DNA methylation signatures of duplicate gene evolution in angiosperms

Gene duplication is a source of evolutionary novelty. DNA methylation may
play a role in the evolution of duplicate genes through its association
with gene expression. While this relationship is examined to varying
extents in a few individual species, the generalizability of these results
at either a broad phylogenetic scale with species of differing duplication
histories or across a population remains unknown. We apply a comparative
epigenomics approach to 43 angiosperm species across the phylogeny and a
population of 928 Arabidopsis thaliana accessions, examining the
association of DNA methylation with paralog evolution. Genic DNA
methylation is differentially associated with duplication type, the age of
duplication, sequence evolution, and gene expression. Whole genome
duplicates are typically enriched for CG-only gene-body methylated or
unmethylated genes, while single-gene duplications are typically enriched
for non-CG methylated or unmethylated genes. Non-CG methylation, in
particular, was characteristic of more recent single-gene duplicates. Core
angiosperm gene families are differentiated into those which
preferentially retain paralogs and ‘duplication-resistant’ families, which
convergently revert to singletons following duplication.
Duplication-resistant families which still have paralogous copies are,
uncharacteristically for core angiosperm genes, enriched for non-CG
methylation. Non-CG methylated paralogs have higher rates of sequence
evolution, higher frequency of presence-absence variation, and more
limited expression. This suggests that silencing by non-CG methylation may
be important to maintaining dosage following duplication and be a
precursor to fractionation. Our results indicate that genic methylation
marks differing evolutionary trajectories and fates between paralogous
genes and have a role in maintaining dosage following duplication.