Increased sensitivity to the antimalarials mefloquine and artemisinin is conferred by mutations in the pfmdr1 gene of Plasmodium falciparum



The declining efficacy of chloroquine and pyrimethamine/sulphadoxine in the treatment of human malaria has led to the use of newer antimalarials such as mefloquine and artemisinin. Sequence polymorphisms in the pfmdr1 gene, the gene encoding the plasmodial homologue of mammalian multidrug resistance transporters, have previously been linked to resistance to chloroquine in some, but not all, studies. In this study, we have used a genetic cross between the strains HB3 and 3D7 to study inheritance of sensitivity to the structurally unrelated drugs mefloquine and artemisinin, and to several other antimalarials. We find a complete allelic association between the HB3-like pfmdr1 allele and increased sensitivity to these drugs in the progeny. Different pfmdr1 sequence polymorphisms in other unrelated lines were also associated with increased sensitivity to these drugs. Our results indicate that the pfmdr1 gene is an important determinant of susceptibility to antimalarials, which has major implications for the future development of resistance.


The emergence and spread of resistance to chloroquine and pyrimethamine/sulphadoxine in Plasmodium falciparum has led to the use of newer antimalarials, including the arylaminoalcohol drugs mefloquine and halofantrine and derivatives of the endoperoxide drug artemisinin. Sporadic clinical resistance to mefloquine and halofantrine has already been reported in Africa and a high prevalence of resistance to mefloquine has been demonstrated in parts of SE Asia (Nosten et al., 1991).

Malaria parasites are haploid during erythrocytic infection, the stage at which the arylaminoalcohol and endoperoxide drugs are active. There is evidence that parasite resistance to mefloquine involves amplification of the pfmdr1 gene, which encodes the P-glycoprotein homologue 1 (Pgh1) protein, the plasmodial homologue of mammalian multidrug resistance transporters (Barnes et al., 1992; Peel et al., 1993; Wilson et al., 1993; Cowman et al., 1994). Sequence polymorphisms in pfmdr1 have previously been associated with increased resistance to chloroquine in some (Foote et al., 1990; Basco et al., 1995; von Seidlein et al., 1997), but not all (Awad-el-Kariem et al., 1992, Wilson et al., 1993), studies. Heterologous expression of the pfmdr1 gene in mammalian cells led to increased susceptibility to chloroquine, which was abrogated by the introduction of mutations in the pfmdr1 gene found in natural isolates (van Es et al., 1994). Similarly pfmdr1 expression in yeast was able to restore a mating phenotype, which was lost by introduction of the same mutations (Volkman et al., 1995). The role of these polymorphisms in resistance to mefloquine and artemisinin remains to be determined.

The present study was stimulated by the observation that the parasite lines HB3 and 3D7 had different sensitivities to artemisinin and mefloquine, which was accompanied by sequence polymorphism in pfmdr1. A genetic cross between HB3 and 3D7, carried out by Walliker et al. (1987), was used to test the hypothesis that increased sensitivity to these drugs is associated with mutation of the pfmdr1 gene in the progeny. Genetic crosses in plasmodia have previously identified determinants of resistance to chloroquine and sulphadoxine (Su et al., 1997; Wang et al., 1997; Carlton et al., 1998). In the present study, a complete allelic association between the HB3-like pfmdr1 genotype and sensitivity to mefloquine, artemisinin, halofantrine, lumefantrine, artemether and arteflene in the progeny suggests that the pfmdr1 gene is indeed a determinant of resistance to these drugs, and increased sensitivity is conferred by sequence changes in the gene.


Drug sensitivity of parents

The parent 3D7 was significantly less sensitive than the other parent HB3 to the arylaminoalcohol drugs mefloquine, halofantrine and lumefantrine, and to the endoperoxide drugs artemisinin, arteflene and artemether (Table 1). No difference was noted in sensitivity to chloroquine, quinine or atovaquone. IC50 values with dihydroartemisinin were found to be highly variable, which may reflect poor stability of the drug (Bustos et al., 1994).

Table 1. Comparison of in vitro sensitivity of HB3 and 3D7, the parent clones of the genetic cross. IC50 values and 95% confidence intervals are shown.
 IC50 nM (95%CI)
IC50 nM (95%CI)

  • a

    . Significance level of Student's t-test comparing HB3 and 3D7 IC50 values.

  • b

    . Ratio of IC50 values of 3D7 and HB3.

  • NS, not significant.

