Abstract

Background and Aims: The expression of pancreatic-duodenal homeobox 1 ( PDX1 ) in gastric cancer is aberrantly reduced. The aim of this study was to elucidate the regulation of DNA methylation and histone acetylation at the promoter for PDX1 silencing in gastric cancer. Methods:PDX1 expression in response to demethylation and acetylation was detected in human gastric cancer cell lines by reverse transcription–polymerase chain reaction (PCR) and western blot. Four CpG islands within the 5′-flanking region of PDX1 gene were analyzed with their transcription activities being detected by dual luciferase assay. Promoter hypermethylation was identified in gastric cancer cell lines and cancer tissues by methylation-specific PCR or bisulfite DNA sequencing PCR analysis. Histone acetylation was determined by chromatin immunoprecipitation (ChIP) assay. Results: Demethylation by 5′-aza-2′-deoxycytidine (5′-aza-dC) and/or acetylation by trichostatin A (TSA) restored PDX1 expression in gastric cancer cells. Hypermethylation was found in four CpG islands in six of seven cancer cell lines. However, only the distal CpG island located in the promoter fragment of PDX1 , F383 (c.−2063 to −1681 nt upstream of the ATG start codon) displayed significant transcriptional activity that could be suppressed by SssI methylase and increased by 5′-aza-dC and TSA. More than 70% of the single CpG sites in F383 were methylated with hypermethylation of F383 fragment more common in gastric cancerous tissues compared with the paired normal tissues ( P  < 0.05). ChIP assay showed F383 was also associated with low hypoacetylation level of the histones. Conclusion: Promoter hypermethylation and histone hypoacetylation contribute to PDX1 silencing in gastric cancer.

Introduction

In mammals, epigenetic regulation including DNA hypermethylation and histone modification represents the major epigenetic mechanisms implicated in regulation of gene transcription ( 1 ). DNA methylation within the promoter of tumor suppressor genes is a common phenomenon in cancer cells. It results in the transcriptional silencing of these genes and promotes tumor development ( 1–4 ). Promoter hypermethylation has been reported to induce aberrant reduction of some homeobox genes such as aristaless-like homeobox-4 ( 5 ), CDX1 ( 6–8 ) and CDX2 ( 9 ) in cancers including squamous esophageal cancer, hematologic malignancies and colorectal cancers.

Histone acetylation/deacetylation alters the status of open chromatin domains and thus affects gene transcription. This process is modulated by histone acetyltransferases and histone deacetylases (HDACs). Loss of histone H3 and H4 acetylation attributes to the imbalanced recruitment of HDAC and results in transcription repression of tumor suppressors in cancers ( 4 ).

Pancreatic duodenal homeobox-1 ( PDX1 ) is a homeobox gene that belongs to the ParaHox subfamily ( 10 , 11 ). PDX1 plays a critical role in embryogenesis and tissue differentiation ( 12–17 ). During the development of mice embryos, deletion or mutation of pdx1 gene allele resulted in pancreatic malformation ( 16 , 17 ). In pdx1 -null mice embryos, the malformations of the stomach–duodenum junction caused the lack of gastric emptying and subsequent stomach distension ( 15 ). The cell-type specific expression of PDX1 is controlled by 5′-flanking sequences that contain several conserved promoter regions ( 18–20 ). Areas (−2852 to −2547 nt) and II (−2247 to −2071 nt) direct endocrine cell expression ( 18 , 19 ), whereas area III (−1973 to −1694 nt) directs the transient expression in β-cells ( 21 , 22 ). Our previous work has implicated PDX1 as a putative tumor suppressor with reduced expression in gastric cancer ( 23 , 24 ).

In this project, we investigated the epigenetic regulatory mechanism of PDX1 gene in gastric cancer. We showed that both promoter hypermethylation and histone deacetylation accounted for PDX1 silencing. Unlike the responsive sequences around the translation starting codon identified in β-cells ( 25 ), a functional fragment, F383, located at area III (−2063 to −1681 nt) mediated the epigenetic regulation of PDX1 in gastric cancer cells.

Methods and materials

Cell lines and tissue samples

Human gastric cancer cell lines AGS, KATOIII and SNU1 were from the American Type Culture Collection (Manassas, VA). SGC7901 was from the Academy of Military Medical Science (Beijing, China) and TMK1, MKN45 and BCG823 were maintained in our laboratory and described in the previous study ( 23 , 26 , 27 ). All cell lines were grown as described previously ( 23 ). Thirty pairs of human gastric cancerous tissues with the adjacent normal tissues were collected from the specimen archives of the Department of Gastroenterology, the Affiliated Hospital of Sun Yat-sen University. CpGnome universal methylated DNA (Invitrogen, Carlsbad, CA) was used as the positive control for methylation.

Reverse transcription–polymerase chain reaction

As per the manufacturer’s protocol, Mini-RNease RNA extraction kit (Qiagen, Hilden, Germany) and Thermoscript RT system reagent (Gibco BRL, Gaithersburg, MD) were used for RNA extraction and reverse transcription. Polymerase chain reaction (PCR) was performed using 2 μl complementary DNA template, 0.2 U Hot-start Taq DNA Polymerase, forward and reverse primers and deoxynucleoside triphosphates mix in a final volume of 20 μl. The sequences of PDX1 primers were shown in our previous publication ( 23 ). And PCR was carried out for 35 cycles with 94°C denaturation for 30 s, 56°C annealing for 30 s and 72°C elongation for 45 s.

