E-cadherin as a predictive marker of brain metastasis in non-small-cell lung cancer, and its regulation by pioglitazone in a preclinical model
- First Online:
- Cite this article as:
- Yoo, J.Y., Yang, S., Lee, J.E. et al. J Neurooncol (2012) 109: 219. doi:10.1007/s11060-012-0890-8
- 474 Views
It remains unclear whether patients with non-small-cell lung cancer (NSCLC) develop brain metastasis during or after standard therapy. We attempted to identify biological markers that predict brain metastasis, and investigated how to modulate expression of such markers. A case–control study of patients who were newly diagnosed with NSCLC and who had developed brain metastasis during follow-up was conducted between 2004 and 2009. These patients were compared with a control group of patients who had NSCLC but no evidence of brain metastasis. Immunohistochemical analysis of expression of Ki-67, p53, Bcl-2, Bax, vascular endothelial growth factor, epidermal growth factor receptor, caspase-3, and E-cadherin was conducted. The methylation status of the genes for O6-methylguanine-DNA-methyltransferase, tissue inhibitor of matrix metalloproteinase (TIMP)-2, TIMP-3, and death-associated protein-kinase was also determined, by use of a methylation-specific polymerase chain reaction. A significantly increased risk of developing brain metastasis was associated with the presence of primary tumors with low E-cadherin expression in patients with NSCLC. We also investigated the effects of pioglitazone, a peroxisome proliferator-activated receptor γ-activating drug, in tumor-bearing mouse models. We found that E-cadherin expression was proportional to pioglitazone exposure time. Interestingly, pioglitazone pretreatment before cancer cell inoculation prevented loss of E-cadherin expression and reduced expression of MMP9 and fibronectin, compared with the control group. E-cadherin expression could be a predictor of brain metastasis in patients with NSCLC. Preventive treatment with pioglitazone may be useful for modulating E-cadherin expression.
Lung cancer is the leading cause of cancer deaths in the United States and Europe, and 80 % of cases are diagnosed as non-small lung cancer (NSCLC) [1, 2]. The incidence of brain metastasis seems to be increasing as a result of factors which include an increase in the aging population, improvements in systemic chemotherapy, and increased awareness of the warning signs. However, it is impossible to accurately predict which patients are likely to develop brain metastasis from NSCLC after initial therapy. Only a few studies have analyzed biological markers of brain metastasis from lung cancer [3–5].
The mechanism of metastasis is most likely to be a multistep process combining both genetic and epigenetic changes that activate/inactivate tumor-related genes. Gene silencing by CpG island promoter methylation participates in the genesis of highly malignant cell clones with metastatic capacity . We investigated predictive markers for NSCLC brain metastasis by use of a methylation-specific polymerase chain reaction (PCR) of lung cancer specimens, for example O6-methylguanine-DNA-methyltransferase (MGMT), tissue inhibitor of metalloproteinase-2,3 (TIMP-2,3), and death-associated protein (DAP) kinase [7–10]. We also conducted a study to determine whether immunohistochemical analysis of Ki-67, p53, bcl-2, Bax, caspase-3, vascular endothelial growth factor (VEGF), epidermal growth factor receptor (EGFR), and E-cadherin in patients with NSCLC is associated with a risk of brain metastasis. Ki-67 was chosen as a marker of cell proliferation . P53, bcl-2, Bax, and caspase-3 were chose as markers of apoptosis and cell cycle regulation [12, 13]. The VEGF family of proteins modulates angiogenesis, which is essential for tumor growth and metastasis . EGFR is overexpressed in 40–80 % of NSCLC and is frequently correlated with an adverse prognosis . E-cadherin is expressed in most normal epithelial tissues. Loss of E-cadherin is the characteristic feature of epithelial-to-mesenchymal transition (EMT), a cellular event that occurs during normal embryo development and also may be associated with tumor cell invasion and metastasis .
Peroxisome proliferator-activated receptor γ (PPARγ) is a ligand-activated transcription factor belonging to the nuclear hormone receptor super family. Patients who receive PPARγ-activating drugs (used to treat several million patients with type 2 diabetes) are at significantly lower risk of lung cancer . Activation of PPARγ in NSCLC inhibits growth of NSCLC cells in vitro and in xenograft models [18–21]. Interestingly, combined treatment with the PPARγ agonist pioglitazone and the histone deacetylase inhibitor valproic acid inhibits growth and invasion of prostate cancer through E-cadherin upregulation in prostate cancer xenograft mouse models .