CQ15.1 (12.7–17.5)15.7 (14.6–16.8)NS1.03
QUIN152 (110–194)126 (124–129)NS0.83
MEF24 (21.9–26.1)42.6 (39.6–45.6)< 10−61.78
HAL5.14 (4.56–5.72)10.8 (8.92–12.6)0.0012.09
QHS9.35 (8.5–10.2)22 (18.5–25.5)< 10−42.35
AFL45.2 (42.2–48.2)104 (85.5–123)< 10−32.3
DHA6.49 (1.93–11.1)5.27 (2.46–8.08)NS0.81
ARM4.63 (4.44–4.82)10.8 (8.8–12.8)0.0022.33
ATOV0.97 (0.67–1.27)1.33 (0.98–1.68)NS1.37
LMF34.6 (30.7–38.5)87 (62–112) 0.0012.51

Identification of progeny

Parents and progeny were typed at 10 different loci to confirm that they had unique recombinant genotypes. Twelve progeny with unique genotypes were chosen, of which six had HB3-like pfmdr1 polymorphisms and six had 3D7-like pfmdr1 polymorphisms (Table 2).

Table 2. Genotypes of the parent clones HB3 and 3D7 and progeny (x2, X4, X5, X6, X11, X30, X39, X40, X41, X50, X53, X67) of the genetic cross. Allelic status was assigned according to parental type (H, HB3-like genotype; D, 3D7-like genotype).
  1. ND, not determined; chr, chromosome.

9 eba175(size)7HDDDDDDDHHDHHH
10 glurp(size)10HDDHDHHDHHDHHH

Drug sensitivity of progeny

The progeny were placed into two groups depending on whether they had the HB3 or 3D7-like genotype at the pfmdr1 locus. There was a clear allelic association between possession of the HB3-like pfmdr1 genotype and increased sensitivity to mefloquine, halofantrine, lumefantrine, artemisinin, artemether and arteflene (Fig. 1). When ranked for IC50, progeny with the 3D7-like genotype had higher IC50 values than those with HB3-like genotypes. When IC50 values were compared with those of the parents, progeny with the HB3-like genotype had IC50 values significantly lower than those of the 3D7 parent, whereas the progeny with the 3D7-like genotype had IC50 values significantly higher than the HB3 parent.

Figure 1.

Associations between pfmdr1 allelic status in HB3, 3D7 and the progeny and sensitivity to the arylaminoalcohols mefloquine, halofantrine, and lumefantrine and to the endoperoxides artemisinin, arteflene and artemether. IC50 values in nM with 95% confidence intervals are shown. Order of clones from left to right: (HB3-like pfmdr1 genotype as open diamonds) HB3, X2, X39, X40, X41, X53, X67 (3D7-like pfmdr1 genotype as closed circles) 3D7, X4, X5, X6, X11, X30, X50.

All of the progeny had similar IC50 values for chloroquine (data not shown).

Modelling of transmembranes 3 and 11

The two variant amino acids between HB3 and 3D7 in the Pgh1 protein were predicted to lie in transmembrane (TM) domains − 184 in TM3 and 1042 in TM11 (Foote et al., 1989). In an alignment of transmembrane helix 11 of Pgh1, with the homologous segment in the mammalian MDR3 protein, we found that the 1042 residue is homologous to a residue where mutation reduced the ability of transfected mdr3 to confer multidrug resistance (Hanna et al., 1996). TM11 forms an amphipathic helix, with the side-chain of the 1042 residue falling on the hydrophilic side (Fig. 2). Modelling was carried out using the HyperChem release 3 software, with charge assignment and geometry optimization in MM+.

Figure 2.

Modelling of the alpha helical configuration of transmembrane 11 of Pgh1, the protein encoded by pfmdr1, showing the presence of distinct hydrophobic and hydrophilic faces. Amino acid changes S1034C and N1042D, previously linked to resistance with chloroquine (Foote et al., 1990), are located on the hydrophilic face.

In a similar alignment with TM3 it was found that residue 184 of Pgh1 is a conserved aromatic residue. TM3 does not form an amphipathic helix (data not shown).

Association in laboratory isolates

A range of laboratory isolates was tested for antimalarial sensitivity and typed for allelic status of pfmdr1. When comparing the laboratory lines with a pfmdr1 gene copy of 1 (erythrocytic stage parasites are haploid), an association was observed between increased sensitivity to the arylaminoalcohols mefloquine, halofantrine and lumefantrine and to the endoperoxides artemisinin, artemether and arteflene and the presence of the pfmdr1 tyr-86 allele (Fig. 3). The American parasite lines HB3 and 7G8, which have different pfmdr1 point polymorphisms (Foote et al., 1990), also exhibit increased sensitivity to these drugs. W2mef, which was selected for resistance to mefloquine in vitro (Oduola et al., 1988), is the only isolate with tyr-86 and elevated IC50 values, but it has an amplified pfmdr1 gene (Peel et al., 1994).

