Introduction

Colon cancer is the fourth most common malignancy worldwide (Parkin et al., 2005). It is rapidly increasing in Hong Kong and represents the second most common malignancy in this area. However, the mechanism leading to colon cancer development remains elusive. Hypermethylation of CpG islands in the promoter region of genes associated with gene silencing becomes crucial to the development of colon carcinogenesis (Grady and Carethers, 2008). These abnormalities caused by promoter methylation repress the binding affinity of transcription factors by interfering with transcription initiation, influence chromatin structure, alter gene expression and thereby modulate cancer pathways involved in cell proliferation, apoptosis and metastasis. Increasing numbers of tumor suppressor genes associated with epigenetic alterations have been identified in colon cancers (Zitt et al., 2007; Grady and Carethers, 2008). The identification of new genes functionally involved in tumor development and progression may help to find alternative approaches for diagnostic and therapeutic evaluation. Through methylation-sensitive arbitrarily primed PCR, we have identified a novel preferentially methylated gene, T-box transcription factor 5 (TBX5) (also known as T-box 5), in human cancer.

TBX5 is a member of a phylogenetically conserved family of genes that share a common DNA-binding domain, the T-box. T-box genes encode transcription factors involved in the regulation of developmental processes, cell–cell signaling, negative regulation of cardiac muscle cell proliferation, induction of apoptosis and negative regulation of cell migration (He et al., 2002; Mori et al., 2006; Plageman and Yutzey, 2006; Maitra et al., 2009). This gene is located on human chromosome 12q24.1. The encoded protein participates in heart development and specification of limb identity. However, genetic mutations of TBX5 have been associated with Holt–Oram syndrome, a developmental disorder affecting the heart and upper limbs (Newbury-Ecob et al., 1996). As TBX5 is expressed in colon tissues and the ectopic expression of TBX5 contribute to induction of apoptosis and inhibition of cell proliferation in osteosarcoma (U2OS) and lung cancer (H1299) cells (He et al., 2002), aberrant inactivation of this developmental regulator may be the cause of tumor development. To that end, the TBX5 gene expression patterns in human colon cancer and its molecular mechanism of tumor suppressor function are worth investigating. The potential relationships of TBX5 expression with patient clinicopathological features and survival data are also investigated.

Results

Identification of TBX5 as a novel gene downregulated or silenced in colon cancer cells

To identify candidate cancer genes with promoter methylation, we performed methylation-sensitive arbitrarily primed PCR. A preferentially methylated CpG island of TBX5 promoter was identified that epigenetically silences TBX5 expression in human cancer. We then examined the mRNA expression of TBX5 in eight colon cancer cell lines. Reverse transcript–PCR (RT–PCR) showed that TBX5 transcript was silenced or reduced in seven (88%) cell lines except LS180 (Figure 1b). In contrast, TBX5 expression was readily detected in normal colon tissues (Figure 1b), suggesting an aberrant gene silencing of TBX5 in colon cancers.

Figure 1
figure 1

(a) A typical CpG island (CGI) spans the promoter region of TBX5 exon 1. Each vertical bar represents a single CpG site. The transcription start site is indicated by a curved arrow. A region for COBRA and BGS is shown. BstUI digestive sites are indicated. (b) TBX5 was frequently silenced or downregulated in colon cancer cell lines, but readily expressed in normal colon tissues by semiquantitative RT–PCR. Methylation of TBX5 was determined by COBRA. The undigested fragment (upper band) represents the unmethylated DNA (U). The digested fragments represent the methylated DNA (M). (c) Detailed BGS analysis confirmed the methylation status of the TBX5 in colon cancer cell lines and in normal colon tissues. Five to six colonies of cloned BGS–PCR products from each bisulfite-treated DNA samples were sequenced and each is shown as an individual row, representing a single allele of the promoter CpG island analyzed. One circle indicates one CpG site. Filled circle, methylated; open circle, unmethylated. (d) The mRNA expression of TBX5 was restored after treatment with demethylation agent 5-aza.

