Single-genome sequencing reveals within-host evolution of human malaria parasites

Abstract

Population genomics of bulk malaria infections is unable to examine intrahost evolution; therefore, most work has focused on the role of recombination in generating genetic variation. We used single-cell sequencing protocol for low-parasitaemia infections to generate 406 near-complete single Plasmodium vivax genomes from 11 patients sampled during sequential febrile episodes. Parasite genomes contain hundreds of de novo mutations, showing strong signatures of selection, which are enriched in the ApiAP2 family of transcription factors, known targets of adaptation. Comparing 315 P. falciparum single-cell genomes from 15 patients with our P. vivax data, we find broad complementary patterns of de novo mutation at the gene and pathway level, revealing the importance of within-host evolution during malaria infections.

Keywords: Parasite mutations; Plasmodium falciparum; Plasmodium vivax; de-novo mutations; genetic diversity; malaria; malaria low parasitaemia; parasite relatedness; single-cell sequencing; within-host evolution.

Figure 1.. Development of a low parasitaemia single cell sequencing (LPSCS) protocol.

(A) The LPSCS protocol. Cryopreserved parasites are thawed and grown ex vivo for <40 hours, after which late stage parasites are enriched in the culture by magnetic purification. Following magnetic enrichment parasites are stained with a live cell dye and late stage parasites are identified by their high DNA content. Magnetic enrichment does not eliminate all uninfected RBCs (blue box) or early stage parasites, these still contribute to the composition of the sample. Correspondingly late stage parasites are targeted for isolation by gating high DNA content cell (red box). After isolation of single cells WGA is performed under highly sterile conditions, and libraries prepared for genome sequencing. (B) The success rate of single cell DNA amplification. The x-axis shows the coverage obtained for each single cell library, the y-axis shows the purity of each sequence. These show the majority of single cells are both clonal, and high coverage (top right corner), while bulk sequences and samples from a published study (Pearson et al., 2016) show a wider distribution of clonality.

Figure 2.. IBD Sharing within and between infections.

(A) A network representation of pairwise IBD. Nodes are individual samples (either bulk or single cell), and are joined by edges when they share >75% of their genomes IBD. Nodes are colored by infection and time point, circles are single cell data and squares bulk data. Data from 82 bulk infections is also included. At this threshold there are few bulk infections which share sufficient IBD for edges to be generated. A range of IBD sharing thresholds is shown in Supplementary Fig. 2. We see that parasites from within infections and across time points form tight clusters indicative of recently ancestry. (B) The distribution of proportion of total shared IBD within infection time points. (C) The distribution of mean IBD tract length in Mb within infection time points. For each infection we see parasites which are more related to one another than the population background.

Figure 3.. Extreme inbreeding can explain clusters of putative mutations.

Population level allele frequencies (PLAF) for each putative de novo mutation (PDNM) for VHX542 (A) and DMA004 haplotype 1 (B). PDNMs are ordered by chromosomal location. For VHX542 all high frequency differences are present in the population and are unlikely to be from de novo mutation. Genotypes for each PDNM in VHX542 (C) and DMA004 haplotype 1 (D). Reference alleles are in beige and alternative alleles are in blue. The order of PDNMs is the same as in (A) and (B). A single cell in (D) contains most variation in the sample (~60 PDNMs), most of these PDNMs are variable in the population. Read pile-ups for PDNMs from regions of inbreeding (E), three cells are shown from each time point. The distance between each PDNM in DMA004 haplotype 1. For each pile-up read depth is >40X and the mutation is supported by >99% of the reads. The number of unique PDMNs observed in each cell is shown for VHX542 (F) and DMA004 haplotype 1 (G).

Figure 4.. Identification of mutations defining recurrences in VHX542.

(A) Hypnozoite stage parasites reside in the liver, and periodically reactivate, causing a recurrence. (B) There is no correlation between the time interval between recurrences and the mean number of mutations accrued by a cell. (C) The proportion of mutations in the P. vivax genome categorized by functional impact. Population denotes all mutations detected in Thailand from Pearson et al, Rare Population is a subset of these mutations present at <1% frequency. (D) The domain structure of the five P. vivax ApiAP2 genes showing mutations in this study.

Figure 5. de novo mutations in P. falciparum are enriched for functional changes in dispensible, and highly expressed genes.

(A) The proportion of mutations in the P. falciparum genome categorized by functional impact. (B) There is a significant enrichment of putative de novo mutations (PDMNs) in genes which are dispensible for blood stage function in vitro. The distribution of Mutagenesis Index Score (MIS) is significantly higher in genes where PDNMs are detected (Mut.) than those where no PDNM was detected (No Mut.). The distribution of MIS is shown for each class of genic mutation. (C) Within each life cycle stage genes where PDNMs were detected (Mut) were expressed significantly higher than other genes (sgSpz – salivary gland sporozoites, ooSpz – oocyst sporozoites, hlSpz – hemolymph sporozoites, ook – ookinetye, gam – gametocytes, asex – asexual blood stage.