During the last decade major research interests of the group were focused on the large-scale conformational properties of double-helical DNA. A typical DNA molecule adopts, with comparable probability, many different conformations in solution. As a result, large-scale properties of the double helix have to be analyzed in terms of probability of different conformations, and this distinguishes DNA from other biological macromolecules. Correspondingly, the questions that we ask should be also formulated in probability terms. What are typical conformations of large DNA molecules? What is the equilibrium probability and the rate of juxtaposition of two specific DNA sites separated along the DNA contour? How does DNA supercoiling affect these properties? What distributions of topological forms should we expect after converting linear molecules to the circular form? These and similar questions are important for many biological processes which involve DNA molecules. Study of DNA properties requires specific theoretical and experimental methods. Such methods are used and developed in my group. The work is based on tight complementation of theoretical and experimental approaches. Properties of topologically restricted molecules, supercoiled, knotted, and linked DNAs, attract our major attention since topological changes can be reliably detected experimentally and efficiently evaluated by computer. The study of the equilibrium fraction of catenanes formed between supercoiled DNA and linear molecules which are converting into circular form in the experiment, gives good example of the approach (click here to see a short description of this work).

We can simulate DNA conformational properties with high accuracy. Therefore, we can use the simulation as a tool to study proteins that interact with two or more DNA sites. The approach should be especially efficient for enzymes that change DNA topology, such as type II DNA topoisomerases and site-specific recombinases. Typically, because of the flexibility of substrate DNA molecules, these enzymes form distributions of different topological forms rather than DNA molecules with a uniqe topology. We can calculate such distributions of reaction products for specific models of an enzyme action. We can also measure the distributions formed by the actual enzymes and substrate DNA molecules. Comparing the simulated results with corresponding experimental data serves as a powerful instrument for testing and designing models of these enzyme action.

Our studies of type II DNA topoisomerases and site-specific recombinases unexpectedly showed that local DNA bends caused the protein binding, strongly affect topological distributions of the reaction products. Therefore, we needed an accurate solution method to measure DNA bends. Such a method, based on cyclization of short DNA fragments, has become a major experimental technique in the lab. The accuracy of the method makes it extremely useful in the studies of both DNA conformational properties and DNA-protein complexes.