Although the molecular basis of resistance to a number of common antimalarial drugs is well known, a geographic description of the emergence and dispersal of resistance mutations across Africa has not been attempted. To that end we have characterised the evolutionary origins of antifolate resistance mutations in the dihydropteroate synthase (dhps) gene and mapped their contemporary distribution.
Methods and Findings
We used microsatellite polymorphism flanking the dhps gene to determine which resistance alleles shared common ancestry and found five major lineages each of which had a unique geographical distribution. The extent to which allelic lineages were shared among 20 African Plasmodium falciparum populations revealed five major geographical groupings. Resistance lineages were common to all sites within these regions. The most marked differentiation was between east and west African P. falciparum, in which resistance alleles were not only of different ancestry but also carried different resistance mutations.
Resistant dhps has emerged independently in multiple sites in Africa during the past 10–20 years. Our data show the molecular basis of resistance differs between east and west Africa, which is likely to translate into differing antifolate sensitivity. We have also demonstrated that the dispersal patterns of resistance lineages give unique insights into recent parasite migration patterns.
Citation: Pearce RJ, Pota H, Evehe M-SB, Bâ E-H, Mombo-Ngoma G, Malisa AL, et al. (2009) Multiple Origins and Regional Dispersal of Resistant dhps in African Plasmodium falciparum Malaria. PLoS Med 6(4): e1000055. https://doi.org/10.1371/journal.pmed.1000055
Editor: Lorenz von Seidlein, Joint Malaria Project, United Republic of Tanzania
Received: May 30, 2008; Accepted: February 13, 2009; Published: April 14, 2009
Copyright: © 2009 Pearce et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study was funded by the Gates Malaria Partnership (GMP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: CQ, chloroquine; CQR, chloroquine resistance; IPTi, Intermittent Preventive Treatment of Malaria in Infants; IPTp, Intermittent Preventive Treatment of Malaria in Pregnancy; SP, sulphadoxine–pyrimethamine
Plasmodium falciparum, a mosquito-borne parasite that causes malaria, kills nearly one million people every year, mostly in sub-Saharan Africa. People become infected with P. falciparum when they are bitten by a mosquito that has acquired the parasite in a blood meal taken from an infected person. P. falciparum malaria, which is characterized by recurring fevers and chills, anemia (loss of red blood cells), and damage to vital organs, can be fatal within hours of symptom onset if untreated. Until recently, treatment in Africa relied on chloroquine and sulfadoxine–pyrimethamine. Unfortunately, parasites resistant to both these antimalarial drugs is now widespread. Consequently, the World Health Organization currently recommends artemisinin combination therapy for the treatment of P. falciparum malaria in Africa and other places where drug-resistant malaria is common. In this therapy, artemisinin derivatives (new fast-acting antimalarial agents) are used in combination with another antimalarial to reduce the chances of P. falciparum becoming resistant to either drug.
P. falciparum becomes resistant to antimalarial drugs by acquiring “resistance mutations,” genetic changes that prevent these drugs from killing the parasite. A mutation in the gene encoding a protein called the chloroquine resistance transporter causes resistance to chloroquine, a specific group of mutations in the dihydrofolate reductase gene causes resistance to pyrimethamine, and several mutations in dhps, the gene that encodes dihydropteroate synthase, are associated with resistance to sulfadoxine. Scientists have discovered that the mutations causing chloroquine and pyrimethamine resistance originated in Asia and spread into Africa (probably multiple times) in the late 1970s and mid-1980s, respectively. These Asian-derived mutations are now common throughout Africa and, consequently, it is not possible to determine how they spread across the continent. Information of this sort would, however, help experts design effective measures to control the spread of drug-resistant P. falciparum. Because the mutations in dhps that cause sulfadoxine resistance only began to emerge in the mid-1990s, they haven't spread evenly across Africa yet. In this study, therefore, the researchers use genetic methods to characterize the geographical origins and contemporary distribution of dhps resistance mutations in Africa.
The researchers analyzed dhps mutations in P. falciparum DNA from blood samples collected from patients with malaria in various African countries and searched the scientific literature for other similar studies. Together, these data show that five major variant dhps sequences (three of which contain mutations that confer various degrees of resistance to sulphadoxine in laboratory tests) are currently present in Africa, each with a unique geographical distribution. In particular, the data show that P. falciparum parasites in east and west Africa carry different resistance mutations. Next, the researchers looked for microsatellite variants in the DNA flanking the dhps gene. Microsatellites are DNA regions that contain short, repeated sequences of nucleotides. Because the number of repeats can vary and because microsatellites are inherited together with nearby genes, the ancestry of various resistance mutations can be worked out by examining the microsatellites flanking different mutant dhps genes. This analysis revealed five regional clusters in which the same resistance lineage was present at all the sites examined within the region and also showed that the resistance mutations in east and west Africa have a different ancestry.
These findings show that sulfadoxine-resistant P. falciparum has recently emerged independently at multiple sites in Africa and that the molecular basis for sulfadoxine resistance is different in east and west Africa. This latter result may have clinical implications because it suggests that the effectiveness of sulfadoxine as an antimalarial drug may vary across the continent. Finally, although many more samples need to be analyzed to build a complete picture of the spread of antimalarial resistance across Africa, these findings suggest that economic and transport infrastructures may have played a role in governing recent parasite dispersal across this continent by affecting human migration. Thus, coordinated malaria control campaigns across socioeconomically linked areas in Africa may reduce the African malaria burden more effectively than campaigns that are confined to national territories.
Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed.1000055.
- This study is further discussed in a PLoS Medicine Perspective by Tim Anderson
- The MedlinePlus encyclopedia contains a page on malaria (in English and Spanish)
- Information is available from the World Health Organization on malaria (in several languages) and on drug-resistant malaria
- The US Centers for Disease Control and Prevention provide information on malaria (in English and Spanish)
- Information is available from the Roll Back Malaria Partnership on its approach to the global control of malaria, and on malaria control efforts in specific parts of the world
- The WorldWide Antimalarial Resistance Network is creating an international database about antimalarial drug resistance
Chloroquine (CQ) and the antifolate combination of sulphadoxine–pyrimethamine (SP) were, until recently, the mainstay of malaria treatment in Africa. Resistance to both drugs is now widespread. In both cases the importation of resistance mutations to Africa from Asia played a decisive role in the establishment of resistance –, but details of where, when, or how resistance genes were introduced in Africa are unknown.
CQ was first used in the 1950s, and chloroquine resistance (CQR) appeared in Asian and south American foci in the early 1960s. CQR did not appear in Africa until 1978, when the initial focus was in east Africa. It subsequently appeared to radiate from that focus, reaching west Africa between 1986 and 1989 , . The major genetic determinant of CQR is now known to be the P. falciparum chloroquine resistance transporter (encoded by the pfcrt gene) , and microsatellites in the flanking sequence around pfcrt resistance alleles show that the Asian lineage of mutant pfcrt is present in P. falciparum populations throughout Africa , . It is probable that resistant pfcrt was introduced to Africa on multiple occasions, as CQR was common throughout Asia by that time , but archived samples from 30 years ago are rare, making a retrospective analysis of the geographical dispersal of resistant pfcrt in Africa intractable.
SP began to be used for treatment of CQR malaria in Africa in the 1980s. Resistance to SP involves adaptations in the target molecules, dihydrofolate reductase (DHFR) (the target of pyrimethamine) ,  and dihydropteroate synthase (DHPS) (the target of sulphadoxine) , . The triple-mutant dhfr (containing mutations N51I+C59R+S108N) confers a significant component of resistance to SP . It is found throughout Africa – and is derived from a single ancestor, which originated in Southeast Asia . Analysis of archived parasite specimens show that the triple-mutant lineage was present at least as early as 1985 in Cameroon  and at least as early as 1988 in Kenya . Like pfcrt, resistant dhfr may have been imported on multiple occasions, but historical samples are limiting and since the Asian-type sequence is now common throughout Africa it is not possible to ascertain any geographical detail about its arrival and dispersal on the continent more than 20 years ago.
Although the pan-African distribution of Asian-derived pfcrt and dhfr lineages today suggests that the P. falciparum populations of Africa form one large continuous whole, the geography of their introduction 20–30 years ago and their subsequent dispersal has not been characterised. The emergence of dhps resistance alleles has occurred more recently than either pfcrt or dhfr, creating an opportunity to directly observe resistance dispersal events while they are in progress.
