Culture adaptation of malaria parasites selects for convergent loss-of-function mutants

Abstract

Cultured human pathogens may differ significantly from source populations. To investigate the genetic basis of laboratory adaptation in malaria parasites, clinical Plasmodium falciparum isolates were sampled from patients and cultured in vitro for up to three months. Genome sequence analysis was performed on multiple culture time point samples from six monoclonal isolates, and single nucleotide polymorphism (SNP) variants emerging over time were detected. Out of a total of five positively selected SNPs, four represented nonsense mutations resulting in stop codons, three of these in a single ApiAP2 transcription factor gene, and one in SRPK1. To survey further for nonsense mutants associated with culture, genome sequences of eleven long-term laboratory-adapted parasite strains were examined, revealing four independently acquired nonsense mutations in two other ApiAP2 genes, and five in Epac. No mutants of these genes exist in a large database of parasite sequences from uncultured clinical samples. This implicates putative master regulator genes in which multiple independent stop codon mutations have convergently led to culture adaptation, affecting most laboratory lines of P. falciparum. Understanding the adaptive processes should guide development of experimental models, which could include targeted gene disruption to adapt fastidious malaria parasite species to culture.

Figure 1. Culture of new P. falciparum clinical isolates and sampling for genome sequencing.

(A) Flowchart showing numbers of isolates cultured and those selected for analysis. (B) Circles indicate culture time points of six single genotype isolates sampled for whole genome sequencing. (C) Genome-wide SNP haplotypes of these six isolates at the start and end of the culture periods. Each thin vertical line is a SNP (blue if it is identical to the 3D7 reference genome sequence, red if it is the alternative allele), shown in a concatenated linear scheme for all 14 chromosomes; for visual clarity, only one out of every 20 SNPs is shown. At the start of culture, the isolates were unrelated (each distinct from the others by between 5000 and 7000 detectable SNP differences). Five of the lines show unique genome-wide haplotypes that are unchanged throughout culture, but Line 3 was evidently contaminated by Line 4 (time course of replacement is shown in Fig. S1).

Figure 2. Selection of novel SNP alleles during culture adaptation of three unrelated P. falciparum clinical isolates in The Gambia.

(A) Allele frequencies at each sampled time point were determined by alternative sequence read counts for each SNP, and novel alleles reaching a frequency of more than 20% over the time course are plotted in colour (dashed lines indicate nonsense mutant alleles). The temporal frequency changes for these five SNPs attained genome wide significance (P < 10⁻⁹ for each; read counts at each timepoint are given in Table S2). (B) Gene models with an arrow indicating the mutation position for each emerging SNP identified in panel A. Green boxes indicate predicted functional domains in each gene: AP2 domains in an ApiAP2 transcription factor gene, the catalytic domain in the serine/threonine protein kinase gene SRPK1, and Calcium/lipid-binding C2 domains in DOC2. The numbers of codons in each gene are indicated underneath each scheme.

Figure 3. Nonsense mutations in long-term laboratory adapted P. falciparum strains are clustered in a few specific genes.

(A) Genome-wide map with red dots indicating positions of genes having nonsense SNPs only in laboratory-adapted strains. These three genes (two ApiAP2 genes and Epac) have nonsense SNP alleles detected among 11 laboratory strains (listed in Supplementary Tables S3 and S4) but not in sequences from uncultured clinical samples (blue dots indicate genes with nonsense SNPs in 1% or more of 2483 uncultured samples from the Pf3K project release 3.1 at www.malariagen.net/pf3k). The subtelomeric regions and internal non-core chromosomal regions (containing genes that are absent in most malaria parasite species) of the 3D7 reference genome are shaded in darker gray. (B) Models of the three genes with nonsense SNP alleles only in laboratory-adapted strains. Downward arrows show the positions of the premature stop codon alleles identified in each strain, while upward arrows indicate previously detected nonsense or frameshift (fs) mutations in two strains additional to those analysed here as well as the 3D7 strain reference genome sequence version 3.1 (http://www.genedb.org/Homepage/Pfalciparum). Green boxes indicate predicted catalytic and AP2 domains, and the numbers of codons in each gene are indicated underneath each scheme.