RNA sequences that conform to the consensus sequence of 5 splice sites but are not used for splicing occur frequently in protein coding genes. Mutational analyses have shown that suppression of splicing at such latent sites may be dictated by the necessity to maintain an open reading frame in the mRNA. Here we show that stop codon frequency in introns having latent 5 splice sites is significantly greater than that of introns lacking such sites and significantly greater than the expected occurrence by chance alone. Both observations suggest the occurrence of a general mechanism that recognizes the mRNA reading frame in the context of pre-mRNA.
Keywords: latent 5 splice sites; splice site selection; stop codons; suppression of splicing (SOS); translatabilityIn a recent paper, Zhang et al. (2003) presented a computational analysis that attempted to challenge the idea that the translatability of an exon can influence its splicing. This idea has been recently supported by a number of studies: Wang et al. (2002) and Mendell et al. (2002) demonstrated the occurrence of nonsense-associated alternative splicing in mutated genes, and a study from our laboratory demonstrated that splicing at intronic latent 5Ј splice sites is suppressed in wild-type genes when in-frame stop codons occur between the latent and the upstream normal (authentic) 5Ј splice site.Latent 5Ј splice sites are defined as sequences that conform to the canonical 5Ј splice site consensus sequence but, normally, are not used for splicing. Latent sites appear to be highly abundant in the genome, particularly in introns of protein-coding genes (see below). To explain the phenomenon of suppression of splicing (SOS) at latent 5Ј splice sites, we proposed that the necessity to maintain the translatability of mRNAs, by avoiding the inclusion of premature termination codons in them, could serve as a criterion that differentiates normal 5Ј splice sites from latent ones (Miriami et al. 1994). We substantiated this proposal by showing, in two gene systems, that an intronic latent 5Ј splice site can be activated if all upstream stop codons, which are in the reading frame of the upstream exon, are eliminated by point or frame-shift mutations . We also validated, by three criteria, the generality of the translatability hypothesis in a computerized genomic survey of a human database of 2206 introns, 1496 of which contain a total of 10,490 latent 5Ј splice sites and 710 of which are devoid of such sites. First, of the 1496 introns with latent 5Ј splice sites, 1359 (90.8%) have at least one in-frame stop codon upstream of the most 3Ј latent site. Second, in-frame stop codons occur upstream of 10,045 (95.8%) of the 10,490 latent 5Ј splice sites. Third, the density of in-frame stop codons in the 1496 introns with latent 5Ј splice sites is significantly higher than that in the 710 introns devoid of latent sites (0.0484 versus 0.0338 per effective number of codons; P < 0.001; Miriami et al. 2002). Zhang et al. (2003) challenged the translatability concept because th...