Clinical treatment failure with SP was first reported in Tanzania in 1995 , , and the timing of its emergence in Africa broadly coincides with the first appearance of mutant dhps against a pre-established background of resistant dhfr. In east Africa, mutations at codons A437G and K540E of dhps, together with the triple mutations of dhfr were shown to be a significant predictor of SP treatment failure in Kenya, Malawi, and Uganda –. In west Africa, where the K540E mutation is rare, an association of treatment failure with the A437G plus the triple-mutant dhfr has been reported by studies in Gabon , Ghana , The Gambia  and Congo (Brazzaville) . In contrast, a study in Ghana by Marks et al. reported no association of A437G with treatment outcome .
The spatial distribution of dhps mutations in Africa has not previously been mapped. To obtain the highest resolution possible, we generated new SNP data for 20 countries and combined this with additional data from seven published studies to cover a total of 50 sites in the 27 countries shown in Figure 1. To investigate the evolutionary origins of dhps mutations we examined diversity at microsatellite markers flanking the gene and used this to characterise lineages of common ancestry that have been subject to recent selection. In this way we were able to describe the dispersal dynamics of resistance alleles currently under selection and to generate new hypotheses about the geography of malaria migration in modern Africa.
Materials and Methods
Study sites are listed in Table 1. The details of sample collection procedures and ethical permissions at every site for which new data are described here are detailed in Text S1. Literature searches were done during June 2007 and updated in October 2007 using the National Library of Medicine search engines, Pub Med and Medline (details given in Text S2). We identified 20 published studies in which dhps point mutation haplotypes including codons 436, 437, and 540 had been reported in P. falciparum isolates sampled in Africa since 1997 (Table 1).
In all studies finger-prick blood spots were taken from infected individuals and parasite DNA was extracted from the blood spots using the Chelex method, and the polymorphic region of dhps was PCR-amplified prior to sequence-specific oligonucleotide probing (SSOP) for polymorphism at codons 436, 437, and 540. The DNA extraction, PCR amplification, and SSOP dot-blotting procedures have been described previously . Since blood-stage P. falciparum is haploid, the determination of allelic haplotypes is straightforward when an infection consists of a single genotype, because only one form of sequence at every SNP is seen. When infections are composed of multiple genotypes, however, the mixture of different sequence variants makes inference of point mutation haplotypes within that infection more difficult. A sample was considered to have a single haplotype when only one sequence variant was found at each locus. In mixed-genotype infections, if one genotype was substantially in the majority (i.e., the hybridisation signal of the minority sequence was less than half the intensity of the majority), then the majority haplotype was recorded. One haplotype only was counted from each infection, and those mixed infections for which haplotypes could not be resolved were omitted from the calculation of haplotype frequencies (numbers of mixed infections excluded from each sample are given in Table S1). It should be noted that the rate of detection of mixtures is dependent on the extent of polymorphism at a given locus. Furthermore, sampling at various study sites was not standardised for factors known to affect rates of mixture, namely, patient age, rates of self treatment before attendance at the health facility, and transmission intensity itself.
Maps of Africa
Maps of the distribution of dhps alleles were constructed using MapInfo (MapInfo Limited, Windsor, United Kingdom).
Microsatellite loci flanking dhps at 0.8 kb, 4.3 kb, and 7.7 kb from the 3′ end of the gene were amplified in samples from 20 sites across 19 countries. Full primer sequences and cycling conditions can be found in . The amplification products were run diluted 1∶100 and run with LIZ-500 size standard on the ABi 3730 DNA analyser (Applied Biosystems, Foster City, California, United States) and analysed using the software Genemapper (Applied Biosystems). In samples where more than one allele was present at a locus the data was considered missing at that locus for that sample.
Gene diversity values were calculated as , where He is expected heterozygosity, n is the number of samples, and pi is the frequency of the ith allele in the sample set. Heterozygosity of microsatellites flanking each of the five key dhps alleles was calculated for each separate study site. Geographic sites at which the sample size was less than ten were not included. The box plots were constructed in the statistical package R , which calculates the median and interquartile intervals. The total range of the distribution of heterozygosity values was described by the upper and lower extreme values, provided they fell within a range that was calculated as 1.5× the interquartile range below the first and above the third quartiles. Extreme values that fell outside this distribution were identified as outliers and plotted separately. The statistical significance of the loss of diversity around mutant dhps alleles was calculated by Wilcoxon's rank sum test which compared the variability of He among geographical sites with equivalent He values for the same microsatellite flanking the wild-type allele (haplotype SAK at codons 436, 437, and 540).
Population Comparison of Resistance Allele Sharing
To examine the extent to which resistance alleles were exchanged among the populations at 20 sites we used a pairwise population measure of resistance allele sharing. For the analysis we expressed the combination of point mutations and linked microsatellite alleles at the 0.8 kb and 4.3 kb loci as a single allelic haplotype, and estimated DPS for a single locus. DPS was calculated as 1−ps (where ps is the sum of the minima of the relative frequencies of all alleles shared between compared population samples) .(1)where is the frequency of allele a in populations i and j.
Resistance allele sharing among the 20 geographical sites is summarised in the neighbour joining tree generated in the neighbour package of Phylip .
Analysis of Diversity Flanking Wild-Type Alleles
To compare heterozygosity at the 0.8 kb microsatellite locus linked to wild-type alleles with haplotypes SAK and AAK at codons 436, 437, and 540, we used a method described by Nash et al. . Significance was determined by comparing the ratio of heterozygosity (He SAK/He AAK) observed in each geographical region with the ratio of heterozygosities from 10,000 simulated datasets in which microsatellite alleles were reshuffled amongst all parasites. To obtain the level of significance for the difference in He, the number of occasions that the simulated ratio of heterozygosities exceeded the observed was counted and converted to the proportion of the 10,000 simulated datasets.
For assessing the relationship between the sensitive chromosomes and the single-mutant AAK chromosomes, Nei's standard genetic distance  was calculated for all pairwise comparisons in Phylip . The significance of the observed standard genetic distance between allelic populations was determined by comparison to genetic distance values from 10,000 simulated datasets in which the alleles at each locus were reshuffled among all parasites. To obtain the level of significance, the number of occasions that the simulated distance exceeded that for the observed data was counted, then converted to the proportion of the 10,000 simulated datasets. The statistical package R  was used to permutate the datasets.
The Geographical Distribution of dhps Mutation Haplotypes
To map the distribution of dhps mutations in Africa we collected and typed P. falciparum DNA extracted from finger-prick samples from 3,761 malaria patients at 31 sites in 20 African countries. The details of individual studies are described in full in Text S1. Where possible these data were supplemented with previously published data from the literature (the complete list of survey sites is given in Table 1). We combined original and published data to obtain the point mutation haplotypes for codons 436, 437, and 540 of dhps for 5,493 unmixed isolates collected between 1997 and 2007 at 50 unique geographical locations in Africa. Blood stage parasites are haploid, so where infections consist of a single genotype it is possible to determine complete haplotype information. This determination is not possible with blood samples from patients with mixed-genotype infections. Among the pooled samples from all geographic locations together we found five major point mutation haplotypes; SGE, AGK, SGK, AAK, and SAK. In addition there were rare single-mutant haplotypes coding for alternative substitutions at position 436. We found 67 examples of the 436F mutation dispersed across 14 geographical populations, seven examples of 436C in five countries, and 13 isolates in Cote d'Ivoire with a 436Y substitution which were previously described by Djaman et al. .
The SAK is generally regarded as the ancestral wild type, and alleles that contain the S436A alone (AAK) are considered alternative wild types . All haplotypes containing the A437G substitution (SGE, AGK, SGK) are known to confer resistance to sulphadoxine in vitro , and on that basis we classified the five haplotypes as either wild type (SAK and AAK) or resistant (SGE, AGK, and SGK). Figure 2 shows the geographical distribution of resistant alleles SGK, AGK, and SGE and wild-type alleles AAK and SAK. A complete listing of the frequencies of all haplotypes at every site is given in Table S1.
Resistant alleles; the upper map shows the relative proportions of the three major resistance alleles, SGK, AGK, and SGE. Wild-type alleles; the lower map shows the ratio of SAK and AAK alleles among wild-type dhps alleles. In both cases the diameter of the pie is proportional to the combined frequencies of the alleles represented in the total population.
