To identify the major determinants of the DNA binding specificity of nuclear transcription factors,
the Comparative Binding Energy (COMBINE) analysis has been performed for two datasets.
In Tomic et al.,1 COMBINE QSAR models were derived for a set of 320 complexes of DNA and
glucocorticoid receptor mutants. Here, we derive COMBINE QSAR models for a set of 32 complexes.
This set differs from the larger one in two aspects. The complexes have additional mutation sites
in the DNA binding domain and, instead of just activity measurements, both activity and binding
affinity measurements are available. Models of better predictive ability were obtained with the
smaller, but experimentally better characterized, dataset. The parameters important for determining
binding specificity are nevertheless similar for both datasets: the electrostatic interaction energies
between the mutated nucleotides and mutated residue(s) as well as some charged amino acid residues
(Arg-447, Arg-470, Arg-477), and the solvation free energies of the mutated base(s). However, the
relative importance of these parameters is different in the two datasets.
Key words:
Quantitative Structure-Activity Relationship (QSAR), COMBINE analysis, molecular modeling,
gene regulation, molecular mechanics.
References: 1. S. Tomic, L. Nilsson, and R. C. Wade, J. Med. Chem. 43 (2000) 1780-1792.
CROATICA CHEMICA ACTA CCACAA 74 (2) 295-314 (2001)
ISSN-0011-1643 CCA-2733 Original Scientific Paper