The central role of protein kinases in signal transduction pathways has generated intense interest in targeting these enzymes for a wide range of therapeutic indications. Here we report a method for identifying and quantifying protein kinases in any biological sample or tissue from any species. The procedure relies on acyl phosphate-containing nucleotides, prepared from a biotin derivative and ATP or ADP. The acyl phosphate probes react selectively and covalently at the ATP binding sites of at least 75% of the known human protein kinases. Biotinylated peptide fragments from labeled proteomes are captured and then sequenced and identified using a mass spectrometry-based analysis platform to determine the kinases present and their relative levels. Further, direct competition between the probes and inhibitors can be assessed to determine inhibitor potency and selectivity against native protein kinases, as well as hundreds of other ATPases. The ability to broadly profile kinase activities in native proteomes offers an exciting prospect for both target discovery and inhibitor selectivity profiling.

Summary Protein kinases are intensely studied mediators of cellular signaling, yet important questions remain regarding their regulation and in vivo properties. Here we use a probe-based chemoprotemics platform to profile several well studied kinase inhibitors against more than 200 kinases in native cell proteomes and reveal new biological targets for some of these inhibitors. Several striking differences were identified between native and recombinant kinase inhibitory profiles, in particular, for the Raf kinases. The native kinase binding profiles presented here closely mirror the cellular activity of these inhibitors, even when the inhibition profiles differ dramatically from recombinant assay results. Additionally, Raf activation events could be detected upon live cell treatment with inhibitors. These studies highlight the complexities of protein kinase behavior in the cellular context and demonstrate that profiling with only recombinant/purified enzymes can be misleading.

The interaction of a drug with its target is critical to achieve drug efficacy. In cases where cellular environment influences target engagement, differences between individuals and cell types present a challenge for a priori prediction of drug efficacy. As such, characterization of environments conducive to achieving the desired pharmacologic outcome is warranted. We recently reported that the clinical CDK4/6 inhibitor palbociclib displays cell type-specific target engagement: Palbociclib engaged CDK4 in cells biologically sensitive to the drug, but not in biologically insensitive cells. Here, we report a molecular explanation for this phenomenon. Palbociclib target engagement is determined by the interaction of CDK4 with CDKN2A, a physiologically relevant protein inhibitor of CDK4. Because both the drug and CDKN2A prevent CDK4 kinase activity, discrimination between these modes of inhi-bition is not possible by traditional kinase assays. Here, we describe a chemo-proteomics approach that demonstrates high CDK4 target engagement by palbociclib in cells without functional CDKN2A and attenuated target engagement when CDKN2A (or related CDKN2/INK4 family proteins) is abundant. Analysis of biological sensitivity in engineered isogenic cells with low or absent CDKN2A and of a panel of previously characterized cell lines indicates that high levels of CDKN2A predict insensitivity to palbociclib, whereas low levels do not correlate with sensitivity. Therefore, high CDKN2A may provide a useful biomarker to exclude patients from CDK4/6 inhibitor therapy. This work exemplifies modulation of kinase target engagement by endogenous proteinaceous regulators and highlights the importance of cellular context in predicting inhibitor efficacy.

Palbociclib is a cyclin-dependent kinase (CDK) 4/CDK6 inhibitor approved for breast cancer that is estrogen receptor (ER)-positive and human epidermal growth factor receptor 2 (HER2)-negative. We profiled palbociclib in cells either sensitive or resistant to the drug using an ATP/ADP probe-based chemoproteomics platform. Palbociclib only engaged CDK4 or CDK6 in sensitive cells. In resistant cells, no inhibition of CDK4 or CDK6 was observed, although the off-target profiles were similar in both cell types. Prolonged incubation of sensitive cells with the compound (24 h) resulted in the downregulation of additional kinases, including kinases critical for cell cycle progression. This downregulation is consistent with cell cycle arrest caused by palbociclib treatment. Both the direct and indirect targets were also observed in a human tumor xenograft study using the COLO-205 cell line in which phosphorylation of the retinoblastoma protein was tracked as the pharmacodyanamic marker. Together, these results suggest that this probe-based approach could be an important strategy toward predicting patient responsiveness to palbociclib.

Hsp90 is an ATP-dependent chaperone of widespread interest as a drug target. Here, using an LC-MS/MS chemoproteomics platform based on a lysine-reactive ATP acyl phosphate probe, several Hsp90 inhibitors were profiled in native cell lysates. Inhibitor specificities for all four human paralogs of Hsp90 were simultaneously monitored at their endogenous relative abundances. Equipotent inhibition of probe labeling in each paralog occurred at sites both proximal to and distal from bound ATP observed in Hsp90 cocrystal structures, suggesting that the ATP probe is assaying a native conformation not predicted by available structures. Inhibitor profiling against a comprehensive panel of protein kinases and other ATP-binding proteins detected in native cell lysates identified PMS2, a member of the GHKL ATPase superfamily as an off-target of NVP-AUY922 and radicicol. Because of the endogenously high levels of Hsp90 paralogs in typical cell lysates, the measured potency of inhibitors was weaker than published IC₅₀ values. Significant inhibition of Hsp90 required inhibitor concentrations above a threshold where off-target activity was detectable. Direct on- and off-target engagement was measured by profiling lysates derived from cells treated with Hsp90 inhibitors. These studies also assessed the downstream cellular pathway effects of Hsp90 inhibition, including the down regulation of several known Hsp90 client proteins and some previously unknown client proteins. Overall, the ATP probe-based assay methodology enabled a broad characterization of Hsp90 inhibitor activity and specificity in native cell lysates.