A fibrous form of poly d(A):poly d(T) has a heteronomous secondary structure which is the first to be confirmed for a polynucleotide duplex: although both chains are 10(1) helices, mutually hydrogen-bonded in the standard (Watson-Crick) fashion, each has a quite different conformation. One chain -- probably poly d(A) -- has C3'-endo-puckered furanose rings characteristic of the A family of polynucleotide secondary structures while the other -- probably poly d(T) -- has the C2'-endo-puckered rings of the B family. Since analogous heteronomous structures could be assumed by DNA-DNA or DNA-RNA duplexes containing more general base sequences the polymorphic range of polynucleotide double-helices may be even greater than we have come to suppose.

The B form of poly d(GC):poly d(GC) in orthorhombic microcrystallites in oriented fibers has a secondary structure in which a dinucleotide is the repeated motif rather than a mononucleotide as in standard, smooth B DNA. One set of nucleotides (probably GpC) has the same conformations as the smooth form but the alternate (CpG) nucleotides have a different conformation at C3'-O3'. This leads to a distinctive change in the orientation of the phosphate groups. Similar perturbations can be detected in other poly d(PuPy):poly d(PuPy) DNAs such as poly d(IC):poly d(IC) and poly d(AT):poly d(AT) in their D forms which have tetragonal crystal environments. This suggests that such perturbations are intrinsic to all stretches of duplex DNA where purines and pyrimidines alternate and may play a role in the detection and exploitation of such sequences by regulatory proteins.