Prevalence of Plasmodium falciparum anti-malarial resistance-associated polymorphisms in pfcrt, pfmdr1 and pfnhe1 in Muheza, Tanzania, prior to introduction of artemisinin combination therapy

Background A report of the chloroquine and amodiaquine resistance pfcrt-SVMNT haplotype in Tanzania raises concern about high-level resistance to the artesunate-amodiaquine combination treatment widely employed in Africa. Mutations in the pfmdr1 multi-drug resistance gene may also be associated with resistance, and a highly polymorphic microsatellite (ms-4760) of the pfnhe1 gene involved in quinine susceptibility has not been surveyed in Tanzania. Methods A total of 234 samples collected between 2003 – 2006 from an observational birth cohort of young children in Muheza, Tanzania were analysed. In these children, 141 cases of P. falciparum infections were treated with AQ and 93 episodes were treated with QN. Haplotypes of pfcrt and pfmdr1 were determined by a Taqman assay, and ms-4760 repeats in pfnhe1 were assessed by nested PCR amplification and direct sequencing. Parasite population diversity was evaluated using microsatellite markers on five different chromosomes. Results The pfcrt-CVIET haplotype was present alone in 93.6% (219/234) of the samples over the study period; the wild-type chloroquine- and amodiaquine-sensitive haplotype pfcrt-CVMNK was present in 4.3% (10/234) of the samples; and both haplotypes were present in 2.1% (5/234) of the samples. No significant change in wild-type pfcrt-CVMNK prevalence was evident over the 4-year period of the study. The pfcrt-SVMNT haplotype associated with high-level amodiaquine resistance was not detected in this study. The pfmdr1 locus was genotyped in 178 of these samples. The pfmdr1-YYNY haplotype predominated in 67.4% (120/178) of infections and was significantly associated with the pfcrt-CVIET haplotype. All samples carried the wild-type pfmdr1-N1042 codon. The ms-4760 repeat on pfnhe1 locus displayed 12 distinct haplotypes with ms-4760-1 predominating in the population. Analysis of these haplotypes showed no association of a particular haplotype with quinine treatment outcome. Conclusion The pfcrt-CVIET chloroquine resistance haplotype dominated in the collection of P. falciparum samples from Muheza. The pfcrt-SVMNT haplotype, which threatens the efficacy of amodiaquine and was reported in the same time period from Korogwe, Tanzania, 40 Km from Muheza, was not detected. Relative low prevalence of pfcrt-SVMNT in Africa may result from genetic or other factors rendering P. falciparum less supportive of this haplotype than in South America or other regions. Trial registration Trial Protocol Number: 08-I-N064. Electronic supplementary material The online version of this article (doi:10.1186/s12936-015-0642-2) contains supplementary material, which is available to authorized users.


Background
Parasite resistance to anti-malarials including artemisinins [1,2] poses a great challenge to Plasmodium falciparum malaria control efforts. Monitoring parasite haplotypes that predict susceptibility to major anti-malarials can guide treatment policies. Chloroquine (CQ) and amodiaquine (AQ) resistance are linked to mutations in the P. falciparum CQ resistance transporter gene pfcrt [3][4][5]. Two major haplotypes associated with these forms of drug resistance are pfcrt-CVIET in Africa and regions of Asia, and pfcrt-SVMNT in regions of South America, Asia and The Pacific. The pfcrt-CVIET haplotype swept the African continent in the wake of widespread CQ use [6].
Quinine (QN) remains an efficacious alternative treatment for malaria, including severe malaria and malaria in pregnancy. QN resistance is not widespread in malaria endemic areas possibly due to the multigenic nature of its resistance mechanism. Polymorphisms in both pfcrt [4,35] and pfmdr1 [25,36] have been associated with QN susceptibility in vitro. The P. falciparum homologue sodium-hydrogen ion exchanger gene (pfnhe1) located in chromosome 13 has been linked to moderate QN resistance: polymorphisms in one of three microsatellite regions of pfnhe1, namely two or more repeats of DNNND in ms-4760, were associated with a higher QN IC 50 in the progeny of a genetic cross (HB3 × Dd2), in cultureadapted clinical isolates and in laboratory lines [37]. Data from patient isolates tested in vitro for their QN response showed an association of two or more DNNND of the pfnhe1-ms-4760 with higher IC 50 values in some [38][39][40][41] but not all studies [42][43][44]. In a review of the genetic markers of QN resistance, Okombo and colleagues have grouped the ms-4760 into blocks of particular repeats (block I to VI) [45]. A recent worldwide analysis of ms-4760 sequences identified a high diversity of haplotypes as well as an apparent absence of some of these haplotypes from endemic regions of Africa [46].
Reports of the pfcrt-S agt VMNT (AGT codon for serine) haplotype in Tanzania [47] and Cameroon [48], in addition to the pfcrt-S tct VMNT (TCT codon for serine) haplotype in Angola [49], raise concern whether AQ treatment failures may increase, particularly if the P. falciparum parasites also carry pfmdr1 N86 and 1042D codons that, together with pfcrt-SVMNT, are associated with the high levels of AQ resistance found in South America [5]. Spread of the pfcrt-SVMNT and pfmdr1 N86/1042D combination would pose a major threat to AS/AQ combination currently used in 25 African countries for uncomplicated malaria [50]. In light of these reports, a survey for the presence of pfcrt-SVMNT and/or pfmdr1 N86/1042D in samples collected from children treated for malaria in Muheza, located approximately 40 kilometres west of Korogwe District where the observation of pfcrt-SVMNT was reported [47], was undertaken in this study. In addition, the diversity of pfnhe1 ms-4760 in children treated with QN in this holoendemic region of Tanzania was investigated.

