Whole genome sequencing of multiple Leishmania donovani clinical isolates provides insights into population structure and mechanisms of drug resistance

Abstract

Visceral leishmaniasis is a potentially fatal disease endemic to large parts of Asia and Africa, primarily caused by the protozoan parasite Leishmania donovani. Here, we report a high-quality reference genome sequence for a strain of L. donovani from Nepal, and use this sequence to study variation in a set of 16 related clinical lines, isolated from visceral leishmaniasis patients from the same region, which also differ in their response to in vitro drug susceptibility. We show that whole-genome sequence data reveals genetic structure within these lines not shown by multilocus typing, and suggests that drug resistance has emerged multiple times in this closely related set of lines. Sequence comparisons with other Leishmania species and analysis of single-nucleotide diversity within our sample showed evidence of selection acting in a range of surface- and transport-related genes, including genes associated with drug resistance. Against a background of relative genetic homogeneity, we found extensive variation in chromosome copy number between our lines. Other forms of structural variation were significantly associated with drug resistance, notably including gene dosage and the copy number of an experimentally verified circular episome present in all lines and described here for the first time. This study provides a basis for more powerful molecular profiling of visceral leishmaniasis, providing additional power to track the drug resistance and epidemiology of an important human pathogen.

Figure 1.

The phylogenomic context of functional variation in clinical L. donovani lines. (A) A neighbor-joining phylogenomic tree of Leishmania species. The relative similarity is shown (1% was equivalent to 108,000 mutations). The first column indicates the fraction of the known genomes orthologously aligned to L. donovani and the second is the number of SNPs identified (Table 2). (B) Principle component analysis (PCA) of genomic SNP variation in the 17 L. donovani lines resistant (red) and susceptible (blue) to SGG. The two most significant PCs distinguished three main groups, accounting for 63.3% of the total variation. The PCA L. donovani line numbers correspond to those shown in C, a median-joining phylogenetic network of genome-wide nonsynonymous sites variation (396 SNPs) for samples resistant (red) and susceptible (blue) to SSG from India (beginning with BHU) and Nepal (BPK). The branch lengths displayed are proportional to the number of differences between lines. The position of the ancestral node for L. infantum (black) was shortened: For comparison, the L. donovani:L. infantum branch length was 138 times that between the most divergent L. donovani (1, BHU568/0cl1; and 10, BPK035/0cl1).

Figure 2.

The neutrality of coding-site allele-frequency spectra. The allele-frequency spectrum for SNPs at nonsynonymous (black line) and synonymous (dashed line) sites among the 17 samples for the genome. Allele frequencies were computed using read-depth coverage values and were folded because the selective process may act on both the ancestral alleles inferred from L. infantum as well as the derived version with the L. donovani population. The ratio of nonsynonymous to synonymous SNPs approaches one as the allele frequencies increase, suggesting that higher frequency variants are subject to stronger purifying selection to remove deleterious alleles.

Figure 3.

Aneuploidy in natural populations of L. donovani. The heatmap shows the copy-number status of the 36 chromosomes for the 17 lines as disomic (1, blue), trisomic (1.5, green), tetrasomic (2, yellow), and pentasomic (2.5, ivory). The color key shows the normalized chromosome read-depth and the distribution frequency.

Figure 4.

Copy-number variation spanning MAPK-locus on chromosome 36. Copy number per cell in L. donovani (red, SSG-R; blue, SSG-S) was much higher for the MAPK locus than the chromosome 36 median for L. donovani (black dashed line) and L. infantum (black). The amplified region is marked in gray and their white edges indicate the location of two direct repeats. Genes within and adjacent to the amplified region are given in black and light gray, respectively. The MAPK (mitogen-activated protein kinase) gene is labeled as LMPK (LdBPK_366760). Recombination events may occur through a highly homologous 220-bp region starting at the start codon of the tartrate-sensitive acid phosphatase (LdBPK_366740) and histidine secretory acid phosphatase (LdBPK_366770). LdBPK_366750 encodes a protein of unknown function. Two RIME (L. donovani ribosomal mobile elements; light-blue) flank the duplicated region. Read-depth values were computed using 1-kb sliding windows, and error bars are shown for each kilobase.