Developing Target Identification Platforms Using Profiled Kinase Inhibitors

Abstract

Kinases are important enzymes in cellular signaling with their expression and activity tightly regulated. Dysregulated kinase activity can lead to numerous disease states such as cancer. Inhibiting aberrant kinase activity can slow cancer cell growth or cause cancer cell death. Thus, kinase inhibitors are well-validated drugs for cancer treatment. To date, nearly all kinase inhibitors approved for cancer treatment have been discovered using hypothesis driven target-based approaches. This is in sharp contrast to other cancer drug classes which have recently seen an increase in approvals and new chemical entities whose leads were discovered through phenotypic-based approaches. Phenotypic screening enables the discovery of novel mechanisms of action. Furthermore, cancer drug discovery is steadily moving toward strategically combining target- and phenotypic-based approaches with success in multiple drug classes. Kinase inhibitor cancer drugs lag behind other drug classes in this regard due, in part, to the use of poor phenotypic models. Cancer cell lines, the most common model, do not recapitulate cells found in tumors, and kinase signaling pathways are very sensitive to the context of cellular environment. For kinase inhibitor drugs to benefit from integrating target- and phenotypic-based approaches, creative strategies combining kinase target data with clinically relevant models will be needed. Versatile small molecule probes will be needed to investigate kinase targets identified from such approaches.

Herein, I describe a library of profiled kinase inhibitors with diverse chemistries and biochemical activities for use in phenotypic assays. I use a machine learning-based algorithm to relate the compound inhibition profiles across 237 kinases to their cell-based activities. This approach enables the identification of important kinases in multiple cell lines of sarcoma, a class of rare and understudied cancers. In these screens I identified Protein Kinase D (PRKD) as a putative novel target in synovial sarcoma. A synergy screen of a synovial sarcoma cell line in the presence of a PRKD inhibitor vastly changed the targets identified. These new targets, such as Cyclin Dependent Kinase (CDK) and AKT, displayed synergism when inhibited along with PRKD. I then apply this framework in advanced models of triple negative breast cancer (TNBC). Here, I use ten TNBC patient-derived xenografts (PDXs) to create short-term ex vivo 3D cell cultures from harvested tumors that are amenable for high-throughput screening. The profiled kinase inhibitor screen of these cultures identified multiple kinases broadly important in TNBC. Two identified kinase groups, FES/FER and MARK/SIK, have early emergent genomic evidence as potential targets in TNBC. My pharmacologically-based findings suggest these kinases as actionable targets. Also, I cluster these PDXs using the kinase target scores obtained. Lastly, I describe the development of an irreversible small molecule fluorescent probe for use in localization studies. This probe was found to exhibit a signal in fluorescent microscopy specific to c-SRC, a kinase shown to be a TNBC target in previous studies and in the above PDX screens. I found that this probe displayed turn-on fluorescence, could be used in live-cell microscopy, did not require washing, and was compatible with live-cell super-resolution stimulated emission depletion (STED) microscopy. I then use this probe to interrogate c-SRC localization in multiple TNBC cell lines and track localization changes in response to drug treatment. This work highlights that understanding kinase chemical biology on both molecular and global levels will be needed to continue investigating these bona fide cancer targets.