The size of the pie charts in the upper map of Figure 2 indicates the proportion of the total sample that was resistant. Sites where resistance allele frequencies were high relative to wild-type alleles are indicated by large pie charts, while those where resistance alleles were less abundant than wild type are indicated by small pie charts. The abundance of resistance alleles varied from site to site presumably in accordance with local malaria drug treatment practices. Selection for resistance has clearly been high in both east and west Africa. Although there was no obvious geographical trend in the ratios of resistant to wild-type alleles, there was a very striking difference in the type of resistance alleles that were prevalent in east and west Africa, illustrated by the coloured segments of the pie charts in the upper map of Figure 2. In east Africa, the A437G and K540E substitutions occurred together as a double-mutant allele SGE (shown in red). In west Africa, the A437G substitution was found alone as either an SGK (blue) or an AGK (green) allele, and the SGE allele was rare or absent.
The SGE allele was prevalent in a number of east African sites, and its frequency exceeded 95% of the total parasite population in Kenya 2006 (this study), Uganda 2005 (this study), Sudan 2003 (this study), and three sites in Ethiopia 2004 (this study and , ). The SGE allele was rare (0%–9%) in central and southwest African samples (Gabon 2007, Cameroon 2004, Congo 2004, Central African Republic 2004, Angola 2004, and Namibia 2005) and absent in many west African samples (Nigeria 2005, Cote d'Ivoire 2001, Burkina Faso 2002, Burkina Faso 2003, The Gambia 2004, Mauritania 1998, Mali 1997, Senegal 2003, and Senegal 2004) although small numbers of SGE were detected in three west African samples: Guinea 2004–2005 (7%); Navrongo, Ghana 2003 (<1%); and Hoehoe, Ghana 2005 (<1%).
In west, central, and southwest Africa, instead of SGE, the SGK and AGK resistance alleles prevailed, often at high frequency; for example, in the samples from Nanoro, Burkina Faso 2003 (98%); Hohoe, Ghana 2005 (94%); Angola 2004 (92%); Navrongo, Ghana 2003 (85%); Mutengene, Cameroon 2004 (84%); Namibia 2005 (78%); Congo 2004 (72%); Gabon 2007 (69%); Guinea 2004–2005 (59%); and Bousse, Burkina Faso 2002 (57%). The combined frequency of SGK and AGK resistance alleles was intermediate or low in other samples such as Senegal 2004 (50%), Nigeria 2005 (47%), The Gambia 2004 (46%), Senegal 2003 (40%), Cote d'Ivoire 2001 (39%), Central African Republic 2004 (15%), Mauritania 1998 (18%), and Mali 1997 (0%).
The frequency of wild-type alleles in the total population inevitably reflects the history of recent antimalarial drug use at any given site, because where the drug selection has been intense, wild-type alleles are increasingly displaced by resistant alleles. In the lower map of Figure 2 the size of the pie chart at each site indicates the proportion of wild-type alleles in the total population as indicated in the scale. The largest pie charts indicate populations in which the majority of parasite isolates were found to be wild type, and the smallest pie charts indicate where the wild types were rare. The relative proportions of SAK and AAK alleles within the wild-type population are indicated by the yellow and cyan segments. The AAK allele was most common in central Africa, accounting for more than half of the total wild-type alleles in Gabon 2007; Burkina Faso 2002; Guinea 2004–2005; Ghana 2005; Nigeria 2005; Mutengene, Cameroon 2004; Yaounde, Cameroon 2004; and Central African Republic 2004. Moving out from this area, the relative proportion of wild-type alleles that were AAK decreased. In the west; 28% of wild-type alleles were AAK in The Gambia 2004, 22% in Senegal 2003, 34% in Senegal 2004, and 53% in Mauritania 1998. Moving south, the proportion of AAK becomes increasingly rare, accounting for just 13% of wild-type alleles in Tanzania,6% in South Africa, and 5% in Mozambique; and absent in Namibia (0%) and Angola (0%).
Selective Sweeps around Resistant dhps Alleles
To confirm that resistant alleles had been subject to selection and to classify alleles according to shared ancestry we examined microsatellite diversity in the flanking region of dhps. Three microsatellite loci at 0.8 kb, 4.3 kb, and 7.7 kb from the 3′ end of dhps were successfully analysed in 1,674 unmixed samples from 20 geographical sites (the raw data are listed in full in the Table S2). A loss of diversity around resistant alleles, compared to that found surrounding the wild-type alleles, is evidence of directional selection and often referred to as a selective sweep. The expected heterozygosity (He) around each dhps allele at each geographical site was calculated extending outwards from the dhps gene and values are presented for loci at 0.8 kb, 4.3 kb, and 7.7 kb. In Figure 3 the diversity around wild-type (SAK and AAK) and resistant (AGK, SGK, and SGE) alleles are compared. The range of He values around the median are illustrated by box distributions, which show the interquartile range, and the whiskers, which show upper and lower extremes of the distribution.
The expected heterozygosity (He) at flanking loci 0.8 kb, 4.3 kb, and 7.7 kb from the dhps gene was calculated for each geographical site (provided the number of observation ≥10), and box plots show the median, interquartile ranges, and the upper and lower extremes of the distribution of He values among geographical sites. Where there are statistical outliers, these are indicated by small squares.
Diversity around wild-type alleles was universally high with little variation among geographical sites. Also there was no significant change in diversity with distance along the chromosome from either the SAK or the AAK dhps alleles, which is consistent with expectations for a locus that has not been subject to recent selection. The median values for He around the SAK alleles were 0.867 (0.8 kb), 0.778 (4.3 kb), and 0.905 (7.7 kb). Similarly, the diversity around AAK was high: 0.756 (0.8 kb), 0.667 (4.3 kb), and 0.824 (7.7 kb).
Contrasting with wild-type dhps alleles there were clear signatures of selection around all three alleles coding for the A437G substitution, confirming that these alleles have been subject to selection. Loss of diversity was most pronounced at loci flanking the SGE double-mutant allele, where He values were 0.116 (0.8 kb), 0.093 (4.3 kb), and 0.206 (7.7 kb). These values are significantly different from the He flanking the SAK haplotype (p<0.001 at all three sites when compared using Wilcoxon's rank sum test). There was also a loss of diversity around the AGK alleles, where equivalent He values were 0.653 (p = 0.005), 0.603 (p = 0.003), and 0.600 (p = 0.01), respectively, and SGK alleles, where He values were 0.561 (p = 0.0007), 0.532 (p = 0.0002), and 0.775 (p = 0.0115), respectively. The variability of He values among geographical populations was more pronounced with AGK and SGK alleles than for SGE alleles; this difference may be due to multiple lineages occurring within individual populations. “Soft” selective sweeps are found where multiple lineages are superimposed within a single population , causing He to be higher than in populations where a single lineage is present. We went on to examine how many lineages could be identified and to examine their geographical distribution among populations.
Multiple Origins of Resistant dhps
Resistance mutations that have common ancestry can be identified on the basis of flanking microsatellite polymorphism, because closely linked neutral markers are carried with the selected allele by hitch-hiking. Haplotypes of the two most closely linked markers (0.8 kb and 4.3 kb) were ranked, first according to allele size at locus 0.8 kb and then by allele size at locus 4.3 kb. In Figure 4 the ranked microsatellite haplotypes are listed along a common x-axis in three bar charts, which show their frequency among sensitive (Figure 4A), single 437 mutant alleles SGK and AGK (Figure 4B), and double-mutant 437+540 SGE alleles (Figure 4C). A complete list of microsatellite haplotypes is available in the Table S2.
In the bar graphs all the microsatellite haplotypes observed have been ranked first according to allele size at locus 0.8 kb and then by allele size at locus 4.3 kb along a common x-axis. The association of specific microsatellite haplotypes with different dhps alleles is apparent from the frequencies of each haplotype shown in the individual charts. (A) haplotypes linked to SAK AAK wild-type alleles, (B) haplotypes linked to AGK SGK single-mutant alleles, and (C) haplotypes linked to SGE double-mutant alleles.
The microsatellite haplotypes associated with SAK and AAK alleles (Figure 4A) were largely unique to every isolate, which is consistent with the expectation that they have not been under recent selection. In contrast, many of the resistance alleles were found to share a common flanking microsatellite haplotype, indicating that they have been under recent selection and that they were derived from the same ancestral mutant lineage. Inspection of the microsatellite haplotypes associated with SGK and AGK alleles (Figure 4B) shows that multiple lineages have emerged, and within these the S436A mutation appears to have been gained or lost on multiple occasions. Three lineages of an AGK/SGK allele were identified, shown in Figure 4B as AGK/SGK 1, AGK/SGK 2, and AGK/SGK 3. These lineages were defined on the basis of a shared allele size at the most proximal (0.8 kb) microsatellite locus. Occasional recombination in the flanking region at sites more distant from the gene accounted for some variability at the 4.3 kb locus, and isolates contained within these clusters of related haplotypes are highlighted by underlining in Figure 4B.
