Transcriptional silencing of the DLC-1 tumor suppressor gene ... - Nature

Oncogene (2003) 22, 3943–3951

& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

Transcriptional silencing of the DLC-1 tumor suppressor gene by epigenetic mechanism in gastric cancer cells Tai Young Kim1, Hyun-Soon Jong1, Sang-Hyun Song1, Alexandre Dimtchev2, Sook-Jung Jeong2, Jung Weon Lee1, Tae-You Kim1,3, Noe Kyeong Kim3, Mira Jung2 and Yung-Jue Bang*,1,3 1 National Research Laboratory for Cancer Epigenetics, Cancer Research Institute, Seoul National University College of Medicine, Seoul 110-744, Korea; 2Department of Radiation Medicine, Georgetown University Medical Center, Washington, DC 20007, USA; 3 Department of Internal Medicine, Seoul National University College of Medicine, Seoul 110-744, Korea

DLC-1 (deleted in liver cancer) gene is frequently deleted in hepatocellular carcinoma. However, little is known about the genetic status and the expression of this gene in gastric cancer. In this study, Northern and Southern analysis showed that seven of nine human gastric cancer cell lines did not express DLC-1 mRNA, but contained the DLC-1 gene. To identify the mechanism of the loss of DLC-1 mRNA expression in these cell lines, we investigated the methylation status of DLC-1 gene by using methylation-specific PCR (MSP) and Southern blot, and found that five of seven DLC-1 nonexpressing gastric cancer cell lines were methylated in the DLC-1 CpG island. Treatment with 5-aza-20 -deoxycytidine (5Aza-dC) induced DLC-1 mRNA expression in the gastric cancer cell lines that have the methylated alleles. Studies using SNU-601 cell line with methylated DLC-1 alleles revealed that nearly all CpG sites within DLC-1 CpG island were methylated, and that the in vitro methylation of the DLC-1 promoter region is enough to repress DLC-1 mRNA expression, regardless of the presence of transcription factors capable of inducing this gene. In all, 29 of 97 (30%) primary gastric cancers were also shown to be methylated, demonstrating that methylation of the DLC-1 CpG island is not uncommon in gastric cancer. In addition, we demonstrated that DLC-1 mRNA expression was induced, and an increase in the level of acetylated H3 and H4 was detected by the treatment with trichostatin A (TSA) in two DLC-1 nonexpressing cell lines that have the unmethylated alleles. Taken together, the results of our study suggest that the transcriptional silencing of DLC-1, by epigenetic mechanism, may be involved in gastric carcinogenesis. Oncogene (2003) 22, 3943–3951. doi:10.1038/sj.onc.1206573 Keywords: DLC-1; methylation; histone deacetylation; gastric cancer

*Correspondence: Y-J Bang, Department of Internal Medicine, Seoul National University College of Medicine, 28 Yongon-dong, Chongrogu, Seoul 110-744, Korea; E-mail: [email protected] Received 13 December 2002; revised 5 March 2003; accepted 7 March 2003

Introduction A newly identified gene frequently deleted in liver cancer (DLC-1), which was simultaneously cloned as a hot potato (HP) by another group (unpublished, GenBank Accession no. AF026219), shares high sequence similarity with rat p122RhoGAP (Yuan et al., 1998), a GTPase activating protein (GAP) that serves as a negative regulator of Rho proteins by stimulating its intrinsic GTPase activities (Morii et al., 1993). The Rho family proteins function as important regulators of the organization of the actin cytoskeleton (Ridley and Hall, 1992) and are known to be involved in Ras-mediated oncogenic transformation (Qiu et al., 1995). Whereas guanine nucleotide exchange factors (GEFs) like Dbl and Vav were discovered to be oncogenic proteins by activating Rho proteins (Hart et al., 1991; Crespo et al., 1997), GAPs may function as tumor suppressors by inactivating Rho proteins. A recent report showed that transfection of DLC-1 cDNA into hepatocellular carcinoma (HCC) cell lines deficient in DLC-1 gene induced significant tumor growth inhibition and reduced colony formation (Ng et al., 2000), suggesting that DLC-1 is a tumor suppressor gene. The DLC-1 gene was found to be located at 8p21.3– 22 (Yuan et al., 1998), a region frequently deleted in various cancer cells including hepatocellular, lung, colorectal, prostate, and breast cancer (Emi et al., 1992; Fujiwara et al., 1994; Yaremko et al., 1995; Wistuba et al., 1999), and has been reported to be deleted in HCC cell lines and in primary tumors (Yuan et al., 1998; Ng et al., 2000). However, the mechanism of the loss of DLC-1 expression without genomic deletion has not been explored. Epigenetic mechanisms that involve DNA methylation and alteration of chromatin structure are an important way of transcriptionally silencing many genes, especially tumor suppressor genes, acting as an alternative to genetic defects in human cancers. Recently, several studies have shown that methylation of CpG island is associated with gene silencing in tumors, which include pRb, p15Ink4b, p16Ink4a, hMLH1, and TGFb type I receptor (Herman et al., 1996b, 1998; Stirzaker et al., 1997; Kang et al., 1999; Song et al., 2000). Genes silenced by DNA methylation can be reactivated by

