Hyperdiverse gene cluster in snail host conveys resistance to human schistosome parasites

Abstract

Schistosomiasis, a neglected global pandemic, may be curtailed by blocking transmission of the parasite via its intermediate hosts, aquatic snails. Elucidating the genetic basis of snail-schistosome interaction is a key to this strategy. Here we map a natural parasite-resistance polymorphism from a Caribbean population of the snail Biomphalaria glabrata. In independent experimental evolution lines, RAD genotyping shows that the same genomic region responds to selection for resistance to the parasite Schistosoma mansoni. A dominant allele in this region conveys an 8-fold decrease in the odds of infection. Fine-mapping and RNA-Seq characterization reveal a <1Mb region, the Guadeloupe Resistance Complex (GRC), with 15 coding genes. Seven genes are single-pass transmembrane proteins with putative immunological roles, most of which show strikingly high nonsynonymous divergence (5-10%) among alleles. High linkage disequilibrium among three intermediate-frequency (>25%) haplotypes across the GRC, a significantly non-neutral pattern, suggests that balancing selection maintains diversity at the GRC. Thus, the GRC resembles immune gene complexes seen in other taxa and is likely involved in parasite recognition. The GRC is a potential target for controlling transmission of schistosomiasis, including via genetic manipulation of snails.

Fig 1. Resistance phenotypes.

Standard errors of proportions are indicated by vertical bars. (A) Susceptibility declined rapidly over five generations in snail lines exposed to 10 (R10) or 30 (R30) miracidia, when only uninfected snails were allowed to contribute to the next generation (N = 46–100). A control line (GUA) that was not exposed to the parasite showed no comparable change (N = 46–65). (B) Among 289 snails, the R allele at the GRC locus grc1 is strongly correlated with resistance in a dominant fashion. There are no significant differences in resistance among genotypes with R, nor among genotypes without R.

Fig 2. F_ST values between the unselected control line (GUA) and each selected line (R10 and R30).

Three null RAD markers, aligned with BWA [65] to Scaffold1, all showed unusually high F_ST in one or both comparisons (red). These markers conform to perfect linkage disequilibrium with the haplotypes as determined by Sanger sequencing, with the different F_ST values owing only to the error inherent in estimating allele frequencies from null markers. An additional unaligned marker with high F_ST in both comparisons was identified with Stacks [35] and found to reside on Scaffold4 (blue).

Fig 3. Pairwise divergence of GRC genes.

Genes are aligned to their approximate genomic position on the x-axis, staggered slightly for ease of visualization. Scaffold7477 has been inserted into its inferred position within Scaffold1. Scaffold4 has been inverted relative to its arbitrarily designated reference genome orientation to indicate that only the end of this scaffold is part of the GRC. Pairwise divergence for all three haplotype combinations (indicated by color) is shown in the left y-axis. Gene symbols vary for ease of distinguishing adjacent genes, and to show divergence type: solid symbols represent nonsynonymous divergence (d_N) in coding genes, open symbols represent silent (synonymous and noncoding) divergence in coding genes (d_S), and other symbols represent divergence in noncoding genes. Only the six TM1 genes in the center of the region of association (grctm2–7) show high (>1%) d_N. The names of the two most promising candidates, grctm5 and grctm6, are highlighted in red. Brown lines and squares indicate the boundaries of the GRC region of statistical association as determined by Sanger sequencing markers: for these markers, the right y-axis indicates linkage disequilibrium correlation (r; Table 1) with marker locus grc1 (labeled). The gap between Scaffold1 and Scaffold4 (dotted line) is of unknown size, but contains no expressed genes with sequence or expression differences among haplotypes, as these would have been detected in our RNA-Seq analysis.

Fig 4. Allele-specific nonsynonymous substitution among alleles of the seven TM1 genes.

Alleles of the same gene are aligned to each other (gaps indicated by whitespace), but across genes only the transmembrane domain (TM) is aligned (extracellular (N-terminal) regions are to the left of the TM). If a haplotype includes two copies of a gene that is single-copy on the other haplotypes, both copies are shown (e.g. gene grctm5 is duplicated on the R haplotype, so we label those sequences Ra and Rb). For each allele, we calculated allele-specific nonsynonymous substitution (i.e. divergence from the inferred ancestral sequence) in 75bp sliding windows, indicated by color (“allele-specific d_N”). Regions with no sequence similarity are indicated in red. Substitution across a 75bp window could not be calculated in black sections. Fibronectin III domains (FN3) are shown. The premature stop codon in grctm1 S2 is shown with an asterisk. Nonsynonymous substitution is extremely high across the TM1 genes, exceeding 15% in some windows for all genes except grctm1, and occasionally reaching over 30%. Only grctm5 and grctm6 show high nonsynonymous substitution specific to R alleles.