Whole-genome sequencing and microarray analysis of ex vivo Plasmodium vivax reveal selective pressure on putative drug resistance genes

Abstract

Plasmodium vivax causes 25-40% of malaria cases worldwide, yet research on this human malaria parasite has been neglected. Nevertheless, the recent publication of the P. vivax reference genome now allows genomics and systems biology approaches to be applied to this pathogen. We show here that whole-genome analysis of the parasite can be achieved directly from ex vivo-isolated parasites, without the need for in vitro propagation. A single isolate of P. vivax obtained from a febrile patient with clinical malaria from Peru was subjected to whole-genome sequencing (30× coverage). This analysis revealed over 18,261 single-nucleotide polymorphisms (SNPs), 6,257 of which were further validated using a tiling microarray. Within core chromosomal genes we find that one SNP per every 985 bases of coding sequence distinguishes this recent Peruvian isolate, designated IQ07, from the reference Salvador I strain obtained in 1972. This full-genome sequence of an uncultured P. vivax isolate shows that the same regions with low numbers of aligned sequencing reads are also highly variable by genomic microarray analysis. Finally, we show that the genes containing the largest ratio of nonsynonymous-to-synonymous SNPs include two AP2 transcription factors and the P. vivax multidrug resistance-associated protein (PvMRP1), an ABC transporter shown to be associated with quinoline and antifolate tolerance in Plasmodium falciparum. This analysis provides a data set for comparative analysis with important potential for identifying markers for global parasite diversity and drug resistance mapping studies.

Fig. 1.

Pileup alignment of Illumina sequencing reads across the P. vivax genome. Regions are colored by read depth, with regions covered by less than 20 reads in black, 20–100 reads in red, and more than 100 reads in blue. Subtelomeric and internal regions containing repetitive regions or multicopy highly variable genes indicated by circles correlate with lower read depth. Asterisks indicate regions with loss of hybridization by microarray that contain highly variable surface protein genes (Fig. 2).

Fig. 2.

Genome-wide detection of polymorphisms in a patient-derived isolate of P. vivax. The log₂ ratios of intensities of IQ07 versus SalI were displayed along each chromosome in the P. vivax genome and are colored by the running mean over 500 bp. Loss of hybridization (black) in IQ07 relative to the reference isolate indicates highly variable genes and correlates with subtelomeric and internal loci of multigene families that are involved in host–pathogen interactions such as msp3, msp7, sera, and vir. The loss of hybridization on chromosomes 6 and 7 (marked with asterisks) that contain hypothetical genes of unknown function corresponds to syntenic regions in P. falciparum that contain highly variable surface protein genes, including msp genes.

Fig. 3.

Predicted polymorphisms in drug resistance genes in a Peruvian patient-derived P. vivax isolate. The result from the microarray-based prediction algorithm is plotted in blue above the gene models. The log₂ ratios of the intensities of IQ07 and Salvador I were displayed below and are colored by the running mean over 25 bp using the same scale as displayed in Fig. 2. (A) We detected a mutation near codon 58 in pvdhfr. (B) We detected a strong mutation signal near codon 205 in pvdhps, which has not been previously described as polymorphic. (C) IQ07 contained mutations near codons 958 and 1022 in pvmdr1. (D) We detected a mutation ~500 bp upstream of the start codon that was in the 5′ UTR of pvcrt. We detected a mutation ~100 bp into the first intron of the pvcrt gene but did not detect any mutations in exons. (E) We detected multiple mutations by microarray in the putative ABC transporter pvmrp.