Increased pfmdr1 gene copy number and the decline in pfcrt and pfmdr1 resistance alleles in Ghanaian Plasmodium falciparum isolates after the change of anti-malarial drug treatment policy

Background With the introduction of artemisinin-based combination therapy (ACT) in 2005, monitoring of anti-malarial drug efficacy, which includes the use of molecular tools to detect known genetic markers of parasite resistance, is important for first-hand information on the changes in parasite susceptibility to drugs in Ghana. This study investigated the Plasmodium falciparum multidrug resistance gene (pfmdr1) copy number, mutations and the chloroquine resistance transporter gene (pfcrt) mutations in Ghanaian isolates collected in seven years to detect the trends in prevalence of mutations. Methods Archived filter paper blood blots collected from children aged below five years with uncomplicated malaria in 2003–2010 at sentinel sites were used. Using quantitative real-time polymerase chain reaction (qRT-PCR), 756 samples were assessed for pfmdr1 gene copy number. PCR and restriction fragment length polymorphism (RFLP) were used to detect alleles of pfmdr1 86 in 1,102 samples, pfmdr1 184, 1034, 1042 and 1246 in 832 samples and pfcrt 76 in 1,063 samples. Merozoite surface protein 2 (msp2) genotyping was done to select monoclonal infections for copy number analysis. Results The percentage of isolates with increased pfmdr1 copy number were 4, 27, 9, and 18% for 2003–04, 2005–06, 2007–08 and 2010, respectively. Significant increasing trends for prevalence of pfmdr1 N86 (×2 = 96.31, p <0.001) and pfcrt K76 (×2 = 64.50, p <0.001) and decreasing trends in pfmdr1 Y86 (×2 = 38.52, p <0.001) and pfcrt T76 (×2 = 43.49, p <0.001) were observed from 2003–2010. The pfmdr1 F184 and Y184 prevalence showed an increasing and decreasing trends respectively but were not significant (×2 = 7.39,p=0.060; ×2 = 7.49, p = 0.057 respectively). The pfmdr1 N86-F184-D1246 haplotype, which is alleged to be selected by artemether-lumefantrine showed a significant increasing trend (×2 = 20.75, p < 0.001). Conclusion Increased pfmdr1 gene copy number was observed in the isolates analysed and this finding has implications for the use of ACT in the country although no resistance has been reported. The decreasing trend in the prevalence of chloroquine resistance markers after change of treatment policy presents the possibility for future introduction of chloroquine as prophylaxis for malaria risk groups such as children and pregnant women in Ghana.


Background
Malaria is a childhood killer disease in sub-Saharan Africa. Consequently, the absence of an effective malaria vaccine and low uptake of currently available preventive tools makes chemotherapy a strong pillar of the strategies to control malaria in disease-endemic areas of the world. However, the emergence and spread of clones of Plasmodium falciparum which are resistant to most available anti-malarial drugs makes the control of the disease difficult to achieve. Since the first report of chloroquineresistant P. falciparum in Thailand [1] and the rapid spread of resistance worldwide, followed by the recommendation by WHO to use artemisinin-based combination therapy (ACT), many countries in malaria-endemic areas have been monitoring anti-malarial drug resistance. Active implementation of malaria surveillance and the exploitation of molecular as well as phenotypic methods to characterize parasite drug susceptibility profiles will expedite the clarification of spatio-temporal drug sensitivity. This will enable early detection of resistance to the artemisinin derivatives and more efficient use of other available anti-malarial drugs.
