Genome-wide SNP analysis of Plasmodium falciparum shows differentiation at drug-resistance-associated loci among malaria transmission settings in southern Mali

Abstract

Plasmodium falciparum malaria cases in Africa represent over 90% of the global burden with Mali being amongst the 11 highest burden countries that account for 70% of this annual incidence. The persistence of P. falciparum despite massive global interventions is because of its genetic diversity that drives its ability to adapt to environmental changes, develop resistance to drugs, and evade the host immune system. Knowledge on P. falciparum genetic diversity across populations and intervention landscape is thus critical for the implementation of new strategies to eliminate malaria. This study assessed genetic variation with 12,177 high-quality SNPs from 830 Malian P. falciparum isolates collected between 2007 and 2017 from seven locations. The complexity of infections remained high, varied between sites, and showed a trend toward overall decreasing complexity over the decade. Though there was no significant substructure, allele frequencies varied geographically, partly driven by temporal variance in sampling, particularly for drug resistance and antigen loci. Thirty-two mutations in known drug resistance markers (pfcrt, pfdhps, pfdhfr, pfmdr1, pfmdr2, and pfk13) attained a frequency of at least 2% in the populations. SNPs within and around the major markers of resistance to quinolines (pfmdr1 and pfcrt) and antifolates (pfdhfr and pfdhps) varied temporally and geographically, with strong linkage disequilibrium and signatures of directional selection in the genome. These geo-temporal populations also differentiated at alleles in immune-related loci, including, protein E140, pfsurfin8, pfclag8, and pfceltos, as well as pftrap, which showed signatures of haplotype differentiation between populations. Several regions across the genomes, including five known drug resistance loci, showed signatures of differential positive selection. These results suggest that drugs and immune pressure are dominant selective forces against P. falciparum in Mali, but their effect on the parasite genome varies temporally and spatially. Interventions interacting with these genomic variants need to be routinely evaluated as malaria elimination strategies are implemented.

Keywords: differentiation; drug resistance; genetic variation; malaria; positive selection.

FIGURE 1

Geographic and temporal malaria parasites sampled across different endemic sites in Mali between 2007 and 2017, showing, (A) sampling locations in Mali (marked with black points), and bars of number of samples per location from different years as color-coded. (B) Distribution of the F _WS inbreeding coefficient for infection complexity determined for isolates from different locations sampled.

FIGURE 2

Population structure and ancestry of P. falciparum from Mali. (A) Plot of the first two components (Axis1 vs. Axis2) of principal component analysis (PCA) using genotype data from samples in Mali. (B) Scatter plot of first two dimensions of multidimensional scaling (MDS) of identity by state between pairs of isolates. Each point represents an isolate, color-coded by sites of sampling. (C) Barplot of ancestry proportions per sample and site of collection shown on the x-axis. Each ancestral population is indicated by a different color, (K1 (red), K2 (turquoise), and K3(blue)) and each vertical bar represents a single sample, colored according to the estimated ancestry proportions.

FIGURE 3

Genome-wide SNP differentiation between geo-temporal populations of P. falciparum from Mali. (A) Manhattan plot of genome-wide Jost’s D differentiation index per SNP between all sampling locations (geo-temporal populations) in Mali. The genomic location of drug-resistance-associated genes (pfdhfr, pfcrt, and pfdhps) is indicated with arrows. (B) Allele frequency proportions (pie charts) for the PFCRT A220S locus across different sampling locations. (C) Manhattan plot of genome-wide Jost’s D differentiation index per SNP between samples from Faladje collected in 2007 and 2015–2017 and (D) Faladje 2007 against 2013. (E) Temporal changes in proportion of alleles and haplotypes of drug-resistance-associated loci in pfdhfr, pfcrt, and pfdhps.

FIGURE 4

Heatmap of linkage disequilibrium (LD) between SNPs in drug resistance markers, labeled by gene name, genomic position, and amino acid allele. Each cell is color shaded by intensity of LD, with the r ² value between the loci indicated.

FIGURE 5

Signatures of positive selection in the genomes of temporal and geo-temporal populations of P. falciparum in Mali. (A) Manhattan plots of genome-wide −log10 q-values of standardized integrated haplotype score (iHS) for temporal populations from Faladje, showing positions of genes for drug-resistance-associated loci, pfdhfr, pfcrt, and pfdhps, and the antigen pftrap. Each row represents the year of sampling, showing 2007, 2013, and combined 2015–2017 samples. (B) Heatmap of -log10 q-values of cross population extended haplotype homozygosity test between pairs of geo-temporal populations shown in rows with abbreviated names for Kolle (kol), Bougoula-Hameau (boug), Faladje (fala), Kenieroba (ken), Bamako (bko), Dangassa (dang), and Nioro (nioro). (C) Combined Rsb scores of SNPs across the P. falciparum genome over all pairwise comparison of seven geo-temporal populations. The top genomic regions harboring SNPs with a mean Rsb >5 are labeled alphabetically from chromosome 1–14. The genomic coordinates and genic loci within these regions are given in Table 2.