Workflow for E3 Ligase Ligand Validation for PROTAC Development

Abstract

Proteolysis targeting chimeras (PROTACs) have gained considerable attention as a new modality in drug discovery. The development of PROTACs has been mainly focused on using CRBN (Cereblon) and VHL (Von Hippel-Lindau ligase) E3 ligase ligands. However, the considerable size of the human E3 ligase family, newly developed E3 ligase ligands, and the favorable druggability of some E3 ligase families hold the promise that novel degraders with unique pharmacological properties will be designed in the future using this large E3 ligase space. Here, we developed a workflow aiming to improve and streamline the evaluation of E3 ligase ligand efficiency for PROTAC development and the assessment of the corresponding "degradable" target space using broad-spectrum kinase inhibitors and the well-established VHL ligand VH032 as a validation system. Our study revealed VH032 linker attachment points that are highly efficient for kinase degradation as well as some of the pitfalls when using protein degradation as a readout. For instance, cytotoxicity was identified as a major mechanism leading to PROTAC- and VHL-independent kinase degradation. The combination of E3 ligase ligand negative controls, competition by kinase parent compounds, and neddylation and proteasome inhibitors was essential to distinguish between VHL-dependent and -independent kinase degradation events. We share here the findings and limitations of our study and hope that this study will provide guidance for future evaluations of new E3 ligase ligand systems for degrader development.

Figure 1

Workflow for validating E3 ligase for PROTAC design using promiscuous kinase inhibitors as POI ligands. (A) Schematic representation of the developed workflow. (B) Structures of the kinase inhibitors 1 (and methylated derivative 2) and 3. (C) Cocrystal structure of 2 with STK17B (PDB ID: 3LM0) reveals the solvent exposed pyrimidine moiety as a suitable linker attachment point (red box). (D) General chemical structure of PROTACs used highlighting the promiscuous kinase ligand (green), the linker (beige), and the VHL ligand (purple). Two structurally diverse kinase ligands were used based on 1(59) and 3(60) (left). Kinase ligands based on 1 were coupled via either a piperazine spacer (Inhib-1) or a carboxylic acid (Inhib-2) attached to the central pyrimidine moiety, enabling varying linker coupling chemistry. Inhib-3 was coupled via its terminal piperazine moiety (left bottom). Kinase ligands were coupled to the commonly used VHL ligand VH032-NH₂ (right top) or VH032-OH (right bottom) using alkyl (mid top) or PEG (mid bottom) linkers.

Figure 2

Mapping of the kinome target space. (A) Chemical structures of all kinase parent ligands 4–6 and linker conjugates 4-L and 5-L. (B) Kinobeads data of kinase parent ligands 4–6 and the linker conjugates 4-L and 5-L visualized as radar plots (pK_d^app). (C) Bar chart (left panel) representing the number of kinases engaged with a K_d^app below 1 μM threshold in a Kinobeads assay (black), number of target kinases for which K_d^app fell below 1 μM threshold upon linker attachment (dotted), and number of kinases for which K_d^app increased above 1 μM threshold upon linker attachment (gray) of kinase parent inhibitors 4 and 5 and their corresponding linker conjugates 4-L and 5-L. Venn diagram (right panel) showing the number of kinases engaged with a K_d^app below 1 μM by each individual kinase parent ligand and target overlaps.

Figure 3

Overview of all synthesized PROTACs. (A) Chemical structures of all synthesized PROTACs. (B) Pie charts summarizing and highlighting the structural characteristics across the PROTACs set: (I) distribution of kinase parent inhibitors 4–6; (II) distribution of the employed linker coupling chemistry on the kinase ligand; (III) distribution of used alkyl and PEG linkers; and (IV) distribution of utilized VHL ligand exit vectors.

Figure 4

Kinase parent inhibitors, the respective linker conjugates, and the resulting promiscuous kinase PROTACs show cellular target engagement of model kinases. (A) Exemplary NanoBRET dose–response curves of the parent inhibitor 4, the linker conjugate 4-L, and the PROTACs 4-a–4-d measured using nanoLUC full-length AAK1 in intact (left) and permeabilized (right) cells. (B) IC₅₀ values were calculated as the means of duplicates of a 10-point dose–response curve with errors calculated using the standard deviation. Compound sets based on the same kinase parent inhibitor are clustered in colored boxes (4 represented in dark green, 5 represented in light blue, and 6 represented in light green). ^aEstimated IC₅₀ values are based on extrapolation. ^bIC₅₀ values were outside the assay window.

Figure 5

Level of promiscuity correlates with compound cytotoxicity. Exemplary CellTiterGlo dose–response curves of Jurkat cells treated with kinase parent inhibitors 4–6, linker conjugates 4-L and 5-L, and the resulting promiscuous kinase PROTACs after 24 h of incubation. Compounds are shown according to the used kinase parent inhibitor. Bottom right: calculated IC₅₀ values are shown as the mean of triplicate measurements in μM. Errors were calculated using the standard deviation. Compounds were grouped based on their parent ligand as shown in the figure caption. ^aIC₅₀ values were outside the assay window. ^bEstimated IC₅₀ values are based on extrapolation in GraphPad. ^cEstimated IC₅₀ values as corresponding dose–response curves were not sigmoidal.

Figure 6

PROTACs 4-a and 6-b induce VHL-dependent degradation of multiple kinases across the entire kinome. Jurkat cells were treated with 1 μM of PROTAC for 6 h. Nonkinase proteins that showed no significant changes in expression levels are represented by light-blue dots; kinases that displayed no significant expression level alterations are represented by green dots; nonkinase proteins that exhibited significant changes in expression level are represented by dark gray dots; and kinases that were significantly up- or downregulated are represented by enlarged cyan dots and are labeled. (A) Volcano plots of PROTACs 4-a and 6-b. (B) Volcano plots of the negative controls 4-a^neg and 6-b^neg. (C) Chemical structures of the negative controls 4-a^neg and 6-b^neg harboring the inactive hydroxyproline epimer of VH032-NH₂ precluding interaction with VHL.

Figure 7

Validation of proteomic hits by Western blots and HiBiT experiments. (A,B) Western blots of Jurkat cells treated with different concentrations of the specified compounds for 6 h. ITK (A) and AURKA protein levels (B) were compared with Jurkat cells treated with DMSO. GAPDH was used as a loading control. (C,D) ITK (C) and AURKA (D) protein levels were based on luciferase measurements. MV4–11 cells, expressing tagged AURKA^HiBiT protein, and Jurkat cells, expressing tagged ITK^HiBiT, were treated with different concentrations of the specified compounds for 6 h. Following cell lysis, the resulting lysates were complemented with the largeBiT fragment, and luciferase activity was measured. Top panel: the activity of PROTACs 4-a and 6-b was compared to that of their respective kinase parent inhibitors (4 and 6) and their corresponding negative controls (4-a^neg and 6-b^neg). Bottom panel: cotreatment experiments of PROTACs 4-a and 6-b were conducted in the presence of either the neddylation inhibitor MLN4924 or the proteasome inhibitor MG132, respectively.