The growing understanding of the key roles of miRNAs in plant development and physiology (reviewed in ref. 1 and elsewhere) has led to an interest in the forces shaping miRNA evolution and the possible contributions of miRNAs to phenotypic diversity in plants.2 We and others recently used a molecular population genetics approach to study miRNAs in Arabidopsis thaliana (hereafter referred to as Arabidopsis)3–5 to determine the extent of sequence variation at these regulatory loci. In general, these papers show that mature miRNAs, the ∼21 nt portions of each miRNA that actually bind to target mRNAs to trigger their cleavage, are under strong constraint.

In our resequencing survey of 66 miRNAs from 16 miRNA families using 24 accessions, we identified only two SNPs in mature miRNAs, both of which are low frequency-derived mutations that are unlikely to abolish miRNA-mRNA pairing.3 This low level of polymorphism at miRNAs is significantly below the mean level of polymorphism observed across the genome and indicates a stronger evolutionary constraint at miRNAs than is observed at coding genes.3 All three studies also found that regions of the double-stranded precursor (pre)-miRNA that are important to maintaining the structural integrity of the pre-miRNA are constrained as well, albeit to a lesser degree than the mature miRNAs.3–5 Overall, these results suggest that most miRNAs are under intense negative selection, and are unlikely to contribute much to the large amount of gene expression and quantitative trait variation that exists within Arabidopsis.

Based on the recently published whole genome resequencing of 20 Arabidopsis accessions using microarray data,6 Zeller et al., also found that miRNAs that are not conserved outside of Arabidopsis have higher polymorphism than those that are.5 This latter result, along with an observed increase in flanking variation at nonconserved miRNAs, suggests that these miRNAs may actually segregate as functional or nonfunctional in Arabidopsis, although this finding could also be due to the misidentification of nonconserved miRNAs.5 The possibility of miRNA functionality segregating within species is compelling as it helps to connect the pattern of miRNAs seen within species to what is seen across species.

Indeed, although many miRNAs are conserved across plant species,7,8 a large number of plant miRNAs have also been shown to be species or clade-specific.7,9–12 The evolution of such nonconserved miRNAs could have reshaped gene networks during the history of the diversification of plants to promote the development of novel, derived characteristics. Mechanistically, it has been shown that miRNAs can evolve de novo from tandem gene duplicates that are inverted relative to each other and that such miRNAs can subsequently target their progenitors by sequence similarity.13,14 Gene duplication is a common occurrence in plants15 and, hence, novel miRNAs may have arisen at a regular rate within plant species during the evolutionary history of plants, with some fraction actually fixing. A recent paper in which massively parallel sequencing was used to identify miRNAs across several Drosophila species suggests that such a model for the birth and death of novel miRNAs could be realistic in other phyla as well.16

Another mechanism of miRNA evolution that could generate phenotypic diversity is the duplication and divergence of miRNAs. Many miRNA families that are conserved across multiple plant species exhibit either expansion or contraction of miRNA copy number across species (based on miRNA annotations in miRBase17,18), suggesting that these miRNA families must have undergone copy number gains and losses across plant evolutionary history. After such miRNA duplications, one of the duplicates may be free to neofunctionalize,19 a possibility supported by the extensive sequence differences that exist between miRNA family members outside of the mature miRNA. Little is known about the extent to which these sequence differences that exist between members of the same miRNA family reflect spatiotemporal functional differences. However, it is likely that in some cases spatiotemporal expression differences between such miRNA family members do exist. For instance, Warthmann et al. found that by using transgenic Arabidopsis lines overexpressing miR319 family members from across the Brassicaceae, functional differences could be detected between Brassica oleraceai's miR319a and miR319c.4

We have attempted to highlight exciting, recent research on miRNA evolution in plants. With the upcoming completion of more plant genomes, in particular Arabidopsis' relatives Arabidopsis lyrata and Capsella rubella, as well as the whole genome resequencing projects that are likely to come for some plant species, the resources to more fully explore miRNAs and their effects on plant phenotypic diversity will exist. Much remains to be determined about the role of miRNAs in the diversification of plants.