Ligand docking and scoring in DNA oligonucleotides. Binding of doxorubicinand modeled analogs to optimize sequence specificity

New approaches to optimizing ligand docking and scoring in a complex 2-dime nsional problem (varying ligand and "receptor" sequence) are described. We have explored, in detail, the sequence specificity for the potent anticance r drug doxorubicin in 64 DNA quartet base-pair sequences. The molecular mod eling program HINT was used to calculate quantitative binding scores that c an be simply related to free energy. The optimum scoring sequence, d(CAATAT TG)(2), was then docked with 24 doxorubicin analogs. These CAAT models were , in turn, "mutated" to CAAG, CGAT and CGAG and re-optimized. Close examina tion of this series of models suggests a rational approach to identifying t he sequence specificity of doxorubicin analogs and may lead to design of ne w agents.