Full size image

Promoter methylation of TBX5 was correlated with transcriptional silencing

To elucidate whether silencing of TBX5 is due to the epigenetic inactivation, TBX5 methylation status was examined by combined bisulfite restriction analysis (COBRA). Seven cell lines with decreased or silenced TBX5 expression displayed methylated promoter, including DLD-1, HT-29, LOVO, SW480, SW620, CaCO2 and CL14, whereas no methylation was detected in the normal colon tissues (Figure 1b). We further validated the COBRA results by cloned bisulfite genomic sequencing (BGS) (Figure 1c). The BGS results were consistent with those of COBRA, in which dense methylation was found in methylated cell lines (SW620, HT-29), but not in unmethylated cell line (LS180) and normal colon tissues (Figure 1c).

Demethylation treatment with 5-aza-2′-deoxycytidine restored TBX5 expression

To further reveal whether methylation directly mediates TBX5 silencing, we randomly treated four methylated cell lines that showed silencing or downregulation of TBX5 with 5-aza-2′-deoxycytidine (5-aza). This treatment resulted in the restoration of TBX5 expression in all cell lines examined (Figure 1d), inferring that promoter methylation directly contributed to the TBX5 silencing.

Genetic deletion and mutation of TBX5 were not detected in colon cancer cell lines

Further genetic deletion and mutation analyses of TBX5 coding exons by DNA direct sequencing did not show any homozygous deletion or mutation in eight colon cancer cell lines and 20 primary colon cancers, suggesting that genetic alteration does not contribute to the silencing or downregulation of TBX5 gene in colon cancer.

Ectopic expression of TBX5 suppressed colon cancer cell growth

The frequent silencing of TBX5 mediated by promoter methylation in colon cancer cells but not in normal colon tissue implicated that TBX5 may have a role in tumor growth. To test this speculation, we examined the growth inhibitory effect through ectopic expression of TBX5 in SW620 and CaCO2, which showed no TBX5 expression. Re-expression of TBX5 in the transient transfected SW620 and CaCO2 was confirmed by RT–PCR (Figures 2a1 and a2), which caused a significant decrease in cell viability in both SW620 (P<0.05) and CaCO2 (P<0.05) (Figure 2a3). The suppressive effect on cancer cell proliferation was further confirmed by colony formation assay. A significant reduction of colony numbers was observed in cells stably transfected with pcDNA3.1-TBX5, compared with empty vector in monolayer culture (down to 54–65% of vector controls, P<0.001) (Figure 2b). Thus, TBX5 exhibits growth inhibitory ability in tumor cells and functions as a potential tumor suppressor.

Figure 2
figure 2

(a) TBX5 inhibited tumor cell viability. Colon cancer cell lines (SW620 and CaCO2) were transiently transfected with pcDNA3.1 (Vector), pcDNA3.1-TBX5 (TBX5) or green fluorescent protein plasmid for 48 h. (a1) The transfection efficiency was estimated by green fluorescent protein under a uorescence microscope. (a2) Ectopic expression of TBX5 in SW620 and CaCO2 cell lines was evidenced by RT–PCR. (a3) Ectopic expression of TBX5 in SW620 and CaCO2 significantly suppressed cell viability. (b) TBX5 suppressed tumor cell clonogenicity. (b1) The effect of ectopic TBX5 expression on cancer cell growth was further investigated by colony formation assay. Cells were transfected with pcDNA3.1-TBX5 or control vector, and selected with G418 for 10–14 days, and stained with Gentian Violet. (b2) Ectopic expression of TBX5 in SW620 and CaCO2 was determined by RT–PCR. (b3) Quantitative analysis of colony formation. Values are expressed as the mean±s.d. from three independent experiments. *P<0.05; **P<0.001.

Full size image

TBX5 induced apoptosis

We examined the contribution of apoptosis to the observed growth inhibition of TBX5-transfected cells. Apoptosis was investigated using two-color fluorescence-activated cell sorting analysis. Our results indicated an increase in the numbers of both early apoptotic cells (12.33±0.70 vs 9.97±1.02%, P<0.05) and late apoptotic cells (16.2±1.48 vs 10.9±0.95%, P<0.01) in TBX5-transfected SW620 cells than those in vector-transfected SW620 cells (Figures 3a1 and a2). Induction of apoptosis was further confirmed by the analysis of the expression of apoptosis-related proteins. As shown in Figure 3a3, re-expression of TBX5 enhanced the protein level of the active form of caspase-3, caspase-7 and nuclear enzyme poly-(ADP-ribose) polymerase in the stably transfected SW620 cells.