In this study, a candidate marker of brain metastasis was assayed in a tumor-bearing mouse model treated with the PPARγ-activating agent pioglitazone to evaluate the possibility of preventing or therapeutically treating brain metastasis.
Materials and methods
Patients and tumor samples
We identified patients with NSCLC between 2004 and 2009 who had received whole-brain radiotherapy (WBRT) or underwent a craniotomy with tumor resection for treatment of brain metastasis at a single institution, and for whom primary lung cancer pathology samples were available. The inclusion criteria for the study were diagnosis of lung adenocarcinoma or squamous cell carcinoma, adjuvant therapy with WBRT, no other previous or synchronous malignant tumors, and no metastasis to the spine or spinal cord on initial diagnosis.
We identified patients who were newly diagnosed with NSCLC during the same period, did not develop brain metastasis during follow-up, and who had available primary lung cancer pathology samples. Only patients with a negative neurological examination and negative brain imaging studies were included in the control group.
Lung cancer specimens were obtained by needle biopsy or open surgery, including video-assisted thoracoscopy. All tumors were fixed in formalin and embedded in paraffin for routine histopathological examination. Approval for the study was obtained from the Institutional Review Board of the Catholic University of Korea, College of Medicine.
Lung cancer tissue blocks were cut to 4-μm thickness. The sections were deparaffinized, submerged in methanol containing 3 % hydrogen peroxide, blocked, and immunostained by use of the avidin–biotin–peroxidase complex method with diaminobenzidine as the label (Vectastain ABC Elite Kit, Vector Laboratories, Burlingame, CA, USA), in accordance with the supplier’s instructions. The procedure for staining with the Ki-67, p53, bcl-2, Bax, VEGF, EGFR, caspase-3, and E-cadherin antibodies has been described elsewhere [11, 12]. The antibodies used for immunohistochemistry of Ki-67 p53, bcl-2, and Bax were clone PP-67 (1:100; Abcam Cambridge, MA, USA,), clone PAb 240 (1:200; Abcam), clone 100/D5 (1:50; Abcam), and mouse monoclonal antibody to Bax (1:300; Abcam), respectively. Mouse monoclonal antibodies to VEGF (1:100; Santa Cruz Biotechnology, Santa Cruz, CA, USA), EGFR (1:50; AnaSpec, Freemont, CA, USA), and caspase-3 (1:50; AnaSpec) were also used for expression analysis. The antibody used for the E-cadherin immunohistochemistry was cone 36/E-cadherin (1:50; BD Biosciences, Sparks, MD, USA). Counterstaining was performed for 1 min with Mayer’s hematoxylin. Images were acquired by use of an Olympus (Tokyo, Japan) BX41 microscope with a Camedia digital camera.
The immunostained slides were examined by light microscopy by one of the authors (Yoo, JY). Ki-67 and caspase-3 were recorded as labeling indices by counting the number of immunopositive nuclei among 100 tumor cells in at least five representative high-power fields across the slide. Results from p53, bcl-2, and Bax staining were simplified as positive (at least some stained cells) or negative. The immunohistochemical results for VEGF, EGFR, and E-cadherin were scored semi-quantitatively using a four-point scale: 0, no immunoreaction; 1+, faint or equivocal immunoreaction in <20 % of cells; 2+, unequivocal, strong immunoreaction in 20–50 % of cells; and 3+, unequivocal, strong immunoreaction in >50 % of cells. Tumors with 1+, 2+, and 3+ expression were interpreted as positive, and tumors with no expression were interpreted as negative.
DNA preparation and methylation-specific polymerase chain reaction (PCR)
A representative area of the tumor on the block was chosen on the basis of the respective hematoxylin and eosin-stained tissue section. DNA extraction was performed by use of a modified single-step method of DNA extraction. Genomic DNA was chemically modified with sodium bisulfite to covert unmethylated cytosine to uracil, leaving methylcytosine unaltered, by use of the EZ DNA Methylation Kit (Zymo Research, Irvine, CA, USA). The converted DNA was subjected to methylation-specific PCR using two primer sets designed for amplifying the methylated or unmethylated allele of the promoter. PCR was conducted as described elsewhere .
Inoculation of intracranial cancer cells and experimental design
NCI-H358, a human lung adenocarcinoma cell line, was obtained from the American Tissue Culture Collection (Manassas, VA, USA) and maintained in RPMI 1640 media supplemented with 10 % fetal bovine serum (FBS) under conventional tissue-culture conditions.
