PEOPLE: CURRENT LAB MEMBERS

Carbonic Anhydrase

Peter Kutchukian
Graduate Student
Department of Chemistry and Chemical Biology
Harvard University
12 Oxford St.
Cambridge, MA 02138
Tel: (617) 384-7393   
Fax: (617) 384-9228
kutchuk@fas.harvard.edu
web site

Research Summary

The in silico design of drug candidates continues to challenge computational chemists. One current approach is to screen virtual libraries of compounds. This might entail judging compounds drug-like characteristics (for example, by invoking the Lipinski “rule of five” criterion), comparing them to a pharmacophore template derived from known ligands or from a crystal structure of the binding site, or docking the compounds into an active site and then evaluating their complementarity with a scoring function – such as a force field. These or similar methods are often used in sequence, weeding out poor candidates in the early stages, and spending more computational time (e.g., for docking) on interesting candidates. An alternative route is to grow molecules into the binding site of a target. This option merits consideration as it is not limited by the size or and structural limitations of a virtual database. A major goal of our research program is to develop an approach to computational drug design that addresses issues that are dealt with in the above methods, but with an efficient procedure solidly based in statistical mechanics.

Our Combinatorial Small Molecule Growth (CombiSMoG) package couples an accurate knowledge-based potential (SMoG2001) with a Monte Carlo (MC) small molecule growth algorithm, allowing one to grow small molecules into the binding pocket of a target protein (c). Since our scoring function is extremely fast, and our library of common organic fragments that are utilized in growth is sufficiently large (100), a rapid number of potential leads that cover a vast amount of chemical space can be assessed in a short period of time (~100,000/day on Octane UNIX workstation). The accuracy of our potential – which compares favorably with other current scoring functions (b) - then allows us to select the most promising candidates for synthesis and experimental evaluation. We were pleased to see that our method was validated during our first drug design effort when we discovered the best-known inhibitor (K d ~ 30 pM) of human carbonic anhydrase (d).

SH3 LP pocket

Encouraged by our early success, we have continued to improve our potential (a), and now plan to refine our growth algorithm. One shortcoming of the molecules that CombiSMoG produces is that their synthetic feasibility must be assessed by a synthetic chemist as a final “screening” process. That is, CombiSMoG lacks any chemical intuition. We plan to incorporate simple rules of organic synthesis into the growth algorithm, in order to grow molecules that are in the realm of synthetic space. We also aim to refine our growth algorithm so that the molecules it produces are biased towards “drug-like” or “natural product-like” compounds. This will be done by statistically analyzing how classes of fragments are connected in large databases of compounds. For example, how likely is a carbonyl group connected to an aromatic moiety. Using these connectivity probabilities, we will weight the likelihood that certain fragments will be connected to other fragments as molecules are grown, thus biasing our final compounds toward drug-like frameworks. In principle, we could also use this approach to grow ligands that resemble any class of compounds. For example, we could train our growth algorithm with a database of natural product compounds. We might even desire a more specific training set, such as a class of natural products such as polyketides. With such improvements to our growth algorithm, we will be able to bias the growth of synthetically feasible compounds that are similar in nature – if desired – to training set of compounds.

a) Dominy, B., Shakhnovich, E. J. Med. Chem.2004, 47, 4538-4558.

b) Ishchenko, A., Shakhnovich, E. J. Med. Chem.2002, 45, 2770-2780.

c) Grzybowski, B., Ishchenko, A., Shimada, J., Shakhnovich, E. Acc. Chem. Res.2002, 35, 261-269.

d) Grzybowski, B., Ishchenko, A., Kim, C., Topalov, G., Chapman, R., Christianson, D., Whitesides, G., Shakhnovich, E. Proc. Natl. Acad. Sci., 2002, 99, 1270-1273.