AKT1 (protein kinase B) is the cellular homolog of the v-akt oncogene, and represents one of three isoforms of a multigene protein-serine/threonine kinase family. Gene amplification and/or overexpression of one or more Akt isoforms have been noted in cancers of the stomach, brain, breast, prostate, ovary and pancreas. High AKT activity is particularly evident in tumors with mutations or deletion of the tumor suppressor, PTEN, which functions primarily as a phosphoinositide 3-phosphatase. Expression of wild-type PTEN in human tumor cell lines or an AKT2 antisense cDNA in a human pancreatic carc/noma xenograft dramatically reduced cell and tumor growth, respectively, suggesting the utility of AKT as a therapeutic target. The broad objective of this application, therefore, is to develop isoform-specific inhibitors of AKT through the use of molecular modeling and virtual screening of compound libraries coupled with a unique screening assay based on homodimerization of the/soform-specific N-terminal donaain (AH domain) of AKT. Lead drugs identified in this screen will be assessed for their ability to inhibit tumor cell growth, induce apoptosis and block AKT activation via transphosphorylation. This application will address the hypothesis that interruption of AKT dimerization will inhibit its activity and downstream effector pathways leading to inhibition of tumor cell proliferation and activation of apoptosis. This hypothesis will be addressed by the following Specific Aims: 1) To refine our AKT1 model and to use it in construction of the homodimer resulting from interaction of the AH domains, 2) To identify structural regions in the AKT AH domain located at the interface of the AKT homodimer and to use virtual screening of compound libraries (NCI, ACD) to identify small molecules capable of interacting with these dimerization regions, and 3) To test those compounds showing the best scoring function for binding to the AH domain for their ability to inhibit AKT dimerization, and thus, activation of AKT in vitro. Based upon the latter results, structure-activity relationship (SAR) data will be generated by chemical modificationm of the lead molecules in order to improve compound activity. Modeling of active structures to the AKT1 AH domain binding pocket will be used in an iterative manner to guide compound synthesis as well as to suggest possible compound libraries to be created using combinatorial chemistry methods.
