Pharmacological Inhibition of KIT Activates MET Signaling in Gastrointestinal Stromal Tumors

Abstract

Gastrointestinal stromal tumors (GIST) are the most common adult sarcomas and the oncogenic driver is usually a KIT or PDGFRA mutation. Although GISTs are often initially sensitive to imatinib or other tyrosine kinase inhibitors, resistance generally develops, necessitating backup strategies for therapy. In this study, we determined that a subset of human GIST specimens that acquired imatinib resistance acquired expression of activated forms of the MET oncogene. MET activation also developed after imatinib therapy in a mouse model of GIST (KitV558del/+ mice), where it was associated with increased tumor hypoxia. MET activation also occurred in imatinib-sensitive human GIST cell lines after imatinib treatment in vitro. MET inhibition by crizotinib or RNA interference was cytotoxic to an imatinib-resistant human GIST cell population. Moreover, combining crizotinib and imatinib was more effective than imatinib alone in imatinib-sensitive GIST models. Finally, cabozantinib, a dual MET and KIT small-molecule inhibitor, was markedly more effective than imatinib in multiple preclinical models of imatinib-sensitive and imatinib-resistant GIST. Collectively, our findings showed that activation of compensatory MET signaling by KIT inhibition may contribute to tumor resistance. Furthermore, our work offered a preclinical proof of concept for MET inhibition by cabozantinib as an effective strategy for GIST treatment.

Figure 1. MET is activated in a subset of human GISTs

(A) Phospho-RTK array densitometry of untreated GIST882, HG129, and HG209 cells. *, p<0.05, ANOVA. (B) Quantitative RT-PCR of MET mRNA expression in imatinib-sensitive (Sensitive) and imatinib-resistant (Resistant) human GIST cell lines relative to GIST-T1 cells. *, p<0.05, t-test. (C) Western blot of GIST882 and HG209 cell lines. (D) Western blot of human GIST specimens that were untreated (UT), sensitive (Sens.), or resistant. Clinical risk score for primary gastric GISTs was calculated using the Miettinen classification (42). Tumor specimen from which HG209 cell line was derived is indicated (↑). (E) Western blot and pMET:MET pixel densitometry of human GIST cell lines treated with control or imatinib 0.5 µM for 24h, representative of 3 independent experiments. For A, C, and E, cells were maintained in serum-free media. Bars, mean ± SEM.

Figure 2. KIT inhibition induces MET activation in a GIST mouse model

(A) Kit^V558del/+ mice were treated with vehicle or imatinib and Met mRNA expression was measured by quantitative RT-PCR. n=4–5 mice per group. *, p<0.05, t-test. (B,C) Kit^V558del/+ mice were treated with vehicle, imatinib, or sunitinib. (D) Representative FACS histogram of MET expression (blue) compared to isotype (solid gray) (left) and percent MET positive (right) of intratumoral CD45⁺, CD45⁻KIT⁻, and CD45⁺KIT⁺ cells of untreated Kit^V558del/+ mouse tumors. *, p<0.05, ANOVA. (E) Kit^V558del/+ mice were treated with vehicle or imatinib for 2 weeks and tumor HGF concentration was measured by ELISA and compared to vehicle-treated Kit^V558del/+ mouse liver. *, p<0.05, t-test. (F) Kit^V558del/+ mouse tumors were treated, and Hif1a mRNA expression was measured by quantitative RT-PCR. n=4–5 mice per group. *, p<0.05, t-test. (G) Kit^V558del/+ mice were treated for 4 weeks and then injected with pimonidazole 1h prior to sacrifice. Representative immunofluorescent images of tumors. Green, pimonidazole. Blue, 4',6-diamidino-2-phenylindole (DAPI). Scale bar, 100 µM. n=4 mice per group. Bars, mean ± SEM.

Figure 3. MET inhibition increases the effects of imatinib in human GIST cell lines

(A) Cell viability of serum-starved human GIST cell lines treated with vehicle or HGF 100 ng/mL for 72h. *, p<0.05, t-test. (B) Cell viability of HG209 human GIST cell line treated with imatinib or crizotinib at the indicated concentrations. Using nonlinear regression, IC₅₀ values for imatinib and crizotinib were determined for HG209 (>1000 vs. 109 nM). (C) HG209 cells were transfected with non-target control siRNA (Neg siRNA) or ON-TARGET plus SMARTpool siRNA (MET siRNA). Cell viability 72h after transfection is shown. *, p<0.05, t-test. Western blot demonstrating efficiency of MET knockdown. (D) Cell viability of GIST882 cells treated with sterile water control (CTRL), HGF 100 ng/mL, or imatinib 50 nM (IM) for 72h. *, p<0.05, t-test. (E) Viability of human GIST cell lines treated with DMSO control (CTRL), crizotinib (Criz), or imatinib (IM) for 72h. *, p<0.05, ANOVA. (F) GIST882 and GIST-T1 cells were transfected with non-target control siRNA (Neg siRNA) or ON-TARGET plus SMARTpool siRNA (MET siRNA). GIST882 cells were then treated with sterile water control or imatinib 0.1 µM for 48h, and GIST-T1 cell were treated with sterile water control or imatinib 0.05 µM for 48h, and cell viability was measured. *, p<0.05, t-test. Western blot demonstrating efficiency of MET knockdown. (G) NSG mice with established HG129 s.q. xenografts were treated for 42 days as indicated. *, p<0.05, t-test. n=4–5 mice per group. Data from B, C, D, E, and F are representative of ≥3 independent experiments. Bars, mean ± SEM.

Figure 4. MET inhibition increases the effects of imatinib in Kit^V558del/+ mice

Kit^V558del/+ mice were treated for 2 weeks with vehicle (Veh), imatinib (IM), or crizotinib (Criz) and then (A) tumors were weighed and (B) subjected to western blot, and (C) immunostaining for KIT and Ki67. Scale bar, 100 µM. Data are representative of 2 independent experiments, n=5–7 mice per group. Bars, mean ± SEM. *, p<0.05.

Figure 5. Cabozantinib, a dual MET and KIT small molecule inhibitor, is effective in multiple preclinical models of GIST

(A) Viability of human GIST cell lines treated with imatinib or cabozantinib at the indicated concentrations. Using nonlinear regression, IC₅₀ values were calculated for imatinib and cabozantinib in GIST882 (80 vs. 30 nM), GIST-T1 (37 vs. 15 nM), and HG129 (38 vs. 14 nM) cells and for cabozantinib in HG209 (10 nM) cells. (B–E) NSG mice with established HG129 or HG209 subcutaneous xenografts tumors were treated for 15 days. Representative gross appearance of tumors (scale bar, 1 cm), tumor growth curves, and representative KIT and Ki67 immunohistochemistry (scale bar, 100 µM) are shown. n=3–9 mice per group. (F) Kit^V558del/+ mice were treated for 2 weeks and then tumors were weighed and (G) stained for KIT, Ki67, and CD31 (representative images, scale bar, 100 µM). n=6–7 mice per group. (H) Kit^V558del/+ mice were treated for 8 weeks, and tumors were stained for KIT and Ki67 (representative images, scale bar, 100 µM). n=2 mice per group. Data from A, F and G are representative of 3 independent experiments. Bars, mean ± SEM. *, p<0.05.