The world of small RNAs just got bigger. In Caenorhabditis elegans, two small temporal RNAs produced by the ribonuclease Dicer have previously been shown to be involved in regulating developmental timing. These RNAs — lin-4 and let-7, 22 and 21 nucleotides in length, respectively — act as antisense repressors of messenger RNA translation and, until recently, they were the only known RNAs of this class. But three papers published in Science now show that lin-4 and let-7 probably belong to a large family of single-stranded RNAs, 20 - 24 nucleotides in length, called microRNAs (miRNAs). These results indicate that post-transcriptional regulation by small RNAs is more common than previously believed.

Lagos-Quintana and colleagues used complementary DNA libraries constructed from a size-fractionated RNA population to identify 14 new miRNAs in Drosophila melanogaster and 19 new miRNAs in humans. Lau et al. created a cDNA library enriched for Dicer products, distinguished from other oligonucleotides by their small size, 5'-monophosphate group and 3'-hydroxyl group, to identify 54 novel miRNAs in C. elegans. Finally, using size-selected cDNA cloning together with computational methods, Lee and Ambros identified 15 miRNAs in C. elegans, 11 of which matched those identified by Lau and co-workers. In all cases, they showed the miRNAs were not simply the breakdown products of mRNAs or structural RNAs.

These papers identified 91 different miRNAs in total, about 12% of which have been conserved through evolution. Moreover, Lau and colleagues found that ∼85% of the miRNAs identified in C. elegans had homologues in the 90%-complete C. briggsae genome sequence.

miRNAs are produced through processing, probably by Dicer, of a ∼70-nucleotide precursor stem-loop structure. Either the 5' or the 3' arm of the precursor can be released to form the miRNA, with one exception. miR-56, identified by Lau et al., exists in two forms, resulting from processing of both the 5' and 3' arms of the precursor stem. How miRNA excision occurs has yet to be defined.

The mir genes often cluster together in the genome; for example, Lagos-Quintana and colleagues showed that mir-3, -4, -5, and -6 form a gene cluster in the Drosophila genome. The mir gene clusters investigated so far are co-expressed, and Lau and co-workers predicted that, in C. elegans, the gene cluster mir-35 to -41 is transcribed to form a single RNA precursor, which is processed to produce miR-35 to -41. Some mir genes have multiple genomic copies, and some miRNAs are highly homologous.

All three groups investigated the expression of miRNAs and found that, in some cases, it was both stage- and tissue-specific. For example, Lee and Ambros found that mir-1 is expressed stage-specifically in mouse embryogenesis and tissue-specifically in the human heart. These regulated expression patterns indicate an involvement in developmental control.

miRNAs have been proposed to function as "riboregulators", regulating gene expression by binding sequence-specifically to mRNAs, thereby blocking translation. The challenge now is to define the potential targets of miRNAs and their exact functions. There are probably many miRNAs yet to be identified and, if they are found to be as numerous and diverse as the miRNAs identified in these papers, they could have a range of regulatory functions. These authors seem to have discovered a small fortune, and the world of small RNAs could turn out to be very big indeed.

Story by Rachel Smallridge, Nature Reviews Molecular Cell Biology