Function of the c-Met receptor tyrosine kinase in carcinogenesis and associated therapeutic opportunities

The binding of c-Met by its selective ligand HGF can induce structural changes in c-Met [26]; specifically, its intracellular protein tyrosine kinase (PTK) domain becomes activated, resulting in exposure of the multisubstrate docking site (MDS). Grb2 is then recruited to this site [27]. After autophosphorylation of the PTK domain, it can bind the SH2/SH3 domain of Grb2 [28], which subsequently recruits downstream guanine nucleotide exchange factors (GEFs) such as SOS. Downstream SOS can recruit Ras-GTP from the cell matrix to the membrane and convert it to activated Ras-GTP. Ras successively activates Raf, MEK, MAPKs, ERK, JNK (Jun N-terminal kinase), and p38 (HOG), among others, and the activated MAPKs then enter the cell nuclei to activate transcription factors (e.g. Elk1, Etsl, c-Myc) through phosphorylation. This, in turn, can interfere with the cell cycle and induce cell transformation, consequently promoting carcinogenesis. MAPKs also induce the degradation of proteins and matrix, promote cell migration, and sustain tumor proliferation (Fig. 2) [29, 30].

In tumor cells, the mutation rate of the Ras gene is approximately 25%, whereas in pancreatic cancer and colon cancer, the mutation rates could be 85 and 40%, respectively. Such mutations are predominantly point mutations and gene amplifications [27]. Mutations occur in codons 11, 12, 13,18, 59, and 69, which affect the interaction between Ras and GAP. Upon mutation, its intrinsic GTPase activity is inhibited, which can lead to malignant cell transformation through sustained activation of Ras2GTP (Fig. 2).

When HGF binds c-Met and induces autophosphorylation, the phosphorylated residue acts as a docking site for the heterodimeric PI3K-p85 subunit. Here, the p85 subunit of phosphatidylinositol-3-kinase (PI3K) binds to the adaptor protein at the SH2/SH3 domain, using the same phosphorylated site. When PI3K recruits enough activated receptors, it initiates the phosphorylation of many phosphatidylinositol intermediates. Especially, in many tumor-associated signaling cascades, PI3K can convert phosphatidylinositol-4, 5-diphosphate (PIP2) to phosphatidylinositol-3,4,5-trisphosphate (PIP3). Phosphorylated RTKs can bind the SH2 domain of p85, and subsequently recruit the p85-p110 complex to cell membranes to activate the complex. Activated PI3K accelerates the conversion of PIP2 to PIP3. The association between PIP3 and signaling proteins containing a PH domain, namely, AKT and PDK1, facilitates the phosphorylation of AKT at Thr-308 and at Ser-473 by PDK1 [27]. Activated AKT, which later translocates to cell nuclei, modulates downstream transcription factors like FKHRL1, NF-κB, and Bcl-2, and inhibits the expression of tumor suppressor genes. AKT also phosphorylates GSK-3 and mammalian target of rapamycin (mTOR) or a series of inhibitory proteins such as p21CIP1 and p27KIP1; these, in turn, separately upregulate the expression of Cyclin D, shorten the cell cycle, and ultimately contribute to tumorigenesis [31]. In addition to this, RTKs might also activate the PI3K/AKT pathway through Ras (Fig. 2).

One study found that mTOR can regulate the degradation of extracellular matrix in cancer cells and influence the synthesis and secretion of matrix metalloproteinase; through this mechanism, this protein can also promote the invasion and metastasis of tumor cells [32]. Activated AKT might also phosphorylate nitric oxide synthase to produce NO, which positively regulates angiogenesis (Fig. 2).

The PI3K/AKT/mTOR pathway can modulate the expression of vascular endothelial growth factor (VEGF) and hypoxia inducible factor-1 (HIF-1) through the activation of human double minute 2 (HDM2) (Fig. 2) [33].

In addition, PTEN (phosphatase and tension homology deleted on chromosome 10) negatively regulates phosphorylation in the PI3K pathway. Specifically, this protein facilitates the dephosphorylation of PIP3, converting PIP3 to PIP2. Hence, it relieves the negative regulation of the downstream PI3K components AKT and mTOR. In tumor cells, mutations or deletions in PTEN are common, and enable the increased activation of the PI3K/AKT/mTOR pathway; this leads to aberrant activation of this pathway (Fig. 2).

HGF/c-Met is closely related to Wnt/β-catenin signaling, and promotes tumor proliferation, invasion, and metastasis by modulating this signaling pathway [34]. Studies have shown that in colon cancer and glioblastoma, c-Met expression can enhance Wnt/β-catenin signal transduction, and prevent GSK3β from phosphorylating β-catenin; this, in turn, promotes the translocation of β-catenin to the nucleus, facilitating tumorigenesis. Accordingly, it has been shown that c-Met inhibitors can inhibit Wnt pathway activity in tumor cells [35, 36]. Meanwhile, it has been found that in breast cancer cells undergoing osteolytic bone metastasis, the activation of HGF/c-Met signaling can promote β-catenin translocation to the nucleus and enhance its transcriptional activity. Therefore, HGF/c-Met can exert its biological function through the Wnt signaling pathway (Fig. 2) [37].

