Recombination hotspots and population structure in Plasmodium falciparum

Abstract

Understanding the influences of population structure, selection, and recombination on polymorphism and linkage disequilibrium (LD) is integral to mapping genes contributing to drug resistance or virulence in Plasmodium falciparum. The parasite's short generation time, coupled with a high cross-over rate, can cause rapid LD break-down. However, observations of low genetic variation have led to suggestions of effective clonality: selfing, population admixture, and selection may preserve LD in populations. Indeed, extensive LD surrounding drug-resistant genes has been observed, indicating that recombination and selection play important roles in shaping recent parasite genome evolution. These studies, however, provide only limited information about haplotype variation at local scales. Here we describe the first (to our knowledge) chromosome-wide SNP haplotype and population recombination maps for a global collection of malaria parasites, including the 3D7 isolate, whose genome has been sequenced previously. The parasites are clustered according to continental origin, but alternative groupings were obtained using SNPs at 37 putative transporter genes that are potentially under selection. Geographic isolation and highly variable multiple infection rates are the major factors affecting haplotype structure. Variation in effective recombination rates is high, both among populations and along the chromosome, with recombination hotspots conserved among populations at chromosome ends. This study supports the feasibility of genome-wide association studies in some parasite populations.

Figure 1. Inferred Population Structure of Global Parasite Isolates

Each vertical bar represents an individual parasite with its names given at the top of the panels. Predefined numbers of populations (K = 2–8) were run ten times each, using Structure 2. (A) Population partitions using SNPs from Chromosome 3 (K = 4–6, linkage model). At K = 5, the most consistent membership coefficients were obtained. (B) Population partition using SNPs from 37 putative transporters (admixture model). Similar clustering was obtained with that of Chromosome 3, including Camp and T2/C6 (no data for K1) with African parasites and 106/1 with SE Asian parasites. However, African parasites were partitioned into chloroquine-resistant and -susceptible parasites. Note: Only 81 are available for the transporter data. *, parasites resistant to chloroquine; ?, parasites from which drug data are not available.

Figure 2. The Distribution of Detectable Recombination Events on Chromosome 3 of P. falciparum

In (A) and (B), each panel shows, for two populations, a minimum number of recombination events (assuming an infinite-sites model) between each pair of segregating sites, scaled by physical distance to identify regions of high and low recombination. (A) African (upper) and American (lower) populations. (B) SE Asian (upper) and PNG (lower) populations. The color bar unit is recombination event/kb. (C) Estimates of population recombination rate variation for African (red line), SE Asian (blue), American (black), and PNG (purple) samples using the RJMCMC method with a jump penalty of five. Shown on top are the locations of genes on the plus (top) and minus (below) strands, with known cell-surface genes in red [34].

Figure 3. LD and Haplotype Blocks across Chromosome 3 of P. falciparum

(A) LD decays with increasing distances between variable sites along Chromosome 3 in populations from Africa, SE Asia, S America, and PNG. The LD index r² (square of correlation coefficient) were calculated considering all pair-wise values for SNPs with minor allele frequency > 5%. Parasite isolates K1, Camp, T2/C6, 105/7, 3D7, HB3, Haiti, JAV, ECP, 569, and 1905 are excluded from the LD analyses because these parasites were placed in clusters different from locations they were isolated from or have different genetic backgrounds. (B) Haplotype blocks defined as regions where all pairs of sites have D′ ≥ 0.8. Values below each block are the number of htSNPs required to capture all haplotype variants when haplotype blocks are characterized by complete association among variants (r² = 1). Bars on the top are locations of assayed SNPs. Triangles denote a site in the middle of a recombination hotspot, and the width of the triangles represents the region hotspot spans.