Plasmodium malariae structure and genetic diversity in sub-Saharan Africa determined from microsatellite variants and linked SNPs in orthologues of antimalarial resistance genes

Malaria elimination programs and tools are focused on eliminating the main malaria parasites P. falciparum and P. vivaxbut other malaria parasites species such as P. malariae are co-transmitted in malaria endemic regions and warrant attention for achieving elimination. As knowledge of the diversity and effect of elimination tools on these minor species have not had much academic or public health focus, this study took advantage of the Pathogen Diversity Network Africa (PDNA), to collate a small but the widest sample set of P. malariae isolates across seven African countries and one from Asia, describing population structure, high genetic diversity, genotypic richness and evenness in the species of malaria parasite.

A high level of genetic diversity is essential for the long-term survival of populations, and the extent of variation determines the ability of the species to adapt to environmental challenges imposed by nature or control interventions. The high diversity in P. malariae found here is similar to those previously reported in Kenya and Malawi2,3, despite the variable and small number of samples analyzed from some of the countries. Malaria transmission intensity and history of interventions vary across the different countries represented in this study and could have affected the results. High transmission results in frequent heterologous recombination of the parasite in the mosquito vector, breaking down linkage disequilibrium between variable loci and increasing the genetic diversity within populations. In general, the probability of outcrossing in parasite populations varies from a high to low malaria transmission gradient in the West, Central and Eastly direction in sub-Saharan Africa20. This notwithstanding, heterozygosity was high and genetic distance was low between isolates despite the differences in malaria transmission intensity, except for the three isolates from Cameroon. The low genetic distance observed in this study requires further validation since there are no studies on it P. malariae for direct comparison. However, the results obtained are consistent with a recent one P. falciparum study in Nigeria21 howbeit in sharp contrast with an older study in Senegal22, suggesting relatively high levels of panmixia in the current sampled populations despite the varying transmission patterns. Indeed, multi-locus genotypes were rare across all populations, an indication of recombination and absence of the clonal expansion observed in some P. falciparum populations with low or seasonal malaria transmission23.

Recurrent gene flow promotion between parasite populations across countries via human or vector migration, could have led to a lack of differentiation according to geographic origin3. This was evident in low microsatellite differentiation indices between countries, although the small number of isolates per country population limits the accuracy of the inferred indices. It is also possible that gene flow alone does not explain the high genetic diversity or lack of geographic differentiation observed in P. malariaeother factors such as a lack of bottleneck event or intervention to reduce local diversity could also be considered.

Population structure analysis with neutral microsatellite loci (ie SSRs) identified five clusters, each with isolates from different countries. This is further indication of high intrapopulation variability in these markers and absence of population specific selection of these loci that may drive population differentiation. This structure is not consistent with isolation by distance seen before P. falciparum, where genetic clusters can be allocated to geographic populations in the west, central and eastern African regions. neither P. malariae mostly occurs as coinfections with P. falciparum, the drivers of such independent substructure may therefore be different, or it is possible that this could have been established by an earlier event preceding any population drift due to demography or isolation. Using the admixture model in STRUCTURE, an optimum of three ancestral clusters were determined and this also showed all isolates with components of each ancestry regardless of the country of origin. STRUCTURE implements a Bayesian algorithm to identify groups of individuals at Hardy–Weinberg and linkage equilibrium. However, its robustness was shown to be impacted by small uneven sample sizes between subpopulations and/or hierarchical levels of population structure24,25.

The SNP data from selected P. malariae orthologues of P. falciparum drug resistance genes also clustered isolates into 5 subpopulations, although only 3 less distinct clusters were retained after stringent filtration processes, and membership of the subgroups did not overlap fully with those determined by microsatellites. The distribution between the SNP and microsatellite distance differed, with wider distances between smaller numbers of isolates using SSR data. This is expected as SSRs are multi-allelic, likely neutral, and more likely to differ between pairs of isolates. Thus, further investigation into the possible drivers of population differentiation for this parasite species will improve understanding of its complexity, particularly with regard to control and elimination strategies. While it appeared that most infections had mixed genomes (polygenomic) as indicated by Fws, this was affected by the denoising and filtration pipelines used for analysis. Thus, further investigation using appropriate sample size will be necessary to clarify if co-transmission of different clones and increased possibility of recombination exists for this Plasmodium species. High level of infection complexity is important to note, given that it is one of the indices for monitoring the effect of interventions. Unique for P. falciparum, the complexity did not seem to be higher in relatively higher malaria transmission settings and may be part of the unique biology of this species that needs further investigation. As drugs and other interventions drive down populations, selection and changes in complexity should be monitored for the species.

Drugs have been a major selective force on P. falciparum, with resistance associated with mutations in several genes and signatures of positive selection across the genomes. We identified 20 P. malariae mutations in the orthologous drug resistant genes by combining different sequence alignment and variant calling algorithms. These putative variants have not been described in previous targeted or genome scans, probably because of differences in isolates used or the methods applied. Here we retained only high-quality variants supported by combinations of two mapping and two SNP calling algorithms. Most of the candidate variants were synonymous but there were several nonsynonymous SNPs across seven genes, especially in Pmcytbwhose P. falciparum orthologue drives resistance against atovaquone. Atovaquone is a member of the quinolines, to which resistance in P. falciparum have been associated with mutations in multidrug resistance gene (Pfmdr1), chloroquine resistance transporter (Pfcrt) and an amino acid transporter (Pfaat1). Most of the LD observed with the unfiltered dataset were not reproducible with the denoised and filtered dataset, with the exception of the LD observed in the SNPs of the mitochondrial gene, either due to common ancestry or selection of dominant haplotype by drugs or other factors. Additional candidate variants were seen in Pmdhfr and Pmdhps, orthologues of antifolate resistance in P. falciparum. The antifolate antimalarials, sulfadoxine-pyrimethamine are still in wide use for chemoprevention against malaria in pregnancy and combination with amodiaquine for seasonal malaria chemoprevention in West Africa. These together could be selecting for the identified variants. While the nonsynonymous SNPs reported here occur in low frequencies, further verification, characterization and association of these SNPs will require increased genomic surveillance and phenotype association studies from in vivo and ex vivo therapeutic efficacy tests.

A limitation of this study, which warrants cautious interpretation of the results, is the small number of samples analyzed across the different countries and the lack of bio and clinical data of the samples. Larger population studies for P. malariae with appropriate epidemiology or clinical data are required to validate the findings of smaller studies as reported here. Another limitation is the use of standard bioinformatics pipelines designed for P. falciparum. While these may be acceptable for preliminary analysis, custom pipelines that take possible amplification and sequencing errors into consideration may be better for P. malariae, particularly because of the scarcity of population data with high quality confirmed variants from this parasite species. The different sample types analyzed could also be a limitation to this study, the dried blood spot samples were more amenable to producing poor quality results, possibly due to the low prevalence and low parasite density of the non-falciparum species. Therefore, venous blood sampling and more robust molecular techniques that take these into consideration will be beneficial in future molecular surveillance of P. malariae.

The current drive for malaria elimination needs innovative strategies to target all malaria parasites. One approach can be by integrating genomic surveillance of all Plasmodium species into malaria control and elimination programs in sub-Saharan Africa, learning from the experience with COVID-19, to refine approaches as new variants are identified and monitored. This study has established the relevance of this in P. malariae.