Selection and mutation on microRNA target sequences during rice evolution

Abstract

Background: MicroRNAs (miRNAs) posttranscriptionally down-regulate gene expression by binding target mRNAs. Analysis of the evolution of miRNA binding sites is helpful in understanding the co-evolution between miRNAs and their targets. To understand this process in plants a comparative analysis of miRNA-targeted duplicated gene pairs derived from a well-documented whole genome duplication (WGD) event in combination with a population genetics study of six experimentally validated miRNA binding sites in rice (O. sativa) was carried out.

Results: Of the 1,331 pairs of duplicate genes from the WGD, 41 genes (29 pairs) were computationally predicted to be miRNA targets. Sequence substitution analysis indicated that the synonymous substitution rate was significantly lower in the miRNA binding sites than their 5' and 3' flanking regions. Of the 29 duplicated gene pairs, 17 have only one paralog been targeted by a miRNA. This could be due to either gain of a miRNA binding site after the WGD or because one of the duplicated genes has escaped from being a miRNA target after the WGD (loss of miRNA binding site). These possibilities were distinguished by separating miRNAs conserved in both dicots and monocot plants from rice-specific miRNAs and by phylogenetic analysis of miRNA target gene families. The gain/loss rate of miRNA binding sites was estimated to be 3.0 x 10(-9) gain/loss per year. Most (70.6%) of the gains/losses were due to nucleotide mutation. By analysis of cultivated (O. sativa; n = 30) and wild (O. rufipogon; n = 15) rice populations, no segregating site was observed in six miRNA binding sites whereas 0.12-0.20 SNPs per 21-nt or 1.53-1.80 x 10(-3) of the average pairwise nucleotide diversity (pi) were found in their flanking regions.

Conclusion: Both molecular evolution and population genetics support the hypothesis that conservation of miRNA binding sites is maintained by purifying selection through elimination of deleterious alleles. Nucleotide mutations play a major role in the gain/loss of miRNA binding sites during evolution.

Figure 1

Flow chart for the methods used in this study. Two approaches, molecular evolution and population genetics were used to analyze a selection of the miRNA target genes and the gain or loss of miRNA binding sites. a: the rice genome annotation by the TIGR; b: empirical parameters: no mismatch at positions 10 and 11; no more than one mismatch at positions 2–12; no more than two consecutive mismatches downstream of position 13; c: 76 miRNAs from miRBase (Release 11.0, ) and 38 newly identified miRNAs [3]; d: see Additional file 8; e: see Additional file 4.

Figure 2

The loss of miR397 binding sites in a gene family. A: Phylogenetic tree of gene family of L-ascorbate oxidase precursor. The members predicted to be targets of miR397 are labelled with asterisks and the branch node where the WDG event occurred is indicated in blue bold line and the corresponding WGD gene pairs are shown in red lines. Os: Oryza sativa; At: Arabidopsis thaliana and Pt: Pinus taeda; B: The alignment of miR397 binding sites of four target genes from the WGD. The numbers of mismatch between miR397 and its binding sites are shown in parentheses. The mutation sites among the four genes are boxed.