Western blot

Whole cell lysates were prepared and assayed for protein expression by western blot as reported previously ( 23 ). Immunoblots were developed and visualized by the enhanced chemiluminescence system (Amersham, Piscataway, NJ) according to the manufacturer’s protocol. Glyceraldehyde-3-phosphate dehydrogenase was used as the internal loading control.

5-Aza-2′-deoxycytidine and trichostatin A treatment

Cancer cells were seeded in six-well plates at a density of 2 × 10 5 cells per well 12 h before the drug treatment. The cells were treated with the following drugs alone or in combination: 5′-aza-2′-deoxycytidine (5′-aza-dC) (5, 10 or 20 μM; Sigma, St Louis, MO) and trichostatin A (TSA) (500 nM; Sigma). Cells were harvested after 24 h. Double-diluted water was used as the dissolvent control.

Construction of putative promoter fragments

To indentify functional CpG islands and evaluate the association of DNA methylation and promoter transcription, the CpG islands of the PDX1 promoter had to be identified. We searched PDX1 genomic sequence at −2000 to +600 nt using Promoter Database ( http://rulai.cshl.edu/cgi-bin/TRED/tred.cgi?pid=10745&process=refreshSeq&upstream_offset=-2000&downstream_offset=600&reset=Refresh ) and found four enriched CpG islands by MethPrimer ( http://www.urogene.org/methprimer/index1.html ). The four islands were, respectively, located at −1975 to −1704 nt (island 1), −1226 to −1066 nt (island 2), −818 to −694 nt (island 3) and −112 to +521 nt (island 4) ( Figure 2a ). Then, four fragments containing these putative CpG islands were amplified and inserted into the firefly luciferase reporter vector, pGL3basic (Promega, Madison, WI). A XhoI site was added into the 5′ terminus of the forward primers and HindIII site was added to the reverse primer. With the upstream nucleotide adjacent to the ATG translation start codon, henceforth defined as c.1, the four reporter constructs were named F383 (−2063 to −1681 nt), F314 (−1320 to −1007 nt), F283 (−864 to −583 nt) and F724 (−170 to +545 nt), respectively. The cloned fragments were all verified by restriction enzyme digestion and sequencing. All primers for construction of the PDX1 promoter were designed according to the genomic sequence of the PDX1 gene (accession no. NC_000013.9) published in GenBank with their sequences being listed in supplementary Table 1 , available at Carcinogenesis Online.

Transient transfection, in vitro methylation and luciferase reporter assay

Reporter constructs F383, F314, F283 and F724 (0.8 μg/well) were transfected into cells pre-seeded into 24-well tissue culture plate by Lipofectamine 2000 (Invitrogen). Empty pGL3basic vector was used as control, whereas PRL-CMV (0.008 μg/well; Promega) was used to normalize the reporter gene. Twenty-four hours after transfection, cells were harvested and firefly luciferase activities were measured using the Dual-Luciferase reporter assay system (Promega) with a model TD-20/20 Luminometer (EG&G Berthold, Australia). The value of firefly luciferase activity was normalized to that of renilla luciferase. Promoter activity was presented as the fold change of relative luciferase unit (RLU) compared with the basic vector control. RLU equals the value of the firefly luciferase unit, divided by the value of the renilla. Each experiment was performed in triplicate, and at least three sets of independent transfection experiments were performed. SssI methylase, which is capable of non-specifically methylating all CpG dinucleotides, was used for the in vitro methylation of putative promoter fragments. In each case, 3 μg of plasmid DNA (triplicates in 24-well plate), pGL3basic and all reporter constructs were incubated in a 20 μl reaction system containing 4 U of SssI methylase, 2 μl of 20× S -adenosylmethionine and 2 μl of 10× NEBuffer for 4 h at 37°C followed by 20 min at 65°C to stop the reaction.

Genomic DNA isolation, bisulfite DNA sequencing PCR analysis and methylation-specific PCR analysis

Genomic DNAs were isolated from cells or resected specimens using DNeasy Blood & Tissue Kit (Qiagen) according to the manufacturer’s instructions. DNA samples were treated with sodium bisulfite to convert cytosine to uracil. Briefly, 1.0 μg of genomic DNA from each sample was denatured with 2 mol/l NaOH at 37°C for 10 min, followed by incubation with 3 mol/l sodium bisulfate (pH 5.0; Sigma) at 50°C for 16 h, Two microliters of 20 μl bisulfite-treated DNA was amplified at 95°C for 20 min and then 40 cycles at 94°C for 30 s, 52°C for 30 s and 72°C for 45 s and a final extension at 72°C for 10 min using the bisulfate-treated DNA sequencing PCR primers ( supplementary Table 2 is available at Carcinogenesis Online). For cell lines, PCR products were purified and then inserted into pGEM-T vector (Promega); 10–12 white clones were selected for each sample and then sequenced to determine the aberrant methylation of each CpG island. Moreover, two normal gastric tissues were used as normal controls.

Methylation status of human gastric cancerous tissues and paired normal tissues was evaluated by methylation-specific polymerase chain reaction (MSP) analysis. After isolation and modification with sodium bisulfite, genomic DNA samples were amplified with methylation- or unmethylation-specific primers with their sequences listed in supplementary Table 2 , available at Carcinogenesis Online. The total 25 μl PCR reaction contained 10 pmol/l primers, 25 μmol/l deoxynucleoside triphosphates, 2 μl of bisulfate-treated DNA, 0.5 U of hot-start Taq polymerase and the respective buffers. The condition for PCR amplification was as follows: 20 min hot start at 95°C followed by 40 cycles at 94°C for 30 s, 52°C for 30 s, 72°C for 45 s and a final extension at 72°C for 10 min. PCR products were visualized on a 2% agarose gels stained with ethidium bromide. Moreover, a normal gastric tissue was used as normal control.

