UC Berkeley

One of the foremost limitations of drug development efforts is that the majority of disease-driving proteins are considered to be “undruggable” by conventional definitions, which look for well-defined binding pockets. Broadly, the work outlined in this doctoral dissertation addresses the challenge of expanding the scope of druggability by applying chemoproteomic methods to map and profile reactive nucleophilic amino acids within proteins. Using this approach to map ligandable sites in the context of human disease we discover novel ligand binding sites that do not fall within the parameters for conventional drug targets, but still allow for functional modification of pharmacologically relevant proteins. One approach to discovering such sites is to map the protein targets of biologically active natural products. Natural products have long been known to be valuable therapeutic agents due to their variety of biological activity. There exists among these active compounds a subset of covalently acting molecules, which are well-suited for chemoproteomic profiling. Identifying the nucleophilic residues targeted by these covalently acting natural products not only yields better understanding of their mechanism of action but can also reveal unique reactive nodes which they access.

In one example of the utility of this approach I employed chemoproteomic profiling to discover that the anti-cancer natural product nimbolide, covalently reacts with a cysteine within an intrinsically disordered region of the E3 ubiquitin ligase RNF114. I demonstrated that covalent modification of RNF114 by nimbolide leads to impaired ubiquitination of the tumor suppressor p21 through a nimbolide-dependent destabilization of the RNF114-p21 interaction, causing the anti-cancer effects of this natural product through stabilization of p21. The discovery that nimbolide targets a substrate recognition domain within RNF114 also suggested that nimbolide could be used as a novel recruiter for the E3-activity of RNF114 in the design of proteolysis targeting chimeras (PROTACs). Consistent with this premise, I showed that a PROTAC linking nimbolide to the BRD4 inhibitor JQ1 led to degradation of BRD4 in breast cancer cells. Subsequent studies demonstrated the importance of identifying nimbolide as a novel E3 recruiter, as a nimbolide-dasatinib degrader led to preferential degradation of oncogenic BCR-ABL over c-ABL, a preference not seen with other known E3-recruiters. Finally, having identified the reactive hotspot on RNF114 which is targeted by nimbolide, activity-based screening approaches have enabled us to identify a fully synthetic alternative to nimbolide, EN219, to recapitulate its anti-cancer effects and to ease development of further RNF114-recruiting PROTACs.

In a separate effort to pharmacologically recapitulate the effects of the natural product withaferin A, chemoproteomic profiling of this anti-cancer natural product identified a unique druggable hotspot, which sits at an interface within the protein phosphatase 2A (PP2A) tumor-suppressor complex. Interestingly, targeting this site on the regulatory subunit PPP2R1A, leads to activation of PP2A activity, inactivation of AKT signaling, and impaired breast cancer cell proliferation. Using activity-based screening I identified and optimized a more synthetically tractable cysteine-reactive covalent ligand, JNS 1-40, capable of recapitulating these effects and impairing in vivo tumor growth by selectively targeting this site. Discovery of this synthetically scalable PPP2R1A ligand has enabled its use as a PP2A activating tool compound in cancer studies and opens opportunities to recruit activated PP2A complex in heterobifunctional therapies in the future.