Frequent alternative splicing of human genes

Abstract

Alternative splicing can produce variant proteins and expression patterns as different as the products of different genes, yet the prevalence of alternative splicing has not been quantified. Here the spliced alignment algorithm was used to make a first inventory of exon-intron structures of known human genes using EST contigs from the TIGR Human Gene Index. The results on any one gene may be incomplete and will require verification, yet the overall trends are significant. Evidence of alternative splicing was shown in 35% of genes and the majority of splicing events occurred in 5' untranslated regions, suggesting wide occurrence of alternative regulation. Most of the alternative splices of coding regions generated additional protein domains rather than alternating domains.

Figure 1

Alternative acceptor sites of exon 3 of somatotropin and somatotropin variants. (V) Presomatotropin (V00520); (K) presomatotropin variant (K00470); (A) EST contig (THC207918); (B) EST contig (THC195752); (C) EST contig (THC195753). (<, >) annotated and observed sites (resp. donor and acceptor); (]) annotated but not observed acceptor sites; [)] observed but not annotated acceptor sites; (uppercase letters) mismatching nucleotides; (===) intron shadows.

Figure 2

Alternative splicing of pulmonary surfactant protein C gene. (J) Genomic sequence (J03890); (A, B, C, D) EST contigs (THC211006, THC211005, THC173453, and THC173454, respectively); (E) EST(N75529). (<, >) known sites (donor and accceptor, respectively); [(] new sites; other notation as in Fig. 1.