Clinical samples
Samples analysed in this study had been collected from children aged 3 -33 months as part of the Mother-Offspring Malaria Study performed in Muheza, northeastern Tanzania, between 2003 and 2006 ( Figure 1). The details of this observational birth cohort study were approved by Tanzanian and US ethical review boards, and signed informed consent was obtained from all mothers for children to participate in the study, as previously reported [51,52]. Confirmation of infection with P. falciparum was performed by microscopic examination of blood smears. Blood samples from children who presented with symptoms consistent with clinical malaria (headache, nausea, vomiting, temperature ≥ 37.5°C) were stored on filter paper (FP). Parasite levels in the blood were reported as the number of parasites (pa) per 200 white blood cells (pa/200 WBC), and parasitaemia (pa/μl) estimates were calculated by assuming an average of 8000 WBCs/μl. Standard treatment for children with less than 2,500 pa/200 WBC was a 3-day 25-mg/kg AQ regimen (10, 10, 5 mg/kg/day). Children who failed AQ treatment or whose parasitaemia was above 2,500 pa/200 WBC were treated with QN at 10 mg/kg every 8 hours for 7 days [51]. Children unable to take the drugs orally were treated with intravenous QN twice daily until the child was able to take QN orally and complete the 7-day regimen.

DNA extraction and genotyping
An area of approximately 7 mm 2 from FP containing dry blood was cut with a pair of scissors that were decontaminated with 70% ethanol between samples. The FP was soaked overnight in 20% Chelex®. The supernatant was removed to a clean tube and the FP was washed twice in 1 ml 1X PBS. DNA was extracted using the QIAamp DNA mini kit (Qiagen Inc., USA) per the manufacturer's recommendations and eluted in 50 μl of AE elution buffer. Samples from a non-template control (NTC) obtained by spotting parasite negative AB+ human blood on FP and from a piece of FP not containing any blood were included in each extraction batch (30 samples) to assure absence of contamination.
Pfmdr1 haplotypes (codons 86, 184, 1042, and 1246) were determined by a Taqman assay employing the primers and probes listed in Table 1. Multiplexed reactions were run for each pair of codons in a single tube in a Rotor Gene 3000 (Qiagen Inc., USA) with four detection channels. The first tube contained primers and probes specific for codons 86 and 1246 and the second tube  controls representing the described codon polymorphisms of P. falciparum clones 3D7, Dd2, HB3 and 7G8 were included in each experiment. Reactions with NTC and FP extractions (described above) and with deionized water substituted for template DNA were employed as negative controls in each experiment. Direct sequencing of the Taqman assay products from 45 samples was also performed to verify the polymorphisms in pfmdr1.
A 350 bp region of the pfnhe1 gene was amplified in 94 of the 99 QN treated samples using nested PCR, as previously described [41]. PCR products were purified by EXOSAP-it (USB, Ohio, USA) and bidirectionally sequenced.
Diversity assessment of 100 randomly selected samples was performed with six polymorphic microsatellite markers present in distinct chromosomes; polyα, TA81, TA1, TA60 and TAA87 [54,55], and 299812 [56]. The reverse primer for each marker was fluorescently labeled and PCR products were analysed in an ABI 3730XL Genetic Analyzer (Life Technologies, Carlsbad, CA, USA). Microsatellite data were analysed with Gene Mapper software (v3.2). A mixed microsatellite haplotype was assigned when the minor signal was at least 33.3% of the major signal [57]. Polyclonality was scored if mixed signals were observed from at least two microsatellites.

