Reactivation of the homeotic tumor suppressor gene CDX2 by 5-aza-2′-deoxycytidine-induced demethylation inhibits cell proliferation and induces caspase-independent apoptosis in gastric cancer cells

The DNA methylation inhibitor 5-aza-2′-deoxycytidine (5-aza-CdR) is widely used as an anticancer drug for the treatment of leukemia and solid tumors. Gastric cancer (GC) patients who were positive for caudal type homeobox transcription factor 2 (CDX2) expression showed a higher survival rate compared with those who were CDX2 negative, which suggests that CDX2 performs a tumor suppressor role. However, the molecular mechanisms leading to the inactivation of CDX2 remain unclear. In the present study we demonstrated that the expression levels of CDX2 and DNA methyltransferase enzyme 1 (DNMT1) mRNA were significantly higher in GC compared with distal non-cancerous tissue. The expression of CDX2 mRNA was significantly correlated with Lauren classification, TNM stage and lymph node metastasis. DNMT1 mRNA expression was significantly correlated with TNM stage, pathological differentiation and lymph node metastasis. The expression of CDX2 mRNA was inversely correlated with that of DNMT1 mRNA in GC. Hypermethylation of the CDX2 gene promoter region, extremely low expression levels of CDX2 mRNA and no expression of CDX2 protein were the characteristics observed in MKN-45 and SGC-7901 GC cell lines. Following the treatment of MKN-45 cells with 5-aza-CdR, the hypermethylated CDX2 gene promoter region was demethylated and expression of CDX2 was upregulated, while DNMT1 expression was downregulated. Furthermore, a concentration- and time-dependent growth inhibition as well as increased apoptosis were observed. Caspase-3, −8 and −9 activities increased in a concentration-dependent manner following exposure to different concentrations of 5-aza-CdR. Therefore, our data show that the overexpression of DNMT1 and methylation of the CDX2 gene promoter region is likely to be responsible for CDX2 silencing in GC. 5-Aza-CdR may effectively induce re-expression of the CDX2 gene, inhibit cell proliferation and enhance the caspase-independent apoptosis of MKN-45 cells in vitro.


Introduction
Gastric cancer (GC) is a worldwide health problem with >600,000 cases reported annually. The highest rates occur in Japan, China, Eastern Europe and South America, with 42% of worldwide cases occurring in China (1,2). The occurrence of GC is associated with gene mutations, deletions and other genetic and epigenetic mechanisms, which are the result of the interaction between genetic and environmental factors. Epigenetic modification controls gene expression via DNA methylation, histone modification, chromatin remodeling and non-coding RNAs. Abnormal epigenetic modifications may lead to tumorigenesis (3,4), and their role in the process of tumor formation is likely to be of increasing interest. DNA methylation is a type of reaction using S-adenosylmethionine (SAM) as a methyl donor for converting cytosine into S-methyl cytosine using the DNA methyltransferase enzyme (DNMT) (5). Mizuno et al (6) showed that the expression of DNA methyltransferase 1 (DNMT1) in tumor cells is 4-12-fold higher than that in normal cells, confirming that the DNMT1 increase is involved in tumorigenesis. 5-Aza-2'-deoxycytidine (5-aza-CdR) is a nucleoside analog methylation inhibitor that forms a covalent complex with DNMT1 to inhibit its methyltransferase activity, resulting in less methylation (7).
Caudal type homeobox transcription factor 2 (CDX2), a member of the caudal-related homeobox gene family, plays a key role in early mammalian intestinal development and the maintenance of intestinal epithelia via regulation of intestine-specific gene transcription (8,9). CDX2 expression in normal tissue mainly exists in the small intestinal and colonic mucosa, not in the normal gastric mucosa, but it appears ectopic expression from intestinal metaplasia to the intestinal-type GC (10). Numerous studies indicate that CDX2 performs a tumor suppressor role in human colorectal carcinogenesis (11)(12)(13)(14). Recently, several studies reported that the survival rates for CDX2 expressionpositive GC were significantly higher than those for the negative group, suggesting a possible tumor suppressor role for CDX2 (15,16). However, the molecular mechanisms leading to the inactivation of CDX2 remain unclear.
In the present study, the expression and association of CDX2 and DNMT1 mRNA in GC tissues and normal tissues distal to the GC were analyzed. To further elucidate the molecular mechanisms behind the inactivation of the CDX2 gene, we detected the promoter methylation status by methylation-specific PCR (MSP) and the expression of CDX2 mRNA and protein by real-time fluorescence quantitative polymerase chain reaction (RFQ-PCR) and western blotting in the GC cell lines AGS, MKN-45 and SGC-7901. We then investigated the ability of 5-aza-CdR to induce CDX2 gene re-expression and its effects and mechanisms in GC cell proliferation and apoptosis by treating MKN-45 cells with the demethylating agent 5-aza-CdR in vitro. Our study reveals that the loss of CDX2 function in GC may be attributed to promoter hypermethylation, and that the reactivation of CDX2 by 5-aza-CdR inhibits cell proliferation and induces caspase-3-independent apoptosis in GC cells.