Figure 3
figure 3

(a) TBX5 induced apoptosis of SW620 cells. Apoptosis was measured by flow cytometry analysis of Annexin V-FITC double-labeled SW620 cells transfected with pcDNA3.1-TBX5 (TBX5) or empty vector (Vector). (a1) Flow cytometry profile represents Annexin V-FITC staining in x axis and propidium iodide in y axis. Dual staining of cells with Annexin V-FITC and propidium iodide enabled categorization of cells into four regions. Region Q1 shows the necrotic cells, Q2 shows the late apoptotic cells, Q3 shows the live cells and Q4 shows the early apoptotic cells. (a2) The experiment was repeated three times and data represent the average of the early apoptotic and late apoptotic cells. *P<0.05; **P<0.001. (a3) Protein expressions of apoptotic-related genes were determined by western blot. β-Actin was used as internal control. (b) TBX5 reduces the migration rates of SW620 cells. Representative image of cell migration of SW620 stable cell lines in scratch wound-healing assay. pcDNA3.1-TBX5 or empty vector-transfected SW620 cells grew up to 90% confluence and were wounded. Photographs were taken at 0, 48 and 96 h, respectively, after the wound was made.

Full size image

TBX5 reduced the migration rates of SW620 cells

We tested TBX5 for another tumor suppressor activity, namely the ability to inhibit cell migration by wound-healing assay. As shown in Figure 3b, a decrease of the cell migration ability in wound closure was observed in SW620 cells transfected with TBX5 as compared with cells transfected with empty vector, as measured after 48 and 96 h of wound healing, indicating that the migration of SW620 cells was inhibited by TBX5.

Identification of genes modulated by TBX5

To gain insights into the molecular basis underlying the tumor suppressive effect of TBX5, gene expression profile in TBX5 stably transfected SW620 was analyzed by cDNA microarray. This array contains 84 functionally well-characterized genes involved in human tumorigenesis. When compared with empty vector-transfected SW620 cells, the antitumorigenesis effect by TBX5 was mediated by regulating important genes in apoptosis, cell proliferation and metastasis (Supplementary Table 1). TBX5 increased the expression of pro-apoptotic extrinsic pathway-related genes, including tumor necrosis factor alpha (TNFα), tumor necrosis factor receptor superfamily member 10b (TNFRSF10B), TNFRSF1A, TNFRSF25, apoptosis executor caspase-8 and apoptotic activator BCL2-associated X protein (BAX). TBX5 also exerted antiproliferative effect by increasing cyclin-dependent kinase inhibitor 2A (CDKN2A) expression and dampening the oncogene synuclein gamma (SNCG) expression. Moreover, TBX5 greatly induced the expression of antimetastasis gene, metastasis suppressor 1 (MTSS1), and downregulated the expression of the pro-metastatic gene, metastasis associated protein 1 family member 2 (MTA2), both of which contributed to the antimetastatic effect of TBX5.

Frequent TBX5 methylation in primary colon cancers

We next examined the TBX5 methylation status in 105 primary colon cancers, 20 adjacent non-tumor tissues and 15 normal colon tissues. Frequent methylation was detected in primary colon cancers (71/105, 68%), but less in paired non-tumor tissues (6/20, 30%) (Figure 4a1) (P=0.005). Further detailed BGS methylation analysis showed densely methylated promoter alleles in tumor, which are rare in paired non-tumor tissue (Figure 4a2). Fifteen normal colon tissues showed no methylation, indicating that TBX5 methylation was tumor specific.

