TOWARDS A CELLULAR CODE
Arianne J. Matlin‡, Francis Clark* and Christopher W. J. Smith‡
Abstract | In violation of the ‘one gene, one polypeptide’ rule, alternative splicing allows individual
genes to produce multiple protein isoforms — thereby playing a central part in generating
complex proteomes. Alternative splicing also has a largely hidden functionin quantitative
gene control, by targeting RNAs for nonsense-mediated decay. Traditional gene-by-gene
investigations of alternative splicing mechanisms are now being complemented by global
approaches. These promise to reveal details of the nature and operation of cellular codes that
are constituted by combinations of regulatory elements in pre-mRNA substrates and by cellular
complements ofsplicing regulators, which together determine regulated splicing pathways.
Pre-mRNA splicing is necessitated by the split nature of eukaryotic genes, in which the exons that will make up the mRNA product are interrupted by non-coding introns in the DNA and in the initial pre-mRNA transcript. Intron removal, and the concomitant joining of exons, is orchestrated by the spliceosome — amacromolecular ribonucleoprotein complex that assembles on the pre-mRNA in a series of complexes (E, A, B and C)1,2 BOX 1. Complex assembly is guided by consensus sequences at the ends of the introns. The importance of accurate splicing is illustrated by the fact that at least 15%, and perhaps as many as 50%, of human genetic diseases arise from mutations either in consensus splice site sequences or in themore variable auxiliary elements known as exon and intron splicing ENHANCERS (ESEs and ISEs, respectively) and SILENCERS (ESSs and ISSs, respectively)3–6. These auxiliary
elements are involved in defining both constitutive and alternative exons (FIG. 1A). Alternative splicing, in which different combinations of splice sites can be joined to each other (FIG. 1B), has assumed a high profilerecently, owing to the dual realization that there are fewer human genes than originally anticipated, and that alternative splicing is more the rule than the exception. Analyses of expressed sequence tag (EST) and cDNA datasets conservatively estimated that ~40–60% of human genes are alternatively spliced (reviewed in REF. 7), and this number increased to 73% when alternative splicing microarray data werecombined with ESTs8. This has been independently corroborated by ‘GENOME TILING’ MICROARRAYS across chromosomes 21 and 22, which indicated that alternative splicing occurs in >80% of genes9. Global comparisons of human and mouse alternative splicing reveal both conserved and species-specific events, with the balance between the two classes depending on the methodology used10–13. Conservedalternative exons are often flanked by more conserved intronic sequences than constitutive exons, and this has been used as an independent predictor of alternative splicing14–16. Most alternative splicing events affect the coding sequence, with half of these altering the reading frame17 and a third apparently leading to NONSENSEMEDIATED DECAY (NMD) of the RNA product18. Alternative splicing ofuntranslated regions can also have important regulatory consequences, including NMD, even though the open reading frame is unchanged. The Drosophila melanogaster Dscam gene exemplifies the extreme structural diversity that is achievable by alternative splicing. Dscam is a cell surface protein that is involved in axon guidance in the developing brain, and exists as up to 38,016 alternatively splicedisoforms19. Pairwise interactions between Dscam variants show a preference for homophilic interactions only between identical isoforms. This ability to distinguish self from non-self could be crucial in correctly establishing neuronal connections20 At the other end of the complexity spectrum is the alternative splicing of the Sex-lethal (Sxl) and transformer (tra) genes in D. melanogaster. In each...