Materials and methods
Tissue samples. Sixty pairs of tissue specimens of GC and the matching distal non-cancerous gastric mucosal tissues (the distance from normal tissue to tumor was >5 cm) were obtained from 60 patients. All patients underwent surgery without preoperative radiation or chemotherapy at the Surgery Department of the Affiliated Hospital of Nantong University (Nantong, China) between January 2009 and August 2011. The tumor and the distal non-cancerous gastric mucosal tissues were snap frozen in liquid nitrogen (N 2 ) and stored at -80˚C until use. All the specimens were diagnosed separately by two pathologists to determine the pathological classification of GC according to the 7th edition of the AJCC cancer staging manual for stomach cancer (17). The detailed profiles of clinical and pathological variables, including age, gender, tumor size, Lauren classification, differentiation and lymph node metastasis status of the patients, are listed in Tables I  and II. The tissue samples were collected with the informed consent of all patients and approval of the ethics committee of the Affiliated Hospital of Nantong University. The mRNA was amplified for 40 cycles and the cycling parameters were: 95˚C for 30 sec, 95˚C for 5 sec and 60˚C for 34 sec. Each measurement was performed in triplicate and the average was calculated. For relative quantification, 2 -∆∆Ct was calculated and used as an indication of the relative expression levels (18).
Western blotting. Western blotting analysis of the CDX2 and internal control β-actin proteins was performed as described previously (19). Monoclonal antibody to CDX2 protein (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was diluted at 1:100.
Analysis of caspase activities. Caspase activities were measured using caspase activity assay kits C1115, C1151 and C1157 (Beyotime) according to the manufacturer's instructions. Briefly, cells were washed with PBS, resuspended in lysis buffer and left on ice for 15 min. The lysate was centrifuged at 20,000 x g at 4˚C for 15 min. The activities of caspase-3, -8 and -9 were measured using substrate peptides acetyl-Asp-Glu-Val-Asp p-nitroanilide (Ac-DEVD-pNA), acetyl-Ile-Glu-Thr-Asp p-nitroanilide (Ac-IETD-pNA) and acetyl-Leu-Glu-His-Asp p-nitroanilide (Ac-LEHD-pNA), respectively. The release of p-nitroanilide (pNA) was qualified by determining the absorbance with a microplate reader (Model 550, Bio-Rad, Hercules, CA, USA) at 405 nm. Each plate contained multiple wells of a given experimental condition and multiple control wells.
Statistical analysis. Results were presented as the mean ± standard deviation. All data were analyzed using statistical software Stata version 11.0 (Stata, College Station, TX, USA). Statistical differences between the groups were analyzed using either one-way ANOVA or Student's t-test. Linear regression was calculated between the expression levels of CDX2 and DNMT1 mRNA in tissues. P<0.05 was considered to indicate a statistically significant result.