Figure 4
figure 4

(a) TBX5 was frequently methylated in primary colon cancers. (a1) Representative gel images of TBX5 methylation in primary colon cancers and paired adjacent non-tumor tissues by COBRA. M, methylated; U, unmethylated. (a2) Detailed BGS analysis confirmed dense methylation in tumor tissues, but less in paired normal tissues. Filled circle, methylated; open circle, unmethylated. (b) Kaplan–Meier survival curves show that colon cancer patients with TBX5 methylation had poorer survival than those without TBX5 methylation. This difference is statistically significant based on log-rank test (P<0.001).

Full size image

TBX5 methylation was associated with poor survival of colon cancer patients

The association of TBX5 methylation status and clinicopathological characteristics including clinical outcome was analyzed in 80 colon cancer patients with known survival data (median survival time is 54.4 months, range from 2.3 to 65.6 months). There was no correlation between TBX5 methylation and clinicopathological features of colon cancer patients, such as age, gender, tumor staging, lymph node metastasis, distant metastasis and overall tumor, node, metastasis (TNM) staging (Greene et al., 2002) (Table 1).

Table 1 Clinicopathological features of TBX5 methylation in colon cancer
Full size table

The characteristics of colon cancer patients that related to the survival status are shown in Table 2. As expected, tumor staging (P=0.009), lymph node metastasis (P=0.001), distant metastasis (P=0.004) and overall TNM staging (P=0.0004) are significant prognostic factors. Moreover, TBX5 methylation was found to be significantly associated with death (P=0.0003). In univariate Cox regression analysis (Table 3), TBX5 methylation was associated with a significantly increased risk of cancer-related death (relative risk (RR), 6.21; 95% confidence interval, 2.39–16.2; P=0.0002). After the adjustment for potential confounding factors (TNM staging), multivariate Cox regression analysis showed that TBX5 methylation was a predictor of poorer survival of colon cancer patients (RR, 5.42; 95% confidence interval, 2.06–14.4; P=0.0007) (Table 4). As shown in the Kaplan–Meier survival curves, colon cancer patients with TBX5 methylation had significantly shorter survival than others (P<0.001, log-rank test) (Figure 4b).

Table 2 Distribution of patient characteristics by survival status
Full size table
Table 3 Univariate Cox regression analysis of potential poor prognostic factors for colon cancer patients
Full size table
Table 4 Multivariate Cox regression analysis of potential poor prognostic factors for colon cancer patients
Full size table

Discussion

In this study, we used methylation-sensitive arbitrarily primed PCR to identify genes that were differentially methylated in human cancer, in the hope that several of them would prove to be novel candidate tumor suppressors. We showed for the first time that TBX5 is widely distributed in normal colon epithelium, while absent or very lowly expressed in colon cancer cells because of high methylation of the promoter region. Bisulfite sequencing of the TBX5 promoter region showed it to be densely methylated in colon cancer cell lines, whereas there was no methylation in the normal colon DNA. The silencing of TBX5 can be reversed by pharmacological demethylation, but no genetic deletion/mutational inactivation of TBX5 was found in all tumor cell lines examined, inferring methylation is the predominant mechanism for the downregulation of TBX5 gene. These results suggested that TBX5 could be a potential tumor suppressor and its downregulation could have some role in the development of colon cancer.

We characterized the putative tumor suppressor function of TBX5 in human colon cancer cell lines. We cloned a full-length human TBX5 cDNA into pcDNA3.1 vector, and transduced it into silenced colon cancer cells. Restoration of TBX5 in the colorectal cancer CaCO2 and SW620 cells significantly inhibited colony formation and cell proliferation and induced apoptosis. Furthermore, ectopic expression of TBX5 in colon cancer cells decreased migration ability. These results indicate for the first time that TBX5 functions as a tumor suppressor in colon cancer. TBX5 belongs to the member of the T-box superfamily. Members of the T-box gene family have important and diverse roles in development and disease (Papaioannou and Silver, 1998; Finotto, 2008; Nemer, 2008). Mutations in T-box genes are the cause of several congenital diseases and are implicated in cancer (Cai et al., 2005). Our findings have highlighted the importance of TBX5 as a potential tumor suppressor in colon cancer.