Chromatin immunoprecipitaion and quantitative real-time reverse transcription–PCR

Evaluation of histone acetylation levels was performed using a chromatin immunoprecipitation (ChIP) assay with a commercially available kit (Upstate Biotechnologies, Charlottesville, VA) according to the manufacturer’s protocol. Approximately 2 × 10 6 cells were treated with 1% formaldehyde to cross-link the protein to DNA for 15 min at 37°C. After washing, the cell pellets were resuspended in lysis buffer and sonicated to yield an average DNA size of ∼500 bp. The sonicated cells were subsequently deposited by centrifugation and diluted with ChIP dilution buffer. Twenty microliters of the diluted lysates were left as the input control. Samples were incubated with protein A-Sepharose and sonicated salmon sperm DNA for 1 h at 4°C, then incubated with 5 μg specific antibodies against acetyl-histone H3 (lys 9 and 14) and acetyl-histone H4 (lys 5, 8, 12 and 16) (Upstate Biotechnology) or normal rabbit IgG (Santa Cruz Biotechnology, CA) overnight at 4°C followed by incubation with Protein A-Sepharose for 1 h. After washing, the sepharose–antibody–histone complexes were eluted with elution buffer (1% sodium dodecyl sulfate and 0.1 M NaHCO 3 ). The cross-linking was then reversed by 5 M NaCl at 65°C for 4 h. The DNAs were extracted with phenol–chloroform, precipitated with ethanol and applied for quantitative analysis by quantitative real-time reverse transcription–PCR.

Quantitative real-time reverse transcription–PCR were performed for 60 cycles with 94°C for 15 s, 58°C for 15 s and 72°C for 40 s by using Applied Biosystems Sequence Detection System 7900 (ABI Prism 7900HT; Applied Biosystems, Foster City, CA) in 20 μl mixture including 10 μl power SYBR GREEN PCR Master Mix (Applied Biosystems). Inspection of melt curves and Ct values generated were used for quantification of the copy numbers of four fragments and the results were expressed as the fold change compared with that obtained with IgG. The primer sequences are listed in supplementary Table 3 , available at Carcinogenesis Online. The amount of immunoprecipitated DNA was normalized to the input DNA. Each treatment was set up in triplicate, and two independent ChIP experiments were performed.

Statistical analysis

The results of luciferase activities are expressed as mean ± SEM. Student’s t -test and Fisher’s exact test were used to determine the statistical significance between different groups by the statistical software, SPSS13.0. A P value of <0.05 was considered significant.

Results

Restoration of PDX1 expression in cancer cells by demethylation and acetylation

We have reported that PDX1 expression in gastric cancer cells is downregulated ( 23 ). To examine if epigenetic regulation was involved in PDX1 silencing, we first treated seven gastric cancer cell lines with 5′-aza-dC. As shown in Figure 1A , the endogenous messenger RNA expression of PDX1 in gastric cancer cells was weak and partly upregulated by 5′-aza-dC in these cell lines. Similarly, PDX1 expression could be upregulated by the HDAC inhibitor, TSA ( Figure 1B ). Furthermore, 5′-aza-dC and TSA could increase PDX1 expression in a dose-dependent manner ( Figure 1C ). To evaluate the interaction of 5′-aza-dC and TSA on PDX1 expression, we randomly chose the two cell lines TMK1 and KATOIII. As shown in Figure 1D and E , a slightly synergistic effect was observed when cells were treated with combination of 5′-aza-dC and TSA. These results suggested that PDX1 expression in gastric cancer cells was modulated by epigenetic regulation that included both DNA methylation and histone deacetylation.

Fig. 1.

Restoration of PDX1 expression in cancer cells by demethylation and acetylation. ( A ) Gastric cancer cell lines were treated with double-diluted water , the dissolvent control (C) or 5 μM 5′-aza-dC (A) for 24 h. ( B ) Gastric cancer cell lines were treated with double-diluted water (C) or 500 nM TSA (T) for 24 h. PDX1 messenger RNA expression was detected by reverse transcription–PCR with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) being used as the internal control. ( C ) Cancer cells were treated with the indicated concentration of 5′-aza-dC or TSA for 24 h. PDX1 messenger RNA expression was detected by reverse transcription–PCR with GAPDH being used as the internal control. ( D and E ) TMK1 and KATOIII cells were treated with 5′-aza-dC and TSA alone or in combination for 24 h, PDX1 messenger RNA and protein expressions were detected by reverse transcription–PCR (D) and western blot (E). GAPDH was used as the internal control. All of these pictures were representatives of three to four independent experiments with similar findings.

Fig. 1.

Restoration of PDX1 expression in cancer cells by demethylation and acetylation. ( A ) Gastric cancer cell lines were treated with double-diluted water , the dissolvent control (C) or 5 μM 5′-aza-dC (A) for 24 h. ( B ) Gastric cancer cell lines were treated with double-diluted water (C) or 500 nM TSA (T) for 24 h. PDX1 messenger RNA expression was detected by reverse transcription–PCR with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) being used as the internal control. ( C ) Cancer cells were treated with the indicated concentration of 5′-aza-dC or TSA for 24 h. PDX1 messenger RNA expression was detected by reverse transcription–PCR with GAPDH being used as the internal control. ( D and E ) TMK1 and KATOIII cells were treated with 5′-aza-dC and TSA alone or in combination for 24 h, PDX1 messenger RNA and protein expressions were detected by reverse transcription–PCR (D) and western blot (E). GAPDH was used as the internal control. All of these pictures were representatives of three to four independent experiments with similar findings.

