Analysis of Single-Nucleotide Polymorphisms in the crt-o and mdr1 Genes of Plasmodium vivax among Chloroquine-Resistant Isolates from the Brazilian Amazon Region

Abstract

Plasmodium vivax parasites with chloroquine resistance (CQR) are already circulating in the Brazilian Amazon. Complete single-nucleotide polymorphism (SNP) analyses of coding and noncoding sequences of the pvmdr1 and pvcrt-o genes revealed no associations with CQR, even if some mutations had not been randomly selected. In addition, striking differences in the topologies and numbers of SNPs in these transporter genes between P. vivax and P. falciparum reinforce the idea that mechanisms other than mutations may explain this virulent phenotype in P. vivax.

Plasmodium vivax is the most widely distributed human malaria parasite, causing approximately 80 to 300 million clinical cases of malaria each year (17). Numerous factors indicate that this burden will increase due to the emergence and spread of chloroquine-resistant parasites (3, 17).

More than 50% of the malaria cases in Latin America occur in Brazil, and P. vivax predominates as the causative agent (16, 21). Notably, failures of chloroquine treatment of P. vivax malaria in the Brazilian Amazon city of Manaus have been reported recently (1). The local confirmation of the presence of active P. vivax parasites resisting chloroquine at the proposed minimal effective concentration in plasma for sensitive strains is a public health concern deserving attention.

Point mutations in two digestive-vacuole membrane proteins of P. falciparum, the P. falciparum chloroquine resistance transporter (PfCRT) and multidrug resistance 1 protein (PfMDR1), have been associated with chloroquine resistance (CQR), albeit to different extents (2, 10). Orthologues of these proteins in P. vivax (P. vivax CRT-O [PvCRT-O] and PvMDR1) have been identified previously (6, 15, 18), and recently, pvmdr1 mutant alleles were suggested to be associated with both in vitro and in vivo CQR in Southeast Asia (6, 20).

Here, we report a single-nucleotide polymorphism (SNP) analysis of pvmdr1 and pvcrt-o genes in P. vivax isolates from chloroquine-treated patients with and without recrudescent disease in the Brazilian Amazon region. In addition to complete coding sequences, we analyzed sequences from 5′ flanking regions and introns.

Field isolates were collected during a 28-day in vivo chloroquine efficacy study conducted in the city of Manaus, Brazil (8). Plasmatic chloroquine levels in all volunteers were measured by high-performance liquid chromatography on day 3 to confirm adequate dosing and good absorption of the oral chloroquine intake (three doses of 10, 7.5, and 7.5 mg/kg of body weight in 150-mg tablet form at 24-h intervals). Clinical treatment failure was defined as the occurrence of a positive blood smear result (confirmed by PCR diagnostic analysis) on day 14, 21, or 28 and the presence of parasites in peripheral blood (collected on the same day as the positive blood smear) containing >10 ng/ml of chloroquine as determined by high-performance liquid chromatography (7). Measurements of chloroquine and its active metabolite desethylchloroquine in whole blood were not obtained, as plasma samples were collected and processed at and transported from remote field sites. The presence of drug-resistant isolates in plasma samples with a mean chloroquine concentration ± standard deviation of 356.6 ± 296.1 ng/ml, however, undoubtedly confirms CQR (4). Using these stringent criteria, we selected seven samples (four obtained prior to treatment [on day 0] from patients with nonrecrudescent disease and three obtained after treatment [on days 21 and 28] from other patients with recrudescent disease). Different sets of primers, PCR conditions, and algorithms were used to amplify coding and noncoding regions and to generate and analyze sequences (see the supplemental material).

Analyses of the complete coding sequences of the pvmdr1 gene demonstrated that these sequences contained 24 SNPs and a single conserved microsatellite sequence. Notably, 17 (73%) of the 24 SNPs detected were nonsynonymous; 11 were contained in predicted extracellular loops in the parasite digestive-vacuole cytosol, and 5 were present in transmembrane domains (TMDs) (Fig. 1; also see Table S2 in the supplemental material). Despite the high frequency of SNPs, however, none were found in ABC conserved motifs (see Table S3 in the supplemental material).

FIG. 1.

Predicted structures of and representative polymorphisms in PvMDR1 and PfMDR1. PvMDR1, like PfMDR1, has two hydrophobic homologous domains, each with six transmembrane α-helices, and a cytosolic domain harboring nucleotide-binding domain 1 (NBD₁) and NBD₂, each containing an ATP-binding site with characteristic Walker motifs A and B and the S signature (ATP-binding cassette) of these transporters. In this figure, boxes with dotted lines in the diagram of PvMDR1 delimit predicted NBD locations. In the PfMDR1 illustration, closed dots represent point mutations associated with CQR in P. falciparum. In the PvMDR1 illustration, open dots represent SNPs described previously (5, 6, 11, 18, 20), open triangles represent SNPs identified in this study, and patterned dots represent SNPs described in other studies (5, 6, 11, 18, 20).