DLC-1 silencing in gastric cancer cells TY Kim et al

3944

treatment with 5-aza-20 -deoxycytidine (5-Aza-dC), which is a well-established inhibitor of DNA methyltransferase (Jones and Taylor, 1980). Inactive chromatin by histone deacetylation is also known to be other pathway of gene silencing, which include p21Waf1, hTERT, and hLHR (Sowa et al., 1997; Takakura et al., 2001; Zhang and Dufau, 2002). The level of histone acetylation depends on the activity of histone acetyltransferase (HAT) and histone deacetylase (HDAC). Trichostatin A (TSA), a histone deacetylase inhibitor, is reported to activate genes whose expression is silenced by the inhibition of HDAC activity (Van Lint et al., 1996). In the present study, we investigated the possibility that the deletion of DLC-1 gene could inactivate DLC-1 expression in gastric cancer, because the high frequency of allelic losses of 8p21–22 was also found in gastric cancer (Baffa et al., 2000). Our data demonstrate that DLC-1 mRNA expression is lost in a majority of gastric cancer cell lines and primary tumors by DNA methylation and histone deacetylation, not by genetic defects, suggesting that epigenetic inactivation of DLC-1 gene may play a role in gastric carcinogenesis.

Results Loss of DLC-1 expression without gene deletion in gastric cancer cell lines Of the nine gastric cancer cell lines examined by Northern blot analysis, SNU-484 and SNU-638 cell lines expressed 7.5 and 4.5 kb transcripts of the DLC-1 gene, which is consistent with an earlier report (Yuan et al., 1998), but the other cell lines did not (Figure 1a). A recent report showed that the deletion of DLC-1 gene was associated with the loss of the gene expression in HCC (Yuan et al., 1998; Ng et al., 2000). To determine whether the loss of DLC-1 expression was because of genomic deletion in gastric cancer cell lines, the DLC-1 gene was checked by Southern blot analysis. As shown in Figure 1b, the genomic DNA of DLC-1 was detected in all gastric cancer cell lines, and this was found to be not significantly different from its level in normal human gastric tissue DNA. Therefore, the observed loss of DLC-1 mRNA expression was not caused by genomic deletion in the gastric cancer cell lines. Methylation status of the DLC-1 CpG island in gastric cancer cell lines Methylation of CpG island is an alternative way of causing the gene silencing of diverse tumor suppressor genes. Structurally, the 663-bp fragment of the 50 upstream region from the starting codon of DLC-1 gene (GenBank Accession no. AF404867) has a high G þ C content (73%) and CpG : GpC ratio of 0.92, thus satisfying the criteria for a CpG island. To explore the potential role of CpG island methylation in the transcriptional silencing of DLC-1 gene, we checked the methylation status in the gastric cancer cell lines by methylation-specific PCR (MSP), in which methylated Oncogene

