- Top of page
- Materials and Methods
c-Met is often overexpressed in non-small cell lung cancer, but it remains unsolved whether its overexpression leads to its activation. We used an antibody specific to phospho-c-Met (Tyr1235) to investigate c-Met activation immunohistochemically in 130 surgically resected lung adenocarcinomas. The expression of c-Met and hepatocyte growth factor (HGF) was also investigated. Phospho-c-Met was positive in 21.5% (28/130) of cases. c-Met was positive in 74.6% of cases (97/130) and was expressed at high levels in 36.1% of cases (47/130). HGF was expressed at high levels in 31.5% of cases (41/130). Phospho-c-Met was correlated with high levels of HGF (P =0.0010) and high levels c-Met expression (P = 0.0303), but it was also found to be positive in 12 cases with little to no HGF expression. Phospho-c-Met expression was significantly associated with tumor differentiation (P = 0.0023) and papillary histology (P = 0.0011), but not with pathological stage, lymph node metastasis or survival. High levels of c-Met and HGF were also associated with papillary histology (P = 0.0056 and P = 0.0396, respectively), but not with tumor differentiation. Phospho-c-Met was correlated with phospho-Akt (P = 0.0381), but not with phospho-Erk or phospho-Stat3. Phospho-Akt expression was marginally correlated with the expression of phospho-epidermal growth factor receptor (EGFR) (P = 0.0533) and, importantly, it was strongly correlated with the expression of either phospho-c-Met or phospho-EGFR (P = 0.0013). The data suggest that in lung adenocarcinoma tissue, c-Met activation may take place either ligand-dependently or ligand-independently via c-Met overexpression. c-Met activation may play special roles in the papillary subtype and in well differentiated lung adenocarcinomas. (Cancer Sci 2007; 98: 1006–1013)
Lung cancer is the leading cause of cancer death in many developed countries, including the USA and Japan.(1,2) Lung adenocarcinoma, one of the major histological subtypes, is rising in incidence.(3) Patients with lung carcinoma have a poor prognosis: those with stage I disease have a 5-year survival of only 70%.(4) Molecular-targeted therapy provides a new therapeutic modality in this setting and is now under intense investigation. Among the various molecular targets, receptor-type tyrosine kinases such as epidermal growth factor receptor (EGFR) and c-Met show promise for therapeutic intervention.(5)
The c-Met gene was originally identified as a cellular counterpart of the chemically induced oncogene tpr-met isolated from a human osteosarcoma cell line.(6) The c-Met gene encodes a high-affinity receptor for hepatocyte growth factor (HGF).(7) HGF binding augments the intrinsic tyrosine kinase activity of c-Met, resulting in autophosphorylation of several tyrosine residues within the intracellular region.(8,9) The phosphorylation of each tyrosine residue initiates distinct signal transduction cascades involving signaling molecules such as Erk, Akt and Stat3.(8,9)
Several studies have documented overexpression and/or amplification of c-Met in various cancers(8,9) including those of the stomach, colon, liver, prostate, ovary and lung.(10–12) Addition of HGF stimulates cell proliferation, motility and invasion of carcinoma cells.(8,9) Moreover, the abrogation of c-Met signaling by ribozymes, small interfering RNA (siRNA) and pharmacological inhibitors decreases the invasion and metastasis of tumor cells.(8,9) Thus, c-Met appears to be a promising molecular target for the treatment of cancer patients, including those with lung cancer.(8,9)
Although culture studies indicate that c-Met overexpression induces a ligand-independent activation of c-Met,(13) it remains to be seen whether c-Met overexpression leads to c-Met phosphorylation and activation in primary tumor tissues.(14) Inoue et al. recently examined c-Met expression and activation in normal and cancerous gastric tissues.(14) To do so they used a newly developed rabbit polyclonal antiphospho-c-Met antibody that specifically recognizes the phosphorylation of Tyr1235, a tyrosine residue known to play a crucial role in the kinase-induced activation of c-Met.(15) Although c-Met was expressed in both normal gastric mucosa and gastric cancer cells, Inoue et al. detected phospho-c-Met expression only in the latter.(14)
In the present study we used this same antibody to determine the prevalence of c-Met activation in a series of surgically resected lung adenocarcinomas. We then examined the expression of c-Met and HGF by immunohistochemistry, and analyzed the correlations among these three parameters and their correlations with clinicopathological variables. Lastly, we examined the expression of phospho-Stat3, phospho-Akt and phospho-Erk, which are downstream signaling molecules that mediate the biological actions of the HGF/c-Met system(9) and investigated their correlation with phospho-c-Met.