Two major clusters of microsatellite haplotypes were associated with the SGE double-mutant allele (Figure 4C). Within each cluster the microsatellite haplotypes all share the same allele at the closest microsatellite locus (0.8 kb), but may vary at locus 4.3 kb because of the increasing likelihood of recombination events with distance from the site of selection. Lineages defined on the basis of variation at the 0.8 kb locus are highlighted by underlining and named SGE1 and SGE2 in Figure 4C.
The Geographic Dispersal of Resistant Lineages
To examine the extent to which resistance alleles were exchanged among the populations at the 20 sites, we used pairwise population measure of resistance allele sharing. For the analysis we expressed the combination of point mutations and linked microsatellite alleles at the 0.8 kb and 4.3 kb loci as a single haplotype, and estimated DPS for a single locus.
Resistance allele sharing among the 20 geographical sites is summarised in the neighbour joining tree in Figure 5. Resistance allele sharing among the 20 sites reveals five regional clusters. Within these clusters the same lineages are common, but between them few or none of the resistance allele lineages were shared. The geographical distribution of resistance allele lineages and their representation in each of the 20 geographical sites are illustrated in the map in Figure 5. The geographical distribution of AGK/SGK 1, AGK/SGK 2, and AGK/SGK 3 lineages were each unique, with the concentration of each lineage implying their likely site of origin. The lineage AGK/SGK 1 was found predominantly in central and southwest African sites, Namibia, Angola, Congo, and Gabon, AGK/SGK 2 was found predominantly in the west African sites Senegal, Guinea, Burkina Faso, Ghana, and Nigeria, and AGK/SGK 3 was found predominantly in Cameroon. It is clear from the frequencies of resistance lineages expressed in the pie chart map in Figure 5 that there has been dispersal throughout west and central Africa from their original foci, with Cameroon at the confluence of west, central, and southwest African gene pools.
The distribution of the five major lineages among the geographic sites is indicated in the map. Resistance alleles whose flanking microsatellite haplotypes did not conform to a defined major lineage are shown in grey. Sharing of resistance allele lineages among the African populations is shown in a cladogram based on pairwise comparison of allele sharing (DPS), which includes all the flanking haplotypes identified. Closely related populations cluster in large geographic regions that supercede national boundaries.
The flanking haplotypes associated with SGE alleles revealed the existence of two lineages SGE1 and SGE2. A previous study has shown that SGE mutants from South Africa and northern Tanzania were derived from a one ancestral lineage . We found that the geographic range of that lineage (SGE1) not only includes South Africa and Tanzania but extends through Kenya, Uganda, Democratic Republic of Congo (DRC), Mozambique, and Zambia. Interestingly, the SGE alleles in Ethiopia and northeastern Sudan are descendants of an independently derived and regionally distinct lineage (SGE2).
Microsatellite Diversity Flanking Wild-Type dhps Alleles
Polymorphism at codon 436 is generally considered to be the ancestral state, and the high levels of diversity measured at the microsatellite loci flanking the SAK and AAK alleles when all geographical populations were compared was consistent with the ancestral sensitive state. There was, however, some evidence of weak selection acting upon the AAK allele when parasites from individual geographical sites are examined separately. One flanking haplotype was found at relatively high frequency among the samples with the AAK allele at dhps (this lineage is defined as type H73 in Table S2) in Cameroon (Yaounde n = 9, Mutengene n = 3), Gabon (n = 5), and Nigeria (n = 2), indicating a recently selected expansion of this allele in that region. We infer that selection was attributable to the mutation at codon 436. We tested for additional mutations at codons 581 and 613 in these samples and none were found, but we cannot entirely exclude the possibility that other linked adaptations  might be involved.
If the 436A mutation consistently confers a selective advantage, we would predict that the reduction of diversity around AAK would be detectable in other geographic regions. To test this prediction we compared the diversity in the 0.8 kb microsatellite locus linked to the SAK and AAK alleles in each region. The northeast Africa region was excluded from this analysis because AAK and SAK alleles were very rare in the Ethiopia and Sudan samples. Table 2 shows a significant loss of heterozygosity at the 0.8 kb locus among AAK alleles relative to SAK alleles in all regions except the southeast. The margin of difference was greatest in Cameroon and in southwest Africa (p-values are shown in Table 2).
The significant reduction in diversity around AAK relative to SAK in each region suggests that multiple lineage expansion events have occurred independently in the different regions. If this were the case we would expect to see greater regional differentiation among the AAK-linked microsatellite haplotypes than among those linked to sensitive SAK alleles. Pairwise genetic identity among SAK-linked microsatellite haplotypes from the four regions was calculated using all three flanking microsatellites. Table 3 indicates that they are all of a similar level of identity, whereas pairwise comparisons of populations using genetic identity among AAK-linked microsatellites in the three populations have much lower identity (Table 3). The greater dissimilarity among AAK was a significant departure from expected (test by permutation).
We examined the contemporary distribution of dhps resistance mutations and found that single, codon 437–mutant (AGK/SGK) alleles were predominant in west and central Africa while the double, codon 437– and 540–mutant (SGE) alleles prevailed throughout east Africa. Flanking sequence analysis showed multiple origins of both single- and double-mutant alleles: three major AGK/SGK lineages and two major SGE lineages. All had been subject to recent selection and each had a highly distinctive regional geographic distribution. In southeast Africa SGE 1 was predominant, while in Northeast Africa SGE 2 prevailed. The three AGK/SGK lineages predominated in different regions: AGK/SGK1 in the southwest, AGK/SGK2 in west Africa, and AGK/SGK3 in central Africa (Cameroon).
Explaining the Geographical Distributions
The difference between east and west African parasites can most economically be explained as a consequence of the limited number of resistant dhps lineages. Each emerged in distinct geographical foci and subsequently became dispersed across a wide region. It has previously been observed with reference to pyrimethamine and chloroquine that the rate of emergence of mutant parasite lineages is far less frequent than might be predicted based on the mutation rate and the number of parasites in each human infection , and our findings in relation to the origins of resistant dhps are consistent with this observation.
A review of the literature indicates that emergence of all the major resistance lineages took place in the early- to mid-1990s. The A437G substitution was widespread throughout west and central Africa by 1995, reported at prevalences of 37% (27 of 72) in Cameroon in 1995 , 28% (10 of 36) in Gabon in 1995 , 25% (12 of 48) in Mali 1995 , and 25% (8 of 32) in Mali again in 1995 . The distribution of these early observations could imply that all three of the AGK/SGK lineages had emerged and increased in frequency to detectable levels prior to 1995, although it will be necessary for flanking microsatellite analysis to be performed on archived samples to establish this with certainty. The predominance of AGK/SGK lineages hints at their region-specific origins in southwest, west, or central Africa, but the extent of allele sharing between regions demonstrates that parasite migration across these large geographic distances in the intervening years was extensive.
Molecular studies suggest the emergence of SGE 1 and SGE 2 lineages in east Africa also occurred in the mid 1990s. SGE was first reported in samples collected in Kenya during 1993–1995 , in Tanzania in 1995 , and Malawi in 1995–1996  while in KwaZulu-Natal, South Africa it was absent in 1995–1996 but had appeared by 1999 . We found only the SGE 1 lineage in these countries and infer therefore that these reports describe the emergence of that lineage. The emergence of the SGE 2 lineage occurred in Ethiopia and northeastern Sudan at around the same time. A time series of dhps analyses in northeastern Sudan shows the SGE was absent in 1993 but had appeared by 1998 . Although there are no molecular studies of dhps in Ethiopia in the same era, three studies in 2004 at widely dispersed geographical sites (this study and , ) all found that the allele was fixed or almost fixed. This indicates that it had been subject to strong drug selection pressure and perhaps that Ethiopia was the site of its first emergence. Ethiopia officially changed the first-line treatment to SP in 1999 but prior to that SP was widely used for treatment for at least 5 years.
Contrasting with the situation in West and Central Africa we found no mixing of SGE1 and SGE2 between the sites sampled in northeast and southeast Africa. Although the P. falciparum populations of Africa are often considered a continuous and largely homogeneous whole, it is likely that there are geospecific factors that promote or restrict the dispersal of mutations through migration of parasites. In this case it seems likely that political instability and civil war placed greater restrictions on travel across conflict zones during this period of dhps dispersal. More detailed spatial genetic analysis on the margins of lineage distributions would provide a more precise indication of the forces that govern the dispersal of resistance in these areas.
