Application of the GrowMol Computer Program to the Aspartic Acid Pepsin: Results of the Design and Synthesis of a Novel Inhibitor

Tuesday, April 8, 1997 - 3:30pm - 4:30pm
Keller 3-180
Regine Bohacek (ARIAD Pharmaceuticals Inc.)
The conceptual basis for structure-based drug design was formulated 100 years ago by Emil Fisher. His lock and key hypothesis is a constantly recurring theme in modern drug design. Previously, knowledge of the lock, i.e. the biochemical target, could only be inferred by the structures of a variety of keys (ligands) all of which fit the lock. However, now, thanks to the advances in protein expression, X-ray crystallography and NMR, more and more often we now actually have the 3-dimensional structure of the lock. And, thus, the new challenge is how to utilize this information to rapidly discover novel, potent molecules which will exquisitely fit the lock.

A number of computational methods are emerging that can use these 3-dimensional structures to design, de novo, molecular structures able to bind with high affinity to a target. We are developing a computer program called GrowMol, which generates organic structures that are both spatially and chemically complementary to the target binding site. By growing molecules an atom at a time to fill the various nooks and crannies of a binding site, GrowMol can generate structures with exquisite complementary to the host. At each step, the position and type of atom to be added are randomly selected using Boltzmann statistics to bias acceptance toward atoms that can form favorable interactions with the binding site. Previous work has demonstrated that GrowMol could generate known inhibitors of thermolysin as well as large number of novel, diverse structures complementary to the thermolysin binding site [1]. We will describe the results of applying GrowMol to aspartic acid pepsin and the discovery of a novel, low molecular weight inhibitor.

Another exciting area of drug design is that of combinatorial chemistry. If the structure of the target is known, then both the scaffold and the building blocks can be screened on the computer prior to synthesis. Knowledge of which of the typically thousands of available building blocks may form the most favorable interactions with the binding site, can improve the probability of finding hits! Examples of the use of GrowMol for designing libraries will be described.

This is joint work with Colin McMartin, Peter Glunz and Daniel H. Rich.


Bohacek, R. S., McMartin, C., SIAM J. Math Anal. 1995, 116, 147--179.