To relate these polymorphisms to protein function, we mapped selective constraints throughout coding sequences by calculating a K_d/K_s (ω) ratio for each amino acid substitution (see Fig. S1 in the supplemental material). These results revealed ω values of ≥1, indicating higher accelerated rates of nonsynonymous substitutions than expected to result from chance at the sites of most SNPs (P < 0.00001). Indeed, an ω value of 1.80811 (P < 0.00001), evidencing positive selection, was calculated for nucleotide positions 2722 to 2727, which correspond to amino acids 907 and 908 in PvMDR1. Presently, however, it is difficult to ensure that this positive selection is advantageous for the CQR phenotype and not for another metabolic aspect(s) related to the function of the PvCRT-O and PvMDR1 transporters. Like other authors (5, 6, 11, 18), we did not find SNPs at homologous positions of PfMDR1. We did, however, find a polymorphism in PvMDR1 at amino acid position 89, which is located very near and in the same intravacuolar loop, between TMDs I and II, as an SNP at the corresponding position in PfMDR1 (amino acid position 86), which has partial correlation with the CQR phenotype (12, 20). In addition, SNP Y976F, proposed previously as an early marker of CQR (6, 20), was found in only one isolate from a patient with nonrecrudescent disease. It is thus clear that the value of these polymorphisms as markers of CQR in P. vivax needs to be further explored.

Complete sequencing of the 14 exons of pvcrt-o revealed the presence of one synonymous transition and five nonsynonymous substitutions (Fig. 2; see also Table S4 in the supplemental material). Interestingly, the lysine (K) insertion at amino acid 10 of PvCRT-O, highly prevalent in Thai isolates (20), was also found in a sample from a chloroquine-treated patient from Brazil with recrudescent disease. Moreover, the rate of nonsynonymous substitutions in TMD VII was significantly higher than expected to result from chance (P < 0.05) (see Fig. S2 in the supplemental material). Despite these differences, however, we were unable to find an association between chloroquine treatment failure and the SNPs detected. Similar results following SNP analyses of pvcrt-o in monkey-adapted strains and human isolates have been reported previously (5, 15, 20).

FIG. 2.

Predicted structures of and representative polymorphisms in PvCRT-O and PfCRT. Like PfCRT, PvCRT-O has 10 predicted transmembrane helices, with C- and N-terminal domains located in the parasite cytoplasm. In the diagram of PvCRT-O, open dots represent SNPs detected in chloroquine-resistant and chloroquine-sensitive samples by Nomura et al. (15), and open triangles represent (named) SNPs described in this work. Closed dots in the PfCRT illustration represent point mutations that have been strongly associated with CQR in P. falciparum.

The lack of mutations in coding sequences of pvmdr1 and pvcrt-o associated with treatment failure prompted us to look for these associations in introns and 5′ flanking sequences, as they influence gene expression in malaria (7, 13). Interestingly, 6 (46%) of 13 introns contained SNPs, nine of which were transitions and one of which was a transversion (see Table S5 in the supplemental material). The remaining introns contained conserved microsatellites (see Table S6 in the supplemental material). Yet differences in length at the microsatellite in intron 12 (see Fig. S3 in the supplemental material) among samples sharing the same evolutionary spatial and temporal distributions suggest that this microsatellite marker may be a good candidate for the study of this locus. Moreover, analyses of 5′ flanking sequences from pvmdr1 and pvcrt-o revealed a single SNP (A→G) at nucleotide position 566 upstream from the ATG start codon in pvcrt-o. This high degree of conservation is consistent with the results of another study which analyzed untranslated regions of pfmdr1 in six reference strains with different chloroquine phenotypes (14). Together, these results suggest the existence of functional constraints at these genome loci that may play an important role in gene regulation.

In conclusion, we have analyzed the complete coding and noncoding sequences of the pvmdr1 and pvcrt-o genes from Brazilian P. vivax isolates that fulfilled rigorous criteria for chloroquine sensitivity and resistance. Although we did not genotype these isolates to determine if CQR isolates represent clonal as opposed to complex populations, it is clear that CQR isolates are already circulating in the Brazilian Amazon basin. Therefore, these sequences can be used as a baseline for future prospective studies of drug resistance in this region. Our analysis revealed, however, that there was no correlation between CQR and pvmdr1 and pvcrt-o mutations in these specific isolates, even if some mutations had been not randomly selected. The striking differences in the topologies of SNPs in the MDR1 and CRT transporter genes between P. vivax and P. falciparum thus indicate that mechanisms other than mutations may be implicated in the appearance of CQR in these two human malaria parasites (likely candidates are gene amplification and changes in expression levels [9, 11, 19]) or that other genes are associated with this virulent phenotype in P. vivax.

Nucleotide sequence accession numbers.

Sequence data generated in this study were deposited in the National Center for Biotechnology Information (NCBI) database under accession numbers EU333967 to EU333979.