Hypermethylation of PDX1 promoter in gastric cancer cell lines

To define the putative-responsive sequence for epigenetic regulation, we first searched a 2600 bp genomic sequence including the ATG translation starting codon (−2000 to +600 nt) of the PDX1 gene using Methprimer software ( http://www.ucsf.edu/urogene/methprimer ). Four putative CpG islands were found located at −1975 to −1704 nt, −1226 to −1066 nt, −818 to −694 nt and −112 to +521 nt, respectively ( Figure 2A ). We then examined the methylation status of these CpG islands in seven gastric cancer cell lines by MSP assay. Complete methylation in all four CpG islands was found in SGC7901, AGS, TMK1 and SNU1 cells, whereas partial methylation was found in all CpG islands except F383 in BCG823 and MKN45 cells. However, we showed no DNA methylation of all four islands examined in KATOIII cells as well as in normal gastric tissue ( Figure 2B ). To confirm the accuracy of the MSP methodology, we treated TMK1 cells with 5′-aza-dC. We found that the complete methylation in all four CpG islands was successfully reversed into partial methylation ( Figure 2C ). These results suggested that hypermethylation of PDX1 promoter existed in most gastric cell lines.

Fig. 2.

Hypermethylation of CpG islands of the PDX1 gene in cancer cell lines. ( A ) The schematic diagram of putative CpG islands within the PDX1 promoter identified by bioinformatics analysis. The axis is the percentage of the dinucleotide, guanine and cytosine. The genomic DNA sequences (−2000 to +600 nt) of PDX1 was analyzed by Methprimer software. Locations of four putative CpG islands were displayed in areas with gray shadow. ( B ) MSP result of four putative CpG islands in seven gastric cancer cell lines and one normal gastric tissue. M, methylated; U, unmethylated. ( C ) Alteration of methylation status in TMK1 cells induced by 5′-aza-dC.

Fig. 2.

Hypermethylation of CpG islands of the PDX1 gene in cancer cell lines. ( A ) The schematic diagram of putative CpG islands within the PDX1 promoter identified by bioinformatics analysis. The axis is the percentage of the dinucleotide, guanine and cytosine. The genomic DNA sequences (−2000 to +600 nt) of PDX1 was analyzed by Methprimer software. Locations of four putative CpG islands were displayed in areas with gray shadow. ( B ) MSP result of four putative CpG islands in seven gastric cancer cell lines and one normal gastric tissue. M, methylated; U, unmethylated. ( C ) Alteration of methylation status in TMK1 cells induced by 5′-aza-dC.

Identification of a functional CpG island within the 5′-flanking region of PDX1 gene

To determine if the sequences contained by CpG islands were critical for PDX1 transcription, we cloned these fragments into pGL3basic vector to generate four luciferase reporter constructs, F383, F314, F283 and F724. After transient transfection into cells, a dual luciferase assay showed that F383, the most distal fragment displayed substantial promoter activity in all three cell lines tested ( Figure 3A ). This fragment was included in promoter area III described previously ( 18 , 19 ). The RLU of F383 in AGS, TMK1 and KATOIII cells were 14.37 ± 3.38, 14.51 ± 4.12 and 11.32 ± 0.17, respectively. On the contrary, the transcription activities of F314, F283 and F724 were not significantly different from that of pGL3basic vector ( Figure 3A ).

Fig. 3.

Identification of a functional CpG island within the 5′-flanking region of the PDX1 gene. ( A ) The four reporter constructs were transfected transiently into AGS, TMK1 and KATOIII cells and dual luciferase assays were performed to assess their transcriptional activities. Firefly luciferase activity was normalized to renilla activity. Promoter activity was presented as the fold change of RLU compared with the pGL3basic vector. RLU equals the value of the firefly luciferase unit, divided by the value of the renilla. This result was expressed as the mean of triplicate treatments ± standard deviation. ( B ) Reporter construct F383 was transfected into cells in absence or presence of SssI methylase (4 U) and 5′-aza-dC (5 μM). Transcriptional activities were assessed by dual luciferase assays (* P  < 0.05, ** P  < 0.01, *** P  < 0.001 versus control). The RLUs of SssI methylase and 5′-aza-dC were normalized to that of the non-treatment control (Ctrl). All of these experiments were repeated for three to four times with identical results.

Fig. 3.

Identification of a functional CpG island within the 5′-flanking region of the PDX1 gene. ( A ) The four reporter constructs were transfected transiently into AGS, TMK1 and KATOIII cells and dual luciferase assays were performed to assess their transcriptional activities. Firefly luciferase activity was normalized to renilla activity. Promoter activity was presented as the fold change of RLU compared with the pGL3basic vector. RLU equals the value of the firefly luciferase unit, divided by the value of the renilla. This result was expressed as the mean of triplicate treatments ± standard deviation. ( B ) Reporter construct F383 was transfected into cells in absence or presence of SssI methylase (4 U) and 5′-aza-dC (5 μM). Transcriptional activities were assessed by dual luciferase assays (* P  < 0.05, ** P  < 0.01, *** P  < 0.001 versus control). The RLUs of SssI methylase and 5′-aza-dC were normalized to that of the non-treatment control (Ctrl). All of these experiments were repeated for three to four times with identical results.