We showed the molecular basis that TBX5 exerts the tumor suppressor property in colon cancer through induction of apoptosis, control of cell proliferation and inhibition of cell migration using a cDNA microarray and western blot. We observed that induction of TBX5-mediated apoptosis occurs by the modulation of extrinsic apoptosis pathway and upregulation of pro-apoptotic BAX and Granzyme A. TBX5 induced the expression of key genes mediating extrinsic apoptosis pathway, including extracellular death ligand TNFα, death receptors TNFRSF10B, TNFRSF1A, TNFRSF25 and downstream apoptosis executors caspase-8, caspase-7, caspase-3 and nuclear enzyme poly-(ADP-ribose) polymerase. The extrinsic apoptosis pathway is initiated by the binding of extracellular death ligands such as TNFα to transmembrane death receptors. Engagement of death receptors with their cognate ligands provokes the recruitment of adaptor proteins such as Fas-associated death domain protein (Schütze et al., 2008), which in turn recruits and aggregates several molecules of caspase-8, thereby promoting its autoprocessing and activation. Activation of caspase-8 processes other effector caspase members, including caspase-3 and caspase-7 to initiate a caspase cascade. These effectors further initiate the proteolytic cleavage of the nuclear enzyme poly-(ADP-ribose) polymerase, which caused loss of DNA repair, cellular disassembly and finally undergo apoptosis (Figure 5). Apoptosis induced by extrinsic pathways has been considered to be an important antitumor mechanism (Johnstone et al., 2008; Balkwill, 2009; Holoch and Griffith, 2009). Another important pro-apoptotic activator induced by TBX5 is BAX. It was reported that BAX promotes mitochondrial membrane permeability and cytochrome c release, thereby triggering the formation of the apoptosome and cell death (Taylor et al., 2008). Moreover, ectopic expression of TBX5 leads to the upregulation of Granzyme A (a tryptase) that can induce caspase-independent cell death (Chowdhury and Lieberman, 2008; Cullen et al., 2010). Collectively, these results suggested that TBX5-mediated growth inhibition occurs by modulation of apoptosis by various apoptosis-regulating pathways (Figure 5).

Figure 5
figure 5

Schematic diagram for the molecular basis of TBX5 as a tumor suppressor gene in colon cancer. Ectopic expression of TBX5 suppressed colon cancer cell growth was associated with several biological effects: (1) TBX5 induced tumor cell apoptosis through cross-talk of extrinsic apoptosis pathway, apoptotic BAX and Granzyme A signaling cascades. TBX5 induced the expression of key genes mediating extrinsic apoptosis pathway, including extracellular death ligand TNFα, death receptors TNFRSF10B, TNFRSF1A, TNFRSF25 and downstream apoptosis executors caspase-8, caspase-7, caspase-3 and nuclear enzyme poly-(ADP-ribose) polymerase (PARP); (2) increasing the expression of CDKN2A and inhibiting oncogene SBCG, which in turn suppressed cell proliferation; and (3) upregulating antimetastatic MTSS1 and downregulating pro-metastatic MTA2.

Full size image

The antiproliferative effect caused by TBX5 in colon cancer cells is at least associated with upregulation of CDKN2A, a critical cell cycle inhibitor. It was indicated that CDKN2A abrogates the binding of CDK4/6 kinases to cyclin D, thus preventing the inactivation of retinoblastoma family members and leading to the inhibition of cell proliferation (Kim and Sharpless, 2006). Moreover, the suppression of SNCG by TBX5 contributed to reduced cell proliferation and metastasis. Aberrant overexpression of SNCG had been observed in multiple human cancers, especially in metastatic tumors. Overexpression of SNCG, a potential oncogene, stimulates cancer cell proliferation and metastasis by modulating mitogen-activated protein kinase pathways (Pan et al., 2002; Gupta et al., 2003a, 2003b; Yanagawa et al., 2004; Liu et al., 2007). SNCG is considered to be an important prognostic marker for cancer progression and metastasis in human cancers including colon cancer (Liu et al., 2005; Hu et al., 2009). Inhibition of cancer cell migration by TBX5 may result from upregulating MTSS1 and downregulating MTA2. Similar findings were reported by others that antimetastatic MTSS1 and pro-metastatic MTA2 were correlated with the metastatic potential of certain types of cancers (Hicks et al., 2006; Parr and Jiang, 2009; Toh and Nicolson, 2009). Thus, through modulating these important genes, TBX5 exerted antitumorigenesis effect by promoting apoptosis and suppressing cell proliferation and metastasis. In this regard, upregulation of CDKN2A, MTSS1 and downregulation of SNCG and MTA2 by TBX5 may offer the explanation for the growth and metastasis inhibition in colon cancer cells (Figure 5).

