Identification of a DNA structural motif that includes the binding sites for Sp1, p53 and GA-binding protein

We have analyzed predicted helical twist angles in the 21-bp repeat region of the SV40 genome, using a semi-empirical model previously shown to accurately predict backbone conformations. Unexpectedly, the pattern of twist angles characteristic of the six GC-boxes is repeated an additional five times at positions that are regularly interspersed with the six GC-box sequences. These patterns of helical twist angles are associated with a second, imperfectly-repeated sequence motif, the TR-box 5'-RRNTRGG. Unrelated DNA sequences that interact with trans-acting factors (p53 and GABP) exhibit similar twist angle patterns, due to elements of the general form 5'-RRRYRRR that occur as interspersed arrays with a spacing of 10-11 bp and an offset of 4-6 bp. Arrays of these elements, which we call pyrimidine sandwich elements (PSEs), may play an important role in the interaction of trans-acting factors with DNA control regions. In 13 human proto-oncogenes analyzed, we identified 31 PSE arrays, 11 of which were in the 5'-flanking regions of the genes. The most extensive array was found in the promoter region of the K-ras gene. Extending over 80 bp of DNA, it contained 16 PSEs that showed an average deviation from the SV40 criterion pattern of angles of only 1.2 degrees.