook P, Chanthavanich P, Poonpanich Y, Mookmanee D, Chongsuphajaisiddhi T, Sadler B M, Hussein Z, Canfield C J, Hutchison D B A. Efficacy and pharmacokinetics of atovaquone and proguanil in children with multidrug-resistant Plasmodium falciparum malaria. Trans R Soc Trop Med Hyg. 1998;92:201–206. [PubMed]
23. Safwat G, Rathod P K. Frequency of drug resistance in Plasmodium falciparum: a nonsynergistic combination of 5-fluoroorotate and atovaquone suppresses in vitro resistance. Antimicrob Agents Chemother. 1996;40:914–919. [PMC free article] [PubMed]
24. Srivastava I, Morriaey J M, Darrouzet E, Daldal F, Vaidya A B. Resistance mutations reveal the atovaquone-binding domain of cytochrome b in malaria parasites. Mol Microbiol. 1999;33:704–711. [PubMed]
25. Srivastava I K, Rottenberg H, Vaidya A B. Atovaquone, a broad spectrum antiparasitic drug, collapses mitochondria membrane potential in a malaria parasite. J Biol Chem. 1997;272:3961–1966. [PubMed]
26. Sukwa T Y, Mulenga M, Chisdaka N, Roskell N S, Scott T R. A randomized, double-blind, placebo-controlled field trial to determine the efficacy of Malarone (atovaquone/proguanil) for the prophylaxis of malaria in Zambia. Am J Trop Med Hyg. 1999;60:521–525. [PubMed]
27. Syafruddin D, Siregar J E, Marzuki S. Mutations in the cytochrome b gene of Plasmodium berghei conferring resistance to atovaquone. Mol Biochem Parasitol. 1999;104:185–194. [PubMed]
28. Thaitong S, Beale G H. Resistance of ten Thai isolates of Plasmodium falciparum to chloroquine and pyrimethamine by in vitro tests. Trans R Soc Trop Med Hyg. 1981;75:271–273. [PubMed]
29. Torres R A, Weinberg W, Stansell J, et al. Atovaquone for salvage treatment and suppression of toxoplasmic encephalitis in patients with AIDS. Clin Infect Dis. 1997;24:422–429. [PubMed]
30. Trager W, Jensen J B. Human malaria parasites in continuous culture. Science. 1976;193:673–675. [PubMed]
31. Vaidya A B, Lashgari M S, Pologe L G, Morrisey J. Structural features of Plasmodium falciparum cytochrome b that may underlie susceptibility to 8-aminoquinolines and hydroxynaphthoquinones. Mol Biochem Parasitol. 1993;58:33–42. [PubMed]
32. Van Vugt M, Edstein M D, Proux S, Lay K, Ooh M, Looareesuman S, White N J, Nosten F. Absence of an interaction between artesunate and atovaquone-proguanil. Eur J Clin Pharmacol. 1999;55:469–474. [PubMed]
33. Walker D J, Wakefield A E, Dohn M N, Miller R F, Baughman R P, Hossler P A, Bartlett M S, Smith J W, Kazanjian P, Meshnick S R. Sequence polymorphisms in the Pneumocystis carinii cytochrome b gene and their association with atovaquone prophylaxis failure. J Infect Dis. 1998;178:1767–1775. [PubMed]
34. Zhang Z, Huang L, Shulmeisters V M, Chi Y I, Kim K K, Huang L W, Crofts A R, Berry E A, Kim S H. Electron transfer by domain movement in cytochrome bc1. Nature. 1998;392:677–684. [PubMed]

Articles from Antimicrobial Agents and Chemotherapy are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...