PROTEIN-DNA RECOGNITION

Several general principles emerge from the studies of Cro, lambda repressor, and CAP. The DNA-binding sites are recognized in a form similar to B-DNA. They do not form cruciforms or other novel DNA structures. There seem to be proteins that bind left-handed Z-DNA (87) and DNA in other conformations, but it remains to be seen how these structures are recognized or how proteins recognize specific sequences in single-stranded DNA. Cro, repressor, and CAP use symmetrically related subunits to interact with two-fold related sites in the operator sequences. Many other DNA-binding proteins are dimers or tetramers and their operator sequences have approximate two-fold symmetry. It seems likely that these proteins will, like Cro, repressor, and CAP, form symmetric complexes. However, there is no requirement for symmetry in protein-DNA interactions. Some sequence-specific DNA-binding proteins, like RNA polymerase, do not have symmetrically related subunits and do not bind to symmetric recognition sequences. Cro, repressor, and CAP use alpha-helices for many of the contacts between side chains and bases in the major groove. An adjacent alpha-helical region contacts the DNA backbone and may help to orient the "recognition" helices. This use of alpha-helical regions for DNA binding appears to be a common mode of recognition. Most of the contacts made by Cro, repressor, and CAP occur on one side of the double helix. However, lambda repressor contacts both sides of the double helix by using a flexible region of protein to wrap around the DNA. Recognition of specific base sequences involves hydrogen bonds and van der Waals interactions between side chains and the edges of base pairs. These specific interactions, together with backbone interactions and electrostatic interactions, stabilize the protein-DNA complexes. The current models for the complexes of Cro, repressor, and CAP with operator DNA are probably fundamentally correct, but it should be emphasized that model building alone, even when coupled with genetic and biochemical studies, cannot be expected to provide a completely reliable "high-resolution" view of the protein-DNA complex. For example, the use of standard B-DNA geometry for the operator is clearly an approximation.(ABSTRACT TRUNCATED AT 400 WORDS)