To ascertain the clinical application of TBX5 in colon tumorigenesis in vivo, we examined the promoter methylation of TBX5 by COBRA in 105 primary colon cancers, 20 adjacent non-cancer tissues and 15 normal controls. We found that the TBX5 gene promoter was highly methylated in the majority (68%) of colon cancer tissues compared with adjacent non-tumor tissues (30%), but not methylated in normal controls. Like other cancers, many candidate tumor suppressor genes are reported to be regulated by epigenetic mechanisms (Barton et al., 2008). This study identified that TBX5 methylation is a common event in colon cancer. It is thus worthy further exploring the possible application of TBX5 tumor-specific promoter methylation as an epigenetic biomarker for colon cancer patients. Colorectal cancer varies greatly in clinical outcome, depending on the growth status and aggressiveness of individual tumors. Currently, TNM staging is the most important clinical predictor of patient outcome. However, many patients succumb to disease recurrence. Thus, additional prognostic biomarkers are required to provide better risk assessment. Recognizing the biological functions of TBX5, the inactivation of this gene by promoter methylation would favor tumor progression and a worse outcome. In this regard, we examined the influence of TBX5 methylation on outcome of colon cancer patients. Our results indicated that TBX5 methylation was associated with poorer survival independent of patient characteristics. Our data support an adverse effect of loss of TBX5 expression regulated by promoter methylation on survival of colon cancer patients, providing an additional evidence for the role of TBX5 as a novel candidate tumor suppressor gene in the development of colon cancer.

In conclusion, we have identified a novel functional tumor suppressor gene TBX5 inactivated by promoter methylation in colon cancer lines and found that promoter methylation of TBX5 is frequent and specific in primary colorectal cancers. TBX5 induces tumor cell apoptosis through cross-talk of extrinsic apoptosis pathway, apoptotic BAX and Granzyme A signaling cascades. TBX5 also suppresses tumor cell proliferation and metastasis through upregulation of CDKN2A, MTSS1 and downregulation of SNCG and MTA2, inferring them to be potential targets for the antitumor strategy in colorectal cancer. TBX5 methylation may serve as a potential epigenetic biomarker to predict outcome for colorectal cancer patients.

Materials and methods

Tumor cell lines

Eight colon cancer cell lines (DLD-1, HT-29, LOVO, SW480, SW620, CaCO2, LS180 and CL14) were obtained from the American Type Culture Collection (Manassas, VA, USA). They were cultured in RPMI 1640 medium (Gibco BRL, Rockville, MD, USA) supplemented with 10–20% fetal bovine serum (Gibco BRL) and incubated in 5% CO2 at 37 °C.

Primary tumor and normal tissue samples

A total of 105 colorectal cancers were obtained at the time of operation. The median age of patients was 69 years (range, 36–91 years). Tumor was staged according to the TNM staging system. Twenty of their corresponding adjacent non-cancerous tissues, which were at least 5 cm away from the tumor edge, were obtained from colon cancer patients at the time of surgery. In addition, 15 age- and gender-matched normal colon mucosae from healthy subjects were collected as normal control. The specimens were snap-frozen in liquid nitrogen and stored at −80 °C for molecular analyses. All subjects were given informed consent for obtaining the study materials. The study protocol was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong.