Figure 3.

Associations between pfmdr1 allelic status and sensitivity of unrelated laboratory lines to the arylaminoalcohols mefloquine, halofantrine, and lumefantrine and to the endoperoxides artemisinin, arteflene and artemether. IC50 values in nM with 95% confidence intervals are shown. Pfmdr1 status (gene-copy number and sequence polymorphism) is indicated for each parasite line.


A P. falciparum genetic cross between the clones HB3 and 3D7 was used to study inheritance of sensitivity to a number of antimalarials including the arylaminoalcohol drugs mefloquine, halofantrine and lumefantrine as well as the endoperoxide drugs artemisinin, arteflene and artemether. In vitro cross-resistance, suggestive of an mdr mechanism, has been noted between artemisinin and its derivatives and the arylaminoalcohols, including, mefloquine, halofantrine and lumefantrine (Basco and Le Bras, 1992; Gay et al., 1997; Pradines et al., 1998).

The association found between the HB3-like pfmdr1 allele and increased sensitivity to the arylaminoalcohol and endoperoxide drugs strongly suggests that this allele, or a closely linked determinant, confers sensitivity to these drugs. Considering the presence of progeny with both intermediate and extreme sensitivity values (Fig. 1), it appears that other loci may also contribute to sensitivity, although pfmdr1 is a necessary determinant. The arylaminoalcohols and the endoperoxides are structurally unrelated and, as the pfmdr1 gene is a homologue of mdr transporters, these results suggest that the pfmdr1 gene may indeed participate in a mdr-like phenotype. Phe-184 and/or asp-1042 may confer increased sensitivity by reducing the efficiency or by changing the substrate specificity of the pump. The association between pfmdr1 sequence polymorphism and increased sensitivity was also found with the laboratory isolates, suggesting a more generalized association.

TM11 of the mouse MDR3 protein has been shown to be important in recognition and transport of substrate. Two positions have been found to be very sensitive to mutation by alanine scanning − Y949 and F953 (Hanna et al., 1996). The former mutation is at a position homologous to the 1042-amino-acid polymorphism (asn to asp) of the Pgh1 protein. Modelling of the TM11 of Pgh1 indicates that even although not conserved in sequence, certain structural features are conserved between it and the TM11 of Mdr3 − both have a hydrophobic and hydrophilic side. The amino acid changes associated with sensitivity are on the hydrophilic face. Another position also reported to be associated with resistance to chloroquine, a change at amino acid 1034 from ser to cys, is on the same side. The importance of these mutations in pfmdr1 function is supported by the studies of van Es et al. (1994) and Volkman et al. (1995) in heterologous systems where the introduction of mutations at these positions in pfmdr1 abrogated its function. Pawagi et al. (1994) have concluded from modelling that MDR-mediated transport of hydrophobic drugs across membranes takes place by Π interactions with aromatic residues in the transmembrane alpha helices. However, Hanna et al. (1996) have suggested that a ‘transport path’ for drugs is formed by several helices coming together to form a pore. In agreement with the latter view, the side-chains of amino acids in Pgh1, which are associated with sensitivity to mefloquine and artemisinin, are located on the hydrophilic side of the amphipathic helix.

The amino acid change in TM3 at position 184 in Pgh1 of HB3 is from tyr to phe, that is a change to a more hydrophobic amino acid, which is consistent with it forming a more favourable interaction with an aromatic substrate. Yet, the considerable polymorphism at codon-184 that has been observed in P. falciparum field isolates (Foote et al., 1990; Basco et al., 1995) is difficult to explain. Kwan and Gros (1998) have functionally analysed the first intracellular loop and flanking TM including TM3 of mouse MDR1. An amino acid change at a position homologous to amino acid 184 in MDR3 (phe to leu), expressed in yeast, altered sensitivity to valinomycin and FK506, an antifungal agent, and reduced the frequency of yeast mating. The position is well conserved as an aromatic residue among diverse organisms (Fig. 3) and may therefore be a determinant of substrate specificity. Yet no association has ever been observed between polymorphism at codon-184 and variation in sensitivity to artemisinin, dihydroartemisinin, mefloquine, halofantrine, quinine or chloroquine. Previously it has been shown that amplification of the pfmdr1 gene is associated with resistance to mefloquine, halofantrine and artemisinin. It is likely that in areas where mefloquine is the main drug used, selection for an amplified pfmdr1 gene will occur.

