Proteolysis Targeting Chimera (PROTAC) Linkerology Enhances the CDK-Degradation Selectivity Profile of a Multi-CDK Inhibitor: Discovery of Potent, Anti-Leukemic CDK9 Degraders

Abstract The degradation efficiency of a PROTAC─bivalent compounds that link together a ligand for a protein of interest with a ligand for an E3 ligase─is governed, in part, by the length and flexibility of its linker (“linkerology”), as well as the accessibility of ubiquitinatable surface lysines. Based on the prior serendipitous discovery of the cyclin-dependent kinase 9 (CDK9)-selective PROTAC degrader “THAL-SNS-032”, developed from the multi-CDK2/CDK7/CDK9 inhibitor SNS-032, we speculated that other multi-CDK inhibitors might likewise be transformed into CDK9-selective PROTACs, establishing a “linkerology” platform approach to enhancing CDK family member specificity (and beyond). We tested this hypothesis using AT7519, a multi-CDK2/CDK5/CDK9 inhibitor, and employing cereblon as the targeted E3 ligase. Accordingly, we built candidate PROTACs by linking AT7519 to one of two thalidomide derivatives; those molecules in which AT7519 and the thalidomide derivative were coupled via rigid linkers caused greater degradation of the CDK9 protein than those with more flexible linkers. Western blotting of human acute myeloid leukemia (AML) cells treated with AMC-3-221, our most CDK9-selective PROTAC, resulted in almost-complete reduction of the CDK9 protein, with no detectable degradation of CDK1, CDK2, or CDK5, and minimal degradation of CDK4 and CDK6, corroborated by global proteomics analysis. Subtle shortening or lengthening of AMC-3-221’s rigid linker altered the CDK degradation selectivity profile, suggesting that continued linker variation may further enhance selectivity. Molecular modeling provided insight into the observed selectivity through the analyses of ternary complex stabilities. AMC-3-221 treatment rapidly decreased RNA and protein levels of oncogenic downstream targets of CDK9─MCL-1 and c-/N-MYC─and potently inhibited human AML and other types of human cancer cell lines, with less inhibition of a non-transformed cell line, via an on-target mechanism of apoptosis. AMC-3-221’s CDK9 degradation potency (DC50 < 10 nM) approximated its antileukemic inhibitory potency (IC50 < 50 nM). The maximum tolerated dose in mice of AMC-3-221 was higher than that of another PROTAC, AMC-3-225, consistent with AMC-3-221’s more selective CDK degradation profile. In summary, we demonstrate that converting a multi-CDK inhibitor into a PROTAC with selectivity for CDK9 over other family members is not an isolated phenomenon. Our findings further support the concept that PROTAC linkerology may serve as a platform for enhancing kinase selectivity.