Purifying selection against spurious splicing signals contributes to the base composition evolution of the polypyrimidine tract

Abstract

Among eukaryotes, the major spliceosomal pathway is highly conserved. Long introns may contain additional regulatory sequences, but those in short introns seem to be nearly exclusively related to splicing. While regulatory sequences involved in splicing are well-characterized, little is known about their evolution. At the 3' end of introns, the splice signal nearly universally contains the dimer AG, which consists of purines. The polypyrimidine tract upstream of the 3' splice signal is characterized by over-representation of the pyrimidine bases cytosine and thymine. We hypothesize that the over-representation of pyrimidines in the polypyrimidine tract is caused by avoidance of a premature splicing signal- AG should be most under-represented dimer. DNA-strand asymmetry patterns in fruit flies of the genus Drosophila confirm this prediction. In short introns of Drosophila, a region between the 5' splice signal and the branch point, which at least contains the nucleotides 8-30, is considered to evolve neutrally. Comparing this presumably neutrally evolving region to the polypyrimidine tract, we infer a moderate scaled selection strength below four against 3' splicing signals. Patterns of asymmetry in other eukaryotes indicate that avoidance of premature splicing similarly affects the nucleotide composition in their polypyrimidine tracts.