Rapid identification of genes controlling virulence and immunity in malaria parasites

Abstract

Identifying the genetic determinants of phenotypes that impact disease severity is of fundamental importance for the design of new interventions against malaria. Here we present a rapid genome-wide approach capable of identifying multiple genetic drivers of medically relevant phenotypes within malaria parasites via a single experiment at single gene or allele resolution. In a proof of principle study, we found that a previously undescribed single nucleotide polymorphism in the binding domain of the erythrocyte binding like protein (EBL) conferred a dramatic change in red blood cell invasion in mutant rodent malaria parasites Plasmodium yoelii. In the same experiment, we implicated merozoite surface protein 1 (MSP1) and other polymorphic proteins, as the major targets of strain-specific immunity. Using allelic replacement, we provide functional validation of the substitution in the EBL gene controlling the growth rate in the blood stages of the parasites.

Fig 1. Schematic representation of the multi-crossing LGS approach.

The process starts with the identification of distinct selectable phenotypes in cloned strains of the pathogen population (in this case malaria parasites) and their sequencing, usually from the vertebrate blood stage. A genetic cross between two cloned strains is subsequently produced, in this case inside the mosquito vector. The cross progeny is then grown with and without selection pressure(s). Selection pressure will remove those progeny individuals carrying allele(s) associated with sensitivity to the selection pressure(s), while allowing progeny individuals with the resistant allele(s) to survive. DNA is then extracted from the whole, uncloned progeny for sequencing. SNPs distinguishing both parents are used to measure allele frequencies in the selected and unselected progenies. A mathematical model is then applied to identify and define loci under selection. Regions in these loci are then analyzed in detail to identify potential target polymorphisms underlying the phenotype(s) under investigation. Targeted capillary sequencing can be employed to verify or further characterize polymorphisms. Finally, where applicable, allele replacement experiments can be carried out to confirm the effect of target polymorphisms.

Fig 2. Pure strain growth rates.

(A) Growth rate of Plasmodium yoelii strains 17X1.1pp and CU in CBA mice inoculated with 1 × 10⁶ iRBCs on Day 0. Error bars indicate the standard error of the mean for three mice per group. (B) The relative proportions of CU and 17X1.1pp were measured by Q-RT-PCR targeting the polymorphic msp1 locus at Day 4 post-inoculation with a mixed inoculum containing approximately equal proportions of both strains in naïve mice and mice immunized with one of the two strains. Error bars show the standard error of the mean of five mice per group. *p<0.05, Wilcoxon rank sum test, W = 25, p = 0.0119, n = 5; ** p<0.01, Wilcoxon rank sum test, W = 25, p = 0.0075, n = 5

Fig 3. Genome-wide sequencing data.

(A) Genome-wide Plasmodium yoelii CU allele frequency of two independent genetic crosses grown in (a,b) naïve mice, (c,d) 17X1.1pp immunized mice and (e,f) CU-immunized mice. Light gray dots represent observed allele frequencies. Dark gray dots represent allele frequencies retained after filtering. Dark blue lines represent a smoothed approximation of the underlying allele frequency; a region of uncertainty in this frequency, of size three standard deviations, is shown in light blue. A conservative confidence interval describing the position of an allele evolving under selection is shown via a red bar. Allele frequencies are shown in log scale. (B) Evolutionary models fitted to allele frequency data. Filtered allele frequencies are shown as gray dots, while the model fit is shown as a red line. Dark blue and light blue vertical bars show combined and conservative confidence intervals for the location of the selected allele as reported in Table 3. Numbers in parentheses equate figures with locations in (A). A black vertical line shows the position of a gene of interest.

Fig 4. EBL Amino acid sequence alignment of various malaria species and Plasmodium yoelii strains, and predicted protein structure consequences of the C351Y polymorphism.