Subsequently, we evaluated the effect of in vitro methylation and demethylation on transcription activity of F383. As shown in Figure 3B , in vitro methylation by treatment with SssI methylase reduced the promoter activities of F383 by 0.52 ± 0.05-fold in AGS cells ( P  < 0.01), 0.68 ± 0.001-fold in TMK1 cells ( P  < 0.05) and 0.25 ± 0.03-fold in KATOIII cells ( P  < 0.001). Consistent with this, in vitro demethylation by pretreatment with 5′-aza-dC increased the promoter activities of F383 to 1.47 ± 0.10-fold in AGS cells ( P  < 0.001), 1.02 ± 0.08-fold in TMK1 cells ( P  > 0.05) and 1.01 ± 0.06-fold in KATOIII cells ( P  > 0.05). The promoter activity of F383 was slightly increased in TMK1 cells and might be attributed to non-specificity of 5′-aza-dC treatment. However, the promoter activity of F383 did not change with 5′-aza-dC because none of the four CpG islands studied are methylated in the KATOIII cells. In general, these results indicate that the distal CpG island located at −1975 to −1704 nt, and contained within the F383 fragment, was not only critical for PDX1 transcription but also the responsive sequence for DNA methylation.

Hypermethylaiton of CpG dinucleotides within F383 fragment

Next, we analyzed the methylation status of CpG dinucleotides within the F383 fragment using bisulfite DNA sequencing PCR analysis. As shown in Figure 4A (wild-type sequence), 17 CpG sites were presented in this fragment. All of them were candidate sites for methylation since none of them was altered into T after bisulfite modification ( Figure 1A , modified sequence). Again no methylation was found in all 17 CpG sites in KATOIII cells, and all other cell lines displayed high levels of methylation in the 17 CpG sites. As low as 1.96 ± 1.24% and 0.98 ± 0.97% of methylation were found in the two normal tissues ( P  < 0.0001 compared with cancer cell lines, Figure 4B and C ). The detailed methylation ratios for the 17 CpG dinucleotides in seven gastric cancer cell lines were listed in supplementary Table 4 , is available at Carcinogenesis Online. Only a few showed methylation at the third (8.3%) and the 13th CpG site (16.7%) in the two normal gastric tissues, however, the significant higher methylation ratios for the 17 CpG dinucleotides in seven gastric cancer cell lines were examined ( supplementary Table 4 is available at Carcinogenesis Online).

Fig. 4.

Methylation status of CpG sites within the functional promoter fragment. Fragment F383 was amplified using bisulfite-modified genomic DNA as template and inserted into pGEM-T4 vector. Ten to 12 white clones for each cell line were selected for sequencing. ( A ) Wild-type and modified sequence of F383 in BCG823 cells. The bisulfite DNA sequencing PCR primer sequences were underlined and the methylated CpG dinucleotides that were unable to be altered by bisulfite modification were labeled in bold and italicize. ( B ) Methylation status of the 17 CpG dinucleotides within F383 fragment in cancer cell lines. ( C ) Percentage of methylated CpG sites within F383 in cancer cell lines. 1N and 2N, normal gastric tissues; * P < 0.0001 versus either 1N or 2N.

Fig. 4.

Methylation status of CpG sites within the functional promoter fragment. Fragment F383 was amplified using bisulfite-modified genomic DNA as template and inserted into pGEM-T4 vector. Ten to 12 white clones for each cell line were selected for sequencing. ( A ) Wild-type and modified sequence of F383 in BCG823 cells. The bisulfite DNA sequencing PCR primer sequences were underlined and the methylated CpG dinucleotides that were unable to be altered by bisulfite modification were labeled in bold and italicize. ( B ) Methylation status of the 17 CpG dinucleotides within F383 fragment in cancer cell lines. ( C ) Percentage of methylated CpG sites within F383 in cancer cell lines. 1N and 2N, normal gastric tissues; * P < 0.0001 versus either 1N or 2N.

Aberrant hypermethylation of F383 in gastric cancerous tissues

To validate our in vitro findings, we analyzed the methylation status of F383 in 30 pairs of gastric cancerous tissues and their matched adjacent normal tissues by MSP assay ( Figure 5 ). In adjacent normal tissues, F383 was unmethylated in 16 (53.3%) samples and partially methylated in another 14 (46.7%) samples. In gastric cancer, F383 was completely methylated in three (10%) samples and partially methylated in another 27 (90%) samples (versus adjacent normal control, P  < 0.001). Further analysis indicated that the three gastric cancer tissues with complete methylation were all intestinal type and at advanced stages.

Fig. 5.

Hypermethylation of the PDX1 promoter in gastric cancerous tissues. ( A ) The methylation status of F383 in 30 pairs of gastric cancer (T) and their adjacent normal (N) tissues was analyzed by MSP. M, methylated; U, unmethylated; PC, positive control. These figures were the representatives of three independent experiments using the same DNA sample with repeatable results. ( B ) Percentage of methylated F383 fragment in gastric cancerous tissues and their adjacent normal tissues, P < 0.001 between cancerous and normal tissues.

Fig. 5.