The nude mice were anesthetized by intraperitoneal (i.p.) injection of 12 mg/kg xylazine (Rompun; Cutter Laboratories, Shawnee, KS, USA) and 30 mg/kg ketamine (Ketalar; Parke-Davis, Morris Plains, NJ, USA). The mice were then stereotactically inoculated with 1 × 106 NCI-H358 cells into the right frontal lobe (2 mm lateral and 1 mm anterior to the bregma, at a depth of 2.5 mm from the skull) by use of a sterile Hamilton (Reno, NV, USA) syringe fitted with a 26-gauge needle and a microinfusion pump (Harvard Apparatus, Holliston, MA, USA). Intracranial tumors were confirmed by cranial magnetic resonance imaging (MRI). All MRI experiments were performed on a 4.7 T animal MRI scanner (BioSpec 47/40, Bruker, Germany) with a quadrature volume coil (diameter, 25 mm) at the Korea Basic Science Institute in Ochang, Korea.
Reverse transcriptase-PCR (RT-PCR)
Total RNA from all specimens was extracted by use of a commercial kit (RNeasy Mini Kit; Qiagen). Total RNA (1 μg) was reversely transcribed by use of the RT-premix (M-Biotech, Seoul, Korea). RT-PCR was performed on cDNA samples by use of a DNA thermal cycler (Bio-Rad) with Go Taq Green Master Mix (Promega, Madison, WI, USA), RNase-free water, and the primers: for E-cadherin, 5′-AAGTGACCGATGATGATGCC-3′ (forward) and 5′-CTTCTCTGTCCATCTCAGCG-3′ (reverse); for 18S rRNA, 5′-CGCGGTTCTATTTTGTTGGT-3′ (forward) and 5′-AGTCGGCATCGTTTATGGTC-3′ (reverse). Expression of MMP9 and fibronectin was assessed by RT-PCR using the primers: for MMP9, 5′-AGTTTGGTGTCGCGGAGCAC-3′ (forward) and 5′-TACATGAGCGCTTCCGGCAC-3′ (reverse); for fibronectin, 5′-TTTTGACAAGAAGCATTATCAGATAA-3′ (forward) and 5′-TGATCAAAACATTTC TCAGCTATTGG-3′ (reverse). RT-PCR products were separated on 1.5 % agarose gel containing ethidium bromide, and visualized with UV light.
The distribution of patient characteristics was compared between the study groups and the control group by use of the Wilcoxon rank-sum test and Fisher’s exact test for continuous and discrete variables, respectively. Different immunohistochemical results between groups were analyzed by use of the χ2 test. Overall survival was analyzed by use of the Kaplan–Meier method, and survival data were compared by use of a log-rank test. A p value of <0.05 was considered statistically significant.
Patient characteristics and outcomes
Squamous cell carcinoma
Source of primary tissue
Surgery + radiotherapy
Surgery + chemotherapy
Chemotherapy + radiotherapy
In the case group, four of 15 patients had brain metastases at the time of primary tumor diagnosis whereas the other patients developed brain metastasis during the observation period. The median time from diagnosis of NSCLC to diagnosis of brain metastasis was 4.2 months (range <1–54 months). A craniotomy was performed for five patients to relieve intracranial pressure. Nine patients received chemotherapy and radiotherapy. Three patients were treated by chemotherapy only. Three patients refused further treatment.
Median overall survival of the patients in the control group has not yet been determined. Median overall survival in the case group was 15 months (p = 0.003) (supplemental Fig. 2). The median follow-up period in the case and control groups was 13 and 32 months, respectively.
Methylation-specific PCR for MGMT, TIMP-2, TIMP-3, and DAP-kinase
A summary of the results from methylation-specific PCR analysis for all the tumors analyzed is shown in the supplemental table. No significant difference in epigenetic alterations was found between the two groups.
E-cadherin regulation of pioglitazone in tumor-bearing mice
To investigate the alteration of EMT, expression of mesenchymal markers, including MMP9 and fibronectin, was assessed. As shown in Fig. 3b. MMP9 expression decreased in the pioglitazone-treated group compared with the control group. Reduction of fibronectin expression was proportional to exposure to pioglitazone. Interestingly, pretreatment with pioglitazone caused reduction of expression of MMP9 and fibronectin. These findings are compatible with the results from real-time PCR (supplemental Fig. 3) and Western blotting (supplemental Fig. 4).
