Pan-arthropod analysis reveals somatic piRNAs as an ancestral defence against transposable elements

Abstract

In animals, small RNA molecules termed PIWI-interacting RNAs (piRNAs) silence transposable elements (TEs), protecting the germline from genomic instability and mutation. piRNAs have been detected in the soma in a few animals, but these are believed to be specific adaptations of individual species. Here, we report that somatic piRNAs were probably present in the ancestral arthropod more than 500 million years ago. Analysis of 20 species across the arthropod phylum suggests that somatic piRNAs targeting TEs and messenger RNAs are common among arthropods. The presence of an RNA-dependent RNA polymerase in chelicerates (horseshoe crabs, spiders and scorpions) suggests that arthropods originally used a plant-like RNA interference mechanism to silence TEs. Our results call into question the view that the ancestral role of the piRNA pathway was to protect the germline and demonstrate that small RNA silencing pathways have been repurposed for both somatic and germline functions throughout arthropod evolution.

Figure 1. Genes in small RNA pathways evolve rapidly throughout the arthropods

a, The gain and loss of genes encoding the components of different sRNA pathways during arthropod evolution. Taxa with somatic piRNAs are shown in black, and the colour of the branches is a Bayesian reconstruction of whether somatic piRNAs were present. The posterior probability that the ancestral arthropod had somatic piRNAs is 0.9956. b, Phylogenetic analysis of RdRP genes from arthropods, other animals, plants and fungi. Note S. maritima is more closely related to fungal than animal RdRP (posterior probability at N. crassa - S. maritima node is 1). c, The antisense enrichment (measured as log₂ (antisense/sense) median RNA-Seq read counts) for TEs that produce siRNAs. Species are classified by possession of an RdRP. Note S. maritima (red) lacks an animal RdRP. d, Counts of sense and antisense RNAseq reads of the 60 most highly expressed TEs in H. melpomene (no RdRP; red) and P. tepidariorum (six RdRPs; blue). Among these, the 15 TEs in each species that generate the most siRNAs are shown as filled circles while the remainder are open circles. In H. melpomene siRNAs are associated with antisense transcription.

Figure 2. piRNAs are absent in B. terrestris male germline

The size and 5ʹ nucleotide of sRNAs from testis (a) and ovary (b). Plots show unique reads that map to the genome (where the same sequence occurred more than once, all but one read was eliminated). The inset shows the overlap between sense and antisense 25-29nt sRNAs.

Figure 3. Somatic piRNAs are widespread, and target TEs throughout the arthropods

a, Genes in the piRNA pathway have higher somatic expression in species with somatic piRNAs. For species with multiple copies of a gene, the mean scaled somatic expression level of each duplicate is displayed. The box shows the median and interquartile range (IQR), and the bottom and top whiskers show the range of points no further than 1.5×IQR away from the first and third quartiles respectively. b-d, The size and 5ʹ nucleotide of sRNAs mapping to the TEs from P. tepidariorum, showing 10 bp overlap between sense and antisense 25–29nt sRNAs. piRNAs targeting TEs are evident in the germline (b) and soma (c), and these somatic piRNAs are resistant to sodium periodate oxidation, indicating that they are 3ʹ methylated (d). Plots show all reads that map to the TEs.

Figure 4. Virally-derived sRNAs in three arthropod species

The size and 5′ nucleotide of sRNAs mapping to viral transcripts and genomes reconstructed from RNA-Seq data. Virally-derived piRNAs are evident in A. aegypti (a) and C. sculpturatus (b), and virally-derived siRNAs are found in T. castaneum (c). Only A. aegypti shows the 10 bp overlap between sense and antisense 25–29 nt sRNAs that is diagnostic of Ping-Pong amplification (insets). Reads derived from the sense strand are shown above zero, antisense reads below.

Figure 5. A model of the divergent sRNA pathways silencing TEs in different arthropods

Our data suggest that the mechanisms of sRNA pathways have diverged in two key areas. In some lineages, the piRNA pathway is restricted to the germline (e.g., flies), whereas in most others it is active in the soma and the germline (e.g., spiders). Additionally, in some lineages (e.g., spiders), RdRP may synthesize dsRNA from transcripts produced by RNA polymerase II, amplifying the siRNA response.