Hypermethylation of the PDX1 promoter in gastric cancerous tissues. ( A ) The methylation status of F383 in 30 pairs of gastric cancer (T) and their adjacent normal (N) tissues was analyzed by MSP. M, methylated; U, unmethylated; PC, positive control. These figures were the representatives of three independent experiments using the same DNA sample with repeatable results. ( B ) Percentage of methylated F383 fragment in gastric cancerous tissues and their adjacent normal tissues, P < 0.001 between cancerous and normal tissues.

Role of histone acetylation in PDX1 transcription

To characterize the role of acetylation in PDX1 silencing, we detected the association between PDX1 promoter and acetylated histones using ChIP assay followed by quantitative real-time reverse transcription–PCR. We showed that F314, F283 and F724 fragments, which had no obvious promoter activity ( Figure 3A ), could all bind to acetylated histone H3 and H4, most notably in TMK1 cells ( Figure 6A ). However, the association between these fragments and acetylated histones in KATOIII cells were at low levels ( Figure 6A ). This finding indicated the presence of hypoacetylation in KATOIII cells but not in TMK1 cells in these promoter areas. Regarding F383, its association with acetylated histone H3 or H4 were absent or only at low levels in both cell lines ( Figure 6A ), suggesting F383 is a common fragment for deacetylated regulation in both cell lines. To further confirm this finding, we examined the association between acetylated histones and F383 in cells treated with TSA. We found that the level of histone H3 and histone H4 associated with F383 was significantly increased by TSA. The fold change of acetylated H3 and H4 were 249.60 ± 17.27 and 1.84 ± 0.09 in TMK1 cells and 3.92 ± 0.11 and 4.06 ± 0.11 in KATOIII cells ( P  < 0.05, Figure 6B ). Finally, we treated the reporter constructs of F383 with TSA in vitro . We showed that TSA upregulated the transcription activity of F383 to as high as 8.70 ± 0.04-fold in KATOIII cells ( P  < 0.001, Figure 6C ). These findings suggest that this fragment is an important responsive sequence for histone deacetylation.

Fig. 6.

Role of histone acetylation in PDX1 transcription. ( A ) Cross-linked and sonicated lysates of TMK1 and KATOIII cells were immunoprecipitated with anti-acetyl-histone H3 and H4 antibodies. Different promoter fragments of PDX1 were amplified quantitatively by quantitative real-time reverse transcription–PCR. Rabbit IgG was used as the antibody control. The result was expressed as the fold induction compared with that of IgG (* P  < 0.05, ** P  < 0.01 versus IgG control). ( B ) TMK1 and KATOIII cells were treated with TSA (500 nM) or double-diluted water, the dissolvent control (Ctrl). PDX1 -associated histone acetylation was detected by ChIP and quantitative real-time reverse transcription–PCR (* P  < 0.05, ** P  < 0.01 compared with Ctrl). ( C ) F383 was transfected into KATOIII cells in presence of TSA (500 nM) or the dissolvent Ctrl. Transcriptional activities were assessed by dual luciferase assay (*** P  < 0.001 versus control).

Fig. 6.

Role of histone acetylation in PDX1 transcription. ( A ) Cross-linked and sonicated lysates of TMK1 and KATOIII cells were immunoprecipitated with anti-acetyl-histone H3 and H4 antibodies. Different promoter fragments of PDX1 were amplified quantitatively by quantitative real-time reverse transcription–PCR. Rabbit IgG was used as the antibody control. The result was expressed as the fold induction compared with that of IgG (* P  < 0.05, ** P  < 0.01 versus IgG control). ( B ) TMK1 and KATOIII cells were treated with TSA (500 nM) or double-diluted water, the dissolvent control (Ctrl). PDX1 -associated histone acetylation was detected by ChIP and quantitative real-time reverse transcription–PCR (* P  < 0.05, ** P  < 0.01 compared with Ctrl). ( C ) F383 was transfected into KATOIII cells in presence of TSA (500 nM) or the dissolvent Ctrl. Transcriptional activities were assessed by dual luciferase assay (*** P  < 0.001 versus control).

Discussion

PDX1 is a critical regulator for the development and differentiation of some endocrine tissues. Several transcription factors such as USF1 (upstream stimulatory factor 1) ( 28 ), HNF3β (hepatocyte nuclear factor 3β) ( 18 ) and Ptf1α (Pancreas transcription factor 1α) ( 20 , 21 ) were capable of stimulating PDX1 transcription in β-cells ( 18–22 , 28 ). Here, we reported that promoter methylation and histone acetylation accounted for PDX1 gene silencing in gastric cancer. A functional fragment, F383, located within the PDX1 promoter area III (−2063 to −1681 nt) mediated the epigenetic regulation of PDX1 . PDX1 expression could be effectively restored by reversing promoter methylation and histone deacetylation. This supports our previous finding that PDX1 was a novel tumor suppressor for gastric cancer ( 23 ).

Typical promoter methylation is restricted to cytosines in CpG dinucleotides located within CpG islands. Through bioinformatic analysis, we identified four putative CpG islands. Island 1 (−1975 to −1704 nt), 2 (−1226 to −1066 nt) and 4 (−112 to +521 nt) were all located within or near the three nuclease hypersensitive sites (HSSs) ( 18 , 19 ). In previous studies, the HSSs were located between −2560 nt and −1880 nt (HSS1), −1330 nt and −800 nt (HSS2) and −260 nt and +180 nt (HSS3), respectively ( 18 , 19 ). Thus, these fragments should be the acting fragments for PDX1 transcription. Similarly, MSP analysis also revealed complete methylation in all tumor cell lines, albeit by only a small percentage. This, however, is not seen in normal cell lines. Furthermore, the F383 fragment was shown to be completely unmethylated in all normal cell lines contrary to the tumor cell lines. This suggests that aberrant methylation may be an important aspect in the cancer development continuum and implicates epigenetic regulation as a possible influencing factor. This finding was partially supported by a recent publication in which a proximal CpG island (−360 to +200 nt) was methylated and responsible for the pdx1 permanent silencing in β-cells of rat with intrauterine growth retardation ( 25 ).