Results
Parasitaemia and treatment outcomes of children receiving AQ or QN Samples available from the Muheza Mother-Offspring Malaria Study included filter paper blood spots from 141 children who presented to the clinic with a new episode of P. falciparum parasitaemia, had no history of antimalarial treatment within the preceding 4 weeks, and were prescribed 3 doses of AQ ( Figure 2, see Additional file 1). Of these 141 cases that were treated with AQ, 69 developed microscopic parasitaemia again within four weeks, requiring additional clinical care, while no recurrent parasitaemia was observed in 72 cases. Ninety-three samples were available from cases treated with QN. These samples were from seven cases requiring QN within four weeks of AQ treatment and an additional 86 cases of children presenting with parasite densities ≥ 2,500 pa/200 WBC (est. 100,000 pa/μl; 70 cases) or who were unable to take oral AQ therapy (16 cases; Figure 2).
Presenting parasite densities in the samples from all 141 + 93 = 234 AQ-or QN-treated cases ranged from 40 to 1,119,280 pa/μl with a geometric mean of 30,107 pa/μl (average LN (pa/μl) = 10.313). Geometric means of presenting parasite densities were 11,196 pa/μl (average LN (pa/μl) = 9.323) in the cases treated with AQ and were 134,903 (average LN (pa/μl) = 11.812) in the cases treated with QN (Table 2). Presenting parasite densities ranged from 40 -96,800 pa/μl in cases that showed no recurrence after AQ treatment and from 4,480 -1,119,280 pa/μl in cases with no recurrence after QN treatment, while parasite densities in cases of recurrent parasitaemia were 120-204,480 pa/μl after AQ treatment and 1,680 -744,800 pa/μl after QN treatment. By two-tailed t-tests of LN (pa/μl) average and standard deviation values, there were no significant differences in the geometric mean  -86F  CCT TTT TTT ATA TCT GTG TTT GGT G   Pfmdr1-86R  TTA TTA TCA TGA AAT TGT CCA TCT  presenting parasitaemia of cases that did or did not develop recurrent parasitaemia after treatment with AQ (p = 0.78) or QN (p = 0.37).

