The MSPDBL2 Codon 591 Polymorphism Is Associated with Lumefantrine In Vitro Drug Responses in Plasmodium falciparum Isolates from Kilifi, Kenya

The mechanisms of drug resistance development in the Plasmodium falciparum parasite to lumefantrine (LUM), commonly used in combination with artemisinin, are still unclear. We assessed the polymorphisms of Pfmspdbl2 for associations with LUM activity in a Kenyan population. MSPDBL2 codon 591S was associated with reduced susceptibility to LUM (P = 0.04). The high frequency of Pfmspdbl2 codon 591S in Kenya may be driven by the widespread use of lumefantrine in artemisinin combination therapy (Coartem).

C hemotherapy is central to the treatment and control of Plasmodium falciparum malaria but faces the parasite's intrinsic ability to quickly develop resistance to antimalarials. The combination of artemisinin and lumefantrine (LUM) (Coartem) is the treatment of choice for uncomplicated malaria in much of Africa (1). However, parasites showing reduced LUM susceptibility have been reported in some countries (2)(3)(4)(5)(6). This has been associated with the selection of wild-type parasites at the P. falciparum chloroquine resistance transporter (crt) 76 locus (7,8) and with at least a 2-fold increase in the frequency of the multidrug resistance 1 (MDR1) 86N mutants following treatment (2,4,5,9). Recently, a single nucleotide polymorphism (SNP) in merozoite surface protein Duffy binding-like 2 (MSPDBL2) codon 591 (C591S) was shown to be associated with increased resistance to halofantrine, mefloquine, and LUM (10). Since LUM is now widely used in the treatment of malaria, it is important to understand the mechanisms of resistance to this drug. We therefore investigated the role of Pfmspdbl2 in the response to LUM in vitro using Kenyan isolates and chloroquine (CQ) as a reference drug. Pfmspdbl2 is a member of the MSP3 multigene family, including mspdbl1, msp3, and msp6 (11). The associated proteins are expressed simultaneously (11) and potentially interact with other proteins on the merozoite parasite membrane in the invasion of the erythrocyte (12,13). Thus, all 4 genes were included in our investigation.
Parasite genomic DNA was extracted using a QIAmp DNA blood minikit (Qiagen, United Kingdom). We amplified the fulllength Pfmspdbl1, Pfmspdbl2, Pfmsp3, and Pfmsp6 from 65 in vitro culture-adapted isolates with chemosensitivity data for CQ and LUM (7,14) using the primers and PCR cycling conditions described in Table S1 in the supplemental material. PCR products were sequenced using BigDye Terminator v3.1 chemistry (Applied Biosystems, United Kingdom), and the resultant sequences were assembled and edited using SeqMan and aligned using MegAlign (Lasergene 7; DNASTAR, Madison, WI).
Pfmsp3, Pfmsp6, Pfmspdbl1, and Pfmspdbl2 alleles were defined on the basis of their haplotype structure and associations with in vitro drug responses assessed using the Kruskal-Wallis test and the Bonferroni correction for multiple comparisons (15). Thus, P values of Ͻ0.001 remained significant. We also determined the me-dian 50% inhibitory concentrations (IC 50 s) and the 95% confidence intervals (CIs) for each SNP that showed a significant association (P Ͻ 0.05) with the drugs tested. Only SNPs with a Ͼ5% minor allele frequency were included, and all analyses were conducted using Stata v.11 (StataCorp, College Station, TX).
All msp3 and msp6 SNPs analyzed were in linkage disequilibrium, representing two previously defined alleles, K1 and 3D7 ( Fig. 1A and B, respectively) (16,17), therefore precluding individual SNP analysis. The activity of all the drugs tested did not differ in parasites harboring 3D7 or K1 alleles in the msp3 and msp6 genes ( Table 1). The SNPs of Pfmspdbl1 and Pfmspdbl2 were used to determine the actual loci within the haplotypes associated with changes in drug responses.
The MSPDBL1 haplotypes (n ϭ 36) showed an association with CQ (P Ͻ 0.01) and LUM (P ϭ 0.05) drug activity, while the haplotypes of MSPDBL2 (n ϭ 31) showed an association with LUM (P ϭ 0.03) ( Table 1). Twelve Pfmspdbl1 SNPs (Table 2; see  also Table S2 in the supplemental material) and 4 SNPs of Pfmsp-dbl2 (Table 3; see also Table S3 in the supplemental material) were associated with both CQ and LUM. Notably, Pfmspdbl2 SNP1783 (n ϭ 31) codes for codon 591 (since the SNP followed 3 indels, 12 bp long), of which parasites containing serine were associated with reduced susceptibility to LUM (IC 50 , 97.6 nM; 95% CI, 77.7 to 199.8 nM; P ϭ 0.04) ( Fig. 2; Table 3). Codon 591S was also found at a high frequency (68%) in our population, similar to findings in Senegal (80% frequency) (10). This association of Pfmspdbl2 codon 591S with reduced susceptibility to LUM in a different African population adds support to the findings of the study in Senegal and suggests that codon 591 may be a marker for the surveillance of LUM resistance. Importantly, though, codon 591 is not likely to be the causal variant conferring resistance to LUM. Van Tyne et al. (10) demonstrated that stable integrants containing PfMSPDBL2 C591 were more sensitive to mefloquine, halofan-trine, and LUM than those with the 591S parasite. The extensive use of LUM in Africa may be the major driving force favoring the high frequency of the 591S mutation.
The observed inverse drug relationship of 4 Pfmspdbl1 codons (  (8,18,19). This inverse relationship between drugs shown by SNPs of Pfmspdbl1 most likely occurs on a backdrop of the underlying inverse relationship driven by CQ and LUM, since they are drugs that have been widely used for malaria treatment. Consequently, LUM-artemisinin may select for wildtype crt, associated with CQ sensitivity (20) and implicated as a marker of LUM tolerance (2,7), suggesting that LUM is likely to confer resistance via a different mechanism.
Not surprisingly, none of the Pfmsp3, Pfmsp6, Pfmspdbl1, or Pfmspdbl2 SNPs were in linkage disequilibrium with the K76T crt locus (data not shown). Additionally, the MSP3 gene family contains multiple high-frequency polymorphisms, which would increase the probability of random associations with drug activity.    Furthermore, since Pfmspdbl2 has shown evidence of being under balancing selection and is likely to be under immune pressure (21), its role in immunity cannot be ignored. However, it remains to be determined if the C591S mutation can be used as a surveillance marker of LUM resistance in the field.