“…These attempts all invoke the independent dinucleotide step model, wherein the secondary structure and torsional and bending rigidities associated with a given base-pair step are assumed to be completely independent of the sequence of its flanking DNA, or of any altered state of its flanking DNA that is induced by protein or other ligand binding, or by a B-to-Z transition. However, this assumption has been unequivocally contradicted by numerous published NMR structures of small duplexes, which yield different structural parameters for particular steps, e.g., A-A steps, embedded in different flanking sequences (46), as well as by numerous other published experiments (6,36,(47)(48)(49)(50)(51)(52) and unpublished studies of S. A. Winkle (Florida International University, personal communication, 1997) that were reviewed previously (36). Other evidence indicates that dinucleotide step models for directional permanent bends, or for A d , cannot account satisfactorily for the behavior of all sequences (43,53,54).…”