High-throughput genotyping of Plasmodium vivax in the Peruvian Amazon via molecular inversion probes

Abstract

Plasmodium vivax transmission occurs throughout the tropics and is an emerging threat in areas of Plasmodium falciparum decline, causing relapse infections that complicate treatment and control. Targeted sequencing for P. falciparum has been widely deployed to detect population structure and the geographic spread of antimalarial and diagnostic resistance. However, there are fewer such tools for P. vivax. Leveraging global variation data, we designed four molecular inversion probe (MIP) genotyping panels targeting geographically differentiating SNPs, neutral SNPs, putative antimalarial resistance genes, and vaccine candidate genes. We deployed these MIP panels on 866 infections from the Peruvian Amazon and identified transmission networks with clonality (IBD[identity by descent]>0.99), copy number variation in Pvdbp and multiple Pvrbps, mutations in antimalarial resistance orthologs, and balancing selection in 13 vaccine candidate genes. Our MIP panels are the broadest genotyping panel currently available and are poised for successful deployment in other regions of P. vivax transmission.

Fig. 1. Study sites and sample size.

A Locations of sample collection by community within Peru, scaled by size of circle. Darker gray boundaries identify the regions and departments of Loreto (where all but one of the circles are located) and Junín. Blue rectangle defines Iquitos, expanded in (B) Inset map depicting region around Iquitos, scaled by color of shading. Map created in R using open data from GADM database, v 3.6. www.gadm.org and the Sf package (https://cran.r-project.org/package=sf) with GPL-2 license (https://cran.r-project.org/web/licenses/GPL-2).

Fig. 2. Identity by descent (IBD) networks for five communities within the vicinity of Iquitos.

Only monoclonal samples are included. Each network node is a sample, with the edges representing IBD estimates. Nodes are color coded by collection year. An IBD threshold of 0.99 was used to generate networks. The Belén network shows evidence of clonal transmission in 2013 that is closely related to isolates collected in 2012 and 2014. In all five cities, there are networks connecting closely related isolates across years. A Belén, B Indiana, C Iquitos, D Punchana, and E San Juan Bautista.

Fig. 3. Violin plot depicting the complexity of infection (COI) of P. vivax isolates included in the analysis.

A all P. vivax isolates analyzed. Each dot represents the mean COI estimate for a given isolate. The majority of isolates (91.4%, n = 637/697) are monoclonal, with 44 isolates (6.31%) containing two clones, 14 isolates (2.01%) containing three clones, and two isolates (0.287%) containing four clones. B P. vivax isolates from five communities with largest sample numbers. COI does not differ significantly by community. C P. vivax isolates from the same five communities ordered by collection year. COI does not differ significantly by year.

Fig. 4. Upset plot showing the frequency of copy number variation (CNV) among monoclonal isolates.

The majority of isolates (98.0%) do not exhibit CNV. CNV in Pvrbp2b was observed in twelve (2.01%) samples, with one isolate having three copies, as well as duplications in Pvdbp, Pvrbp1b, Pvrbp2a, and Pvrbp2c. One other isolate showed duplications in both Pvrbp2b and Pvrbp2c. No CNV was detected in either Pvebp or Pvrbp1a.

Fig. 5. Scatter plots depicting Tajima’s D values in 100 bp bins with a 50 bp step size across potential vaccine target genes.

Gray boxes indicate gene regions that were not captured by MIPs. Only genes with Tajima’s D values > 2, i.e., showing strong signatures of balancing selection, are depicted. Dots are placed in the midpoint of each bin, with values >|2| highlighted in red. A Pvama1, B Pvdbp, C Pvebp, D Pvmsp1, E Pvrbp1a, F Pvrbp1b, G Pvrbp2a, H Pvrbp2b, I Pvrbp2c, J Pvtrag11, K Pvtrag19, L Pvtrag22, M Pvtrag36.