Energetics of cooperative protein-DNA interactions: comparison between quantitative deoxyribonuclease footprint...

Using the binding of cI repressor protein to the lambda right and left operators as a model system, we have analyzed the two common experimental techniques for studying the interactions of genome regulatory proteins with multiple, specific sites on DNA. These are the quantitative DNase footprint titration technique [Brenowitz, M., Senear, D. F., Shea, M. A., & Ackers, G. K. (1986) Methods Enzymol. 130, 132-181] and the nitrocellulose filter binding assay [Riggs, A., Suzuki, H., & Bourgeois, S. (1970) J. Mol. Biol. 48, 67-83]. The footprint titration technique provides binding curves that separately represent the fractional saturation for each site. In principle, such data contain the information necessary to determine the thermodynamic constants for local site binding and cooperativity. We show that in practice, this is not possible for all values of the constants in multisite systems, such as the lambda operators. We show how these constants can nevertheless be uniquely determined by using additional binding data from a small number of mutant operators in which the number of binding sites has been reduced. The filter binding technique does not distinguish binding to the individual sites and yields only macroscopic binding parameters which are composite averages of the various local site and cooperativity constants. Moreover, the resolution of even macroscopic constants from filter binding data for multisite systems requires ad hoc assumptions as to a relationship between the number of ligands bound and the filter retention of the complex. Our results indicate that no such relationship exists. Hence, the technique does not permit determination of thermodynamically valid interaction constants (even macroscopic) in multisite systems.