RNA extraction, semiquantitative RT–PCR and real-time PCR analyses

Total RNA was extracted from cell pellets and tissues using TRIzol Reagent (Molecular Research Center Inc., Cincinnati, OH, USA). cDNA was synthesized from 2 μg total RNA using Transcriptor Reverse Transcriptase (Roche Applied Sciences, Indianapolis, IN, USA). For semiquantitative RT–PCR, a 151-bp fragment of the TBX5 gene was amplified using AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA, USA) as reported previously (Yu et al., 2009a), with β-actin as internal control. The primer sequences of TBX5 are as follows: forward, 5′-CTCAAGCTCACCAACAACCA-3′ and reverse, 5′-CAGGAAAGACGTGAGTGCAG-3′. Real-time PCR was performed using SYBR Green master mixture on HT7900 system according to the manufacturer's instructions (Applied Biosystems).

DNA extraction and sodium bisulfite modification

Genomic DNA was extracted from colon cancer cell lines and colon tissues using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) and treated with sodium bisulte using a Zymo DNA Modication Kit (Zymo Research, Hornby, Canada). Sodium bisulfite leads to deamination of unmethylated cytosines to uracils without modifying methylated sites. This allows their differentiation by restriction digestion or sequencing.

Combined bisulfite restriction analysis

The methylation level of the promoter region of TBX5 gene was determined by COBRA, a semiquantitative methylation assay. Hot start PCR amplification with 1.5 μl of bisulfite-treated DNA gives a PCR product of 298 bp, spanning promoter region −902 to −603 bp relative to the transcription start site of TBX5 transcript variant 1. The primer sequences are as follows: forward, 5′-TATTGTAGTTTGGTTGAGAGAAAGGA-3′ and reverse, 5′-CTAAATCTAAACTTACCCCCAACTTC-3′. This region contained 19 CpG dinucleotides and three BstUI restriction sites (Figure 1a). PCR products were digested with BstUI (New England Biolabs, Ipswich, MA, USA) at 60 °C overnight (New England Biolabs). BstUI cleaved the sequence 5′-CGCG-3′, which was retained in the bisulfite-treated methylated DNA, but not the unmethylated DNA. The DNA digests were separated in 10% non-denaturing polyacrylamide gels and stained with ethidium bromide.

Cloned bisulfite genomic sequencing

Cloned bisulfite sequencing was performed to identify the methylation status of 19 CG dinucleotide sites (Figure 1a). The above PCR products for COBRA from three colon cell lines, one colon tumor and its adjacent non-cancer tissue and two normal colon tissues were cloned into pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA, USA). Ten colonies were chosen randomly for plasmid DNA extraction with Qiaprep Spin Mini Kit (Qiagen, Valencia, CA, USA) and were sequenced.

Demethylation with 5-aza

Cells were seeded at a density of 1 × 106 cells per ml. After 24 h, cells were treated with 2 μM of the DNA demethylating agent 5-aza (Sigma-Aldrich, St Louis, MO, USA) for 5 days. 5-Aza was replenished every day. Controlled cells were treated with an equivalent concentration of vehicle (dimethyl sulfoxide). Cells were then harvested for DNA and RNA extractions.

DNA mutation analysis

Genetic deletion and mutation analyses of TBX5 coding exons were performed by DNA direct sequencing. PCR products were purified and sequenced according to the recommendation of the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). Sequence homologies were analyzed using BLAST program of the National Centre for Biological Information (NCBI).

Construction of TBX5 expression vector

Mammalian expression vector pcDNA3.1-TBX5 encoding the full-length open-reading frame of human TBX5 gene was constructed. Briefly, RNA from human tissue (Ambion, Austin, TX, USA) was transcribed into cDNA. Sequence corresponding to the open-reading frame clone of TBX5 was amplified and verified by DNA sequencing. PCR amplified inserts were subcloned into the pcDNA3.1 TOPO TA expression vector (Invitrogen). Plasmids used for transfection were isolated using EndoFree Plasmid Maxi Kit (Qiagen).

Western blot analysis

Total protein was extracted and protein concentration was measured by the DC protein assay method of Bradford (Bio-Rad, Hercules, CA, USA). Thirty micrograms of protein from each sample were used for western blotting. Bands were quantified by scanning densitometry.

