Document Type
Dissertation
Date of Award
Fall 1-31-2006
Degree Name
Doctor of Philosophy in Chemistry - (Ph.D.)
Department
Chemistry and Environmental Science
First Advisor
Carol A. Venanzi
Second Advisor
Joseph W. Bozzelli
Third Advisor
Tamara M. Gund
Fourth Advisor
Michael Recce
Fifth Advisor
Edgardo Tabion Farinas
Abstract
As the three-dimensional structure of the dopamine transporter (DAT) remains undiscovered, any attempt to model the binding of drug-like ligands to this protein must necessarily include strategies that use ligand information. For flexible ligands that bind to the DAT, the identification of the binding conformation becomes an important but challenging task. In the first part of this work, the selection of a few representative structures as putative binding conformations from a large collection of conformations of a flexible GBR 12909 analogue was demonstrated by cluster analysis. Novel structurebased features that can be easily generalized to other molecules were developed and used for clustering. Since the feature space may or may not be Euclidean, a recently-developed fuzzy relational clustering algorithm capable of handling such data was used. Both superposition-dependent and superposition-independent features were used along with region-specific clustering that focused on separate pharmacophore elements in the molecule. Separate sets of representative structures were identified for the superpositiondependent and superposition-independent analyses.
In the second part of this work, several QSAR models were developed for a series of analogues of methylphenidate (MP), another potent dopamine reuptake inhibitor. In a novel method, the Electrotopological-state (B-state) indices for atoms of the scaffold common to all 80 compounds were used to develop an effective test set spanning both the structure space as well as the activity space. The utility of B-state indices in modeling a series of analogues with a common scaffold was demonstrated. Several models were developed using various combinations of 2-D and 3-D descriptors in the Molconn-Z and MOE descriptor sets. The models derived from CoMFA descriptors were found to be the most predictive and explanatory. Progressive scrambling of all models indicated several stable models. The best models were used to predict the activity of the test set analogues and were found to produce reasonable residuals. Substitutions in the phenyl ring of MP, especially at the 3- and 4-positions, were found to be the most important for DATbinding. It was predicted that for better DAT-binding the substituents at these positions should be relatively bulky, electron-rich atoms or groups.
Recommended Citation
Misra, Milind, "Ligand-based design of dopamine reuptake inhibitors : fuzzy relational clustering and 2-D and 3-D QSAR modleing" (2006). Dissertations. 751.
https://digitalcommons.njit.edu/dissertations/751