- Top of page
- Materials and Methods
Several studies have reported the overexpression of HGF or c-Met in lung cancer and have investigated the correlations between this overexpression and clinicopathological parameters.(10–12,22,24,25) Tsao et al., for instance, observed the overexpression of c-Met mRNA and protein in 35% of lung adenocarcinomas and confirmed a correlation between c-Met overexpression and higher tumor differentiation.(11) Takanami et al. reported an association between c-Met overexpression and shorter patient survival.(22) At the present time, however, there is no direct evidence that c-Met overexpression actually leads to receptor activation. Ma et al. recently examined c-Met activation immunohistochemically in a limited number of lung adenocarcinomas using a commercially available antiphospho-c-Met antibody.(26) Their results confirmed c-Met activation in some of the adenocarcinomas, as well as in other types of lung cancers investigated in their study. The number of cases analyzed was rather limited, however, and no analyses were undertaken to identify the clinicopathological features of the tumors with c-Met activation.(26)
In the present study we carried out a series of immunohistochemical analyses of lung adenocarcinoma tissues using an antiphospho-c-Met (Tyr1235) antibody recently developed by Inoue et al.(14) Using this antibody, phospho-c-Met was positive in 28 of the 130 cases (21.5%) included in our study, and phospho-c-Met expression correlated with HGF expression (P = 0.0010). We know that some HGF immunoreactivity may represent the inactive form of HGF, given that HGF is secreted as an inactive precursor that requires extracellular activation by proteolytic cleavage.(27) Overall, however, the correlation between phospho-c-Met and HGF expression supports the notion that the HGF/c-Met loop may be activated in an autocrine fashion in cancer cells.(28,29) We were interested to note that phospho-c-Met expression also correlated with high levels of c-Met expression (P = 0.0303), and that 12 cases (9.2%) were positive for phospho-c-Met among tumors with little to no HGF expression. This suggests that some tumors may acquire mechanisms for the ligand-independent activation of c-Met in the same manner as observed in cell lines,(13) probably via c-Met overexpression. Thus, c-Met activation may occur in either a ligand-dependent (autocrine) or a ligand-independent manner in lung adenocarcinoma tissues. This conclusion is supported by our in vitro experiments that demonstrate the constitutive activation of c-Met by a ligand-independent mechanism in a significant proportion of lung adenocarcinoma cell lines (Nakamura Y et al., unpublished data). The findings of this study have important clinical implications from a therapeutic standpoint. Investigators have envisioned approaches to disrupt HGF/c-Met signaling using strategies such as anti-HGF antibody, an NK4 antagonist, ribozymes, a dominant negative c-Met mutant, pharmacological inhibitors and siRNA.(8,9) Among these strategies, the first two may be of only limited effect against tumors with ligand-independent activation of c-Met.
The ligand-independent activation of the c-Met receptor may occur by gene mutation (usually missense mutation) and c-Met overexpression with or without gene amplification.(8,9) Amplification of the c-Met gene has been described for gastric and colonic adenocarcinomas.(8,9) Missense mutations of c-Met have been found in the kinase domains of papillary-type renal cell carcinoma, ovarian cancer, head and neck squamous cell carcinoma, and childhood hepatocellular carcinomas.(9) Ma et al. recently identified missense mutations in the juxtamembrane and extracellular (Sema) domains of c-Met in small-cell carcinoma and adenocarcinomas of the lung.(26,30) A goal for further studies will be to determine the frequencies of mutation and amplification of the c-Met gene in lung adenocarcinoma.
It is of interest that c-Met and phospho-c-Met overexpression were correlated with histological features of adenocarcinoma, such as papillary histology. In thyroid carcinoma, c-Met expression occurs more frequently in papillary carcinoma than in follicular and undifferentiated carcinoma.(31) Germline and somatic activating c-Met mutations are causally linked with hereditary and sporadic forms of papillary renal cell carcinomas, respectively.(8,9) These observations (including ours) strongly suggest that activation of c-Met leads to formation of papillary structures in carcinomas of diverse origins. Because c-Met is involved in branching tubulogenesis of the lung(32) and other organs(8) it is interesting to investigate whether excessive or inappropriate activation of c-Met converts tubules into papillary structure. c-Met activation was also associated with higher differentiation and absence of vascular invasion. Both facts might seem paradoxical, in view of the role of c-Met in cell growth, invasion and metastasis.(8,9) Interestingly, Belfiore et al. reported similar findings for thyroid carcinomas;(31) high c-Met expression was inversely associated with vascular invasion and negative or low c-Met expression was associated with distant metastasis. One explanation would be that in poorly differentiated tumors, molecular abnormalities other than c-Met overexpression and activation may be responsible for the malignant properties of cancer, such as invasion and metastasis. Alternatively, activation of c-Met may require integrity of cell–cell or cell–matrix adhesion, and therefore may occur less efficiently in poorly differentiated tumors.(33–35) c-Met expression was correlated with higher pathological stage and lymph node metastasis, but it had no influence on patient survival in this study. This somewhat puzzling result could be explained if we assume that the association is not causally linked. In fact, the prognostic impact of c-Met expression was shown only by univariate analysis in previous studies.(12,22)
c-Met and phospho-c-Met showed distinct cellular localization in some cases. c-Met was often preferentially localized in basolateral aspects of the cells, whereas phospho-c-Met tended to localize in the apical portion. In this setting, it is interesting that Vadnais et al. observed a localization of phosphorylated c-Met in lamellipodial protrusions in MDCK cells transformed by c-Met overexpression.(36) Trusolino et al. showed that constitutively active c-Met selectively merged with with α4 integrin to form a complex essential for cell migration.(37) This tempts us to speculate that cancer cells stained positively for phospho-c-Met in apical cell membrane represent HGF-stimulated cells with lamellipodial protrusions.
Also interesting was the finding that phosphorylation of either c-Met or EGFR was strongly correlated with expression of phospho-Akt (P = 0.0013). This suggests that EGFR and c-Met play complementary roles in activating the Akt pathway. Activation of Akt is causally related to antiapoptotic characteristics of tumors.(23) It is therefore conceivable that inhibition of the c-Met pathway may provide an alternative therapeutic approach in lung adenocarcinomas with resistance to EGFR inhibitors.
In summary, the results of the present study suggest that the c-Met activation in lung adenocarcinomas may be induced via ligand-dependent (autocrine) and ligand-independent mechanisms, the latter possibly through c-Met overexpression. c-Met activation may play special roles in the papillary subtype and in well differentiated lung adenocarcinomas. c-Met may be an alternative pathway to EGFR in the activation of Akt in lung adenocarcinomas. Further studies on the mechanisms underlying the ligand-independent activation of c-Met are warranted.