Chloroquine resistance has been linked to single nucleotide polymorphisms (SNPs) in the P. falciparum transporter gene (pfcrt) on chromosome 7 [2]. The level of parasite resistance to this drug is under multilocus/ multigenic control [3]. Sets of SNPs in pfcrt codons 72, 74, 75, 76, 97, 152, 163, 220, 271, 326, 356 and 371 were associated with chloroquine resistance in P. falciparum from Southeast Asia, Africa and South America [4][5][6]. The mutation on codon 76, that results from the substitution of threonine for lysine in the gene sequence is the seminal SNP for producing the resistance phenotype and the most reliable molecular marker of chloroquine resistance among the various mutations identified [5,6]. In addition there is reported link between T76 and AQ resistance, however the association is comparatively weaker to that of the other aminoquinoline chloroquine [7][8][9][10][11]. Another genetic mechanism of chloroquine resistance is the SNPs in the P. falciparum multidrugresistance gene (pfmdr1) on chromosome 5 which encodes a P-glycoprotein homologue-1 multi-drug resistant transporter located in the parasite food vacuole, and is associated with enhanced efflux of the drug from resistant parasites [12]. The pfmdr1 mutations linked to antimalarial drug resistance occur at codons 86, 184, 1034, 1042 and 1246 [12][13][14][15]. However, the mutation that occurs as a result of the substitution of asparagine for tyrosine at position 86 is linked with chloroquine resistance [12,14,[16][17][18][19]. Additionally, multiple pfmdr1 SNPs have been associated with susceptibility profiles of many anti-malarial drugs. These SNPs, as well as the gene amplification, may alter substrate specificity for aminoquinolines and arylaminoalcohols [20] promoting P. falciparum resistance to diverse anti-malarials, including chloroquine, mefloquine, quinine and artemisinin derivatives. The Y184F, N1042D and D1246Y mutations are associated with the chloroquine resistance phenotype from samples in Africa, Asia and South America [3,21,22]. The pfmdr1 haplotype of N86, F184 and D1246 (NFD) were selected in recrudescence samples after artemetherlumefantrine (AL) treatment suggesting that this haplotype conferred a fitness advantage upon AL pressure [23]. With the introduction of AL use in Tanzania and Mozambique, the trends in NFD prevalence have been shown to be increasing [24][25][26] which is supports the fact of AL pressure. In addition, the pfmdr1 N86 and pfcrt K76 wild-type alleles are also selected by AL treatment but this observation was not made in the use of artesunateamodiaquine (AS-AQ) and amodiaquine-sulphadoxinepyrimethamine (AQ-SP) [27]. The pfmdr1 D1246Y and N86Y mutations predict resistance and recrudescence to AQ and quinine in Uganda [28]. The N86Y mutation increases sensitivity to artemisinin and dihydroartemisininpiperaquine (DHAP) [20]. The N86Y, Y184F and N1042D mutations increased susceptibility to aryl-amino-alcohol drugs, including mefloquine, halofantrine, lumefantrine, and artemisinin derivatives [28][29][30][31][32]. In South America, the S1034C, N1042D and D1246Y mutations are associated with quinine resistance [33] and are also implicated in increased sensitivity to artemisinin [34,35]. The S1034C and N1042D mutations were reported to reduce or abolish resistance to mefloquine [30].
The pfmdr1 gene expression levels have been considered in the etiology of the parasite resistance to some anti-malarial drugs and it is being explored in epidemiological studies. Increase in pfmdr1 gene copy number has been linked to P. falciparum diminished susceptibility to anti-malarial drugs, such as mefloquine, AS-MQ and AL combinations [36][37][38]. Although there is no reported correlation of pfmdr1 gene copy number and treatment failure, this marker is important for the prediction of recrudescence with the use of the anti-malarials mentioned above [39]. There is an assertion that pfmdr1 gene copy number rather than the SNPs exercises greater influence in mediating anti-malarial drug resistance to some compounds [40]. This was reported to be due to the fact that many transporter proteins mandate concerted complementary attention to copy number variations (CNV) in mediating anti-malarial activity [40].
