Fluorescence based live cell imaging identifies exon 14 skipped hepatocyte growth factor receptor (MET) degraders

Abstract

Despite ongoing efforts to employ structure-based methods to discover targeted protein degraders (TPD), the prevailing strategy continues to be the synthesis of a focused set of heterobifunctional compounds and screening them for target protein degradation. Here we used a fluorescence based live cell imaging screen to identify degraders that target exon 14 skipped hepatocyte growth factor receptor (MET). MET is a known oncogenic driver. MET exon 14 skipping mutations (METex14Δ) are found in lung cancers and result in the loss of a degron that is required for E3-ligase recognition and subsequent ubiquitination, prolonging the half-life and oncogenicity of MET. Since proteolysis targeting chimeras (PROTACs) are heterobifunctional molecules that promote target degradation by the proteosome, we sought to restore degradation of MET lost with METex14Δ using a MET-targeting PROTAC. We generated a library of sixty PROTACs of which 37 used the MET inhibitor capmatinib as the protein of interest targeting ligand. We screened this PROTAC library for targeted degradation of METex14Δ-GFP using live cell imaging. We benchmarked the MET-targeting PROTACs to that of a previously reported MET-targeting PROTAC, SJF8240. Curve fitting live cell imaging data affords determination of time required to degrade 50% of the target protein (DT₅₀), which was used in determining structure activity relationships. A promising candidate, 48-284, identified from the screen, exhibited classic PROTAC characteristics, was >15-fold more potent than SJF8240, had fewer off targets compared to SJF8240, and degraded MET in multiple cell lines.

Fig. 1. Live cell imaging screen for METex14Δ-GFP degradation. (A) Capmatinib docked into MET kinase domain (PDB 3zbx). Capmatinib is shown in magenta sticks and the kinase domain of MET is shown as green ribbon structure. The quinoline ring of capmatinib mimics the adenine ring of the ATP and the quinoline nitrogen is within hydrogen bonding distance of the N–H of Met¹¹⁶⁰ (hydrogen bond is shown in black line) (B) characterization of HEK293 cells stably transfected with either GFP-vector or GFP-labelled MET-exon-14 skipping mutant. (C) Design of capmatinib based PROTAC library and SJF8240. (D and E) Capmatinib based PROTAC library screened at 10 μM and 1 μM using live cell imaging in HEK293 transfected with GFP-Met-exon-14 skipping mutant. The bar graph shows the green count values over confluence (phase) normalized to time zero of each well at 0 h, 8 h and 24 h post addition. The broken redline indicates activity relative to SJF8240. The bars represent mean ± SD of three independent biological replicates (n = 3).

Fig. 2. Characterization of MET-PROTACs. (A) Time course study to assess the effects of the linkers in PROTACs generated with capmatinib and VHL ligand. The line graph is an average of three independent biological replicates (n = 3), green count values over confluence (phase) normalized to 0 time of each well. The number of linker atoms are indicated in parathesis. Degradation time for 50% reduction in GFP signal was determined through curve-fitting the data (Prism 10.2.1). (B) A dose–response study with capmatinib based PROTAC 48-284 and foretinib based PROTAC SJF8240 (10 000, 5000, 1000, 500, 100 nM) in HEK293 transfected with GFP-Met-exon-14 skipping mutant (n = 3), green count values over confluence (phase) normalized to time zero of each well. (C) A time course study with the capmatinib based PROTAC 48-284 (1 μM) with HEK293 transfected with GFP-Met-exon-14 skipping mutant in the presence and absence of MG132 (10 μM). The images analyzed every 2 h for 24 h post-treatment. The line graph is an average of three independent biological replicates (n = 3), green count values over confluence (phase) normalized to 0 time of each well. (D) Volcano plot depicting changes of protein abundance in HEK293 transfected with GFP-labelled MET-exon-14 skipping mutant cells treated with the PROTAC 48-284 and incubated for 24 hours. The lysates were subjected to label-free proteomic analyses and the volcano plot represents 5106 proteins, with the log2 fold change shown on the x-axis and negative log10 p-values on the y axis. Data are presented as the averages of three independent biological replicates (n = 3). Relevant proteins are labelled. (E) The percent degradation of MET-GFP and GFP-labelled MET-exon-14 skipping is shown for DMSO, and 48-284 with the mean of three replicates shown.

Fig. 3. Time and dose effects of MET PROTACs. (A) Hs746T cell line that carries an endogenous METex14Δ mutation was treated with capmatinib based PROTAC 48-284 at 1.0 μM and MET was assessed at the indicated timepoints by western blots with β-actin controls. These cells were also treated at the indicated doses for 8 hours. The effects on downstream RAS/AKT and RAS/ERK pathway signaling were also assessed. (B) Similarly, Hs746T cell line that carries an endogenous METex14Δ mutation was treated with the foretinib-based PROTAC SJF-8240 at 1.0 μM and MET was assessed at the indicated timepoints by western blots with β-actin controls. These cells were also treated at the indicated doses for 8 hours. The effects on downstream RAS/AKT and RAS/ERK pathway signaling were also assessed. (C) Western blot analysis 3 vehicle treated UW21 xenografts and 3 UW21 xenografts treated with 48-284. (D) We determined the ratios of MET to beta-actin using ImageJ and found that there was significantly greater MET in the untreated group than treated group (**p-value < 0.01).