Results
Expression of CDX2 and DNMT1 mRNA in GC and distal non-cancerous gastric tissues. The expression levels of CDX2 and DNMT1 mRNA in 60 GC tissue samples and the matching non-cancerous gastric mucosa tissue samples was detected by RFQ-PCR. The expression levels of CDX2 and DNMT1 mRNA were significantly higher in the GC tissues than in the non-cancerous tissues (P<0.05). The expression of CDX2 mRNA was significantly correlated with Lauren classification, TNM stage and lymph node metastasis (all P<0.05). DNMT1 mRNA expression was significantly correlated with TNM stage, pathological differentiation and lymph node metastasis (all P<0.05; Tables I and II). Linear correlation analysis showed that the expression of CDX2 mRNA was inversely correlated with that of DNMT1 mRNA in GC (r=-0.385, P<0.05).
Association between methylation of the CDX2 gene promoter and gene expression in GC cells. The expression levels of CDX2 mRNA and protein in the human GC cell lines AGS, MKN-45 and SGC-7901 were detected by RFQ-PCR and western blotting. The results suggested that CDX2 mRNA and protein were strongly expressed in the AGS cell line but  Fig. 1A and B). MSP analysis revealed that the CDX2 promoter region was fully hypermethylated in the GC cell lines MKN-45 and SGC-7901, however, partial methylation status was detected in the GC cell line AGS (Fig. 1C). In the cell line MKN-45, treatment with the demethylating agent 5-aza-CdR for 72 h at different concentrations (0, 2.5, 5 and 10 µmol/l) induced a partial promoter demethylation (Fig. 1D). . The human GC cell line MKN-45 was treated with different concentrations (0, 2.5, 5 and 10 µmol/l) of 5-aza-CdR for 72 h. The expression levels of CDX2 and DNMT1 mRNA were detected by RFQ-PCR. The results showed that the expression level of CDX2 mRNA was increased by 5-aza-CdR in a concentration-dependent manner (24.65±2.23, 33.59±1.99 and 48.53±1.77, at 2.5, 5 and 10 µmol/l, respectively) compared with those in the control group (0 µmol/l; P<0.05). A comparable result was observed for CDX2 protein, the expression levels of which also increased (1.42±0.01 and 1.86±0.02, at 5 and 10 µmol/l, respectively) in a concentration-dependent manner compared with those in the other two groups (0 and 2.5 µmol/l; P<0.05). However, the expression level of DNMT1 mRNA showed a marked concentration-dependent decrease (1.00±0.01, 0.89±0.18, 0.34±0.14, 0.19±0.09, at 0, 2.5, 5 and 10 µmol/l, respectively) in MKN-45 cells following exposure to 5-aza-CdR for 72 h (P<0.05; Fig. 2).

5-Aza-CdR inhibits MKN-45 cell growth by induction of apoptosis.
To investigate the effect of 5-aza-CdR on the growth of human GC, the MKN-45 cell line was treated with various concentrations of 5-aza-CdR for 96 h and cell viability was detected by the CCK-8 assay. A concentrationand time-dependent growth inhibition of cell proliferation was observed in the MKN-45 cells (Fig. 3, Table III).

A B C D
changes typical of apoptosis in the nuclear chromatin using fluorescence microscopy. This result strongly suggests that apoptosis rather than necrosis was the mechanism of 5-aza-CdR-induced growth inhibition in the MKN-45 cells.

