Rapid, field-deployable method for genotyping and discovery of single-nucleotide polymorphisms associated with drug resistance in Plasmodium falciparum

Abstract

Despite efforts to reduce malaria morbidity and mortality, drug-resistant parasites continue to evade control strategies. Recently, emphasis has shifted away from control and toward regional elimination and global eradication of malaria. Such a campaign requires tools to monitor genetic changes in the parasite that could compromise the effectiveness of antimalarial drugs and undermine eradication programs. These tools must be fast, sensitive, unambiguous, and cost-effective to offer real-time reports of parasite drug susceptibility status across the globe. We have developed and validated a set of genotyping assays using high-resolution melting (HRM) analysis to detect molecular biomarkers associated with drug resistance across six genes in Plasmodium falciparum. We improved on existing technical approaches by developing refinements and extensions of HRM, including the use of blocked probes (LunaProbes) and the mutant allele amplification bias (MAAB) technique. To validate the sensitivity and accuracy of our assays, we compared our findings to sequencing results in both culture-adapted lines and clinical isolates from Senegal. We demonstrate that our assays (i) identify both known and novel polymorphisms, (ii) detect multiple genotypes indicative of mixed infections, and (iii) distinguish between variants when multiple copies of a locus are present. These rapid and inexpensive assays can track drug resistance and detect emerging mutations in targeted genetic loci in P. falciparum. They provide tools for monitoring molecular changes associated with changes in drug response across populations and for determining whether parasites present after drug treatment are the result of recrudescence or reinfection in clinical settings.

Fig 1

Representative melting peaks of HRM assays. (A) Example of a single-SNP probe assay; Pfcrt H97 shown. The probe is designed to be a perfect match to the wild type (3D7). Any mutations in this probe region lower the melting peak. Shown here is TM90C6B with the mutation H97Q that results from a nucleotide base change from A to T. This class IV SNP is the most challenging for HRM to distinguish. (B) Example of a single-SNP probe assay with multiple alleles. Pfmdr 86 shows wild-type N86 in 3D7 (gray), mutant N86Y in strains K1 and FCR3 (red), and a recently reported mutation in field isolates, N86F in Dd2 (blue). Note that different sources of Dd2 have varying numbers of copies of pfmdr1. In this instance, Dd2 has copies with different SNP mutations that have been sequenced to verify the presence of both alleles. Both N86Y and N86F derive from an A→T nucleotide change. Additionally, the SNPs tested are adjacent to one another in the genome and can be clearly differentiated by the assay. (C) Representative SNP haplotype probe assay. Shown is the assay developed for pfcrt residues 72, 74, 75, and 76. The probe is designed as a perfect match to the 3D7 allele at each locus (codons 72, 74, 75, and 76); thus, sample 3D7(blue) displays the highest melting peak (T_m) possible with this probe. This peak corresponds to a genotype at these four loci of C,M,N,K. Sample 7G8 (red) is mismatched under the probe at codons 72 (TT mismatch) and 76 (CT mismatch) for a genotype of S,M,N,T. Samples Dd2, V1/S, and FCR3 (gray) are all the same genotype and are mismatched to the probe at codons 75 (GT and AA mismatch; 1st and 3rd bases of codon 75) and 76 (CT mismatch), making the genotype C,I,E,T. If a sample were mismatched with the probe at only a single base site, the melting peak would be somewhere between the red and blue peaks. If a sample were mismatched at all 4 sites, the melting peak would be even lower than the gray peak.

Fig 2

Limit of detection and performance with human genomic material. (A) Limit of detection of assays. A representative assay (pfdhps 436/437) shows the limit of detection. The melting peaks are still distinguishable, with 10⁻⁵ ng of Plasmodium template as the lowest curves. Shown are the normalized derivative melting peaks of both the probe and full amplicon (the higher-temperature melting peaks outside the vertical bars of the probe amplicon). (B) Limit of detection of assays with software normalization. The same assay shown in panel A after software normalization of the probe melting peaks (Idaho Technology, Inc., standard software suite analysis of unlabeled probes) still shows strong peaks for the mutant and wild-type alleles. (C) Performance with an excess of human material. Shown are mock samples of culture-adapted parasite genomic material with an excess of human material added. The Pfcrt K72-76 assay is shown. The limit of detection is 10 pg of parasite template combined with 1 ng of human DNA.

Fig 3

Assay performance with mixtures of genomes and MAAB. (A) Mixture of three genomes showing clear differentiation between fractions: 3D7-7G8-Dd2. The pfcrtK72-76 assay is shown. (B) MAAB. Shown in the pfcrt K72-76 assay, MAAB in the LightScanner-32 increases the likelihood of detecting low-frequency allele fractions in a mixed population. Different proportions of mutant DNA were mixed with wild-type DNA as follows: 100% wild-type DNA (3D7), a 50/50 mixture of mutant DNA (7G8) with wild-type DNA (3D7), a 25/75 mixture of mutant DNA (7G8) and wild-type DNA (3D7), and so on down to less than 1% mutant DNA mixed with wild-type DNA in 1 ng total template concentration.

Fig 4

Detection of emerging and new mutations. (A) New SNP haplotype detected in Senegal patient samples. The pfdhps 436A/437A and 436Y/437G haplotypes have not previously been reported in Senegal, though they have been reported in other regions. (B) Novel mutation in cytB. The plot of melting peaks shows the expected wild-type Y268 and mutant S268 peaks as controls in gray and red, respectively. The new mutation, M270I, confirmed by sequencing, is shown in blue.