The role of nucleotide sequences in splice site selection in eukaryotic pre-messenger RNA

Abstract

Alternative splicing of eukaryotic messenger RNA precursors is now known to be of widespread importance in generating multiple transcripts from a single gene. This phenomenon has emphasized the problem of the way in which splice sites are selected; recent studies have discussed the role of secondary structure¹ or affinity and spatial relationships² in this selection. Splice site sequences vary widely, although a loose consensus has been derived for the 9 bases around the 5′ splice site and for a longer region around the 3′ splice site³. Mutagenesis experiments have defined the sequences essential for a potential 5′ splice site^4,5,6, but, except for some experiments with the E1a gene of adenovirus^5,6, these experiments have not examined 5′ splice site sequences for features responsible for site preference where alternative splicing sites exist. Such tests require a choice of site: an appropriate reference site and a constant position at which test sites are introduced. We have begun a series of experiments designed to show whether splice site sequences can be ranked in a hierarchy of preferential use. Here we show that the archetypal consensus sequence is used efficiently, and characterize the cryptic sites of β-globin: sequences alone can explain why these sites are not normally used. We also show with the E1a gene of adenovirus, a simple example of alternative splicing, that one of the two 5′ splice sites used by this gene is intrinsically stronger. We also demonstrate that tandem repeats and secondary structure influence the choice of sites in vivo. We discuss the mechanism of splice site selection.