Rational structure-based design of anti-breast-cancer drugs targeting the ErbB family of receptor tyrosine kinases

ORRY, ANDREW J.; CAVASOTTO, CLAUDIO N.; ABAGYAN, RUBEN A.

Orlando, FL

Simposio; Era of Hope - DOD Breast Cancer Research Program; 2002

The ErbB2 tyrosine kinase domain (TKD) has been implicated in the growth and progression of the most aggressive forms of breast cancer. Breast cancer cells which produce large amounts of ErbB2 result in the excessive activation of multiple signaling cascades within the cell. To date, the crystal structure of ErbB2-TKD domain has yet to be determined. The purpose of this study was to characterize the structure of ErbB2 and identify potential new inhibitors to this domain by modeling, docking and virtual ligand screening (VLS).  Monoclonal antibodies have already demonstrated potent anti-tumor activity against ErbB2, but this approach gives rise to drug-delivery problems. We aimed to simplify the problem of drug-delivery by identifying orally bioavailable organic compounds that can permeate the cell and inhibit the catalytic activity of ErbB2 and hence arrest the progression of the disease. These compounds can also be tailored so they inhibit two or more closely related RTKs. We have undertaken a unique approach to build a molecular model of the ErbB2-TKD domain using homology modeling with multiple templates, ligand docking and side-chain refinements. The resulting models have been used as ErbB2 structures for VLS. The inhibitors bound to the ATP binding site in the crystal structure of the modeling templates were used as a docking validation tool. The ligands in each of the templates were ?re-docked? into the crystal structure and placed within the ligand binding pocket between 0.2Å and 0.8Å RMSD compared with the crystal structure of each complex. This high accuracy of ?re-docking? ligands into TKDs constitutes a validation of our docking methodology and is in perfect agreement with previous results. A comprehensive library of known ErbB2 and other tyrosine kinase inhibitors was constructed and docked into the ErbB2 models in order to establish a scoring threshold for VLS. A virtual ligand screening of the National Cancer Institute database (more than 127,000 compounds) was performed, and compounds were selected for further analysis. A database enrichment factor of 99% was recorded and is to our knowledge the highest ever-achieved using VLS. These VLS results indicates that our ErbB2 molecular model can provide a foundation for structure-based design of drugs and therefore overcoming the problem of our unsuccessful attempts to solve the crystal structure for this protein. This work demonstrates that accurate molecular modeling and virtual ligand screening can identify potential ?drug-like? inhibitors that can bind to the receptor and potentially facilitate anti-tumor activity in breast cells.