Despite the fact that pfmdr1 mutations have previously been associated with chloroquine resistance (Foote et al., 1990; Basco et al., 1995), this was not the case in the genetic cross, both the parents HB3 and 3D7 and all of the progeny being sensitive to chloroquine. This may reflect the multigenic nature of chloroquine resistance, and the requirement of at least one other allele for resistance. In this respect it is important to note that both parents HB3 and 3D7 have alleles of the cg2 gene linked with sensitivity to chloroquine (Su et al., 1997).

In conclusion, sequence polymorphism in the pfmdr1 gene appears to be an important determinant of sensitivity to the arylaminoalcohol and endoperoxide drugs in the HB3 × 3D7 cross and the laboratory lines. These results have important implications for the development of methods to circumvent resistance to these antimalarial drugs and for the monitoring of the spread of resistance in the field.

Experimental procedures

Parasite lines

The parents and progeny of the P. falciparum HB3 × 3D7 genetic cross carried out by Walliker et al. (1987) have been described elsewhere (Fenton et al., 1985; Walliker et al., 1987; Vaidya et al., 1993; Kerr et al., 1994). The other laboratory isolates used in this study were K1, W2, T994, V1/S, T9/96, FC27, SL/D6, 3D7, HB3, 7G8 and W2mef (obtained from the Edinburgh repository).

Parasites were cultivated at 37°C in human erythrocytes (A+) in 25 ml narrow-neck flat-bottomed culture flasks (Corning) containing medium (RPMI 1640, 25 mM HEPES, 23 mM MNaHCO3, 0.2% glucose, 50 µg ml−1 gentamycin, pH 7.3) and 10% AB serum. The same pooled serum was used in all experiments. Cultures were maintained in a 3% O2, 4% CO2, 93% N2 atmosphere. Parasitaemia was calculated by counting at least 1000 erythrocytes using a ×9 graticule and haematocrit was read using a Neubauer haemocytometer, assuming that 5 × 106 cells µl−1 represented 50%.

Genotyping of parasites

DNA was extracted from cultures of 32 progeny clones. Twelve progeny clones that had undergone recombination and differed from one another were identified by typing at 10 different polymorphic loci using PCR-based methods. The genotypes of each of the progeny were confirmed once the parasites had been recovered into culture and reconfirmed following the drug tests.

Polymorphism at pfmdr1 codon-184 (tyr or phe) was detected by amplification with the primers A4 (5′-aaa gat ggt aac ctc agt atc aaa gaa gag-3′) and A2 (5′-gtc aaa cgt gca ttt ttt att aat gac cat tta-3′) followed by restriction digestion with DraI. Codon-1042 polymorphism (asn or asp) was detected by amplification with the primers 1034f (5′-aga att att gta aat gca gct tta tgg gga ctc-3′) and 1042r (5′-aat gga taa tat ttc tca aat gat aac tta gca-3′) followed by restriction digestion with VspI (Duraisingh et al., 1997). Size polymorphism in the omega repeat of the cg2 gene was detected by amplifying across the repeat region with the primers OmN1A (5′-gat gag gag gat gcc tgg ttt tac tgt ctt-3′) and OmN12A (5′-ttc ttg aat act cct ccc cac aca cct cac-3′). Sequence polymorphism in dhfr codon-108 (asn or ser) and dhps-437 (gly or ala) was determined as described previously (Duraisingh et al., 1998). The polymorphism in pfs48/45 codon-253 was determined by the method described in Drakeley et al. (1996). Size polymorphisms in the msp1, msp2 and csp genes were determined by the methods described in Wooden et al. (1992). The dimorphism in eba175 was determined by the method of Binks and Conway (1999).

Drug sensitivity testing

Parasite lines were tested against the following drugs: chloroquine, quinine, mefloquine, halofantrine, artemisinin, arteflene, dihydroartemisinin, artemether, atovaquone and lumefantrine, using the hypoxanthine-based method described in O'Neill et al. (1985). An inoculum of 0.5% parasitaemia and 2.5% haematocrit was always used. HB3 and 3D7 were tested alongside the progeny at least five times in duplicate with each drug. IC50 values for the progeny clones were compared against those of each parent using the Student's t-test. HB3 and 3D7 were also tested alongside the laboratory lines at least three times in duplicate with each drug.

Statistical analysis

Carlton et al. (1998), based on the method of Lander and Botstein (1989), have previously calculated the number of progeny clones required in a genetic cross if the closest genetic marker is 0 cM away from the real determinant. To achieve > 90% power (probability of detecting a real linkage), while using a significance threshold of 5%, the minimum number of clones needed was nine. We have therefore analysed 12 clones in this study.


We are grateful for funding from the Wellcome Trust and Novartis for M.T.D. and from the PHLS of the UK for D.C.W.