Possible diversifying selection in the imprinted gene, MEDEA, in Arabidopsis

Abstract

Coevolutionary conflict among imprinted genes that influence traits such as offspring growth may arise when maternal and paternal genomes have different evolutionary optima. This conflict is expected in outcrossing taxa with multiple paternity, but not self-fertilizing taxa. MEDEA (MEA) is an imprinted plant gene that influences seed growth. Disagreement exists regarding the type of selection acting on this gene. We present new data and analyses of sequence diversity of MEA in self-fertilizing and outcrossing Arabidopsis and its relatives, to help clarify the form of selection acting on this gene. Codon-based branch analysis among taxa (PAML) suggests that selection on the coding region is changing over time, and nonsynonymous substitution is elevated in at least one outcrossing branch. Codon-based analysis of diversity within outcrossing Arabidopsis lyrata ssp. petraea (OmegaMap) suggests that diversifying selection is acting on a portion of the gene, to cause elevated nonsynonymous polymorphism. Providing further support for balancing selection in A. lyrata, Hudson, Kreitman and Aguadé analysis indicates that diversity/divergence at silent sites in the MEA promoter and genic region is elevated relative to reference genes, and there are deviations from the neutral frequency spectrum. This combination of positive selection as well as balancing and diversifying selection in outcrossing lineages is consistent with other genes influence by evolutionary conflict, such as disease resistance genes. Consistent with predictions that conflict would be eliminated in self-fertilizing taxa, we found no evidence of positive, balancing, or diversifying selection in A. thaliana promoter or genic region.

FIG. 1.—

Sliding window plots (window size = 150 bp, step size = 5 bp) for (A) silent nucleotide diversity (π_sil), (B) divergence (K_sil), (C) diversity-to-divergence ratio (π_sil/K_sil) in Arabidopsis thaliana (A.t.), Arabidopsis lyrata lyrata (A.l.l.), and Arabidopsis lyrata petraea (A.l.p.). (D) Tajima's D (D) was calculated with all sites and a window size of 200 bp (step size = 5 bp). *Regions where Tajima's D is significantly different from zero (P < 0.05) using SCANMS (Ardell 2004). (E) The black boxes correspond with MEA exons and the white box indicates the 5′ gene (At1g02570). Arrows with closed ends indicate regions that may be important for MEA cis regulation (Baroux et al. 2006; Gehring et al. 2006; Jullien et al. 2006). Arrows with open ends indicate regions with high H3K27 methylation in 10-day old seedlings, where MEA is silent (Zhang et al. 2007).

FIG. 2.—

Sliding window plots for K_a/K_s (window size = 200 bp, step size = 5 bp) and K_a (window size = 150 bp, step size = 5 bp) of comparisons between species pairs: Arabidopsis thaliana (A.t.), Arabidopsis lyrata (A.l.), Arabidopsis halleri (A.h.), and Boechera stricta (B.s.). Color coding is the same in both plots. The small letters in the upper plot indicate regions with elevated K_a/K_s discussed in the text. At the top of the figure is a map of exons (1–17) and functional domains (gray boxes): (A) an important region for the interaction between MEA and FIE (Spillane et al. 2000; Yadegari et al. 2000); (B) EZD1 and (C) EZD2 are found in E(Z) homologs but their function is unknown; (D) a SANT domain, which may be involved in histone binding (Boyer et al. 2004); (E) a putative nuclear localization signal (NLS); (F) a cysteine-rich region (CXC), which may be important in protein folding (Sardiu et al. 2007). (G) a SET domain, which has H3K27 methylation activity (Springer et al. 2002; Ng et al. 2007).

FIG. 3.—

Phylogenetic relationship of MEA genes estimated by maximum likelihood from five Brassicaceae species. The numbers above branches are the ratios of nonsynonymous/synonymous substitution (ω) estimated by free-ratio branch model of codeml (Yang 1997). Although branch B had a significantly greater ω than other branches, ω was not significantly greater than 1.

FIG. 4.—

Results of OmegaMap analysis for Arabidopsis lyrata petraea coding region under Prior set A of supplementary table S2, Supplementary Material online. (a) The sitewise mean ω (solid line) and the 95% HPD interval (shaded area) of ω along the gene. The dotted line corresponds to neutrality (ω = 1). (b) Sitewise posterior probability of diversifying selection (ω > 1). The two vertical dotted lines correspond to 95% and 99% probabilities. The map of exon boundaries and conserved functional domains at the top of the figure is the same as in figure 2.