There was also a marked transition between east and west Africa where the margins of the SGE and AGK/SGK lineage distributions meet. High-resolution mapping of parasite genetics would be valuable for understanding the dynamics underlying these observations. A previous study that compared parasites from multiple sites in DRC during 2003–2004  found the prevalence of K540E was 13.3%–19.3% in eastern DRC but declined to 0.9%–3.9% in western DRC, indicating restricted population mobility between east and west, in this case, undoubtedly exacerbated by a longstanding war during 1998–2003.
This snapshot of the emergence and dispersal of resistant dhps within Africa provides an interesting counterpoint to the inferred histories of mutant pfcrt and dhfr which emerged in Africa some 10–20 years earlier. In those cases a single highly resistant lineage was imported to Africa from Asia and became established in populations throughout the continent. Since the emergence of resistant dhps is more recent, it could be argued that the observed distribution of dhps resistance alleles is transitional and that given the equivalent amount of time under selection just one resistance lineage would eventually predominate. The SGE 1 lineage is already found in small numbers in Cameroon, Ghana, Guinea, Namibia, Gabon, and Congo, but not in Angola, The Gambia, Mauritania, Mali, Senegal, or Cote d'Ivoire (or Nigeria in this study, although another study has reported the occurrence of 540E in Nigeria) . It is possible that if strong selection through heavy reliance on SP were continued, these foci of SGE would expand, and eventually displace AGK/SGK. The resulting picture would then perhaps be similar to dhfr. There are mildly resistant dhfr double-mutant lineages believed to be of African origin that were shown to be displaced and outcompeted by the Asian-derived triple-mutant allele , , with the result that one highly resistant lineage now prevails almost everywhere.
Could migration between east Africa and Asia explain the introgression of SGE alleles in east Africa? Analysis of microsatellites flanking dhps in Southeast Asian parasites has not been published but studies in Bangladesh  and India  show that the SGK/AGK and SGE haplotypes are both common, with SGK/AGK in the majority. This contrasts with the situation in east Africa where SGK/AGK was rare or absent, even before the SGE became so highly prevalent. There is therefore no compelling evidence that the SGE in east Africa is due to extensive parasite exchange between east Africa and Asia. It is noteworthy, however, that we observed an exact match of our SGE 1 flanking haplotype with that of our Southeast Asian control (K1) (details are in Table S2). It is clear that global mapping of dhps resistance lineages is needed and the ongoing efforts by local investigators and by the wider research community to assemble a global geography of drug resistance will undoubtedly shed further light on this question .
Parasite Dispersal through Human Migration
The distribution of resistance alleles highlights the importance of human migration in dispersing resistance and parasite infection generally. The regions we defined here on the basis of dhps resistance allele-sharing strongly suggest that the economic and transport infrastructures may indirectly govern movement of parasites in Africa through their influence upon volumes of human migration. The regions of allele sharing broadly correspond to African economic communities, which were established to facilitate and promote trade. It would be interesting to explore the contribution of the migrant work force to the dispersal of specific lineages. For example, economic agreements between Gabon and Senegal during the early to mid 1990s were favourable to migrant workers , provides a possible explanation for the moderately high frequencies of the AGK/SGK 1 in Senegal, and higher frequencies of AGK/SGK 2 in Gabon compared to its southwest African neighbours.
The first eradication campaigns showed that movement of malaria parasites by human migration can quickly undermine the successful interruption of transmission . Therefore, the impact of control interventions will always be maximised when applied at a geographical scale that encompasses regions of significant volumes of parasite exchange. In the new era of elimination it will be important to understand the forces that govern parasite migration. Dispersal patterns of drug resistance mutations followed in real time can uniquely illustrate the extent and direction of contemporary parasite migration, and further mapping of dispersal of resistance mutations at other loci across Africa, for example the double-mutant dhfr lineages which are believed to have emerged de novo in Africa, could confirm the definition of regions of significant parasite exchange. In that context, coordinated campaigns within economic areas such as the Southern African Development Community (SADC) will be more likely to succeed than campaigns within defined national territories that will face an uphill struggle against importation of malaria.
Implications for Prevention and Treatment
In regions where the dhps mutant lineages converged upon the same mutant dhps haplotype, we would predict that the SP resistance phenotype of parasites will be equivalent, provided co-adaptive changes at other loci such as dhfr or GTP-cyclohydrolase 1 (gch1)  are also the same. In Sudan and Ethiopia, where we found the SGE 2 lineage, there are also significant differences in the common point mutation haplotypes at dhfr , , , and a suggestion that there may be a different SP resistance phenotype . Of note, in-vitro studies ,  indicate that there is higher drug tolerance in double-mutant parasites. To establish whether the different resistance genotypes that predominate in the different geographical regions require different approaches to clinical care, a direct comparison of their drug tolerance levels in vitro as well as in practice is needed. In the absence of such data, evaluation of antifolate-based interventions should be carried out in each of the five lineage-defined geographic regions.
In particular, although growing resistance problems have led to the withdrawal of SP as a first-line treatment for malaria, it is still currently recommended in combination with artesunate in areas where SP resistance is low, and as monotherapy for use as Intermittent Preventive Treatment of Malaria in Pregnancy (IPTp) in all malaria-endemic countries in sub-Saharan Africa . There is strong evidence of a continuing efficacy of SP for IPTp even in areas where SP resistance is well established , and SP is also being investigated for use as Intermittent Preventive Treatment of Malaria in Infants (IPTi) . In light of our findings, the continuing assessment of SP efficacy in IPTi, IPTp, and/or clinical treatments that include SP or any other antifolate should be carried out in both east and west African sites.
This study provides a snapshot of the geographical distributions of drug resistance alleles, but these distributions are not static, and it will be important to continue monitoring. By systematic sampling around the boundaries of lineage distributions and in those countries not included so far, a more complete picture of dhps allele distribution can be generated. This information will improve our understanding of the true constraints on dispersal of dhps mutant alleles at the extremes of the geographical distributions outlined here. It is perhaps surprising that so few studies have attempted to synthesise a geographical analysis of genetic variation in African P. falciparum. One difficulty is that research tends to be concentrated into a small number of very well-characterised sites. Another obstacle, when collating data from the published literature, is the absence of a standardised reporting format. In our review of molecular drug resistance data there were many cases in which point mutation data were not presented in a way that allowed the inference of haplotypes—a necessary precondition for inclusion in our analysis. We support the development of a programme to standardise reporting with reference to well-characterised controls, decrease the lag between sampling and publication, and increase the availability of drug resistance data as proposed in the establishment of WARN (World Wide Antimalarial Resistance Network) .
The global movement of resistant malaria has played a decisive role in the establishment of both CQ and pyrimethamine resistance in Africa –, but there are significant gaps in our knowledge of when, where, and how resistant genotypes became established on the African continent. Defining the forces that govern the dispersal and successful establishment of resistance genes in Africa is a significant challenge, but will be imperative if emergent resistance to new drug treatments or vaccines is to be managed effectively.
Frequencies of dhps alleles for sites shown in Figure 2.
(0.04 MB PDF)
Microsatellite allele data.
(0.08 MB PDF)
Details of samples used in this study, including informed consent and ethical approval.
(0.11 MB DOC)
This paper is dedicated to the memory of our colleague and coauthor, Brian Sharp, who died before it could be completed but whose enthusiasm for the project was an inspiration. We thank Brian Greenwood, Geoff Target, and Amit Bhasin of the GMP for their tremendous support.
Brian Sharp and Samuel Dunyo died before publication of this paper. The corresponding author, Cally Roper, has therefore supplied the information regarding their contribution to the manuscript and their competing interests with assistance from Dr David Conway (MRC Gambia) and it is correct to the best of her knowledge.
We also thank: the staff of the Uige Provincial Hospital, Uige, Angola for their part in sample collection in Angola, and Carlo Severini and Michela Menegon of the Department of Infectious, Parasitic and Immunomediated Diseases, Istituto Superiore di Sanità, Rome, Italy for their contributions to the analysis of Angolan samples; the MSF team in Congo for their part in sample collection in Kindamba; Ambachew Medhin from WHO Ethiopia and Manica Balasegaram of MSF London for their part in sample collection in Ethiopia; Bertrand Lell, Julian Gabor, Sunny Oyakhirome, Katja Szywon, Juergen F. J. Kun, and Peter G. Kremsner for their part in the sample collection in Gabon; Davis Nwakanma and David Conway for managing the lab work on the samples in The Gambia; Gayathri Warnasuriya, N. K. O. Ibecheozor, and J. O. Ajobiewe for their part in sample collection in Nigeria.