Dicussion
It is generally accepted that the pathogenesis of GC is a multistage process that often takes years, with each stage influenced by environmental, genetic, social and behavioral factors. The characteristics of GC, as Hanahan et al (22,23) described, include sustaining proliferative signaling, evading growth suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, activating invasion and metastasis, reprogramming energy metabolism and evading immune destruction. Oncogenes and the inactivation of tumor suppressor genes are key molecular factors in the tumor microenvironment in gastric carcinogenesis.   CDX2, as an important nuclear transcription factor, has an essential role in the proliferation and differentiation of intestinal epithelial cells in fetal and adult tissues (24). CDX2 is expressed specifically in colonic and small intestinal mucosa and has been implicated in disorders involving abnormal intestinal differentiation and neoplasia (25,26). The use of CDX2 as an immunohistochemical marker has been described previously in studies of human gastric and colonic cancer (27)(28)(29)(30). Several studies have reported that patients with low expression levels of CDX2 in intestinal metaplasia and dysplasia are more likely to progress into GC, and GC patients who were positive for CDX2 expression showed a higher survival rate than those who were CDX2 negative. CDX2 expression levels also gradually during the progression from gastric dysplasia, to early and advanced GCs. In addition, a negative correlation was observed between CDX2 expression and the depth of tumor invasion and lymph node metastasis, suggesting that CDX2 may serve as a powerful predictor for GC (31)(32)(33)(34)(35)(36)(37). A recent study showed that the overexpression of CDX2 was capable of inhibiting cell growth and proliferation in vitro and effectively inhibited GC progression (38). Therefore, CDX2 acts as a tumor suppressor in the upper gastrointestinal tract. However, the expression of CDX2 gradually declined in the process of intestinal metaplasia-dysplasia-GC (34) and the molecular mechanisms leading to the inactivation of CDX2 remain unclear.
In the present study, our results indicate that the expression levels of CDX2 and DNMT1 mRNA were significantly higher in GC tissues than in non-cancerous tissues. The expression of CDX2 mRNA was correlated significantly with Lauren classification, TNM stage and lymph node metastasis. DNMT1 mRNA expression was significantly correlated with TNM stage, pathological differentiation and lymph node metastasis, which is similar to previous findings reported in the literature (39). Linear correlation analysis showed that the expression of CDX2 mRNA was inversely correlated with that of DNMT1 mRNA in GC. It was then hypothesized that the downregulation of CDX2 in GC was likely to be correlated with the hypermethylation of the CDX2 gene promoter region caused by DNMT1 overexpression.
Recent advances in the field of epigenetics have shown that human cancer cells harbor global epigenetic abnormalities in addition to numerous genetic alterations. Among these epigenetic aberrations, DNA hypermethylation is the one which has been the most extensively studied (40). With regard to our findings, MSP analysis revealed that the CDX2 promoter region was fully hypermethylated in the GC cell lines MKN-45 and SGC-7901, however, partial methylation status was detected in the GC cell line AGS. We conclude that promoter hypermethylation of CDX2 is an epigenetic event in GC that may contribute to epigenetic silencing and result in cancer progression and poor prognosis in patients.
The epigenetic silencing of cancer-related genes has proven to be reversible. Therefore, epigenetic alterations are potential targets of interest for molecular targeted therapy in human malignancies (41). In the current study, we have shown that treatment of the GC cell line MKN-45 with different concentrations of the demethylating agent 5-aza-CdR resulted in partial demethylation of the CDX2 promoter region thus allowing the restoration of a potentially silenced gene expression, while DNMT1 showed a marked concentration-dependent decrease following the exposure of MKN-45 cells to different concentrations of 5-aza-CdR for 72 h. The results of the present study clearly show that the transcriptional inactivation of CDX2 was due to hypermethylation. Notably, MKN-45 cells were treated with 5-aza-CdR at various concentrations and the CCK-8 assay showed that a concentration-and time-dependent growth inhibition of cell proliferation occurred in the MKN-45 cells. Apoptosis analysis revealed that 5-aza-CdR treatment increased the proportion of apoptotoc cells, and morphological changes typical of apoptosis were observed in the nuclear chromatin.
Previous studies have reported (42)(43)(44)(45)) that the intrinsic mitochondrial or death receptor pathways are able to trigger cell apoptosis, and the mitochondrial apoptotic pathway, including caspase-dependent or independent apoptosis. Caspases are able to cleave essential cellular substrates after aspartic residues and are critical for the initiation and execution phases of apoptosis. Caspase-8 is involved in the death-receptor pathway while caspase-9 mediates the mitochondrial pathway. Once activated, caspase-8 and caspase-9 activate downstream caspase-3, triggering cell apoptosis. Therefore, we measured the activities of caspase-3, caspase-8 and caspase-9 and the results showed that they were all activated, suggesting that the mitochondrial and death-receptor pathways were involved in 5-aza-CdR-induced apoptosis. However, compared with caspase-8, the activity of caspase-9 was higher in response to 5-aza-CdR, suggesting that the apoptosis proceeded mainly via a caspase-dependent intrinsic mitochondrial pathway.
In summary, the results of the current study suggest the detection of CDX2 and DNMTl mRNA will be beneficial in predicting the GC histological type and patient progression, and  may also be used as markers in the assessment of the biological behavior of GC. Furthermore, our results confirmed that, in the GC cell line MKN-45, the transcriptional inactivation of CDX2 was due to hypermethylation and the high level of DNMT1. Treatment with a DNMT1 inhibitor rescued the expression of CDX2, inhibited cell proliferation and induced caspaseindependent apoptosis. Therefore, our study further emphasizes the importance of the CDX2 gene in gastric carcinogenesis and progression, and a better understanding of DNA methylation is likely to provide us with a potential therapeutic target for GC.