In this present study, we demonstrated a potent inhibitor of c-Met and its excellent cytotoxic effect on c-Met over-expressed gastric cancer cells. 10–20 % of gastric cancer tissues and 40 % of the scirrhous histological subtype have been known to harbor amplified
c-
met gene [
14,
33,
34]. Our western blot also shows that 27 % (5/18) of gastric cancer cell lines have c-Met overexpression (Fig.
2a). Because cells with c-Met overexpression are known to be addicted to c-Met signaling, the strategy targeting c-Met may be a promising therapeutics for many of gastric cancer patients. Previous studies demonstrated c-Met inhibition, by siRNA or small molecule inhibitor, blocks the proliferation of
c-
met amplified cancer cell [
12,
13,
35,
36]. We synthesized hundreds of compounds and performed in vitro biochemical assay. Our compounds have pyridoxazine instead of quinoline in PF-04217903 [
30]. X-ray crystallography indicates that nitrogen of quinoline in PF-04217903 plays a role in hinge binding in a mono-binding manner. Based upon our molecular docking study, we synthesized triazolopyrazine compounds having pyridoxazine as a dual-binder to hinge region. By introducing pyridoxazine, we escaped the patent conflict with PF-04217903. KRC-00509 and KRC-00715 showed best efficacy in enzyme assay among the synthesized ones (Table
1, Additional file
1: Figure S1). Kinase panel assay demonstrated KRC-00715 is exclusively selective to c-Met like PF-04217903 (Table
2). It didn’t inhibit any tyrosine kinase at all at 1 μM. Therefore we expect our compound may reduce side effects to a minimal level in clinical trial. Our cytotoxic data supports this expectation by showing that cell lines with low level of c-Met weren’t suppressed by our compounds at all even at as high as 5 μM (Fig.
2). However, crizotinib, which is a multi-kinase inhibitor, cause cytotoxic effect on several c-Met negative cell lines to some extent. It implicates that crizotinib may cause side effects in clinical trials, and actually several clinical reports demonstrated it cause serious side effects including dermatitis, visual disturbance, heart problem, and so on [
37,
38]. Table
1 indicates the cytotoxic effects of KRC-00509 and KRC-00715 on Hs746T were as good as crizotinib. Namely, our compounds, which inhibit c-Met exclusively, have the same cytotoxicity on c-Met-addicted cells as crizotinib which inhibits multi-kinases including c-Met. That’s why we think our compound may be a superior therapeutic agent to crizotinib for c-Met targeted therapy. Figure
3 shows why c-Met inhibitors had effects only on c-Met over-expressed cells not on c-Met low-expressed cells. Downstream signals, such as Akt, and Erk, were diminished by c-Met inhibitors only in c-Met over-expressed cells (Figure
3, Additional file
1: Figure S2). In addition, total cellular tyrosine phosphorylations were diminished by c-Met inhibitors only in c-Met over-expressed cells. (Additional file
1: Figure S3) These phenomena can be explained by ‘oncogene-addiction’. That is, c-Met over-expressed gastric cancer cells are ‘addicted’ to c-Met signaling. Therefore, c-Met-targeted therapeutics is quite relevant to gastric cancer patients with c-Met overexpression. One thing interesting is that c-Met inhibitors induced G1/S arrest in c-Met-addicted cells (Fig.
4) All of the successful agents for cancer targeted therapy, such as imatinib, sorafenib and gefitinib, have shown strong G1/S arrest to result in apoptosis [
39‐
41]. If we get to know the exact mechanism how these agents cause G1/S arrest, it may widen our perception of cancer therapy. To see the in vivo efficacy of our compounds, we used mouse Hs746T xenograft model. KRC-00715 shows the significant tumor growth inhibition at doses of 50 mpk without loss of weight. However, we suspended the in vivo experiment with KRC-00509 because of the compound’s severe toxicity to mouse.