Ghana is a malaria-endemic country using ACT (AS-AQ, AL and DHAP) for the treatment of uncomplicated malaria since 2005. Information from molecular investigations, in addition to in vivo and in vitro analysis, is crucial for early detection and prediction of resistance to the ACT. Before the change of the drug policy in Ghana, the prevalence of the pfcrt T76 mutation and the pfmdr1 Y86 mutation associated with chloroquine resistance ranged from 46 to 98% and 42 to 95%, respectively from five sentinel sites [16]. Due to the Malawian and Kenyan experiences where there was significant decrease in the resistant parasite populations when the drug was withdrawn from use [41,42], it is expected that a similar phenomenon may occur in Ghana. Thus the continuous monitoring of the prevalence of these mutations in parasites in the country is important. Such information is crucial for consideration of chloroquine as prophylaxis for malaria risk groups like children and pregnant women. With the use of ACT in the country, pfmdr1 copy number determination will help in the early detection of artemisinin derivatives resistance which has implications for chemoprophylactic drug recommendations for natives and travelers from non-endemic regions. This study investigated the prevalence of mutations in pfcrt and pfmdr1 genes as well as the gene copy number of pfmdr1 in clinical isolates collected from nine sentinel sites in Ghana to detect the trends in the prevalence over seven years.

Study sites
Samples used in this investigation were collected from 2003-2010 from nine sentinel sites in Ghana established for monitoring anti-malarial drug resistance in the country. These sites are located in the three ecological zones of Ghana: Begoro, Bekwai, Hohoe, Sunyani and Tarkwa are located in the tropical forest ecological zone and experience perennial malaria transmission. Navrongo, Wa and Yendi are in the Guinea Savanna ecological zone and experience seasonal malaria transmission. Cape-Coast is in the Coastal Savanna ecological zone and experience perennial malaria transmission.

Study samples
Blood samples were collected from children aged six to 59 months presenting with uncomplicated malaria at the sentinel sites after the parents or guardians of these children gave informed consent for the children to participate in the study [43][44][45]. Filter-paper blood blots were made for each child, air-dried and stored in zip-locked bags with silica gel at room temperature until use.

Determination of pfmdr1 gene copy number
In all, 756 filter paper, blood-blot samples collected from 2003-2010 were used for the pfmdr1 copy number (CN) determination. The real-time quantitative PCR (qRT-PCR) method was used following a published protocol [36]. Parasite DNA was extracted using the QIAamp DNA Blood Mini Kit. The pfmdr1 gene copy number was estimated for all the parasite isolates by the relative ΔΔCt method [36]. The DNA from P. falciparum cell lines 3D7, K1, W2mef and DD2 with copy numbers of 1, 1, 2 and 3-4, respectively were used as references. Samples were run in triplicates for the quantitative PCR and repeated additional times for samples with estimated CN values above 1.5. The estimated CN values were rounded to the nearest integer. For the selection of monoclonal infections from the entire data, the merozoite surface protein 2 (msp2) genotyping analysis was done on all 756 samples for FC27 and IC1 genotypes following published protocols [46,47]. Monoclonal infections were selected samples with only one msp2 genotype; either FC27 or IC1with one PCR band seen on the electrophoresis gel, which were scored by two people.

Detection of pfmdr1 and pfcrt polymorphisms
In all, 1,102 samples were analysed for the pfmdr1 polymorphisms at codon 86 (asparagine to tyrosine) and 832 samples for codons 184 (tyrosine to phenylalanine), 1034 (serine to cysteine), 1042 (asparagine to aspartic acid) and 1246 (aspartic acid to tyrosine). For pfcrt polymorphism at codon 76 (lysine to threonine), 1,063 samples were analysed. DNA was extracted from the filter-paper blood blots using the Tris EDTA buffer extraction method [48]. PCR followed by restriction fragment length polymorphism (RFLP) was used to detect the mutations using published protocols for pfmdr1 [13] and pfcrt [5].

Data analysis
The percentage of isolates with increased pfmdr1 gene copy number was determined for each sentinel site for the time points; 2003-04, 2005-06, 2007-08, and 2010. The data were then pooled for the percentage of Ghanaian isolates with increased copy number and the observed trend determined. The prevalence of mutations for pfmdr1 gene codons 86, 184, 1034, 1042 and 1246 and pfcrt codon 76 were determined for each site and an overall data was pooled from the sites for the Ghanaian isolates analysis. The significance of the observed trends in the prevalence of the alleles over the years from 2003 to 2010 was determined using the Chi-squared test for trends (EpiCalc 2000).

