Targeted and whole-genome sequencing reveal a north-south divide in P. falciparum drug resistance markers and genetic structure in Mozambique

Abstract

Mozambique is one of the four African countries which account for over half of all malaria deaths worldwide, yet little is known about the parasite genetic structure in that country. We performed P. falciparum amplicon and whole genome sequencing on 2251 malaria-infected blood samples collected in 2015 and 2018 in seven provinces of Mozambique to genotype antimalarial resistance markers and interrogate parasite population structure using genome-wide microhaplotyes. Here we show that the only resistance-associated markers observed at frequencies above 5% were pfmdr1-184F (59%), pfdhfr-51I/59 R/108 N (99%) and pfdhps-437G/540E (89%). The frequency of pfdhfr/pfdhps quintuple mutants associated with sulfadoxine-pyrimethamine resistance increased from 80% in 2015 to 89% in 2018 (p < 0.001), with a lower expected heterozygosity and higher relatedness of microhaplotypes surrounding pfdhps mutants than wild-type parasites suggestive of recent selection. pfdhfr/pfdhps quintuple mutants also increased from 72% in the north to 95% in the south (2018; p < 0.001). This resistance gradient was accompanied by a concentration of mutations at pfdhps-436 (17%) in the north, a south-to-north increase in the genetic complexity of P. falciparum infections (p = 0.001) and a microhaplotype signature of regional differentiation. The parasite population structure identified here offers insights to guide antimalarial interventions and epidemiological surveys.

Fig. 1. Source of P. falciparum samples providing genetic data.

Tables indicate the number of samples included in the analysis per province and year for each of the three main regions of the country. Provincial borders are indicated with thick lines. The specific districts providing data for the study are colored. Made with QGIS.

Fig. 2. Molecular markers of P. falciparum sulfadoxine-pyrimethamine (SP) resistance in Mozambique.

Frequency of P. falciparum isolates carrying triple mutations in pfdhfr (a), double mutations in pfdhps (b), and quintuple mutations in pfdhfr/phdhps (c) in 2015 and 2018 in seven provinces from Mozambique. For the pfdhps haplotype 436/437/540 (d), frequencies of the different allelic combinations are shown (n = 1365). Frequencies were calculated after excluding mixed genotypes. Data from Sofala was only available for 2015, and from Inhambane and Zambézia for 2018. The error bars represent a 95% confidence interval for the population proportion.

Fig. 3. P. falciparum population structure by geography in Mozambique.

Microhaplotypes from regions of 150–300 bp in length between long tandem repeats were reconstructed from whole genome sequences and used to test the geographic structure of P. falciparum parasites. a Distribution of the expected heterozygosity at the 8722 microhaplotype loci extracted from whole genome sequences. The y-axis represents the number of microhaplotype loci for a given expected heterozygosity. The red line marks the 75% percentile of the distribution; the 25% most diverse loci were considered for population structure analysis. b Chromosomal locations of the 155 most important microhaplotypes, which contribute to the geographic (North-Central-South) classification model. c Principal coordinates analysis with samples grouped into regions (North-Central-South; n = 1089), considering microhaplotypes at loci with expected heterozygosity in the top 25% percentile. d Principal coordinates analysis with samples grouped into regions considering the 155 top microhaplotypes, with an out-of-bag error rate of classification of 24.89%. e, f Complexity of infection (COI) for samples in different regions of Mozambique in 2015 (e) and 2018 (f), as indicated by the number of genetically distinct clones. Regional assignment of samples: North: C. Delgado; Central: Sofala, Tete, and Zambézia; South: Gaza, Inhambane, and Maputo.

Fig. 4. Regional separation, relatedness, and expected heterozygosity of pfdhps allelic haplotypes.

Identify by state (IBS) and expected heterozygosity (H_e) was calculated using 16 microhaplotypes flanking pfdhps to assess the evolutionary history of pfdhps mutant alleles in Mozambique. a Heatmap of the inter-sub-population IBS matrix among dhps alleles in Cabo Delgado observed in 2015 and 2018 (wild-type in codons 436, 437, and 540 [WT/WT/WT]: n = 20; mutant in codon 436 but wild-type in codons 437 and 540 [MUT/WT/WT]: n = 31; wild-type in codon 436 but mutant in codons 437 and 540 [WT/MUT/MUT]: n = 92). Sixteen microhaplotypes in a 50 kb region around pfdhps were used to calculate the pairwise IBS between samples. b t-distributed stochastic neighbor embedding visualization after 10000 iterations and c Expected heterozygosity calculated from the 16 microhaplotype loci in a 50 kb region around pfdhps in parasites collected from Cabo Delgado. Median and interquartile (IQR) H_e values: 0.1, IQR (0.04–0.26) for double mutants; 0.37, IQR (0.2–0.47) for WT/WT/WT; and 0.28, IQR (0.13–0.4) for MUT/WT/WT. The lower, middle, and upper hinges of the rectangle correspond to the 25% quantile, median, and 75% quantile of the distribution, respectively.