Analysis of ESTs for alternatively spliced 3′ and 5′ splice sites reveals optimal exon size in vertebrates

Abstract

In vertebrates most exons are small surrounded by much larger introns. Due to this architecture, splice sites are initially recognized across the exon in what is known as exon definition. Here we examine the relative usage of 3,000 pairs of alternative 3′ splice sites and 2,000 pairs of alternative 5′ splice sites in the human genome. We find that within an optimal exon length window >50 <250 nts, there is a positive correlation between splice site strength and splice site usage. However, when one splice site is located inside of this exon length and the other is not, the splice site inside of the optimal exon length is preferred regardless of splice site strength. Our analysis reveals that splice sites within an optimal exon size are preferentially used when compared to splice sites outside of this optimal size. We conclude that a constraint of exon size is a fundamental parameter that dictates splice site usage. Because the spliceosome eventually forms across the intron, it is believed that intron definition is the predominant mode of splice site recognition for introns smaller than 250 nucleotides in length. However, in a subset of our alternative splice sites, in which one splice site is exon defined and the other is intron defined, the exon defined splice site is preferentially used, suggesting that exon recognition is the default mode of splice site recognition in vertebrates.NIH