Faculty / Research

Work in our laboratory is centered around the general field of structural biology and bioinformatics. We use computer-assisted molecular modeling and gene sequence analyses. Part of the research is done in collaboration with the Medical Informatics Department where Dr. Eisenberg is the Director. Various projects are currently under way:

The three-dimensional structures of DNA containing specific chemical modifications and of DNA/protein complexes: We have determined the three-dimensional structures of several short oligodeoxynucleotide sequences, 9 to 13 base pairs, each containing a single chemical modification known to be mutagenic. We continue these studies by elucidating the structures of new chemical lesions, as well as previously studied ones in the context of new base sequences. We conduct computer simulations based on molecular mechanics and dynamics algorithms. In addition to energy terms stemming from theoretical physics and chemistry, we include empirical energy terms derived from experimental data obtained by NMR by Dr. de los Santos. This methodology known as NMR restrained molecular dynamics is also used to determine the three-dimensional structure of a mutM-protein/DNA complex. This protein is a glycosylase that specifically removes certain DNA lesions. By comparing sequences from the vast array of genomes with the few glycosylases whose structures are known, we are attempting to determine common structural motifs of DNA repair enzyme families.

The role of hydrogen bonds in the specific recognition of biomolecules: It has long been suggested that the patterns of hydrogen bond donors and acceptors on the complementary surfaces of biological molecules, at the site where they bind to each other specifically, play an important role in the determination of this binding specificity. We have conducted computer-model studies on the existing database of DNA/protein complexes in an attempt to test this hypothesis. We have shown that the DNA base sequence that represents the specific binding place for each of several proteins always ranks as the energetically most favorable of all possible base sequence combinations, when only the energies of hydrogen bonds are taken into consideration. This is also the case when water molecules are included as hydrogen bond bridges. Our results suggest that using the principle of complementarily of hydrogen bond patterns can greatly simplify the complex task of finding appropriate docking conformations between molecules known to bind, but whose exact binding sites are not known.