Introduction

The number of microRNAs appears to be ever-growing, as more intensive sequencing of small RNAs reveals a large population of sequences that are expressed at low abundance, or in a tissue- or stage-specific manner. Computational studies have also predicted the existence of thousands of candidate microRNA precursor hairpin structures throughout mammalian genomes (reviewed in Bentwich, 2005). The number of predicted potential targets per microRNA is also steadily increasing, with the recognition that binding of a 7-mer seed at the 5′-end of a microRNA may be sufficient to regulate a target mRNA functionally (Doench and Sharp, 2004; Farh et al., 2005; Lim et al., 2005; Stark et al., 2005; Sood et al., 2006). But how are microRNAs and their targets coordinated – if at all?

A random model

One recent paper proposes that microRNAs arise whenever a RNA hairpin structure happens to be transcribed, that happens to be competent for processing by Drosha and Dicer (Svoboda and Cara, 2006). Most of these microRNAs will have no function at all, at least not initially: They will bind to a relatively large number of putative target regions at random (a 7-mer sequence will bind randomly every 47 = 16 384 bases on average), and those target regions that happen to be associated with a useful phenotypic response will tend to be retained over evolutionary time whereas those mRNAs that show deleterious responses will become relatively depleted in target sequences (Svoboda and Cara, 2006).