Overexpression of FGFR2 contributes to inherent resistance to MET inhibitors in MET-amplified patient-derived gastric cancer xenografts

Gastric cancer is one of the most malignant diseases and one of the leading causes of cancer-associated mortality worldwide. Although advances have been made in surgical techniques, perioperative management and the combined use of surgery with chemotherapy and/or radiotherapy, patients with advanced stage gastric cancer continue to face poor outcomes. Furthermore, it was reported that MET gene amplification and overexpression predicted the sensitivity to MET inhibitors in gastric cancer. However, the identification of drug-resistant tumors has encouraged the pre-emptive elucidation of the possible mechanisms of clinical resistance. The current study assessed a number of patient-derived gastric cancer models with MET amplification and overexpression, including CNGAS028. The tumor tissues were subjected to microarray analysis (using single nucleotide polymorphism 6.0 and human genome U133 arrays) followed by western blotting. The results demonstrated that CNGAS028 xenograft tumors did not respond to treatment with a selective MET inhibitor. Additional analysis indicated that FGFR2 overexpression contributed to the resistance to MET inhibitors. Furthermore, treatment with a combination of fibroblast growth factor receptor 2 and MET inhibitors inhibited the growth of CNGAS028 xenograft tumors in vivo. In conclusion, the current results aid in understanding the mechanism of inherent resistance to selective MET inhibitors as well as provide important information for patient selection and clinical treatment strategies.


Introduction
Gastric cancer is the second most prevalent cause of cancer-associated mortality worldwide (1). However, over the past 50 years, the total incidence rates of gastric cancer have gradually decreased, particularly in developed countries. Furthermore, the disease most commonly occurs within the male population in developing countries, predominantly East Asia, South America and Eastern Europe (2). Conventional therapeutic strategies for gastric cancer include surgery, chemotherapy and radiation therapy (3). However, as gastric cancer has few symptoms during the early stages, the majority of patients are typically diagnosed once the cancer has progressed to an advanced stage. Despite undergoing surgical resection, tumors recur in a large number of patients, in such cases the median survival time following cytotoxic chemotherapy is <1 year. Therefore, the diagnosis and effective treatment of advanced gastric cancer continues to be a challenge for oncologists (4). Although the use of molecular targeted therapy has been studied in other common types of solid tumors, including non-small cell lung cancer and breast cancer, it has yet to be fully explored in gastric cancer (5).
MET was initially identified as an oncogene encoding the receptor tyrosine kinase (RTK) for hepatocyte growth factor. The MET gene has been identified on chromosome 7q21-q31, where it encodes a single precursor that is digested and glycosylated post-transcriptionally, resulting in an extracellular α-chain (50-kDa) linked to a transmembrane β-chain (140-kDa) via disulfide bonds. Oncogenic activation of MET suppresses apoptosis and promotes cell survival, proliferation, migration and differentiation, as well as gene transcription and angiogenesis (6,7).
Gain-of-function mutations in MET are uncommon in gastric cancer (8), with MET activation predominantly attributed to gene amplification (9). A previous used fluorescence in situ hybridization analysis in order to detect MET amplification, which was reported to occur in ≤4% of patients with gastric cancer (10). Various MET inhibitors have been investigated in clinical trials, which showed promising initial results indicating that MET may be a potential therapeutic target for the treatment of gastric cancer (11,12). An increasing number of pharmaceutical companies are focusing on the identification

Overexpression of FGFR2 contributes to inherent resistance to MET inhibitors in MET-amplified patient-derived gastric cancer xenografts
of novel small molecular c-MET inhibitors, including PF2341066 (Pfizer Ltd., Surrey, UK) and ARQ197 (ArQule Inc., Woburn, MA, USA) (13,14). However, the identification of drug-resistant tumors has encouraged the pre-emptive elucidation of potential mechanisms of clinical resistance. The present study describes a patient-derived gastric cancer model resistant to a selective MET inhibitor and attempts to determine the underlying mechanism. Detection of gene copy number and expression by microarray in established gastric cancer xenograft models. The GeneChip ® genome-wide human single nucleotide polymorphism (SNP) 6.0 and human genome U133 plus 2.0 arrays (Affymetrix, Inc., Santa Clara, CA, USA) were used to analyze the genomic gene copy number and gene expression levels in all established patient-derived gastric xenograft tumors, respectively. MAS5 software (Affymetrix, Inc.) was used to analyze the U133 results. Gene profiling comparison was performed by calculating the fold change of the copy number and gene expression between these tumors. The data were processed using the aroma.affymetrix R package (version 2.13.0; http://www.aroma-project.org/), according to the methods described by Bengtsson et al (15).

Establishment
Efficacy studies in gastric cancer xenograft models with MET amplification and overexpression. Gastric tumors (2-cm diameter) were aseptically resected from established patient-derived gastric cancer xenografts with MET amplification and overexpression, then minced into 3x3x3 mm pieces. Host mice were then anesthetized with isoflurane and a section of tumor was implanted into the left flank of each mouse. Each gastric model that developed tumors reaching 150-200 mm 3 in size were randomized into the following four treatment groups (