Ethics
Ethical approval for this study was received from the Noguchi Memorial Institute for Medical Research Institutional Review Board (NMIMR IRB) and the United States Naval Medical Research Unit No 3 (NAMRU-3) IRB.

Results
Increased pfmdr1 gene copy number in Ghanaian isolates The number of experimental repeats, range and mean ± standard deviation of the estimated copy number (CN) for the P. falciparum clones 3D7, K1, W2-mef and Dd2 are shown in Table 1. The samples with estimated CN values above 1.5 were repeated for confirmation. The number of isolates with increased pfmdr1 gene copy number from the nine sites per year is shown in Table 2    pfmdr1 codon 184 alleles for all the sites per year is shown Additional file 1. The observed increasing trend of F184 and the decreasing trend in the prevalence of Y184 alleles did not reach statistical significance (× 2 = 7.39, p = 0.060; × 2 = 7.49, p = 0.057, respectively) ( Figure 2C). The wild type allele of pfmdr1 1034 (S1034) was predominant in the isolates from all the time points. The prevalence of the isolates with the allele was 100% for 2003-04, ranged from 92-100%, 98-100%, and 95-100% for 2005-06, 2007-08 and 2010, respectively. The highest prevalence of the mutant C1034 was 35% and was seen in isolates from Wa (2010). The prevalence of the pfmdr1 codon 1034 alleles for all the sites per year is shown in Additional file 1. The prevalence of the alleles was stable through the time points ( Figure 2D).
The pfmdr1 N1042 allele, which is the wild type allele, was present in majority of the isolates such that all the samples from 2003-04 had the allele (100%). The prevalence of the allele for 2005-06, 2007-08 and 2010 ranged from 97-100%, 98-100% and 65-100%, respectively. For the mutant allele, D1042, the highest prevalence of 73% was observed in isolates from Begoro (2010). The prevalence of the pfmdr1 codon 1042 alleles for all the sites per year is shown in Additional file 1. There was stability in the prevalence of the pfmdr1 codon 1042 alleles over the years ( Figure 2E).
The codon 1246 wild type allele was present as the majority allele in the isolates for all the time points. The prevalence ranged for 2003-04, 2005-06, 2007-08 and 2010 from 96-100%, 80-100%, 94-100%, and 75-100%, respectively. For the mutant allele, Y1246, the highest prevalence was 35%, which was observed in the isolates from Cape-Coast (2010). The prevalence of the pfmdr1 codon 1246 alleles for all the sites per year is shown in Additional file 1. The prevalence of the alleles was stable over the years ( Figure 2F).

Discussion
In the absence of any reported parasite resistance to ACT in Ghana, the increased pfmdr1 gene copy number observed in this study indicates the possibility of nascent clones of P. falciparum with reduced susceptibility to the artemisinin derivatives in the country. The high frequency of malaria in Ghana, coupled with a solitary reliance on ACT treatment, poses a situation where selection and propagation of drug-resistant lineages is highly possible. The malaria treatment policy was changed in 2005 from the use of chloroquine as first-line drug to the use of AS-AQ, and in 2008, AL and DHAP were added as the firstline drugs for the treatment of uncomplicated malaria in Ghana. The data showed that before the change, 2003-04, only one parasite out of the 25 isolates tested had increased copy number of 2. This outcome, which is expected, reflects the absence of the artemisinin derivatives and mefloquine drug pressure as these drugs were in limited use at that time. However, it is interesting to note that soon after the implementation of the new malaria drug policy in 2005-06, 27% of the isolates showed an increased pfmdr1 gene copy number, which was the highest of the four time points. This observation can be attributed to the drug pressure induced by the increased use of the artesunate, a component in the ACT.
