The Role of Small RNA-Based Epigenetic Silencing for Purifying Selection on Transposable Elements in Capsella grandiflora

Abstract

To avoid negative effects of transposable element (TE) proliferation, plants epigenetically silence TEs using a number of mechanisms, including RNA-directed DNA methylation. These epigenetic modifications can extend outside the boundaries of TE insertions and lead to silencing of nearby genes, resulting in a trade-off between TE silencing and interference with nearby gene regulation. Therefore, purifying selection is expected to remove silenced TE insertions near genes more efficiently and prevent their accumulation within a population. To explore how effects of TE silencing on gene regulation shapes purifying selection on TEs, we analyzed whole genome sequencing data from 166 individuals of a large population of the outcrossing species Capsella grandiflora. We found that most TEs are rare, and in chromosome arms, silenced TEs are exposed to stronger purifying selection than those that are not silenced by 24-nucleotide small RNAs, especially with increasing proximity to genes. An age-of-allele test of neutrality on a subset of TEs supports our inference of purifying selection on silenced TEs, suggesting that our results are robust to varying transposition rates. Our results provide new insights into the processes affecting the accumulation of TEs in an outcrossing species and support the view that epigenetic silencing of TEs results in a trade-off between preventing TE proliferation and interference with nearby gene regulation. We also suggest that in the centromeric and pericentromeric regions, the negative aspects of epigenetic TE silencing are missing.

Keywords: Capsella; epigenetic silencing; gene expression; methylation; purifying selection; transposable elements.

Fig. 1.

—Insertion frequency spectrum (IFS) of siRNA + (orange) and siRNA − (blue) TE insertions. (A) IFS of siRNA+ and siRNA− TEs in centromeric/pericentromeric regions and chromosome arms. Error bars are 95% confidence intervals derived from 1,000 bootstrap replicates of each TE category. (B) ISF of siRNA+ TE insertions split into four different groups based on their position on the chromosome arms (within genes, flanking genes, 1–2 kb away from genes and >2 kb away from genes). (C) ISF of siRNA− TE insertions split into the same four groups. Error bars are 95% confidence intervals derived from 1,000 bootstrap replicates of each TE category.

Fig. 2.

—Proportion of the rarest (frequency < 0.02) siRNA + (orange) and siRNA − (blue) TE insertions found within, flanking, 1–2, 2–3, 3–5 and >5 kb away from genes. Error bars are 95% confidence intervals derived from 1,000 bootstrap replicates of each TE category.

Fig. 3.

—Results of the age-of-allele test of neutrality for TE insertions (A) Observed and expected TE insertion frequencies of the 28 evaluated TE copies in the Capsella grandiflora population, ranked by increasing insertion age. (B) Probability of observing an identical or lower TE insertion frequency, if the TE insertions are evolving neutrally. The dotted line represents the significance cutoff of 0.05.