Finally, we acknowledge the malaria patients, their communities and the staff of the many institutions who participated in this work.
ICMJE criteria for authorship read and met: R Pearce H Pota M Evehe E Ba G Mombo-Ngoma A Malisa R Ord W Inojosa A Matondo D Diallo W Mbacham I van den Broek T Swarthout A Getachew S Dejene M Grobusch F Njie S Dunyo M Kweku S Owusu-Agyei D Chandramohan M Bonnet J Guthmann S Clarke K Barnes E Streat S Katokele P Uusiku C Agboghoroma O Elegba B Cissé I A-Elbasit H Giha S Kachur C Lynch J Rwakimari P Chanda M Hawela B Sharp I Naidoo C Roper. Agree with the manuscript's results and conclusions: R Pearce H Pota M Evehe E Ba G Mombo-Ngoma A Malisa R Ord W Inojosa A Matondo D Diallo W Mbacham I van den Broek T Swarthout A Getachew S Dejene M Grobusch F Njie S Dunyo M Kweku S Owusu-Agyei D Chandramohan M Bonnet J Guthmann S Clarke K Barnes E Streat S Katokele P Uusiku C Agboghoroma O Elegba B Cissé I A-Elbasit H Giha S Kachur C Lynch J Rwakimari P Chanda M Hawela B Sharp I Naidoo C Roper. Designed the experiments/the study: H Pota W Mbacham M Kweku S Owusu-Agyei C Lynch C Roper. Analyzed the data: R Pearce H Pota M Evehe A Malisa M Grobusch S Clarke B Cissé C Lynch C Roper. Collected data/did experiments for the study: R Pearce H Pota M Evehe G Mombo-Ngoma R Ord W Inojosa D Diallo W Mbacham I van den Broek T Swarthout A Getachew S Dejene F Njie S Dunyo M Kweku S Owusu-Agyei D Chandramohan J Guthmann S Clarke K Barnes S Katokele P Uusiku C Agboghoroma O Elegba I A-Elbasit H Giha S Kachur C Lynch P Chanda M Hawela B Sharp I Naidoo. Enrolled patients: W Inojosa A Matondo D Diallo W Mbacham T Swarthout A Getachew S Dejene S Dunyo M Kweku S Owusu-Agyei M Bonnet S Clarke K Barnes S Katokele P Uusiku B Cissé I A-Elbasit H Giha C Lynch P Chanda M Hawela. Wrote the first draft of the paper: R Pearce C Roper. Contributed to the writing of the paper: R Pearce A Malisa D Diallo W Mbacham M Grobusch S Owusu-Agyei D Chandramohan M Bonnet J Guthmann S Clarke K Barnes C Agboghoroma B Cissé I A-Elbasit H Giha S Kachur C Lynch J Rwakimari P Chanda I Naidoo C Roper. Performed microsatellite lab analysis for all populations and typing of point mutations in the dhps gene for some of the populations: R Pearce. Performed lab experiments (DNA extraction PCR and point mutation analysis) for many of the populations involved in the study: H Pota. Collected and genotyped samples from Yaounde, Garoua and Mutengene in Cameroon in the Gates Malaria funded project of W Mbacham.: M Evehe. Collected data on patients in Gabon; after completion of the clinical trial performed DNA extraction, PCR, amplification, and point mutation characterization on those samples: G Mombo-Ngoma. Contributed to the molecular genotyping of samples: R Ord. Enrolled malaria patients and collected samples in the study at Uige Hospital Angola in 2004: W Inojosa. Leading researcher of antimalarial drug studies in DRC and Congo, where blood samples of malaria patients were collected, and provided all the information necessary: I van den Broek. As trial coordinator in Shabunda, Democratic Republic of the Congo, coordinated the trial and participated in enrolling patients and collection of biological samples: T Swarthout. Involved in facilitating in the process of patient enrollment collection of specimen and preservation, including the safe shipment of the specimens to the reference laboratory: A Getachew. Packed and shipped sent samples for further advanced lab analysis: S Dejene. Supervised the study-related field work and patient enrollment in Lambarene, Gabon: M Grobusch. Prepared and provided samples for the study: F Njie. Designed coordinated analyzed and published several of the studies that contributed to this analysis: J Guthmann. Designed and implemented a study to examine drug efficacy and selection for drug resistance (which included the genotyping analysis and interpretation of molecular markers of SP resistance) amongst schoolchildren receiving intermittent preventive treatment in Western Kenya in 2006: S Clarke. Collected samples for the study in Namibia: S Katokele. Responsible for the supervision and funding of all field work in Eastern Sudan, collection of clinical data blood samples, DNA extraction, and part of the genotyping: I A-Elbasit H Giha. Designed studies and collected data from southern Tanzania sites that were included in the analysis presented in this paper: S Kachur. Involved in the Ugandan part of the study, which involved enrollment of patients at two sites in Uganda collection of samples, and data analysis: C Lynch. Conducted data collection for the Zambia samples, including site identification and supervision of field work and mobilised resources to meet financial gaps of the study: P Chanda. Collated literature review data on dhps and facilitated collaboration for sample collection in sites in two countries: I Naidoo.
- 1. Roper C, Pearce R, Nair S, Sharp B, Nosten F, et al. (2004) Intercontinental spread of pyrimethamine-resistant malaria. Science 305: 1124.
- 2. Su X, Kirkman LA, Fujioka H, Wellems TE (1997) Complex polymorphisms in an approximately 330 kDa protein are linked to chloroquine-resistant P. falciparum in Southeast Asia and Africa. Cell 91: 593–603.
- 3. Wootton JC, Feng X, Ferdig MT, Cooper RA, Mu J, et al. (2002) Genetic diversity and chloroquine selective sweeps in Plasmodium falciparum. Nature 418: 320–323.
- 4. Payne D (1989) The History and Development of WHO Standard in vivo and in vitro Test Systems for the Sensitivity of Plasmodium falciparum and other Human Plasmodia to Antimalarial Drugs. London: University of London (London School of Hygiene and Tropical Medicine).
- 5. Charmot G, Amat-Roze JM, Rodhain F, Le Bras J, Coulaud JP (1991) [Geographic approach to the epidemiology of chloroquine-resistance of Plasmodium falciparum in tropical Africa]. Ann Soc Belg Med Trop 71: 187–197.
- 6. Fidock DA, Nomura T, Talley AK, Cooper RA, Dzekunov SM, et al. (2000) Mutations in the P. falciparum digestive vacuole transmembrane protein PfCRT and evidence for their role in chloroquine resistance. Mol Cell 6: 861–871.
- 7. Ariey F, Fandeur T, Durand R, Randrianarivelojosia M, Jambou R, et al. (2006) Invasion of Africa by a single pfcrt allele of South East Asian type. Malar J 5: 34.
- 8. Cowman AF, Morry MJ, Biggs BA, Cross GA, Foote SJ (1988) Amino acid changes linked to pyrimethamine resistance in the dihydrofolate reductase-thymidylate synthase gene of Plasmodium falciparum. Proc Natl Acad Sci U S A 85: 9109–9113.
- 9. Peterson DS, Walliker D, Wellems TE (1988) Evidence that a point mutation in dihydrofolate reductase-thymidylate synthase confers resistance to pyrimethamine in falciparum malaria. Proc Natl Acad Sci U S A 85: 9114–9118.
- 10. Brooks DR, Wang P, Read M, Watkins WM, Sims PF, et al. (1994) Sequence variation of the hydroxymethyldihydropterin pyrophosphokinase: Dihydropteroate synthase gene in lines of the human malaria parasite, Plasmodium falciparum, with differing resistance to sulfadoxine. Eur J Biochem 224: 397–405.
- 11. Triglia T, Cowman AF (1994) Primary structure and expression of the dihydropteroate synthetase gene of Plasmodium falciparum. Proc Natl Acad Sci U S A 91: 7149–7153.
- 12. Plowe CV, Cortese JF, Djimde A, Nwanyanwu OC, Watkins WM, et al. (1997) Mutations in Plasmodium falciparum dihydrofolate reductase and dihydropteroate synthase and epidemiologic patterns of pyrimethamine-sulfadoxine use and resistance. J Infect Dis 176: 1590–1596.