In normal cells lacking Wnt pathway activation, β-catenin is cytoplasmic and is phosphorylated at Ser-31, Ser-37, Thr-4, and Ser-45 by GSK3β and CK1 proteins, which are part of the destruction complex. At the same time, it can be acetylated by acetyltransferase p300/CBP-associated factor (PCAF) at Lys-49. Subsequently, these modified sites are recognized by and associate with the β-TrCP E3 ubiquitin ligase, resulting in its degradation by the proteasome, thereby preventing translocation to the nucleus [38, 39]. However, in tumor cells, aberrant activation of the HGF/c-Met pathway and stimulation of Wnt pathway block phosphorylation and acetylation of β-catenin through different signals. This results in the accumulation of β-catenin in the cytoplasm; it then enters the nucleus to displace Groucho, which has a transcriptional inhibitory effect on T-cell factor/lymphoid enhancer factor (TCF/LEF) transcription factors. β-catenin exerts its functions along with BCL9/LGS and Pygo to promote expression of Myc, Cyclin D1, and MMP-7, which facilitates proliferation, invasion, and metastasis (Fig. 2) [38, 40, 41].

Studies have shown that c-Met and RON (receptor originated from nantes) are overexpressed [42] or aberrantly activated in many epithelial-derived malignant cancers [43‐49]. These proteins can be involved in tumorigenesis by promoting cell proliferation, inhibiting apoptosis, enhancing angiogenesis, and promoting metastasis, among other functions, by acting upstream of these processes [46‐49]. c-Met and RON can be activated by HGF and macrophage stimulating protein (MSP), respectively. Activated signaling depends on the tissue availability of adaptor proteins and signaling intermediates or the tendency of the adaptor proteins and signaling intermediates to undergo homodimerization or heterodimerization [50, 51]. MSP and HGF are highly homologous in sequence and structure [52], and are secreted as inactive single chains by multiple tissues and cells including smooth muscle, fibroblasts, adipose tissue, epithelial-derived tumors, liver, lungs, adrenal glands, placenta, and kidney. They are subsequently activated by proteasomal cleavage and form dimeric peptides consisting of α and β chains. In contrast to HGF, the high-affinity RON-binding site (for MSP) is located in the β chain [51].

The dimerization of these two monomers represents a major regulatory mechanism for the activation of tyrosine kinase receptors [53]. In some cases, the formation of a heterodimeric complex permits interaction and crosstalk between different receptors of the same subfamily. The epidermal growth factor receptor (EGFR) family is the best example of a tyrosine kinase receptor that undergoes homo and heterodimerization [54, 55]. Therefore, it is important to study the dimerization mechanism of PTKs. RON and c-Met are co-expressed in many types of tumors and crosstalk between c-Met and RON has been demonstrated [52]. Analysis of their structural homology suggested that they might interact, and in fact, studies have indicated that c-Met and RON can form heterodimers and phosphorylate each other [56]. One study showed that oncogenic addiction to c-Met requires co-expression of RON in four different tumor cell lines [50]. In these cases, RON was constitutively activated, and this was dependent on transphosphorylation by c-Met [50]. Experimentally, it has been shown that c-Met has stronger kinase activity than RON [57], and thus it is possible that heterodimers might be more efficiently activated than RON-RON homodimers. The fact that oncogenic addiction to c-Met requires RON implies that c-Met-RON heterodimers can promote the activation of diverse signaling cascades through different platforms. However, c-Met and RON possess remarkably similar tyrosine-binding sites that serve as docking sites for signaling molecules, and thus these signaling platforms might also be redundant. However, one study found that these two receptors have different kinase activities. Specifically, c-Met can be activated directly through Grb2 binding, but requires modulation for activation by other platforms [58]; in contrast, RON relies mainly on Grb2-associated binder (Gab1), based on the fact that the binding of Gab2 by RON attenuates the recruitment of Gab1 and represses signal transduction.

Grb2 has a unique role with respect to c-MET-RON heterodimers. Although Grb2 inhibits RON autophosphorylation, it enhances this process with c-MET [59]. Considering heterodimers of the EGFR family, the signaling diversity through heterodimers could depend on the relative abundance of each receptor [54].

RON expression might partially modulate c-Met activity, which can be applied when modeling this receptor. With respect to this, we found that knockdown of RON enhances the level and duration of HGF-mediated activation of MAPK and AKT [53]. Although the functional relevance of c-Met-RON heterodimers has not been fully explored, some studies suggest that general knockdown of RON leads to changes in c-Met signaling. For example, it was found that silencing RON in pancreatic cancer cell lines leads to upregulation of c-Met expression and activity [56]. This suggests that inhibitors that co-target or simultaneously block the kinase activities of both c-Met and RON might be clinically useful. However, most studies have not considered the possibility that separately inhibiting either c-Met or RON might lead to compensation by [60] the other.

