Development of Nicotinic Acetylcholine Receptor nAChR α7 Binding Activity Prediction Model

Abstract

Despite the well-known adverse health effects associated with tobacco use, addiction to nicotine found in tobacco products causes difficulty in quitting among users. Nicotinic acetylcholine receptors (nAChRs) are the physiological targets of nicotine and facilitate addiction to tobacco products. The nAChR α7 receptor subtype plays an important role in addiction; therefore, predicting the binding activity of tobacco constituents to nAChR α7 would contribute to methods assessing the addictive potential of tobacco constituents. Among the receptor-ligand complexes obtained from the Protein Data Bank (PDB), we found that there were two distinct clusters of binding-pocket conformations for nAChR α7. One template was selected from each cluster to construct the 3D structures of human nAChR α7 using homology modeling. The competitive docking model was developed for determination of the structure of nAChR α7 bound with a chemical. To develop a prediction model of human α7 binding activity, we used a training data set of extracted 930 chemicals assessed for human α7 binding from the two databases, PubChem and ChEMBL. We also extracted 1448 chemicals evaluated for rat α7 binding from the same databases as the external testing data set. Based on the competitive docking results, the winning docking scores were partitioned to identify the key residues that play important roles in the receptor-ligand binding for each chemical. Our previously published algorithm, Decision Forest (DF), which combines multiple Decision Tree models was used to train the human α7 binding activity prediction model based on the partitioned docking scores. Five-fold cross validations were conducted to estimate the performance of the DF models. The DF model was tested using the external testing data set and then was used to predict the potential human α7 binding activity for the 5275 tobacco constituents of unknown activity. The human α7 binding activity data for 84 of the 5275 tobacco constituents were experimentally measured to confirm the prediction results. The overall prediction accuracy, sensitivity and specificity were 77.4%, 57.7% and 86.2%, respectively, demonstrating that the developed predictive model of human α7 binding activity could be a useful tool for high-throughput screening of potential addictive tobacco constituents. Disclaimer: The findings and conclusions in this abstract are not a formal dissemination of information by the US Food and Drug Administration (FDA) and do not represent Agency position or policy.