Fig 2: Characterizing the GIST kinome in primary tumors using MIB-MS to identify therapeutic targets.(A) Schematic of experimental approach. MIB-MS was used to quantify the kinase abundance in patients with GISTs (untreated, gastric primary GIST from 3 molecular subtypes: KIT mutant, n = 15; PDGFRA mutant, n = 10; WT GIST, n = 8; and normal gastric tissue, n = 9) to map the proteomic landscape of the kinome and identify targets. Kinase levels in tissues were determined using a combination of LFQ and s-SILAC. (B) Kinome tree depicts fraction of kinome quantitated by MIB-MS and frequency across 42 samples measured. (C) Average number of kinases detected by MIB-MS profiling broken down by tissue type. GIST, gastrointestinal stromal tumor; MIB-MS, multiplexed inhibitor beads and mass spectrometry; LFQ, label-free quantitation; s-SILAC, super-SILAC.

Fig 3: Dexamethasone induces PDGFR mediated survival signalling. (A) Functional characteristics of T98G cells treated with 25 µM DEXA for 18 h revealed by GSEA using GO term and PID gene set collections. (B) qRT-PCR analysis for PDGFRA, PDGFRB and PDGFB in the indicated cell lines represented as mean fold change of triplicates treated with DEXA for 18 h relative to DMSO. (C) Western blot analysis for PDGFRA and PDGFRB in the indicated cell lines either untreated or treated with DEXA for 18 h. Beta-actin served as loading control. (D) Western blot analysis for PDGFRA, PDGFRB and the indicated phospho-specific antibodies in T98G cells either untreated or treated with DEXA for 18 h. The lysate of one experiment was run on several lanes per blot, which were probed in parallel. Beta-actin serves as loading control and representative controls are shown. (E) Western blot analysis for phospho-SFK (using a phospho SRC antibody) and phospho-STAT5 in the indicated cell lines either untreated or treated with DEXA for 18 h. Beta-actin served as loading control. (F) Co-expression analysis of the indicated genes from the TCGA Glioblastoma dataset. (G) qRT-PCR analysis for BCL2L1 and MCL1 in the indicated cell lines represented as mean fold change of triplicates treated with DEXA for 18 h relative to DMSO. (H) qRT-PCR analysis as in (G). Cells were treated with DEXA in the absence or presence of 200 nM ponatinib for 18 h. (I) Quantification of the relative cell number of the indicated cell lines grown in the absence or presence of 25 µM DEXA. Data represent the mean ± SEM (n = 3). (J) Quantification of the relative cell number of the indicated cell lines grown with or without 25 µM DEXA in the absence or presence of 25 nM or 100 nM ponatinib (Pon) or 1 µM or 3 µM sunitinib (Sun). Data represent mean ± SEM (n = 3). (K) Quantification of the relative cell number of T98G cells grown with or without 25 µM DEXA in the absence or presence of 5 µM S63845 (S63), 1 µM navitoclax (Nav) or 5 µM venetoclax (Ven). Data represent mean ± SEM (n = 3). ** p < 0.01, *** p < 0.001.

Fig 4: Mapping the distinct kinome signatures among GIST subtypes.(A–C) PCA, including PC1 vs. PC2 (A), PC1 vs.PC3 (B), and PC2 vs.PC3 (C) of MIB-MS in 3 GIST subtypes (KIT mutant, blue; PDGFRA mutant, red; WT, green) and normal gastric tissue (pink). (D) Volcano plot comparisons of KIT mutant vs. WT, (E) PDGFRA mutant vs. WT, and (F) KIT mutant vs. PDGFRA mutant GIST MIB-MS kinome profiles. Differences in kinase log2 LFQ intensities among tumors and normal tissues determined by paired t test Benjamini-Hochberg adjusted P values at FDR of <0.05 using Perseus software. PCA, principal component analysis; GIST, gastrointestinal stromal tumor; MIB-MS, multiplexed inhibitor beads and mass spectrometry; LFQ, label-free quantitation.

Fig 5: Targeting the mutant-GIST kinome signature identifies WEE1 as candidate target.(A) Volcano plot comparisons of KIT mutant and PDGFRA mutant GIST vs. normal gastric tissue MIB-MS kinome profiles. Differences in kinase log2 LFQ intensities among tumors and normal tissues determined by paired t test Benjamini-Hochberg adjusted P values at FDR <0.05 using Perseus software. (B) Scatter plot depicts overlap in kinases elevated or reduced determined by LFQ or s-SILAC. Regression analysis (R2) among quantitative methods was performed in Perseus software. Differential expressed kinases commonly identified by LFQ and s-SILAC quantitation (FDR <0.05) are labeled. (C) Bar graph depicts high-confident kinases log2 LFQ z scores overexpressed in mutant-GIST determined by LFQ and/or s-SILAC quantitation (FDR <0.05). (D) Associated pathways/functions of kinases overexpressed in KIT mutant and PDGFRA mutant GIST vs. normal tissues determined by quantitative MIB-MS profiling. (E) Heatmap depicting viability scores for siRNA library screen targeting high-confident kinases elevated in KIT mutant and PDGFRA mutant GIST in GIST-T1+Cas9 and GIST-T1+D842V KITKO cell lines as measured by Cell Titer Blue assay. siGL2 was negative control, viability score = 1.0. Two independent replicates were performed per cell line. (F) Quantitative RT-PCR confirmed >70% knockdown of Wee1 (top) and MAP3K3 (bottom) mRNA in both cell lines. Expression levels were normalized to HPRT. Data represent mean ± SD. GIST, gastrointestinal stromal tumor; MIB-MS, multiplexed inhibitor beads and mass spectrometry; LFQ, label-free quantitation; s-SILAC, super-SILAC.