To further assess the biological function of these CpG islands, we evaluated the transcriptional activities of reporter constructs. Only the most distal island (F383) displayed significant promoter activity. F314, F283 and F724 were associated with histone hypoacetylation in KATOIII cells but exerted no transcriptional activities. This suggested that the epigenetic modifications in CpG islands 2–4 were not essential for PDX1 transcription. There were three possible interpretations for this phenomenon. Firstly, islands 2–4 were located beyond the active promoter area of PDX1 , whereas F383 was located within the well-defined promoter fragment, area III, described previously ( 18 , 19 ). Secondly, islands 2–4 were located in the center of HSS2 and HSS3, whereas HSSs were most probably occupied by non-histone proteins (transcription factors). This point was supported by our finding that TSA treatment could only increase the transcriptional activity of F383 but not the other three fragments. Lastly, the transcription factors might antagonize the suppressive effect of epigenetic regulation. Although the transcription factors for islands 2 and 3 were unknown, a positive regulator USF1 could bind to its responsive element located at island 4 to stimulate PDX1 expression in β-cells ( 28 ). A recent study reported recruitment of the HDAC1 and deacetylation of histones H3 and H4 were associated with loss of USF-1 binding at the proximal promoter of Pdx1 and subsequently resulted to pdx1 silencing in the rat fetal with intrauterine growth retardation ( 25 ).

Nevertheless, The CpG island at F383 fragment must not be the only methylated functional sequence of PDX1 gene. No methylation was found in this island in KATOIII cells, and promoter activity of F383 didnot significantly regulated by SssI methylase or 5′-aza-dC. However, PDX1 expression in this cell line could be upregulated as well by 5′-aza-dC and/or TSA ( Figure 1 ). We proposed that the effect may be mediated by the methylation of scatter single CpG sites we have not defined here. Similarly, our previous study reported that a single CpG site, and not CpG island, may attribute to XAF1 transcription silencing in gastrointestinal cancer cells ( 27 ). Further, F383 might not be the only promoter sequence regulating PDX1 expression in KATOIII cells. A recent study reported that the conserved area IV at −6200 to −5670 nt from 5′-flanking sequence also regulated pdx1 transcription ( 20 ). Area IV was an upstream control region in the rat pdx-1 gene located between −6200 and −5670 nt, which is also present in the human gene. So area IV might be a putative promoter, containing a hypermethylated CpG island and regulate PDX1 expression in KATOIII cells.

In conclusion, epigenetic regulation including promoter methylation and histone deacetylation contributed to PDX1 silencing in gastric cancer cells with the predominant responsive sequences (−2063 to −1681 nt) localizing to the PDX1 promoter area III.

Supplementary material

Supplementary Tables 14 can be found at http://carcin.oxfordjournals.org/

Funding

Simon K.Y.Lee endowed professorship Research Fund; Gordon Chiu Stomach Cancer Research Fund; Outstanding Researcher Award Fund of the University of Hong Kong, Hong Kong; National Basic Research Program of China (973 Program) (2010CB529306); Doctoral Program Fund; Guangdong natural science fund; Guangdong General Hospital (9451008004002824).

Abbreviations

    Abbreviations
  • 5′-aza-dC

    5′-aza-2′-deoxycytidine

  • ChIP

    chromatin immunoprecipitation

  • MSP

    methylation-specific polymerase chain reaction

  • HDAC

    histone deacetylase

  • HSS

    hypersensitive site

  • PDX1

    pancreatic-duodenal homeobox 1

  • PCR

    polymerase chain reaction

  • RLU

    relative luciferase unit

  • TSA

    trichostatin A

Conflict of Interest Statement: None declared.