- 13. Roper C, Pearce R, Bredenkamp B, Gumede J, Drakeley C, et al. (2003) Antifolate antimalarial resistance in southeast Africa: A population-based analysis. Lancet 361: 1174–1181.
- 14. Ndiaye D, Daily JP, Sarr O, Ndir O, Gaye O, et al. (2005) Mutations in Plasmodium falciparum dihydrofolate reductase and dihydropteroate synthase genes in Senegal. Trop Med Int Health 10: 1176–1179.
- 15. McCollum AM, Poe AC, Hamel M, Huber C, Zhou Z, et al. (2006) Antifolate resistance in Plasmodium falciparum: multiple origins and identification of novel dhfr alleles. J Infect Dis 194: 189–197.
- 16. Maiga O, Djimde AA, Hubert V, Renard E, Aubouy A, et al. (2007) A shared Asian origin of the triple-mutant dhfr allele in Plasmodium falciparum from sites across Africa. J Infect Dis 196: 165–172.
- 17. Lynch C, Pearce R, Pota H, Cox J, Abeku TA, et al. (2008) Emergence of a dhfr mutation conferring high-level drug resistance in Plasmodium falciparum populations from Southwest Uganda. J Infect Dis 197: 1598–1604.
- 18. Certain LK, Briceno M, Kiara SM, Nzila AM, Watkins WM, et al. (2008) Characteristics of Plasmodium falciparum dhfr haplotypes that confer pyrimethamine resistance, Kilifi, Kenya, 1987–2006. J Infect Dis 197: 1743–1751.
- 19. Ronn AM, Msangeni HA, Mhina J, Wernsdorfer WH, Bygbjerg IC (1996) High level of resistance of Plasmodium falciparum to sulfadoxine-pyrimethamine in children in Tanzania. Trans R Soc Trop Med Hyg 90: 179–181.
- 20. Trigg JK, Mbwana H, Chambo O, Hills E, Watkins W, et al. (1997) Resistance to pyrimethamine/sulfadoxine in Plasmodium falciparum in 12 villages in north east Tanzania and a test of chlorproguanil/dapsone. Acta Trop 63: 185–189.
- 21. Omar SA, Adagu IS, Warhurst DC (2001) Can pretreatment screening for dhps and dhfr point mutations in Plasmodium falciparum infections be used to predict sulfadoxine-pyrimethamine treatment failure? Trans R Soc Trop Med Hyg 95: 315–319.
- 22. Kublin JG, Dzinjalamala FK, Kamwendo DD, Malkin EM, Cortese JF, et al. (2002) Molecular markers for failure of sulfadoxine-pyrimethamine and chlorproguanil-dapsone treatment of Plasmodium falciparum malaria. J Infect Dis 185: 380–388.
- 23. Staedke SG, Sendagire H, Lamola S, Kamya MR, Dorsey G, et al. (2004) Relationship between age, molecular markers, and response to sulphadoxine-pyrimethamine treatment in Kampala, Uganda. Trop Med Int Health 9: 624–629.
- 24. Kun JF, Lehman LG, Lell B, Schmidt-Ott R, Kremsner PG (1999) Low-dose treatment with sulfadoxine-pyrimethamine combinations selects for drug-resistant Plasmodium falciparum strains. Antimicrob Agents Chemother 43: 2205–2208.
- 25. Mockenhaupt FP, Teun Bousema J, Eggelte TA, Schreiber J, Ehrhardt S, et al. (2005) Plasmodium falciparum dhfr but not dhps mutations associated with sulphadoxine-pyrimethamine treatment failure and gametocyte carriage in northern Ghana. Trop Med Int Health 10: 901–908.
- 26. Dunyo S, Ord R, Hallett R, Jawara M, Walraven G, et al. (2006) Randomised trial of chloroquine/sulphadoxine-pyrimethamine in Gambian children with malaria: impact against multidrug-resistant P. falciparum. PLoS Clin Trials 1: e14.
- 27. Ndounga M, Tahar R, Basco LK, Casimiro PN, Malonga DA, et al. (2007) Therapeutic efficacy of sulfadoxine-pyrimethamine and the prevalence of molecular markers of resistance in under 5-year olds in Brazzaville, Congo. Trop Med Int Health 12: 1164–1171.
- 28. Marks F, Evans J, Meyer CG, Browne EN, Flessner C, et al. (2005) High prevalence of markers for sulfadoxine and pyrimethamine resistance in Plasmodium falciparum in the absence of drug pressure in the Ashanti region of Ghana. Antimicrob Agents Chemother 49: 1101–1105.
- 29. Pearce RJ, Drakeley C, Chandramohan D, Mosha F, Roper C (2003) Molecular determination of point mutation haplotypes in the dihydrofolate reductase and dihydropteroate synthase of Plasmodium falciparum in three districts of northern Tanzania. Antimicrob Agents Chemother 47: 1347–1354.
- 30. R_Development_Core_Team (2008) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing.
- 31. Bowcock AM, Ruiz-Linares A, Tomfohrde J, Minch E, Kidd JR, et al. (1994) High resolution of human evolutionary trees with polymorphic microsatellites. Nature 368: 455–457.
- 32. Felsenstein J (1989) PHYLIP - Phylogeny Inference Package (Version 3.2). Cladistics 5: 164–166.
- 33. Nash D, Nair S, Mayxay M, Newton PN, Guthmann JP, et al. (2005) Selection strength and hitchhiking around two anti-malarial resistance genes. Proc Biol Sci 272: 1153–1161.
- 34. Nei M, Roychoudhury AK (1974) Sampling variances of heterozygosity and genetic distance. Genetics 76: 379–390.
- 35. Djaman JA, Mazabraud A, Basco L (2007) Sulfadoxine-pyrimethamine susceptibilities and analysis of the dihydrofolate reductase and dihydropteroate synthase of Plasmodium falciparum isolates from Cote d'Ivoire. Ann Trop Med Parasitol 101: 103–112.
- 36. Nzila AM, Mberu EK, Sulo J, Dayo H, Winstanley PA, et al. (2000) Towards an understanding of the mechanism of pyrimethamine-sulfadoxine resistance in Plasmodium falciparum: genotyping of dihydrofolate reductase and dihydropteroate synthase of Kenyan parasites. Antimicrob Agents Chemother 44: 991–996.
- 37. Triglia T, Wang P, Sims PF, Hyde JE, Cowman AF (1998) Allelic exchange at the endogenous genomic locus in Plasmodium falciparum proves the role of dihydropteroate synthase in sulfadoxine-resistant malaria. EMBO J 17: 3807–3815.
- 38. Gebru-Woldearegai T, Hailu A, Grobusch MP, Kun JF (2005) Molecular surveillance of mutations in dihydrofolate reductase and dihydropteroate synthase genes of Plasmodium falciparum in Ethiopia. Am J Trop Med Hyg 73: 1131–1134.
- 39. Schunk M, Kumma WP, Miranda IB, Osman ME, Roewer S, et al. (2006) High prevalence of drug-resistance mutations in Plasmodium falciparum and Plasmodium vivax in southern Ethiopia. Malar J 5: 54.
- 40. Nair S, Nash D, Sudimack D, Jaidee A, Barends M, et al. (2007) Recurrent gene amplification and soft selective sweeps during evolution of multidrug resistance in malaria parasites. Mol Biol Evol 24: 562–573.
- 41. Dasgupta T, Anderson KS (2008) Probing the role of parasite-specific, distant structural regions on communication and catalysis in the bifunctional thymidylate synthase-dihydrofolate reductase from Plasmodium falciparum. Biochemistry 47: 1336–1345.
- 42. Basco LK, Tahar R, Keundjian A, Ringwald P (2000) Sequence variations in the genes encoding dihydropteroate synthase and dihydrofolate reductase and clinical response to sulfadoxine-pyrimethamine in patients with acute uncomplicated falciparum malaria. J Infect Dis 182: 624–628.
- 43. Diourte Y, Djimde A, Doumbo OK, Sagara I, Coulibaly Y, et al. (1999) Pyrimethamine-sulfadoxine efficacy and selection for mutations in Plasmodium falciparum dihydrofolate reductase and dihydropteroate synthase in Mali. Am J Trop Med Hyg 60: 475–478.
- 44. Curtis J, Duraisingh MT, Warhurst DC (1998) In vivo selection for a specific genotype of dihydropteroate synthetase of Plasmodium falciparum by pyrimethamine-sulfadoxine but not chlorproguanil-dapsone treatment. J Infect Dis 177: 1429–1433.
