'. . . The tyranny of the lattice . . .'

A systematic comparison of crystal structures of nine different B-DNA dodecamers, in three different space groups, with and without A-tracts, shows that crystal packing or lattice forces are of secondary importance for helix axis bending, minor-groove width, and propeller twist. While other local helix parameters may be influenced or even established by crystal packing, the properties just enumerated are determined primarily by base sequence. One and the same crystal packing scheme can accommodate a bend in one of two different directions, or no bend at all. A-tract regions of B-DNA are inherently straight and unbent, with base-pair inclination no different from that of general-sequence B-DNA. Where bends are observed at junctions between G·C and A·T regions, they always involve a roll about base-pair long axes in a direction that compresses the wide major groove and, hence, are 90° away from that necessary for the correctness of the junction model of A- tract bending. The G·C/A·T junction appears to be a flexible hinge, capable of adopting either a straight or a bent conformation under the local influence of weak crystal packing forces. Such forces therefore are a source of information about DNA deformability and not a curse to be deplored. But as an indication of the weakness of crystal packing forces, introduction of a single bromine atom in the major groove is sufficient to eliminate a bend, although brominated and unbrominated crystals are isomorphous.