Deviations from Chargaff's second parity role correlate with direction of transcription

The distribution of deviations from Chargaff's second parity rule was exami ned for overlapping sequence windows of a length (1 kb) predicted to be sui table for detecting correlations with functional features of DNA. For long genomic segments from E. coil, Saccharomyces cerevisiae, and Vaccinia virus , Chargaff differences for the W bases and/or for the S bases correlate wit h transcription direction and gene location. For W-rich genomes, the mRNA-s ynonymous strand contains regions which, if extruded from negatively superc oiled DNA, would fold to generate stem-loop structures with A-rich loops. S imilarly, for S-rich genomes the loops would be G-rich. We suggest that the disposition of genes in nucleic acid sequences arises from their having to adapt to a preexisting mosaic of genomic regions, each distinguished by it s potential to extrude single-strand loops enriched for a particular base ( or two non-Watson-Crick pairing bases). The mosaic would have facilitated t he intrastrand and interstrand accounting required for correction of mutati ons, and would have evolved in the early RNA world before the emergence of protein-encoding capacity. The preexisting mosaic would have determined tra nscription direction since there is pressure for all mRNAs of a cell to hav e purine-rich loops, thus decreasing loop-loop interactions which might lea d to formation of ''self" sense-antisense RNA duplexes. (C) 1999 Academic P ress.