We investigated expression of biomarkers that are reported to be involved in the pathogenesis of brain metastasis. We observed that patients with low E-cadherin in their primary NSCLC tumors were at greater risk of developing brain metastasis than control patients with NSCLC who were diagnosed during the same period. Contrary to previous reports [5, 24, 25], we observed no significant differences in Ki-67, p53, bcl-2, Bax, VEGF, EGFR, or caspase-3 expression between the case and control groups. In addition, promoter methylation status of MGMT, TIMP-2, TIMP-3, and DAP-kinase in paraffin-embedded specimens were no different between the groups. An analysis of 975 patients with early-stage NSCLC revealed that the crude incidence of brain metastasis was 6 % and that brain metastasis developed in 29 % of patients who developed distant metastasis . It is believed that routine brain imaging is not justified because of the low incidence of brain metastasis in asymptomatic patients. Therefore, the need to develop a predictor to identify patients with NSCLC who are at risk of developing brain metastasis has increased.
An epithelial-to-mesenchymal transition (EMT) is required for tumor cells to invade and metastasize . During the EMT, the cell–cell junctions of non-motile, polarized epithelial cells dissolve and the cells are converted into individual, non-polarized, motile, and invasive mesenchymal cells. The function and expression of the epithelial cell–cell adhesion molecule E-cadherin is lost, and expression of the mesenchymal cell–cell adhesion molecule N-cadherin is induced . Loss of the E-cadherin gene or the E-cadherin protein is frequently found during tumor progression in most epithelial cancers . Immunohistochemistry of 54 NSCLC surgical pathology specimens indicated that the percentage of E-cadherin staining is associated with developing brain metastasis (adjusted OR, 3.6) . Together with our findings, this suggests that reduced expression of molecules involved in the EMT, for example E-cadherin, could be a biological predictor enabling identification of patients at high risk for developing brain metastasis.
Overexpression of PPARγ in NSCLC cells has been shown to inhibit metastasis by inducing a differentiated epithelial phenotype . It is reported that activation of PPARγ by its ligands completely inhibits cancer cell EMT in lung and pancreatic adenocarcinoma cell lines, as assessed by expression of epithelial and mesenchymal markers . PPARγ activation prevented transforming growth factor β (TGF-β)-induced transcriptional repression of the E-cadherin promoter and inhibited transcriptional activation of the N-cadherin promoter [33–36]. We hypothesize that preventing the loss of E-cadherin expression by use of PPARγ ligands could be a potential target for blocking the functional phenotype of increased motility and invasion during EMT. Pioglitazone was administered before and after establishing the brain tumor xenograft model, and E-cadherin was expressed in proportion to pioglitazone exposure time. An interesting observation was that pretreatment with pioglitazone before inoculation of cancer cells induced E-cadherin expression compared with the control, and had a synergistic effect on the treatment after cancer cell inoculation. We also showed that expression of the mesenchymal markers MMP9 and fibronectin decreased in the tumor-bearing models after pre and post-treatment with pioglitazone. MMP9, a member of MMPs family, has been correlated with the processes of tumor invasion and metastasis in human cancer, for example degradation of the extracellular matrix and endothelial cell basement membrane. It has been found that fibronectin can induce morphological changes typical of EMT in poorly metastatic lung cancer cells [37, 38]. Reduction of the mesenchymal markers suggests that administration of pioglitazone might prevent or attenuate the EMT process and cell invasiveness. We suggest that brain microenvironments could be affected by pioglitazone pretreatment even before brain metastasis, and that this change could modulate subsequent E-cadherin expression after brain metastasis [39, 40].
Previous trials using PPARγ ligands as monotherapy failed to have a therapeutic benefit for a small number of patients with advanced-stage prostate or breast cancer [41, 42]. Recent studies assessing the use of PPARγ ligands in combination with standard chemotherapeutic agents revealed synergistic effect [43, 44]. NSCLC cell lines with mesenchymal features have been shown to be resistant to gefitinib and erlotinib compared with cells with epithelial features . Taken together with our findings, blocking EMT by PPARγ-activating agents may affect the metastatic capacity of NSCLC in a clinical setting.
This work was supported by a National Research Foundation of Korea Grant funded by the Korean Government (2009-0069346). We thank Se Hoon Kim for the analysis of in-vivo experiments.