Genomic confirmation of hybridisation and recent inbreeding in a vector-isolated Leishmania population

Abstract

Although asexual reproduction via clonal propagation has been proposed as the principal reproductive mechanism across parasitic protozoa of the Leishmania genus, sexual recombination has long been suspected, based on hybrid marker profiles detected in field isolates from different geographical locations. The recent experimental demonstration of a sexual cycle in Leishmania within sand flies has confirmed the occurrence of hybridisation, but knowledge of the parasite life cycle in the wild still remains limited. Here, we use whole genome sequencing to investigate the frequency of sexual reproduction in Leishmania, by sequencing the genomes of 11 Leishmania infantum isolates from sand flies and 1 patient isolate in a focus of cutaneous leishmaniasis in the Çukurova province of southeast Turkey. This is the first genome-wide examination of a vector-isolated population of Leishmania parasites. A genome-wide pattern of patchy heterozygosity and SNP density was observed both within individual strains and across the whole group. Comparisons with other Leishmania donovani complex genome sequences suggest that these isolates are derived from a single cross of two diverse strains with subsequent recombination within the population. This interpretation is supported by a statistical model of the genomic variability for each strain compared to the L. infantum reference genome strain as well as genome-wide scans for recombination within the population. Further analysis of these heterozygous blocks indicates that the two parents were phylogenetically distinct. Patterns of linkage disequilibrium indicate that this population reproduced primarily clonally following the original hybridisation event, but that some recombination also occurred. This observation allowed us to estimate the relative rates of sexual and asexual reproduction within this population, to our knowledge the first quantitative estimate of these events during the Leishmania life cycle.

Figure 1. Genome-wide patterns of polymorphism and divergence show patchy heterozygosity symptomatic of hybrid ancestry.

(A) Plot of single-nucleotide polymorphisms (SNPs) with respect to L. infantum JCPM5 in a section of chromosome 28 across the 12 strains. Orange ‘blocked’ bars indicate heterozygous positions; green, red, blue and black indicate homozygous variant calls of A, C, G and T respectively. Regions derived from one or both parental strains are shown in the block diagram below. CUK strain numbers (left) and chromosomal position (above, nucleotides) are as indicated. (B). Schematic of putative hybridization/inbreeding model, illustrating how a single hybridization event followed by crossing within the offspring could result in patchy heterozygosity in the progeny sampled here. (C) Parentage blocks inferred from mixture model predictions. White corresponds to regions that are similar to the JPCM5 parent (SNP sparse) in all isolates; black similar to the unknown parent (SNP dense) in all isolates; blue are heterozygous in all isolates. Red indicates regions that vary between isolates, and so are either SNP dense and SNP sparse in different isolates, or heterozygous in some isolates.

Figure 2. Phased haplotypes help identify one of the parental strains.

(A) A heterozygous SNP-dense region of chromosomes 22 (boxed) was one of 86 regions that could be successfully phased to infer haplotypes of at least 5 kb in at least one CUK strain. This region could be phased in 9 strains. (B) Phylogenetic analysis of this region confirms that the two haplotypes have different evolutionary histories: one related to L. infantum JPCM5 and the other distantly related to the two L. donovani reference genomes.

Figure 3. Genome-wide divergence between the CUK strains and other L. donovani complex reference genomes.

(A) Divergence between the Turkish population and the L. donovani BPK282/0cl4 and L. infantum JPCM5 reference genomes. Values shown are numbers of SNPs in non-overlapping 10 kb windows, scaled using the divergence to L. major for each window. A higher fraction of blocks show no more difference between the Turkish population and JPCM5 than to BPK282/0cl4, but other blocks are significantly divergent from both references. (B) Frequency distribution of divergence between the CUK strains and the reference genomes, L. infantum JPCM5 (blue solid line) and L. donovani BPK282/0cl4 (red dashed line) for 10 kb windows, scaled using the divergence with L. major. Divergence to JPCM5 shows a significantly bimodal pattern, with sets of very low divergence blocks (mode 0.1 SNPs/kb) and higher divergence (mode 1.7 SNPs/kb), while BPK282/0cl4 divergence cannot reject unimodal high divergence (2.0 SNPs/kb). Genome-wide patterns suggest one ancestor closely related to JPCM5 and a second ancestor not closely related to either of the two references.

Figure 4. Aneuploidy in the CUK strains.

The heatmap shows chromosomal copy number for all 36 chromosomes across the 12 CUK strains, based on normalized chromosome read-depth. Copy number varies from disomic (light yellow) to pentasomic (dark-red).

Figure 5. Changes in SNP coverage depth pinpoint recombination breakpoints.

Chromosome 4 is trisomic in three of the strains, so changes in read depth for derived (non JPCM5-reference) alleles (A) show changes within some otherwise unphased blocks of heterozygous SNPs (B), showing recombination breakpoints not visible from simple plots of variant calls and not identifiable by other summary statistics (C). (A) shows derived allele read depth for the three trisomic strains, where changes in read depth inside the shaded boxes indicate presumed recombination events, where allele calls change from being present in 1/3 or 2/3 homologous chromosome copies. (B) Single-nucleotide polymorphisms (SNPs) with respect to L. infantum JCPM5 in a section of chromosome 4 across the 12 strains. Orange ‘blocked’ bars indicate heterozygous positions; green, red, blue and black indicate homozygous variant calls of A, C, G and T respectively. (C) Summary statistics for variant calls in 5 kb windows.