Although there is no evidence data on the use of artesunate in the country before 2005, due to the low sensitivity of parasites to chloroquine then, there was limited use of the drug as monotherapy for the treatment of uncomplicated malaria (Neils B. Quashie, pers comm). After 2005-06 there was a decline from 27% in 2005-6 to 9% in 2007-08 with the introduction of the other two ACT, AL and DHAP. There was an increase from 9% to 18% of isolates with copy number increase in 2010. The observed trend of the parasites with increased copy number at the designated time points in the surveillance is suggestive that there is circulation of parasites with reduced sensitivity to artemisinin derivatives and mefloquine in Ghana. Of the three ecological zones in Ghana, isolates from the three savanna zone areas had the majority with increased gene copy number; Navrongo for 2003-04, Yendi for 2007-08, Wa and Navrongo for 2010. It is, therefore, not surprising that Abuaku and others reported a delayed clearance of parasites with the use of AL, especially with parasites from the savannah zone [43]. This finding adds credence to the assertion that gene copy number variation (CNV) may be the suitable indicator for incipient delays in parasite clearance with the use of ACT. CNV in this case involves the duplication of multi-drug resistance genes which help the parasite to survive under unfavourable drug conditions and as such the pfmdr1 increased copy number is alleged to confer resistance to some ACT and a decrease in the gene copy numbers may confer sensitivity to some anti-malarials [49][50][51].
The increased pfmdr1 gene copy number being recorded in sub-Saharan Africa can be worrying although it is not a good indicator of ACT sensitivity in SE Asia it is important to monitor the prevalence of this marker in the circulating parasite population. With ACT use and vector control, the expectation of reduction in mortality and morbidity in African children will be dwindled with the emergence and spread of resistant parasites to the newly introduced ACT. Since the first report of ACT resistance in Southeast Asia and the associated pfmdr1 gene copy number, the gradual movement of these parasites into Africa is being seen. Increased pfmdr1 gene copy has been reported in West African countries such as Côte d'Ivoire, Burkina Faso, Togo [52], Senegal [30], but not in Nigeria [23], the Gambia (David J Conway, pers comm), Liberia and Guinea Bissau [53]. Other African countries with reported increased pfmdr1 copy number in isolates includes Kenya [9], Gabon [31] and Sudan [54] but not in Cameroon [27] and Mozambique [55].
Chloroquine was the first-line drug for the treatment of uncomplicated malaria in Ghana before the change to the use of ACT. In 2007 Koram's group reported the association between the molecular markers of chloroquine resistance, pfcrt T76 and pfmdr1 Y86 with drug treatment failure in a clinical study [16]. This report [16] together with that from in vivo [45] and in vitro [56] studies ran in parallel, which was made available to the National Malaria Control Programme (NMCP), became one of the basis for treatment policy change in the country. Since then, there has been continuous monitoring of anti-malarial drug resistance in the country using the three aspects of investigation named above. An increasing trend in the prevalence of the pfcrt K76 wild type allele from 19 to 59% from 2005 to 2010 and a decreasing trend for pfcrt T76 from 88 to 56% for the same period were observed. The general observation depicts a decline of pfcrt T76 with and a gradual takeover by K76 after chloroquine was removed in 2005 as the first choice of anti-malarial treatment. This observed trend in Ghana is consistent with observations of decreasing trend of the mutants in Malawi, where the decrease was from 85 to 13% for 1993 to 2000 [41], Kenya, a decrease from 95 to 60% for 1993 to 2006 [42], Senegal, a decrease from 92 to 37% for 2000 to 2010 [30] and Mozambique, from 96 to 32% for 2006 to 2010 [55]. The significant decrease in the prevalence of pfcrt T76 in Ghana is indicative of the fitness, which is low for the resistant allele after the selective drug pressure has been withdrawn. This same scenario was observed for pfmdr1 N86Y, where there was a decline in the prevalence of the mutant from 46 to 28% for 2005 to 2010 and an increase in the prevalence of the wild type strains N86 from 77 to 86%. Similar observations have been made in other studies conducted in Africa where K76T and N86Y were investigated concurrently [30,41,55]. In this study, there was no significant difference between the prevalence of N86 for 2007-08 and 2010, which were 89 and 86%, respectively. This is indicative of a gradual gaining of stability of these genotypes in the population similar to what was observed by Nzila's group in Kenya [42].