Figure 1 Expression and genomic status of DLC-1 in gastric cancer cell lines. (a) The expression of DLC-1 mRNA was determined in gastric cancer cell lines by Northern blot analysis. The loss of DLC-1 mRNA expression was found in seven of nine cell lines but not in SNU-484 and SNU-638 cell lines, which showed a 7.5 kb major transcript and a 4.5 kb minor transcript. HepG2 cell line was used as a positive control. (b) Genomic DNAs were digested with HindIII and separated by electrophoresis in 0.7% agarose gel. No reduction of DNA intensity was detected compared with human normal gastric genomic DNA (NG-DNA)

and unmethylated alleles can be specifically amplified after chemically modifying the DNA with sodium bisulfite (Herman et al., 1996a). As shown in Figure 2b, two DLC-1-expressing cell lines (SNU-484 and -638) were unmethylated at the DLC-1 CpG island, as was expected, whereas five DLC-1 nonexpressing cell lines were either completely methylated (SNU-1, -16, -601, and -719) or partially methylated (SNU-620). We also analysed the methylation status of the DLC-1 CpG island by Southern blot analysis (Figure 2c). A restriction map of the DLC-1 CpG island is shown in Figure 2a. Genomic DNAs from different gastric cancer cell lines were digested with EcoRI, which generated a 5kb fragment containing the DLC-1 CpG island, and this was further digested with the methylation-sensitive restriction enzymes, BssHII or NaeI. If unmethylated, the 5-kb fragment is cleaved by these enzymes into smaller fragments, which can be detected by Southern blot. Five cell lines showing methylation by MSP revealed the 5-kb methylated fragment, which represents the methylated CpG of the individual methylationsensitive enzyme-recognition sites. Four unmethylated cell lines revealed the digested smaller fragments according to MSP. In particular, SNU-620 cell line exhibited both the 5-kb methylated fragment and cleaved smaller fragments, suggesting that the population was heterogeneous and that partial proportion of the cells had unmethylated DLC-1 alleles. Therefore, data obtained from this analysis agreed well with those obtained by MSP. Taken together, these findings strongly suggest that the methylation status of the


3945

Figure 2 Methylation status of the DLC-1 CpG island in gastric cancer cell lines. (a) A map of the CpG island of the DLC-1 gene spanning 668-bp upstream of the ATG translation start codon. The transcription start site is marked with right angle arrow at þ 1. The locations of the BssHII and NaeI restriction sites and the set of MSP primers that recognize unmethylated (uF and uR) or methylated (mF and mR) DNA are shown. (b) Bisulfite-modified DNA derived from gastric cancer cell lines was amplified with primers specific for unmethylated (U) or methylated (M) DNA. The ‘U’ and ‘M’ products were 178 and 172 bp, respectively, because of differences in the lengths of the primers; Lane H2O, a negative control. (c) Methylation status of the DLC-1 CpG island was examined by digesting with EcoRI and then with the methylation-sensitive restriction enzymes BssHII or NaeI, and this was followed by Southern blot analysis using a 488-bp fragment (138 to þ 310) as a probe

DLC-1 CpG island is related with the transcriptional silencing of DLC-1 gene in gastric cancer cell lines. Studies upon the methylation of the DLC-1 CpG island To obtain more information about the methylation status of the DLC-1 CpG island and its effect on gene expression in detail, we studied SNU-484 cell line with the unmethylated alleles, and SNU-601 cell line with the methylated alleles. The methylation patterns of the DLC-1 CpG island were examined by bisulfide genomic sequencing of individual alleles from these cell lines. The promoter region spanning 35 CpG sites was amplified by PCR using sodium bisulfite-modified DNA as templates, and five PCR clones of each cell line were sequenced. As shown in Figure 3a, we found that all CpG sites were completely unmethylated in the analysed clones derived from SNU-484 cell line, and nearly all CpG sites were extensively methylated in those derived from SNU-601 cell line. In order to determine whether methylation of the DLC-1 promoter region could mediate the transcriptional silencing of DLC-1 gene, a luciferase reporter construct containing the entire DLC-1 CpG island was

methylated using SssI, an enzyme that methylates every CpG dinucleotide. The complete methylation of each reporter construct was confirmed by digestion with HpaII (data not shown). As shown in Figure 3b, the transfection of the methylated reporter construct into SNU-601 cell line resulted in the complete abolition of transcriptional activity measured by luciferase activity. A similar result was obtained in the case of SNU-484 cell line, which expresses DLC-1 endogenously. These results suggest that complete methylation of the DLC-1 CpG island is sufficient to repress the DLC-1 expression regardless of the presence of transcription factors capable of inducing this gene. To investigate whether demethylation could restore DLC-1 mRNA expression in SNU-601 cell line, we treated cells with 5-Aza-dC for 4 days at varying concentrations. As shown in Figure 3c, the induction of DLC-1 mRNA, as demonstrated by reverse transcription PCR (RT–PCR), occurred at a concentration of 0.1 mm 5-Aza-dC and the level of DLC-1 mRNA was increased in a dose-dependent manner. In contrast, treatment of SNU-484 cell line with 5-Aza-dC did not alter the level of DLC-1 mRNA expression as expected. To determine whether this induction was associated with Oncogene