(A) EBL orthologous and paralogous sequences from a variety of malaria species and P. yoelii strains were aligned using ClustalW. Only the amino acids surrounding position 351 are shown. The cysteine in positon 351 in P. yoelii is highly conserved across strains and species, with only strain 17X1.1pp bearing a C to Y substitution. PchAS: Plasmodium chabaudi AS strain; PbANKA: Plasmodium berghei ANKA strain; Py17X/17X1.1pp/CU/YM: P. yoelii 17X,17X1.1pp,CU,YM strains; Pk-DBLα/β/γ: Plasmodium knowlesi Duffy Binding Ligand alpha/beta/gamma (H strain); PvDBP: Plasmodium vivax Duffy Binding Protein (Sal-I strain);PcynB_DBP1/2: Plasmodium cynomolgi Duffy Binding Proteins 1/2 (B strain); Pf3D7_EBA140/175/181: Plasmodium falciparum Erythrocyte Binding Antigens 140/175/181 (3D7 strain). (B) Energy minimized homology model of the wild type P. yoelii (Py17XWT) Erythrocyte Binding Ligand (EBL). Inset depicts the disulfide bond between C351 and C420. (The protein is represented in cyan and the disulfide bonds are in yellow). (C) Energy minimized homology model of the mutant (C351Y) P. yoelii (Py17X1.1pp) Erythrocyte Binding Ligand (EBL). Inset depicts the lack of a disulfide bond between Y351 (substituted C351) and C420. (The protein is represented in cyan and the disulfide bonds are in yellow and Tyr351 [mutated] is represented in magenta).

Fig 5. Localization of EBL.

The C351Y polymorphism does not affect EBL subcellular localization in Plasmodium yoelii. (A) P. yoelii schizonts of wild type and transgenic parasite lines were incubated with fluorescent mouse anti-EBL serum, fluorescent rabbit anti-AMA1 serum, and DAPI nuclear staining. Colors indicate the localization of the Pyebl(green) and AMA-1 (red) proteins, as well as nuclear DNA (blue). 17XL: fast growing 17X clone previously shown to traffic EBL to the dense granules, not the micronemes, 17X1.1pp: 17x1.1pp strain, CU: CU strain, 17X1.1-351Y C: 17X1.1pp strain transfected with the CU allele for Pyebl, CU-351C Y: CU strain transfected with the 17X1.1pp allele of Pyebl. (B) The distance of EBL from AMA1 measured for five parasite strains and for 5–9 schizonts per strain; stars indicate p<0.01 using a Mann-Whitney U test. This indicates a shift in the location of Pyebl occurring in 17XL, but not in any other parasite lines.

Fig 6. Site directed mutagenesis of pyebl AA position 351 reverses the phenotypes of parasites with slow and intermediate growth rates.

(A) Growth rate of P. yoelii strains 17X1.1pp, CU and of the CU-strains transfected with either CU (CU-EBL-351C>C) or 17X1.1 (CU-EBL-351C>Y) Pyebl gene in CBA mice inoculated with 1x106 iRBCs on Day 0. (B) Growth rate of P. yoelii strains 17X1.1pp, CU and of the 17X1.1pp-strains transfected with either 17X1.1 (17X1.1pp-EBL-351Y>Y) or CU (17X1.1pp-EBL-351Y>C) Pyebl gene alleles in CBA mice inoculated with 1x106 iRBCs on Day 0. Transfection with the 17X1.1pp (EBL-351Y) allele produces a significantly increased growth rate in the CU strain (CU-EBL-351C>C vs CU-EBL-351C>Y: p <0.01, Two-way ANOVA with Tukey post-test correction) that is not significantly different from 17X1.1pp growth rate following transfection with its native allele (17X1.1pp-EBL-351Y>Y vs. CU-EBL-351C>Y: p >0.05, Two-way ANOVA with Tukey post-test correction). Conversely, transfection with the CU (EBA-351C) allele significantly reduces growth (17X1.1pp-EBL-351Y>Y vs 17X1.1pp-EBL-351Y>C: p <0.01, Two-way ANOVA with Tukey post-test correction) and produces a phenotype that is not significantly different from CU transfected with its own allele (CU EBL-351C>C vs 17X1.1pp-EBL-351Y>C: p >0.05, Two-way ANOVA with Tukey post-test correction).