Nonrandom patterns of simple and cryptic triplet repeats in coding and noncoding sequences

Abstract

Triplet repeats of the sequence purine, purine, and pyrimidine [RRY(i)] are frequent and often polymorphic in humans. Some RRY(i) are composed predominantly of a continuous repeat of one sequence [simple RRY(i)], but the majority are cryptic RRY(i) that are not obvious until the bases are classified into R or Y before the full extent of the repeat becomes apparent. RRY(i) can be divided into 18 classes based on predominant nucleotides. These classes are highly nonrandom in abundance and in location within genes. In humans, simple or cryptic RRY(i), in which AAT or AAC triplets predominate, are preferentially located 3' of Alu repeats. RRY(i) with a predominance of AGC or GGC show a dramatic enrichment in coding sequence, and GGC also shows a dramatic enrichment in 5' untranslated regions of genes. Characterization of RRY(i) present in coding regions identify 10 protein motifs (An, Dn, Hn, Pn, Qn, Tn, GnS0-3Gm, (G/S)n, (S/G/N)n, and (L/P)n). Six of the protein motifs appear predominantly in DNA-binding proteins/transcription factors. Alignment of homologous protein sequences from other mammals reveals that both simple and cryptic RRY(i) are a major source of deletions or insertions in the genes that contain them. Cryptic RRY(i) may be candidates for triplet repeat genetic diseases and, when mutated in somatic cells, may contribute to carcinogenesis.