Genomic evidence of contemporary hybridization between Schistosoma species

Abstract

Hybridization between different species of parasites is increasingly being recognised as a major public and veterinary health concern at the interface of infectious diseases biology, evolution, epidemiology and ultimately control. Recent research has revealed that viable hybrids and introgressed lineages between Schistosoma spp. are prevalent across Africa and beyond, including those with zoonotic potential. However, it remains unclear whether these hybrid lineages represent recent hybridization events, suggesting hybridization is ongoing, and/or whether they represent introgressed lineages derived from ancient hybridization events. In human schistosomiasis, investigation is hampered by the inaccessibility of adult-stage worms due to their intravascular location, an issue which can be circumvented by post-mortem of livestock at abattoirs for Schistosoma spp. of known zoonotic potential. To characterise the composition of naturally-occurring schistosome hybrids, we performed whole-genome sequencing of 21 natural livestock infective schistosome isolates. To facilitate this, we also assembled a de novo chromosomal-scale draft assembly of Schistosoma curassoni. Genomic analyses identified isolates of S. bovis, S. curassoni and hybrids between the two species, all of which were early generation hybrids with multiple generations found within the same host. These results show that hybridization is an ongoing process within natural populations with the potential to further challenge elimination efforts against schistosomiasis.

Fig 1. Chromosomal synteny between Schistosoma mansoni (v9) and Schistosoma curassoni (V1).

(A) Hi-C contact map of the chromosomes (1–7,Z) and unplaced contigs/scaffolds (U, n = 361), visualised using Juicebox Assembly Tools. (B) Circos plot of genome-wide features. Outer: median GC content, shown as a proportion of 100 kb windows along each chromosome, middle: depth of coverage of PacBio subreads (median of 100 kb windows), inner: density of annotated repeat elements in 1 Mb windows, including LINEs (pink), DNA (green), LTRs (orange) and Penelope (blue). (C) We used MCscan to identify homologous chromosomal regions between our newly sequenced S. curassoni assembly and the S. mansoni (v9) assembly. Per-chromosome synteny showing syntenic paths, structural rearrangements (inversions, translocations and duplications) and un-aligned regions between S. curassoni chromosomes (shown in different colours) and S. mansoni (chromosomes shown in dark grey). Chromosomal lengths (in Mb) are shown at the end of each chromosome.

Fig 2. Population structure of sequenced isolates.

Samples denoted with triangles represent reference samples, circles represent samples sequenced for this study. (A) Principal component analysis (PCA) of genomic differentiation between the 21 isolates sequenced for this study (grey circles) and the 12 samples included as references for each available Haematobium clade species (triangles): S. margrebowiei (yellow), S. mattheei (pink), S. intercalatum (orange), S. guineensis (green) S. haematobium (purple), S. curassoni (red) and S. bovis (blue). The first four principal components represented 75% of the total variance. (C) Maximum-likelihood phylogeny inferred using 23,136 parsimony informative single-nucleotide polymorphisms. Tips are coloured and shaped as in A&B, single-species and outgroups groups are highlighted. (D) ADMIXTURE plots showing the population structure, assuming three populations are present (reference samples for three species were included), we used 10-fold cross-validation and standard error estimations using 1000 bootstraps. Y-axis values show the estimated admixture proportions for each sample.

Fig 3. Evidence of admixture in sequenced samples.

The phylogeny and admixture proportion estimates are identical to Fig 2. We calculated the proportion of heterozygous variants per 50 kb windows along each autosome. The three-population statistic (f₃) was used to test for admixture between each test sample and both reference populations for each species. We defined three populations described by the tree ((P1,P2),P3)). We designated reference populations for S. bovis and S. curassoni as P1 and P2, respectively. Individual samples to be analysed were designated at P3.

Fig 4. Per-sample estimates of ancestry.

Ancestry was estimated using ADMIXTURE in 150 kb nonoverlapping windows along each chromosome. The proportion of S. curassoni ancestry is shown in red, S. bovis in blue. Ancestry estimates are shown in phylogenetic order based on a neighbour-joining phylogeny inferred using identity-by-descent distances showing the relatedness between all samples. The scale represents the mean numbers of substitutions. Excluded from this plot are the six samples used to designate reference alleles for each population (SCUR_01, SBOV_01, SBOV_02, BK16_B3_01, BK16_B8_03, RT15_B6_01).