References

1.
Jones
PA
, et al.  . 
The epigenomics of cancer
Cell
 , 
2007
, vol. 
128
 (pg. 
683
-
692
)
2.
Jones
PA
, et al.  . 
The fundamental role of epigenetic events in cancer
Nat. Rev. Genet.
 , 
2002
, vol. 
3
 (pg. 
415
-
428
)
3.
Ushijima
T
Detection and interpretation of altered methylation patterns in cancer cells
Nat. Rev. Cancer
 , 
2005
, vol. 
5
 (pg. 
223
-
231
)
4.
Esteller
M
Cancer epigenomics: DNA methylomes and histone-modification maps
Nat. Rev.
 , 
2007
, vol. 
8
 (pg. 
286
-
298
)
5.
Ebert
MP
, et al.  . 
Aristaless-like homeobox-4 gene methylation is a potential marker for colorectal adenocarcinomas
Gastroenterology
 , 
2006
, vol. 
131
 (pg. 
1418
-
1430
)
6.
Suh
ER
, et al.  . 
DNA methylation down-regulates CDX1 gene expression in colorectal cancer cell lines
J. Biol. Chem.
 , 
2002
, vol. 
277
 (pg. 
35795
-
35800
)
7.
Pilozzi
E
, et al.  . 
CDX1 expression is reduced in colorectal carcinoma and is associated with promoter hypermethylation
J. Pathol.
 , 
2004
, vol. 
204
 (pg. 
289
-
295
)
8.
Wong
NA
, et al.  . 
Loss of CDX1 expression in colorectal carcinoma: promoter methylation, mutation, and loss of heterozygosity analyses of 37 cell lines
Proc. Natl Acad. Sci. USA
 , 
2004
, vol. 
101
 (pg. 
574
-
579
)
9.
Guo
M
, et al.  . 
Epigenetic silencing of CDX2 is a feature of squamous esophageal cancer
Int. J. Cancer
 , 
2007
, vol. 
121
 (pg. 
1219
-
1226
)
10.
Brooke
NM
, et al.  . 
The ParaHox gene cluster is an evolutionary sister of the Hox gene cluster
Nature
 , 
1998
, vol. 
392
 (pg. 
920
-
922
)
11.
Barucca
M
, et al.  . 
Hox and paraHox genes in bivalve molluscs
Gene
 , 
2003
, vol. 
317
 (pg. 
97
-
102
)
12.
Stoffers
DA
, et al.  . 
Developmental expression of the homeodomain protein IDX-1 in mice transgenic for an IDX-1 promoter/lacZ transcriptional reporter
Endocrinology
 , 
1999
, vol. 
140
 (pg. 
5374
-
5381
)
13.
Guz
Y
, et al.  . 
Expression of murine STF-1, a putative insulin gene transcription factor, in beta cells of pancreas, duodenal epithelium and pancreatic exocrine and endocrine progenitors during ontogeny
Development
 , 
1995
, vol. 
121
 (pg. 
11
-
18
)
14.
Larsson
LI
, et al.  . 
Pancreatic-duodenal homeobox 1 -role in gastric endocrine patterning
Mech. Dev.
 , 
1996
, vol. 
60
 (pg. 
175
-
184
)
15.
Offield
MF
, et al.  . 
PDX-1 is required for pancreatic outgrowth and differentiation of the rostral duodenum
Development
 , 
1996
, vol. 
122
 (pg. 
983
-
995
)
16.
Gannon
M
, et al.  . 
Regulatory regions driving developmental and tissue-specific expression of the essential pancreatic gene pdx1
Dev. Biol.
 , 
2001
, vol. 
238
 (pg. 
185
-
201
)
17.
Gu
G
, et al.  . 
Direct evidence for the pancreatic lineage: NGN3+cells are islet progenitors and are distinct from duct progenitors
Development
 , 
2002
, vol. 
129
 (pg. 
2447
-
2457
)
18.
Wu
KL
, et al.  . 
Hepatocyte nuclear factor 3beta is involved in pancreatic beta-cell-specific transcription of the pdx-1 gene
Mol. Cell. Biol.
 , 
1997
, vol. 
17
 (pg. 
6002
-
6013
)
19.
Gerrish
K
, et al.  . 
Pancreatic beta cell-specific transcription of the pdx-1 gene. The role of conserved upstream control regions and their hepatic nuclear factor 3beta sites
J. Biol. Chem.
 , 
2000
, vol. 
275
 (pg. 
3485
-
3492
)
20.
Gerrish
K
, et al.  . 
Conserved transcriptional regulatory domains of the pdx-1 gene
Mol. Endocrinol.
 , 
2004
, vol. 
18
 (pg. 
533
-
548
)
21.
Wiebe
PO
, et al.  . 
Ptf1a binds to and activates area III, a highly conserved region of the Pdx1 promoter that mediates early pancreas-wide Pdx1 expression
Mol. Cell. Biol.
 , 
2007
, vol. 
27
 (pg. 
4093
-
4104
)
22.
Miyatsuka
T
, et al.  . 
Ptf1a and RBP-J cooperate in activating Pdx1 gene expression through binding to Area III
Biochem. Biophys. Res. Commun.
 , 
2007
, vol. 
362
 (pg. 
905
-
909
)
23.
Ma
J
, et al.  . 
Pancreatic duodenal homeobox-1 ( PDX1 ) functions as a tumor suppressor in gastric cancer
Carcinogenesis
 , 
2008
, vol. 
29
 (pg. 
1327
-
1333
)
24.
Zhu
S
, et al.  . 
Alterations of gastric homeoprotein expression in helicobacter pylori infection, incisural antralisation, and intestinal metaplasia
Dig. Dis. Sci.
 , 
2009
, vol. 
54
 (pg. 
996
-
1002
)
25.
Park
JH
, et al.  . 
Development of type 2 diabetes following intrauterine growth retardation in rats is associated with progressive epigenetic silencing of Pdx1
J. Clin. Invest.
 , 
2008
, vol. 
118
 (pg. 
2316
-
2324
)
26.
Wang
J
, et al.  . 
HSF1 down-regulates XAF1 through transcriptional regulation
J. Biol. Chem.
 , 
2006
, vol. 
281
 (pg. 
2451
-
2459
)
27.
Zou
B
, et al.  . 
Correlation between the single-site CpG methylation and expression silencing of the XAF1 gene in human gastric and colon cancers
Gastroenterology
 , 
2006
, vol. 
131
 (pg. 
1835
-
1843
)
28.
Sharma
S
, et al.  . 
Pancreatic islet expression of the homeobox factor STF-1 relies on an E-box motif that binds USF1
J. Biol. Chem.
 , 
1996
, vol. 
271
 (pg. 
2294
-
2299
)