MET gene amplification and expression in Chinese
pat ient-derived gast ric ca ncer m odels. C h i nese patient-derived gastric cancer models (n=30) were established from 83 gastric cancer specimens. The established models were termed CNGAS001-030. The CNGAS001, CNGAS002 and CNGAS003 mouse models are indicated in Fig. 1A. Microarray data from the SNP 6.0 and U133 plus 2.0 gene chips were used to analyze the genomic gene copy number and gene expression levels of all established models, respectively. The microarray data demonstrated that MET was highly amplified and expressed in 16.7% (5/30) of the Chinese gastric cancer xenograft models ( Fig. 1B and C). From the results, it was observed that MET amplification was positively correlated with MET overexpression in the xenograft models.
High amplification and overexpression of MET predicts response to PHA665752 in patient-derived gastric cancer models. The present study analyzed the efficacy PHA665752 in the patient-derived gastric xenograft models with high MET amplification and overexpression (CNCAS005,  The results demonstrated that four gastric cancer xenograft models were significantly sensitive to PHA665752 treatment (P<0.05). However, the CNCAS028 model was resistant to PHA665752 ( Fig. 2A High FGFR2 amplif ication and expression in the CNCAS028 model. As indicated in Fig. 2A, the tumor growth of CNCAS028 was not inhibited by treatment with 30 mg/kg PHA665752 for 21 days. To explore the mechanism of the resistance to the selective MET inhibitor, the genome-wide gene profiles of CNCAS028 were compared with those of the other four gastric cancer models. It was identified that FGFR2 was highly amplified and expressed in the CNCAS028 model, whereas FGFR2 was expressed at a normal level and not amplified in the PHA665752-sensitive xenografts [normal FGFR2 expression, copy number <5 and expression level <4,000 (as determined by GeneChip ® ); Fig. 3A and B]. These results were confirmed by western blot analysis (Fig. 4).
PHA665752 and NVP-BGJ398 combination treatment significantly inhibits tumor growth in the CNCAS028 model. To validate the association between FGFR2 amplification and overexpression as well as PHA665752 resistance, PHA665752 was combined with a selective pan-FGFR2 kinase inhibitor (NVP-BGJ398) to treat the CNCAS028 model (16). As indicated in Fig. 5, treatment with 30 mg/kg PHA665752 was not able to inhibit tumor growth in the CNCAS028 model. Furthermore, treatment with 15 mg/kg NVP-BGJ398 only marginally inhibited tumor growth. By contrast, combined treatment with these two compounds significantly inhibited tumor growth following 21 days of treatment (P<0.01).
Effect of PHA665752 and/or NVP-BGJ398 on signaling transduction in patient-derived gastric cancer models. To investigate   the effect of PHA665752 and NVP-BGJ398 on downstream molecules of the phosphoinositide 3-kinase and RAS signaling pathways, western blot analysis was used to observe changes in the phosphorylation status and total protein expression levels of the tumor tissues. The results demonstrated that all five patient-derived gastric xenograft models highly expressed p-MET and the CNCAS028 xenograft also highly expressed p-FGFR2 (Fig. 4). Western blot analysis also identified that treatment with PHA665752 inhibited the phosphorylation of MET in all five gastric tumor models. In addition, the expression of p-FGFR2 was markedly inhibited by NVP-BGJ398 or combination treatment in the CNCAS028 model (Fig. 6).

Discussion
Gastric and gastroesophageal cancer affect 1 million individuals worldwide every year and are the second most common cause of cancer-associated mortality (17). Targeted therapies have been developed and incorporated into the standard treatment strategies for other types of solid cancer, such as lung or breast. However, such therapies (including MET inhibitors) are only now being examined in the context of gastric and gastroesophageal cancer (18). The current investigations identified MET gene amplification in 5/30 (16.7%) patient-derived gastric cancer xenografts. These results indicated the therapeutic potential of MET inhibitors in gastric cancer. Additional analysis identified that a selective MET inhibitor (PHA665742) was able to significantly inhibit tumor growth in 4/5 gastric cancer models with MET amplification. These results are consistent with a previous study, which demonstrated that MET amplification was associated with the response of the MKN45 gastric cancer cell line to PHA665752 treatment (19). Aberrant RTK expression produces growth and survival signals that are essential for the pathogenesis and progression of various types of cancer. Furthermore, cancer patients treated with targeted inhibitors of key oncogenic kinase drivers, including imatinib, gefitinib and erlotinib, have exhibited promising clinical outcomes (20). However, based on the precedence set by agents such as imatinib in chronic myeloid leukemia and erlotinib in lung adenocarcinoma, inherent resistance may potentially limit the application of single agent therapies (21). The elucidation of novel oncogenic drivers may have extensive implications for targeted therapy. Corso et al (22) reported that activation of HER family members in MET-addicted cancer cells, subsequent to MET inactivation, resulted in increased cell viability in vitro and recovered tumorigenicity in vivo. In addition, Lee et al (23) reported that a novel SND1-BRAF fusion gene exhibited resistance to the MET inhibitor PF-04217903 in GTL16 cells via RAS/RAF/ERK signaling pathway activation. By contrast, the current study determined that a MET-amplified CNCAS028 model was resistant to MET inhibitor PHA66575 as a result of FGFR2 gene amplification and overexpression. Inhibition of FGFR2 signaling in this xenograft model recovered its sensitivity to PHA665752.
In conclusion, MET and FGFR2 coactivation may increase resistance to targeted therapy, possibly due to activation of multiple growth and survival signaling pathways. These findings indicate that combination therapy with MET and FGFR2 inhibitors may be a promising strategy to treat inherent resistance to MET inhibitors in cases of gastric cancer harboring MET and FGFR2 amplification. Future studies should be performed to investigate whether similar results could be obtained in an acquired resistant model.