Yoshinori Takeda

The structures of three proteins that regulate gene expression have been determined recently and suggest how these proteins may bind to their specific recognition sites on the DNA. One protein (Cro) is a repressor of gene expression, the second (CAP) usually stimulates gene expression, and the third (lambda repressor) can act as either a repressor or an activator. The three proteins contain a substructure consisting of two consecutive alpha helices that is virtually identical in each case. Structural and amino acid sequence comparisons suggest that this bihelical fold occurs in a number of proteins that regulate gene expression, and is an intrinsic part of the DNA-protein recognition event. The modes of repression and activation by Cro and lambda repressor are understood reasonably well, but the mode of action of CAP is still unclear.

We measured quantitatively the binding affinities of purified Cro repressor to the chemically synthesized wild-type and mutant OR1 operators, consisting of all three possible base-pair substitutions and of thymine to uracil substitutions at each base-pair position of the 17-base-pair operator sequence. The sequence-specific interactions between Cro repressor and the operator DNA occur at the symmetrically disposed outer 7-base-pair positions of each half operator and at the central base-pair position. The binding of Cro is almost symmetrical with respect to the pseudo-twofold symmetry of the binding site. The binding free energy changes calculated from the affinity changes are mostly additive for specific Cro binding. Also the binding affinities of Cro to the operators or any other DNA sequences can be predicted by simple addition of free energy changes of single base substitutions. We isolated cro mutants by site-directed mutagenesis and studied their DNA binding to the wild-type and base-substituted mutant operators. The sequence-specific contacts derived from such studies are significantly different from the models proposed by Ohlendorf et al. [Ohlendorf, D. H., Anderson, W. F., Takeda, Y. & Matthews, B. W. (1982) Nature (London) 298, 719-723] and by Hochschild et al. [Hochschild, A., Douhan, J., III, & Ptashne, M. (1986) Cell 47, 807-816].

Results of systematic base-substitution experiments suggest that the lambda repressor dimer, made of identical subunits, recognizes the "pseudo(2-fold)symmetric" operator sequence asymmetrically. Base substitutions within the consensus half of the operator affect binding more than base substitutions within the nonconsensus half of the operator. Furthermore, changing the nonconsensus base pairs to the consensus base pairs does not increase, but decreases, binding. Evidently, the two subunits of the lambda repressor dimer bind to the two halves of the operator differently. This is consistent with the recently determined crystal structure of the complex, which shows that the relative positioning of the amino acids to the DNA bases are slightly different in the two halves of the operator. The sequence-specific interactions indicated by the systematic base-substitution experiments correlate well with the locations of the specific contacts found in the complex. Thus, the amino acids of lambda repressor, mainly of alpha 3-helix and the N-terminus arm, seem to directly read-out the DNA sequence by forming specific hydrogen bonds and hydrophobic contacts to the DNA bases. The observed asymmetric recognition suggests that no recognition code governs amino acids and DNA bases in protein-DNA interactions.