Paternally Expressed Imprinted Genes under Positive Darwinian Selection in Arabidopsis thaliana

Abstract

Genomic imprinting is an epigenetic phenomenon where autosomal genes display uniparental expression depending on whether they are maternally or paternally inherited. Genomic imprinting can arise from parental conflicts over resource allocation to the offspring, which could drive imprinted loci to evolve by positive selection. We investigate whether positive selection is associated with genomic imprinting in the inbreeding species Arabidopsis thaliana. Our analysis of 140 genes regulated by genomic imprinting in the A. thaliana seed endosperm demonstrates they are evolving more rapidly than expected. To investigate whether positive selection drives this evolutionary acceleration, we identified orthologs of each imprinted gene across 34 plant species and elucidated their evolutionary trajectories. Increased positive selection was sought by comparing its incidence among imprinted genes with nonimprinted controls. Strikingly, we find a statistically significant enrichment of imprinted paternally expressed genes (iPEGs) evolving under positive selection, 50.6% of the total, but no such enrichment for positive selection among imprinted maternally expressed genes (iMEGs). This suggests that maternally- and paternally expressed imprinted genes are subject to different selective pressures. Almost all positively selected amino acids were fixed across 80 sequenced A. thaliana accessions, suggestive of selective sweeps in the A. thaliana lineage. The imprinted genes under positive selection are involved in processes important for seed development including auxin biosynthesis and epigenetic regulation. Our findings support a genomic imprinting model for plants where positive selection can affect paternally expressed genes due to continued conflict with maternal sporophyte tissues, even when parental conflict is reduced in predominantly inbreeding species.

Keywords: endosperm; genomic conflict; genomic imprinting; plant evolution; positive Darwinian selection.

Fig. 1.

Summary of scenarios for selection on imprinted plant genes. Schematic of Arabidopsis thaliana seed summarizing the impacts of genomic imprinting on genetic selection as predicted by major hypotheses for genomic imprinting. In each case, the diploid F1 embryo is shown in dark green, surrounded by the triploid F1 endosperm, shown in yellow) in which imprinting occurs, and the diploid seed coat (SC) which is part of the maternal sporophyte, shown in light green. (A) Intragenomic conflict in which antagonism between matrigenes and patrigenes over resource allocation results in physical interactions between iMEGs and iPEGs (Spillane et al. 2007). (B) Coadaptation models predict that any selective pressure should be concentrated on iMEGs which are coinherited with cytoplasmic genomes in A. thaliana (Wolf and Brandvain 2014). (C) Indirect conflict or “Kinship Model” predicts that conflict between iPEGs and genes expressed in maternal tissues (e.g., seed coat, scMEG, or other sporophyte tissues) leads to positive selection on iPEGs (Willi 2013).

Fig. 2.

Size of orthology clusters to which imprinted Arabidopsis thaliana genes belong. Orphans are defined according to (Donoghue et al. 2011); genes present in orthology clusters >6 were considered for further selective pressure variation analysis.

Fig. 3.

Summary of the number of genes under positive selection in the data set. (A) Numbers of imprinted Arabidopsis thaliana genes under site and/or lineage specific PS; (B and C) the percentages of A. thaliana iMEGs and iPEGs subject to lineage-specific (B) or site-specific (C) PS compared with the percentages in control sets of endosperm-expressed (“Endosperm”) or genome-wide (“Genome”) biallelic genes; control gene-sets are listed in supplementary table S4, Supplementary Material online.

Fig. 4.

Phylogeny of the 34 species included in our analyses and the age distribution of iMEGs and iPEGs. (A) This shows the frequency of age class (AC) for the iMEGs and iPEGs tested. AC0, Arabidopsis thaliana specific; AC1, A. lyrata; AC2, Brassicaceae; AC3, Brassicales-Malvales; AC4, Rosid; AC5, Eudicot; AC6, Angiosperm; AC7, Tracheophyte; AC8, Embryophyte; AC9, Viridiplantae. (B) Consensus phylogenetic relationships of all 34 species; the phylogenetic position of the age classes and the known whole genome duplication events for the species included in the study are also highlighted (Vanneste et al. 2014).

Fig. 5.

Distribution of D_N/D_S and P_N/P_S ratios for imprinted genes compared with all protein-coding genes in Arabidopsis thaliana. X-axis depicts P_N/P_S ratios, Y-axis represents D_N/D_S ratios. Green dots denote genes under purifying selection, red dots denote genes under positive selection, yellow dots denote genes under neutral evolution, black triangles denote A. thaliana imprinted genes, blue triangles denote pseudogenes with high D_N/D_S and high P_N/P_S. No clustering was observed.