It has been confirmed that signal transduction between the c-Met and EGFR pathways is closely linked in breast cancer, lung cancer, brain cancer, and other tumors; however, the associated mechanism is still not fully understood [61‐64]. Studies have shown that 70% of EGFR-activating mutations in non-small cell lung carcinoma (NSCLC) are associated with an initial positive response to the EGFR inhibitors gefitinib or erlotinib [65]. However, the vast majority of tumors that respond to EGFR inhibitors achieve acquired resistance [66]. Interestingly, the expression and activation of c-Met are associated with initial resistance and acquired resistance to EGFR inhibitors in patients with NSCLC [66‐68]. Initial resistance might occur through the simultaneous activation of c-Met and EGFR pathways in lung cancer, whereas inhibiting both maximizes the inhibitory effect on the tumor [61]. As such, studies have shown that c-Met might be an effective therapeutic target for overcoming EGFR inhibitor resistance in lung cancer [62].

Possible explanations regarding this mechanism are as follows. One study has already shown that the second mutation in EGFR, T790 M, and the amplification of the MET proto-oncogene will lead to the activation of its downstream ERBB3-initiated PI3K/AKT pathway, resulting in EGFR-TKI acquired resistance [67, 69, 70]. When the c-MET gene is amplified, the two downstream pathways (Grb2/MAPK and PI3K/AKT) are activated by the increase in the number of ERBB3 receptors [69, 70].

In addition, continuous interaction with HGF facilitates c-Met amplification-mediated reversible resistance to EGFR-TKI treatment [66, 70]. When HGF activates Met, it activates MAPK and PI3K/AKT signaling pathways through Gab1, leading to the occurrence of irreversible EGFR-TKI resistance [66].

If EGFR and Met mutations exist simultaneously, drug resistance will be further exacerbated [70]. Therefore, we speculate that c-Met activation of downstream PI3K/AKT and MAPK pathways bypasses EGFR activation because they can both act as tyrosine kinase receptors and activate this pathway (Fig. 3). In addition, c-Met can either directly or indirectly transactivate the PI3K pathway; the fact that c-Met is not activated by this RTK also supports this hypothesis [71].

Another study found that EGFR mutation and Met activation were observed in tumor cells. At the same time, whereas the activation of c-Met was not the result of gene mutation, it resulted in poor prognosis for NSCLC metastasis [68]. In addition, after reversible resistance to EGFR-TKIs in lung cancer cells, HGF can induce an irreversible second mutation (Fig. 3) [66].

HGF/c-Met is activated in approximately 50% of hepatocellular carcinomas (HCC), and expression levels of these proteins are associated with poor clinical prognosis for this disease [72‐75]. Cells with constitutive c-Met activity respond to c-Met inhibition [76]; however, one study found that monotherapy does not completely eliminate tumor growth, suggesting that tumor survival mechanisms that bypass the inhibition of this pathway might be involved in the maintenance of tumor growth in response to these treatments [77].

In previous studies, inhibition of the EGFR pathway was shown to lead to either activation or inhibition of the c-Met pathway, whereas another study showed that c-Met inhibition leads to the activation of the EGFR pathway in a c-Met-positive HCC model [76]. In addition, EGFR inhibitor monotherapies are not significantly effective with respect to in vitro cell viability [76]. c-Met inhibitor monotherapy triggers several survival mechanisms that bypass cell death induced by these agents, including increased expression of the EGFR ligand TGF-α and ErbB3. It has been determined that members of the EGFR family can form homodimers or heterodimers and that different dimers have different signal transduction capabilities; specifically, ErbB3 can heterodimerize with ErbB1 to form one of the most potent dimers [78]. Experiments have shown that c-Met inhibition enhances EGFR signaling by increasing ErbB3 expression [76]. In addition, the increase in TGF-α expression that results from c-Met inhibition, whether this occurs through an autocrine or paracrine mechanism, and its effect on HCC cell survival requires further study.

Onartuzumab is a humanized single-armed specific monoclonal antibody targeting c-Met. The binding of onartuzumab to c-Met is highly specific and this antibody can block c-Met-HGF binding specifically by blocking the HGF α-chain and by forming a complex with the Sema-PSI domain of c-Met [79]; this process occurs without exerting an agonistic activity or triggering c-Met dimerization.

Onartuzumab has been applied as a c-Met inhibitor for the treatment of NSCLC and breast cancer in clinical trials (Table 1) [80], and it proved to be considerably effective. Other studies also found that onartuzumab in combination with erlotinib and placebo is effective for NSCLC. Therefore, this drug might have potential to treat c-Met-overexpressing cancer.