3D-QSAR of anti-mitotic, tubulin binding analogs using comparative molecular field analysis (CoMFA)

The Vinca alkaloids vinblastine and vincristine, and the synthetic derivative vinorelbine, have had great success in the clinical treatment of various forms of cancer. Development of new and more effective Vinca congeners should therefore prove advantageous in cancer treatment. Part of a rational drug development process is developing and understanding the structure-activity relationships of the drug candidates. In Part I of this work we have derived a statistically significant three dimensional quantitative structure-activity relationship model using Comparative Molecular Field Analysis (CoMFA) that describes and predicts the cytotoxicities of a series of vinblastine derivatives. The correlation within the model was excellent (r2 = 0.983 for forty compounds). The predictive ability of this model was substantiated by traditional cross-validated partial least squares analysis (q2 = 0.681) and by the prediction of an 11 derivative external test set (r 2pred = 0.974). Notably, the CoMFA model demonstrated accurate, external prediction of the well-established relative potencies of vincristine, vinorelbine, and vinflunine, all of which are being used clinically, or are currently withstanding clinical trials. Accurate activity prediction of the 11 derivatives in the test set, which are comparatively quite deviant from those composing the training set, substantiates this model's usefulness as a tool in the design of novel Vinca derivatives. The CoMFA generated contour regions of the model were compared with the X-ray crystallographic structure of the vinblastine binding site on tubulin. Significant parallels between the chemical characteristics of the Vinca binding pocket and the steric and electrostatic features of the CoMFA fields were noted. These results indicate a fairly direct correlation between factors that affect the tubulin/Vinca interaction and factors that affect the ability of Vincas to initiate cell death in L1210 cell lines. Future, comparative analysis of CoMFA contour regions generated using a variety of cell lines could potentially aid in the design of Vinca derivatives specific to particular cancer types, as well as provide insight into Vinca resistance.; Colchicine, like vinblastine is an anti-mitotic, tubulin targeting agent. While colchicine is one of the oldest and most well studied of the microtubule binding drugs, its clinical applications are unfortunately limited due to toxicity. However, the unique vasculature-disrupting effects and initial clinical success of the colchicine-binding-site drug combretastatin A-4 has led to a recent revival of interest in colchicine-tubulin structure activity relationships. In Part II of this study we have optimized a colchicine/allocolchicine derivative CoMFA model derived using inhibition of microtubule assembly data. The post-optimization model demonstrates significantly improved predictability and statistical stability in comparison to the original model. In addition, good correspondence was found between the CoMFA generated contour regions of the optimized model and the X-ray crystallographic colchicine binding site of tubulin. Analysis of these contour regions in comparison to the binding site could potentially aid in the identification of those amino acids fundamental to activity, leading to better overall understanding of those ligand/protein interactions responsible for colchicine anti-mitotic behavior, and eventually the development of new and more effective anti-cancer agents.