Bridged Proteolysis Targeting Chimera (PROTAC) Enables Degradation of Undruggable Targets

Abstract

Proteolysis Targeting Chimeras (PROTACs) are attractive therapeutic modalities for degrading disease-causing proteins. While many PROTACs have been developed for numerous protein targets, current small-molecule PROTAC approaches cannot target undruggable proteins that do not have small-molecule binders. Here, we present a novel PROTAC approach, termed bridged PROTAC, which utilizes a small-molecule binder of the target protein's binding partner to recruit the protein complex into close proximity with an E3 ubiquitin ligase to target undruggable proteins. Applying this bridged PROTAC strategy, we discovered MS28, the first-in-class degrader of cyclin D1, which lacks a small-molecule binder. MS28 effectively degrades cyclin D1, with faster degradation kinetics and superior degradation efficiency than CDK4/6, through recruiting the CDK4/6-cyclin D1 complex to the von Hippel-Lindau E3 ligase. MS28 also suppressed the proliferation of cancer cells more effectively than CDK4/6 inhibitors and degraders. Altogether, the bridged PROTAC strategy could provide a generalizable platform for targeting undruggable proteins.

Figure 1.

Discovery of MS28, the first cyclin D1 degrader. (A) Schematic representation of bridged PROTAC. The bridged PROTAC, which binds a bridge protein but not its binding partner (POI), recruits the protein complex into proximity with an E3 ligase, resulting in preferential polyubiquitination and subsequent degradation of the POI by the proteasome over the bridge protein. (B) Chemical structure of MS28. (C) Chemical structures of MS28N1 and MS28N2. (D) Western blot (WB) results of MS28, MS28N1, BSJ-03–204 (BSJ), MS140, and PB on degrading cyclin D1 and CDK4/6. Calu-1 cells were treated with the indicated compound at the indicated concentrations for 8 h. Results shown are representative of at least two independent experiments. (E) Binding affinities of MS28, MS28N1, MS28N2, and PB to CDK4 and CDK6 in biochemical assays. Results shown are the mean ± SD from duplicate experiments.

Figure 2.

MS28 preferentially degrades cyclin D1 over CDK4/6 in Calu-1 cells. (A) Left, MS28 degrades cyclin D1 first and CDK4/6 subsequently. Calu-1 cells were treated with DMSO or 3 μM of MS28 for the indicated time. Results shown are representative of three independent experiments. Right, quantification of relative cyclin D1, CDK4, and CDK6 abundance at each time point, following 3 μM MS28 treatment (four technical repeats from three biological repeats). P-values were calculated for each protein relative to its abundance at 0.5 h. ****P < 0.0001, ***P < 0.001, and **P < 0.01. The line above the bars indicates that all of the encompassed bars share the same p-value above the line. (B) MS28 concentration-dependently degrades cyclin D1. Calu-1 cells were treated with MS28 at the indicated concentration for 8 h. Results shown are representative of two independent experiments. (C) DC₅₀ and D_max values of MS28 in Calu-1 cells (calculated from the WB data in panel B and biological repeat). (D) MS28 did not change the mRNA levels of CCND1, CDK4, and CDK6 in RT-qPCR studies. Calu-1 cells were treated with DMSO, 3 μM MS28, or 3 μM PB for 4 h. mRNA levels are relative to DMSO control. Results are representative of two biological repeats.

Figure 3.

MS28 effectively degrades cyclin D1 in a VHL-, CDK6-, and UPS-dependent manner. (A) Left, VHL KO rescues the MS28-induced degradation of cyclin D1 and CDK4/6. Calu-1 cells were transduced with lentivirus containing VHL-targeting sgRNA or an empty vector. Antibiotic-selected cells were checked for VHL expression (Top: WB results confirm CRISPR-mediated VHL KO in Calu-1 cells), followed by treatment with MS28 for 8 h. Right, quantification of cyclin D1 and CDK4/6 abundance in the indicated experimental groups. P-values for each protein were calculated between the mock and VHL KO groups (indicated by bar) within each concentration. (B) Left, CDK6 KD via siRNA rescues the MS28-induced degradation of cyclin D1. Calu-1 cells were transfected with CDK6-targeting siRNA for 2 days, followed by treatment with MS28 for 8 h. Right, quantification of cyclin D1 abundance in the indicated experimental groups. P-values were calculated between the control and CDK6 KD group within each MS28 concentration. (C) VHL coelutes with cyclin D1-CDK6 in the presence of MS28. His-tagged CDK6 was immobilized on cobalt agarose resin and incubated overnight along with cyclin D1. The VCB complex was added the next day with either DMSO or MS28. Pretreatment with MG132 (D) or MLN4924 (E) rescues the MS28-induced degradation of cyclin D1 and CDK4/6. Quantification of cyclin D1 and CDK4/6 abundance are listed aside. For each protein, p-values were calculated between groups indicated by the lines above the bars. Calu-1 cells were pretreated with MG132 or MLN4924 for 1 h, followed by treatment with MS28 for 8 h. WB results shown in panels A–E are representative of at least two independent experiments. ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P < 0.05.

Figure 4.

MS28 is selective for cyclin D1/D3 and CDK4/6. (A) Left, MS28, but not MS28N1, degrades cyclin D1/3 and CDK4/6, and reduces the cyclin A2 level, but not cyclin D2/B1 and CDK2 in Calu-1 cells (4 h-treatment at 3 μM). WB data shown are representative of two biological repeats. Right, quantification of the relative abundance of cyclins and CDKs analyzed in the WB upon DMSO, MS28, or MS28N1 treatment. P-values were calculated relative to the DMSO control for each protein. ****P < 0.0001, ***P < 0.001, and **P < 0.01. (B) MS28 is selective for CDK6 over a panel of 57 other kinases at 1 μM concentration. Data are the means ± SD from duplicate experiments.

Figure 5.

MS28 effectively suppresses the downstream Rb-E2F pathway, proliferation, and tumorigenesis in cancer cells. (A) MS28 and PB significantly reduced mRNA levels of E2F target genes (CCNA2, CCNE1, CDC6, PCNA, and PLK1) in RT-qPCR studies. Calu-1 cells were treated with DMSO, 3 μM of MS28, or PB for 8 h. P-values were calculated in comparison to DMSO from two biological repeats. ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P < 0.05. (B) MS28 inhibits cell growth much more effectively than PB, BSJ, and MS28N1 in Calu-1 cells (treated with the indicated compound for 5 days). Data shown are the mean values ± SD from two biological repeats (each with three technical repeats). (C) MS28 suppresses clonogenicity of Calu-1 cells in a soft agar assay more effectively than PB, BSJ, and MS28N1. Images were taken at the end of the 20-day treatment. Each treatment group (at 0.3 μM) is representative of three independent experiments, each with at least five technical repeats. (D) MS28, not MS28N1, BSJ, or PB, effectively degrades cyclin D1 in NCI-H2110 cells (treated with the indicated compound at the indicated concentrations for 8 h). Results are representative of two biological repeats. (E) MS28, but not MS28N1, BSJ or PB, potently inhibits the growth of NCI-H2110 cells (5 days treatment, three biological repeats).