Discussion
Related mechanisms of drug action and resistance are reflected in the structural similarities of CQ and AQ as well as the evidence that pfcrt polymorphisms associated with CQ resistance affect AQ efficacy. In a study of pfcrt haplotypes and MDAQ responses in vitro, Warhurst et al. [58] proposed that AQ resistance is associated with the hydrophobicity of pfcrt-SVMNT polymorphisms. Amino acid changes encoded by pfmdr1 that modulate responses of CQ-resistant parasites have also been shown to affect responses to AQ and MDAQ [24]. In the analysis of pfcrt and pfmdr1 of two P. falciparum crosses, Sa et al. [5] found that the highest in vitro IC 50 levels for MDAQ occurred in pfcrt-CVIET or pfcrt-SVMNT progeny carrying either pfmdr1-NFCDY or pfmdr1-NFSDD (codon positions 86, 184, 1034, 1042, 1246), consistent with the association of the pfcrt-SVMNT/pfmdr1-NFCDY combination with AQ failure in South America. These findings together with reports of the pfcrt-SVMNT haplotype in Tanzania, Angola, and Cameroon, raise the concern that if pfmdr1 with the 1042D change spread into Africa it could promote high-level AQ resistance where this drug is widely deployed as a partner drug in ACT combinations [50].
In the present study, 234 samples collected over a fouryear period from Muheza, Tanzania were analysed for the presence of pfcrt polymorphisms. The mutant pfcrt-CVIET there was no statistically significant trend for pfcrt-CVMNK prevalence among these samples over the years. The pfcrt-SVMNT haplotype was not detected in any of the samples analysed in this study. A high level of genetic diversity among all samples was confirmed by six highly polymorphic microsatellites of chromosomes 4, 5, 6 (two markers), 8 and 13. This diversity is an expected outcome in high transmission areas such as Muheza where the entomological inoculation rate has been reported to be 405 infectious bites per year [59].
The pfmdr1-1042D polymorphism that is able to greatly affect MDAQ susceptibility has not been reported in Africa so far; however, three studies on the occurrence of pfcrt-SVMNT in Africa did not provide data for pfmdr1 haplotypes [47][48][49]. The present study included assessments of 178 Muheza samples for pfmdr1 polymorphisms at codon positions 86, 184, 1042 and 1246. All 178 samples were found to encode only pfmdr1-N1042 with no evidence for the presence of pfmdr1-1042D. Consistent with previous suggestions that pfcrt and pfmdr1 polymorphisms in Asia and Africa may be maintained by substantially different mechanisms than in South America [60], the pfmdr1-1042D amino acid change may be slow to rise in African P. falciparum populations because of genetic background or other factors that are less supportive for this haplotype than in South America. The impediment posed by these factors remains open to question, however, as laboratory experiments have shown that the N1042D codon change can be crossed into the African background (as demonstrated by results from the 7G8/Brazil × GB4/ Ghana cross [5]). High prevalence in this study of the major pfmdr1-YYNY haplotype in pfcrt-CVIET-carrying parasites is consistent with previous findings of linkage disequilibrium between pfcrt-76T and pfmdr1-86Y polymorphisms [61][62][63]. The diversity of the pfnhe1-ms-4760 repeat in Muheza is similar to the diversity reported from other regions where this repeat has been studied [41,45,46]. The ms-4760-2 haplotype was not detected, consistent with the absence of this haplotype from a previous report from East Africa [46]. However, ms-4760-4 was observed in ten samples, ms-4760-72 in two samples and ms-4760-79 in one sample; these forms of ms-4760 were not found in previous work from East Africa [46] or in a recent study from Mali [41]. In addition, two haplotypes (ms-4760-102 and ms-4760-103), which have not been previously reported, were observed. A recent report found an association of three DNNND repeats in ms-4760 and higher IC 50 for QN in Western Kenya [64]. The results of the present study did not show this association of three DNNND repeats in cases of returned parasitaemia after QN treatment.
Several possible explanations may account for the frequent recurrence of a parasitaemia within 28 days of treatment and the lack of clear associations to known pfcrt, pfmdr1 and pfnhe1 markers of AQ and QN resistance. These include the high entomological inoculation rates that may have contributed frequent reinfections from emergent liver forms as drug levels wane after treatment. The extent of clearance of moderately-resistant parasites carrying pfcrt-CVIET, pfmdr1-86Y, pfmdr1-1246Y or pfnhe1 haplotypes of ≥ 2 DNNND repeats may have been affected by the immune status of the children [65]. Finally, the authors note that anti-malarial treatments were not fully supervised and plasma drug levels were not followed as part of the parent study to this sub-study; unless disease severity warranted in-patient treatment, children were released from the clinic with medicine and instructions to the parents to complete the prescribed treatment. Compliance and pharmacokinetics of the drugs may have affected treatment efficacy, as has been reported elsewhere [66][67][68][69].
Previous studies have documented the return of the wild-type pfcrt-CVMNK haplotype of CQ-sensitivity in parasite populations where alternative drugs including ACT replaced CQ [7][8][9][10][11][12][13][14][15][16][17][18]. Tanzania changed its malaria treatment policy from CQ to SP in 2001. Although a statistically significant increase of pfcrt-CVMNK in Muheza over the period was not found in this study, the presence of this wild-type form in the region may set the stage for the return of increased prevalence of the pfcrt wild-type, as CQ-sensitive parasites in the absence of selective pressure have been observed elsewhere in Tanzania where CQ therapy has also been removed [11]. Selection of opposite alleles of pfmdr1 by ACT partner drugs, such as AQ and lumefantrine [70], suggest that these drugs could be alternated in order to prolong their efficacy.
Occurrence of the pfcrt-S agt VMNT haplotype in Africa has been reported at low prevalence in countries other than Tanzania: examples include instances of this haplotype in Ghanaian samples collected in 1996-97 [60] and recently from Cameroon [48]. Disparate findings can be found in reports from Angola in 2007 and 2010, where pfcrt-S tct VMNT was detected in 58% of 102 samples from Luanda [49], but there was a complete absence of pfcrt-S tct VMNT from 430 samples collected in Dande [71]. Considering that Luanda and Dande are approximately 90 km apart, additional surveys will be needed to establish the potential prevalence of pfcrt-S tct VMNT in this region.

Conclusions
Although the pfcrt-SVMNT haplotype associated with amodiaquine resistance has been reported from Korogwe, Tanzania, this study of 234 samples from Muheza, approximately 40 kilometers away and in the same time period, did not detect this haplotype. The pfcrt-CVIET haplotype widely responsible for chloroquine resistance was most prevalent in this study and was significantly associated with the pfmdr1-YYNY haplotype. Low levels of the wild-type pfcrt-CVMNK haplotype in Muheza may set the stage for the return of chloroquine sensitive P. falciparum populations, as has been reported in other regions where chloroquine use was stopped. Relative absence of pfcrt-SVMNT from Africa may reflect genetic or other factors affecting P. falciparum that are less supportive for this haplotype than in South America and other continental regions.