NEIGHBORING BASE COMPOSITION IS STRONGLY CORRELATED WITH BASE SUBSTITUTION BIAS IN A REGION OF THE CHLOROPLAST GENOME

Nucleotide sequence from a region of the chloroplast genome is present ed for 12 species spanning four subfamilies of the grass family. The r egion contains the coding sequence for the rbcL gene and the intergeni c spacer between the gene coding the large subunit of ribulose-1,5-bis phosphate carboxylase/oxygenase (rbcL) and the photosystem I gene psaI . This intergenic spacer contains a pseudogene for rpl23 as well as tw o noncoding segments with different A+T contents. Using the sequence o f rbcL a chloroplast phylogeny of this family was constructed by parsi mony. Variable sites of the two noncoding segments were traced onto th e phylogeny to study the dynamics of base substitution, This was also performed for the fourfold-degenerate sites of the rbcL sane, A wide v ariation in transversion/transition is observed between the two noncod ing segments and between the noncoding DNA and the fourfold-degenerate sites of rbcL. This variation is correlated with regional A+T content . As regional A+T content decreases, the ratio of transversions to tra nsitions also decreases, Substitutions were then scored in relation to neighboring base composition. The composition of the two bases immedi ately flanking each substitution is highly correlated with the transve rsion/transition bias. When both the 5' and 3' flanking bases are an A or a T, transversions are observed 2.2 times as frequently as transit ions. When either or both neighbors are a C or a G, the opposite trend is found; transitions are observed 1.5 times more frequently than tra nsversions.