- 45. Abdel-Muhsin AM, Mackinnon MJ, Ali E, Nassir el KA, Suleiman S, et al. (2004) Evolution of drug-resistance genes in Plasmodium falciparum in an area of seasonal malaria transmission in Eastern Sudan. J Infect Dis 189: 1239–1244.
- 46. Cohuet S, Bonnet M, Van Herp M, Van Overmeir C, D'Alessandro U, et al. (2006) Short report: molecular markers associated with Plasmodium falciparum resistance to sulfadoxine-pyrimethamine in the Democratic Republic of Congo. Am J Trop Med Hyg 75: 152–154.
- 47. Happi CT, Gbotosho GO, Folarin OA, Akinboye DO, Yusuf BO, et al. (2005) Polymorphisms in Plasmodium falciparum dhfr and dhps genes and age related in vivo sulfadoxine-pyrimethamine resistance in malaria-infected patients from Nigeria. Acta Trop 95: 183–193.
- 48. van den Broek IV, van der Wardt S, Talukder L, Chakma S, Brockman A, et al. (2004) Drug resistance in Plasmodium falciparum from the Chittagong Hill Tracts, Bangladesh. Trop Med Int Health 9: 680–687.
- 49. Ahmed A, Lumb V, Das MK, Dev V, Wajihullah , et al. (2006) Prevalence of mutations associated with higher levels of sulfadoxine-pyrimethamine resistance in Plasmodium falciparum isolates from Car Nicobar Island and Assam, India. Antimicrob Agents Chemother 50: 3934–3938.
- 50. Plowe CV, Roper C, Barnwell JW, Happi CT, Joshi HH, et al. (2007) World Antimalarial Resistance Network (WARN) III: molecular markers for drug resistant malaria. Malar J 6: 121.
- 51. Gnisci D, Trémolières M, Bossard L (2006) Population series - Migration. Atlas on Regional Integration in West Africa. Available: http://www.oecd.org/dataoecd/41/47/38409521.pdf. Accessed November 2008.
- 52. Prothero RM (1965) Migrants and Malaria. London: Longsmans, Green and Co Ltd.
- 53. Nair S, Miller B, Barends M, Jaidee A, Patel J, et al. (2008) Adaptive copy number evolution in malaria parasites. PLoS Genet 4: e1000243.
- 54. Anderson TJ, Nair S, Jacobzone C, Zavai A, Balkan S (2003) Molecular assessment of drug resistance in Plasmodium falciparum from Bahr El Gazal province, Sudan. Trop Med Int Health 8: 1068–1073.
- 55. A-Elbasit IE, Khalil IF, Elbashir MI, Masuadi EM, Bygbjerg IC, et al. (2008) High frequency of Plasmodium falciparum CICNI/SGEAA and CVIET haplotypes without association with resistance to sulfadoxine/pyrimethamine and chloroquine combination in the Daraweesh area, in Sudan. Eur J Clin Microbiol Infect Dis. Available: http://www.springerlink.com/content/70473nx0q6456q10/. Accessed: 29 March 2008.
- 56. Triglia T, Menting JG, Wilson C, Cowman AF (1997) Mutations in dihydropteroate synthase are responsible for sulfone and sulfonamide resistance in Plasmodium falciparum. Proc Natl Acad Sci U S A 94: 13944–13949.
- 57. Peters PJ, Thigpen MC, Parise ME, Newman RD (2007) Safety and toxicity of sulfadoxine/pyrimethamine: implications for malaria prevention in pregnancy using intermittent preventive treatment. Drug Saf 30: 481–501.
- 58. ter Kuile FO, van Eijk AM, Filler SJ (2007) Effect of sulfadoxine-pyrimethamine resistance on the efficacy of intermittent preventive therapy for malaria control during pregnancy: a systematic review. JAMA 297: 2603–2616.
- 59. Grobusch MP, Egan A, Gosling RD, Newman RD (2007) Intermittent preventive therapy for malaria: progress and future directions. Curr Opin Infect Dis 20: 613–620.
- 60. Tinto H, Ouedraogo JB, Zongo I, van Overmeir C, van Marck E, et al. (2007) Sulfadoxine-pyrimethamine efficacy and selection of Plasmodium falciparum DHFR mutations in Burkina Faso before its introduction as intermittent preventive treatment for pregnant women. Am J Trop Med Hyg 76: 608–613.
- 61. Menard D, Yapou F, Manirakiza A, Djalle D, Matsika-Claquin MD, et al. (2006) Polymorphisms in pfcrt, pfmdr1, dhfr genes and in vitro responses to antimalarials in Plasmodium falciparum isolates from Bangui, Central African Republic. Am J Trop Med Hyg 75: 381–387.
- 62. Nsimba B, Jafari-Guemouri S, Malonga DA, Mouata AM, Kiori J, et al. (2005) Epidemiology of drug-resistant malaria in Republic of Congo: Using molecular evidence for monitoring antimalarial drug resistance combined with assessment of antimalarial drug use. Trop Med Int Health 10: 1030–1037.
- 63. van den Broek I, Kitz C, Al Attas S, Libama F, Balasegaram M, et al. (2006) Efficacy of three artemisinin combination therapies for the treatment of uncomplicated Plasmodium falciparum malaria in the Republic of Congo. Malar J 5: 113.
- 64. Swarthout TD, van den Broek IV, Kayembe G, Montgomery J, Pota H, et al. (2006) Artesunate+amodiaquine and artesunate+sulphadoxine-pyrimethamine for treatment of uncomplicated malaria in Democratic Republic of Congo: A clinical trial with determination of sulphadoxine and pyrimethamine-resistant haplotypes. Trop Med Int Health 11: 1503–1511.
- 65. Berzosa PJ, Puente S, Benito A (2005) Malaria cure with sulphadoxine/pyrimethamine combination in 12 semi-immune adults from West-Central Africa with high rates of point mutations in Plasmodium falciparum dhfr and dhps genes. Parasitol Res 97: 287–289.
- 66. Aubouy A, Jafari S, Huart V, Migot-Nabias F, Mayombo J, et al. (2003) DHFR and DHPS genotypes of Plasmodium falciparum isolates from Gabon correlate with in vitro activity of pyrimethamine and cycloguanil, but not with sulfadoxine-pyrimethamine treatment efficacy. J Antimicrob Chemother 52: 43–49.
- 67. Bonnet M, Roper C, Felix M, Coulibaly L, Kankolongo GM, et al. (2007) Efficacy of antimalarial treatment in Guinea: in vivo study of two artemisinin combination therapies in Dabola and molecular markers of resistance to sulphadoxine-pyrimethamine in N'Zerekore. Malar J 6: 54.
- 68. Kofoed PE, Alfrangis M, Poulsen A, Rodrigues A, Gjedde SB, et al. (2004) Genetic markers of resistance to pyrimethamine and sulfonamides in Plasmodium falciparum parasites compared with the resistance patterns in isolates of Escherichia coli from the same children in Guinea-Bissau. Trop Med Int Health 9: 171–177.
- 69. Bwijo B, Kaneko A, Takechi M, Zungu IL, Moriyama Y, et al. (2003) High prevalence of quintuple mutant dhps/dhfr genes in Plasmodium falciparum infections seven years after introduction of sulfadoxine and pyrimethamine as first line treatment in Malawi. Acta Trop 85: 363–373.
- 70. Wang P, Lee CS, Bayoumi R, Djimde A, Doumbo O, et al. (1997) Resistance to antifolates in Plasmodium falciparum monitored by sequence analysis of dihydropteroate synthetase and dihydrofolate reductase alleles in a large number of field samples of diverse origins. Mol Biochem Parasitol 89: 161–177.
- 71. Eberl KJ, Jelinek T, Aida AO, Peyerl-Hoffmann G, Heuschkel C, et al. (2001) Prevalence of polymorphisms in the dihydrofolate reductase and dihydropteroate synthetase genes of Plasmodium falciparum isolates from southern Mauritania. Trop Med Int Health 6: 756–760.
- 72. van den Broek IV, Gatkoi T, Lowoko B, Nzila A, Ochong E, et al. (2003) Chloroquine, sulfadoxine-pyrimethamine and amodiaquine efficacy for the treatment of uncomplicated Plasmodium falciparum malaria in Upper Nile, south Sudan. Trans R Soc Trop Med Hyg 97: 229–235.