The long-term goal of this project is the development of a geometrical formalism that allows us to design protein backbone geometries from first principles. The main concept is to develop a set of building blocks (geometrical shapes) that can be combined to construct protein scaffolds de novo. Once such a scaffold is built, we can then use algorithms developed previously by us to predict sequences that should stabilize the proposed backbone structure. Ultimately such sequences will be tested experimentally. During the last year there have been three major activities:

1. We have developed algorithms that permit the prediction of sequences involving a large number of amino acids. These algorithms simultaneously optimize protein sequence, amino acid side-chain structure,and placement of ligands by deterministically identifying the global minimum of a semi-empirical potential function.

2. These algorithms have been tested experimentally. We have been able to demonstrate that the methods are sufficiently powerful to predict drastic changes in the ligand-binding specificity of receptors by changing the specificity of five sugar- and amino acid-binding proteins to bind ligands as trinitroluene, L-lactate, and serotonin.

3. We have developed a geometrical method that allows us to treat electrostatic effects in proteins with a reasonable degree of accuracy. The formalism of the treatment is appropriate for incorporation into the protein design calculations.

We are now engaged in exploring the ramifications of these results by experimentally validating a relatively large number of computational designs, and we are developing new approaches for introducing backbone movements into the design methodology.