Despite the comparatively faster decline of T76 in the population, the Y86 seems to reduce slowly in this investigation. A study conducted in Gabon with isolates collected from 2004 and 2009, showed stability in T76 with 94% in 2004 and 96% in 2009 whilst the Y86 increased from 16 to 31% for 2004 and 2009 [57]. The apparent disappearance of the chloroquine-resistant parasites with pfcrt T76 and pfmdr1 Y86 from most of the sub-Saharan Africa malarious areas will help in the re-introduction of the cheapest anti-malarial in combination with an artemisinin derivative. The presence of the mutant parasites with pfcrt T76 and pfmdr1 Y86 is a threat to some of the ACT in use. It is important to note that the prevalence of T76 in 2003-04 was 70% and in the year of implementation of AS-AQ in 2005, the prevalence went up to 88%, which is suggestive that another aminoquinoline, in this case, AQ selected the pfcrt codon 76 mutants. This is similar to observations made in studies conducted in Uganda and Burkina Faso [11,28,32]. A report from an in vitro study conducted in Nigeria showed an association between the T76 mutation and decreased susceptibility to artemether [58]. The increasing trend in the prevalence of N86 observed in this study must be closely monitored since this allele is reportedly linked to decreased sensitivity of parasites to lumefantrine and mefloquine and it is also a marker for lumefantrine tolerance [57]. In addition an increasing trend for K76 will create a future problem for ACT use because it has been seen in recrudescent samples after AL use [59].
The prevalence of the other SNPs of the pfmdr1 gene was also investigated, which included the alleles of codons 184, 1034, 1042 and 1246. The mutant allele of codon 184 is known to be selected after AL use in recrudescent samples [54,60]. From the results, in 2003-04, the prevalence was about 52% and it reduced to 51% in 2005 and then increased to 62% in 2010. Although the increasing trend of the prevalence of F184 from 2003 to 2010 was not significant, there is an indication of selection of this suggested AL resistance allele. Results from a 2010 study conducted in seven communities in southern Ghana revealed the prevalence of F184 to be about 72% [61]. In the West African region, a report from a study conducted in Senegal also showed the prevalence of 68% in samples collected in 2009-2010 [30]. The wild type allele, Y184 was observed to be decreasing in prevalence whilst the other alleles, such as the S1034 wild type allele, were stable over the time points. An increase in the prevalence of the mutant (C1034) was observed for 2010. For the codon 1042 alleles, the wild type N1042 was also stable until a decrease in prevalence (from 100 to 92%) was observed in 2010. The mutant D1042 was also about 19% in 2010 with low to non-existent levels in the previous years. For the codon 1246 alleles, D1246 dropped from 98% in 2003-04 to 90% in 2010 whilst the Y1246 was observed to be gradually increasing 7% in 2003-04 to 14% in 2010. The Y1246 mutant has been linked to quinine failure [57], as such the use of quinine as second-line drug in Ghana for the treatment of uncomplicated malaria may delay the selection of these mutant parasites. There is more stability in the three pfmdr1 codons 1034, 1042 and 1246 compared to the 86 and 184. The haplotype mutant of pfmdr1 C1034-D1042-Y1246 has been linked to increased sensitivity to artemisinin [34,35] whilst C1034-D1042 is linked to sensitivity to mefloquine [30]. Furthermore, the Y86-F184-D1042 mutations have been shown to increase the susceptibility of parasites to aryl-amino-alcohol drugs, artemisinin derivatives, mefloquine, halofantrine and lumefantrine [28][29][30][31][32]. As such the stability of these mutants as observed in this study will help maintain the use of ACT in the country. The N86-F184-D1246 haplotype, which was detected in 35% of the isolates in 2003-04, increased significantly to 53% in 2010. Similar trends in prevalence for the NFD haplotype have been reported in Tanzania and Mozambique after the introduction of AL use [24][25][26]. In addition, this haplotype as well as pfcrt K76 were seen in recrudescent samples after AL use [54,59]. The NFD haplotype therefore needs to be monitored effectively especially in malarious countries using AL for chemotherapy.