In response to letter from Benoit Chabot Sir, The search for new spliceosomal introns has been a frustrating endeavor. The first (near) genome-wide search for intron gains, by one of us and others, identified over one hundred promising candidates, all of which proved on closer inspection to be due to a host of annotation and gene family complications. (1) Subsequent promising published examples have been similarly problematic, reflecting lack of close outgroups, annotational problems or unrecognized paralogy. (2)(3)(4) Recently, Zhuo et al. (5) reported a wealth of predicted human introns with similar 5 0 and 3 0 splice sites. These introns are largely primate-specific, in stark contrast to the lack of lineage-specific mammalian introns in previous studies. (1,6) This general pattern is compatible with either de novo intron creation by genomic duplication (as they argued) or with artifactual deletion by template switching (TS) by the reverse transcriptase (as we argued). ( 7) Chabot et al. (8) now provide three arguments against our interpretation.
The authors first argue that stem-loop structures in the mRNA template are important for TS (consistent with two previously studied examples) (9,10) and thus TS could not explain their results, since it is unlikely that all their predicted introns have associated secondary structures. While stem-