A simple model for gene targeting.

Sequence-specific binding to genomic-size DNA sequences by artificial agents is of major interest for the development of gene-targeting strategies, gene-diagnostic applications, and biotechnical tools. The binding of one such agent, peptide nucleic acid (PNA), to a randomized human genome has been modeled with statistical mass action calculations. With the length of the PNA probe, the average per-base binding constant k(0), and the binding affinity loss of a mismatched base pair as main parameters, the specificity was gauged as a "therapeutic ratio" G = maximum safe [PNA](tot)/minimal efficient [PNA](tot). This general, though simple, model suggests that, above a certain threshold length of the PNA, the microscopic binding constant k(0) is the primary determinant for optimal discrimination, and that only a narrow range of rather low k(0) values gives a high therapeutic ratio G. For diagnostic purposes, the value of k(0) could readily be modulated by changing the temperature, due to the substantial Delta H degrees associated with the binding equilibrium. Applied to gene therapy, our results stress the need for appropriate control of the binding constant and added amount of the gene-targeting agent, to meet the varying conditions (ionic strength, presence of competing DNA-binding molecules) found in the cell.