Proline residues are known to perturb the structure of helices by introducing a kink between the segments preceding and following the proline residue. The distortion of the helical structure results from the avoided steric clash between the ring of the proline at position (i) and the backbone carbonyl at position (i - 4), as well as the elimination of helix backbone H-bonds for the carbonyls at positions (i - 3) and (i - 4). Both the departure from the ideal helical pattern and the reduction in H-bond stabilization contribute to the observed flexibility of a proline-containing alpha-helix. The special local flexibility of the proline kink can confer an important role on the proline-containing helix in the conformational changes related to the function of the protein. As a useful tool in determining and evaluating the role of proline-induced flexibility and distortions in protein function, we present here a protocol to quantify the geometry of the distortion introduced in helices by prolines both as a time-averaged value and for individual 'snapshots' along a molecular dynamics simulation.

A computational method has been developed to predict inhibitor binding energy for untested inhibitor molecules. A neural network is trained from the electrostatic potential surfaces of known inhibitors and their binding energies. The algorithm is then able to predict, with high accuracy, the binding energy of unknown inhibitors. IU-nucleoside hydrolase from Crithidia fasciculata and the inhibitor molecules described previously [Miles, R. W. Tyler, P. C. Evans, G. Furneaux R. H., Parkin, D. W., and Schramm, V. L. (1999) Biochemistry 38, xxxx-xxxx] are used as the test system. Discrete points on the molecular electrostatic potential surface of inhibitor molecules are input to neural networks to identify the quantum mechanical features that contribute to binding. Feed-forward neural networks with back-propagation of error are trained to recognize the quantum mechanical electrostatic potential and geometry at the entire van der Waals surface of a group of training molecules and to predict the strength of interactions between the enzyme and novel inhibitors. The binding energies of unknown inhibitors were predicted, followed by experimental determination of K(i)() values. Predictions of K(i)() values using this theory are compared to other methods and are more robust in estimating inhibitory strength. The average deviation in estimating K(i)() values for 18 unknown inhibitor molecules, with 21 training molecules, is a factor of 5 x K(i)() over a range of 660 000 in K(i)() values for all molecules. The a posteriori accuracy of the predictions suggests the method will be effective as a guide for experimental inhibitor design.

mA formalism is presented for quantifying the similarity between any two molecules. The chemical descriptor used for comparison is the molecular electrostatic potential at the van der Waals surface. Thus, both the spatial properties of a molecule and its chemical features are captured in this approach. For molecules that are geometrically alike, the most useful similarity measure stems from orienting the two species so that their physical surfaces are aligned as well as possible, without regard to chemical patterns. After this alignment is achieved, a single measure sensitive to the spatial distribution of the electrostatic potential is used to rank the electronic similarity. Molecular similarity measures are applied to the enzyme systems AMP deaminase and AMP nucleosidase in order to understand quantitatively why their respective transition-state inhibitors bind more tightly than do their substrates. 0 1996 John Wiley & Sons, Inc.structure is the species along the reaction coordinate most tightly bound to the enzyme. The observed rate enhancement over the analogous uncatalyzed reaction for the conversion of substrate to product reflects that the transition state is typically bound 1010-1015 times more tightly to the

Quantum mechanical molecular electrostatic potential surfaces and neural networks are combined to predict the binding energy for bioactive molecules with enzyme targets. Computational neural networks are employed to identify the quantum mechanical features of inhibitory molecules that contribute to binding. This approach generates relationships between the quantum mechanical structure of inhibitory molecules and the strength of binding. Feed-forward neural networks with back-propagation of error are trained to recognize the quantum mechanical electrostatic potential at the entire van der Waals surface of a group of training molecules and to predict the strength of interactions between the enzyme and novel inhibitors. Three enzyme systems are used as examples in this work: AMP (adenosine mono phosphate) nucleosidase, adenosine deaminase, and cytidine deaminase. Quantum neural networks identify critical areas on inhibitor potential surfaces involved in binding and predict with quantitative accuracy the binding strength of new inhibitors. The method is able to predict the binding free energy of the transition state, when trained with less tightly bound inhibitors. The application of this approach to the study of enzyme inhibitors and receptor agonists would permit evaluation of chemical libraries of potential bioactive agents.

No abstract