Drugs act by altering the activities of particular components (targets) within a cell or organism. Thus, drug discovery campaigns begin by identifying a target, followed by screening this target with compounds to identify leads that can be developed into a drug. Unfortunately, identifying effective drug targets for a given disease, let alone for individual patients (e.g. in highly heterogeneous cancers) is an expensive, time-consuming, and error-prone process. As a result, drugs are frequently developed against incorrect or suboptimal targets, and end up showing no clinical efficacy. Powerful genomic technologies have paved the way for much of the modern understanding of molecular biology, but they have not proven efficient at identifying drug targets. Phenotypic screening, which identifies efficacious drugs by screening compounds directly on cells, has thus regained popularity. In phenotypic screening, however, the targets are typically unknown. We are developing an innovative biotechnology platform that directly identifies effective pharmacological targets from cellular disease models by combining the two approaches, target- based and phenotypic-based screening. This is accomplished with the use of a highly annotated chemical library and sophisticated machine learning algorithms. The compounds are screened in a cell-based assay, and the phenotypic readouts are analyzed in relation to the compounds? biochemical activities, revealing the candidate targets that are mediating the therapeutic activity of effective compounds. This approach can one day be applied at the patient level, for example using patient-derived cancer cells. We have focused our proof of concept studies on the kinase family of drug targets, and hypothesize that our platform can identify kinase dependencies in cancer cells that cannot otherwise be identified using transcriptomic and whole exome sequencing data.
The aims of this Phase I application are to 1) deploy our platform to identify novel kinase targets in DLBCL Lymphoma, and 2) identify kinase inhibitors that could be used to build a compound library that optimizes the performance of the platform. Innovative features of the platform include the combination of target- and phenotypic-based screening, the machine learning algorithm that efficiently detects targets as well as anti-targets, the cell-based screening strategy which uses both tumor and normal cells to detect cancer-specific cytotoxicity, and the unique design features of the compound library. The platform will enable rapid target identification in any area of disease where a clinically relevant cell-based model exists.
Over the past two decades, the cost of developing new drugs has skyrocketed. A major culprit is the difficulty in identifying cellular components that can be engaged by drugs to produce a therapeutic effect. This proposal has two main aims: 1) The first aim is to demonstrate that by using computer algorithms to combine biochemical- and cellular- screening data, effective drug targets can be identified for two different subtypes of DLBCL lymphoma. Importantly, the method also identifies off-targets that, if disturbed, will counteract the desired outcome (i.e. lower or neutralize a drug?s efficacy). The method uses normal blood cells from healthy donors to ensure that the identified drug targets serve to specifically abolish cancer cells without harming normal cells. These key features make our method a valuable complement for currently used target identification technologies, including genomics and proteomics. 2) The second aim is to develop the core components of the methodology into a robust and standalone platform that can be used by drug discovery programs at Truvitech, its partners, and its clients. !