Cell viability assay

Cell viability was determined by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium assay (Promega, Madison, WI, USA). Briefly, cells (4.5 × 103 per well) were seeded in 96-well plates and transient transfected with TBX5-expressing vector (pcDNA3.1-TBX5), empty vector (pcDNA3.1) or green fluorescent protein plasmid (1.6 μg per each). The transient transfection efficiency was estimated by green fluorescent protein under a uorescence microscope. After 48 h of transfection, 20 μl of reaction solution was added to cultured cells in 100 μl culture medium and incubated at 37 °C for 1.5 h. The optical density was measured at a wavelength of 490 nm using a VICTOR3 multilabel counter (Perkin-Elmer, Fremont, CA, USA). The cell viability assay was carried out in triplicate wells for three independent experiments.

Colony formation assay

Cells (2 × 105 per well) were plated in a 12-well plate and transfected with pcDNA3.1-TBX5 or empty vector using Lipofectamine 2000 (Invitrogen). After 48 h of transfection, cells were collected and plated in a six-well plate and selected with 0.6 mg/ml of G418 (Calbiochem, Darmstadt, Germany) for 10–14 days, and then with 0.15 mg/ml of G418 for an additional 1 week to establish stable clones. Surviving colonies (⩾50 cells per colony) were counted after fixed with methanol/acetone (1:1) and stained with 5% Gentian Violet (ICM Pharma, Singapore, Singapore) (Yu et al., 2009a, 2009b). The experiment was carried out in triplicate wells for three times.

Annexin V apoptosis assay

Apoptosis was determined by dual staining with Annexin V:FITC and propidium iodide (Invitrogen). Briefly, pcDNA3.1-TBX5 or empty vector-transfected cells (5 × 105 cells per well) were harvested at 48 h post-transfection. Annexin V:FITC and propidium iodide were added to the cellular suspension according to the manufacturer's instructions, and sample fluorescence of 10 000 cells was analyzed using FACSCalibur System (Becton Dickinson Pharmingen, San Jose, CA, USA). The relative proportion of Annexin V-positive cells was determined using the ModFitLT software (Becton Dickinson, San Diego, CA, USA) and counted as apoptotic cells. The assay was carried out in triplicate for three times.

Wound-healing assay

Cell migration was assessed using a scratch wound assay. Briefly, pcDNA3.1-TBX5 or empty vector stably transfected SW620 cells were cultured in six-well plates (5 × 105 cells per well). When the cells grew up to 90% confluence, a single wound was made in the center of cell monolayer using a P-200 pipette tip. At 0, 48 and 96 h of incubation, the wound closure areas were visualized under phase-contrast microscope with a magnification × 100, and the migrated cells were counted. The experiment was performed in triplicate wells for three times.

cDNA expression array

Gene expression profiles of SW620 cells stably transfected with pcDNA3.1-TBX5 or pcDNA3.1 empty vector were analyzed by a commercial gene expression array system named the Human Cancer PathwayFinder RT2 Profiler PCR Array (SABiosciences, Frederick, MD, USA). This array contains 84 functionally well-characterized genes involved in human tumorigenesis (http://www.sabiosciences.com). Gene expression with fold changes ⩾1.5 or ⩽1.5 was considered to be of biological significance.

Statistical analysis

Data were expressed as mean±standard deviation. Non-parametric data between two groups were computed by χ2 test or Fisher's exact test. Multiple group comparisons were made by one-way analysis of variance after Bonferroni's correction or Kruskal–Wallis test where appropriate. The difference for two different groups was determined by Mann–Whitney U-test. The Fisher's exact test was used for comparison of patient characteristics by methylation status and distributions of methylation and covariates by vital status. Patients’ age (at entry of follow-up) by vital status was compared using t-test. RRs of death associated with TBX5 methylation and other predictor variables were estimated from univariate Cox proportional hazards model first. Multivariate Cox models were also constructed to estimate the RR for TBX5 methylation. Overall survival in relation to methylation status was evaluated by Kaplan–Meier survival curve and log-rank test. All analyses were performed using SAS for Windows, version 9, software (SAS Institute Inc., Cary, NC, USA). P-value <0.05 was taken as statistical significance.