3946

a change in the methylation status of DLC-1, DNA was isolated from 5-Aza-dC-treated SNU-601 cell line and subjected to MSP analysis. The induction of DLC-1 mRNA expression was associated with the presence of unmethylated DLC-1 alleles, which were absent before the 5-Aza-dC was treated, thus confirming that this induction resulted from the demethylation of the DLC-1 CpG island (Figure 3d). To determine whether DNA methylation was responsible for the loss of DLC-1 mRNA expression in other gastric cancer cell lines that have the methylated alleles,

we treated these cell lines with 5 mm 5-aza-dC. Figure 3 showed that the DLC-1 mRNA expression was induced in all five cell lines, as determined by the presence of the predicted 262-bp RT–PCR product. Methylation of DLC-1 CpG island in primary gastric tissues We next examined whether DLC-1 gene is methylated in primary gastric cancer tissues (Figure 4). MSP analysis showed that the DLC-1 CpG island was methylated in 30% (29 of 97) of the primary gastric cancer tissues. In contrast, those islands were not methylated in the corresponding normal gastric tissues, indicating that methylation of DLC-1 CpG island is restricted to cancer cells. It was also observed that gastric cancer samples contained unmethylated DLC-1 alleles. We speculated that these might be because of the invariable contamination of noncancerous cells such as stromal, inflammatory, and vascular cells. From these, we demonstrated that the methylation of DLC-1 CpG island is not an artifact of in vitro experimental system,

Figure 4 Representative MSP analysis of the DLC-1 CpG island in primary gastric cancer tissues. Parallel amplification reactions were performed using primers specific for unmethylated (U) or methylated (M) DNA. Case 2 is a representative pair of unmethylated sample and the remainders are representative pairs of methylated samples chosen from the total of 97 analysed

Figure 3 Studies on the methylation of the DLC-1 CpG island. (a) Bisulfite genomic sequencing of the region (nt. from þ 20 to þ 310) of the DLC-1 CpG island was performed in a DLC-1-expressing cell line (SNU-484) and a DLC-1-nonexpressing cell line (SNU-601), which contained the methylated alleles. Each circle indicates a CpG site and each line of circles represents the analysis of a single-cloned allele. &, an unmethylated CpG site; &, a methylated CpG site. (b) pGL3basic vector containing the promoter region of DLC-1 (nt. from -428 and þ 273) was in vitro methylated using SssI (CpG) methylase and transfected into SNU-484 or SNU-601 cell line. Luciferase activity was normalized versus b-galactosidase activity. Data are representative of three independent experiments performed in triplicate. (c) RT–PCR analysis was performed after treatment with increasing concentrations of 5-Aza-dC for 4 days. b-actin expression demonstrates the efficiency of cDNA synthesis. SNU-484 cell line was not appreciably affected by 5-Aza-dC treatment. (d) MSP analysis was performed upon SNU-601 cell line before and after treatment with 5-Aza-dC. The presence of unmethylated (U) product indicates the presence of the DLC-1 gene demethylation. Parallel amplification reactions were performed using primers specific for U or methylated (M) DNA. (e) Effects of 5-Aza-dC on DLC-1 mRNA expression in nonexpressing gastric cancer cell lines with the methylated alleles. Total RNA was isolated after treating 5 mm 5-Aza-dC for 4 days. RT–PCR was performed using specific primers for DLC-1 as described in ‘Materials and Methods’ Oncogene


3947

but relatively common and cancer-specific event in gastric carcinogenesis. DLC-1 expression is silenced by histone deacetylation A growing body of data indicate the importance of histone acetylation and deacetylation and corresponding chromatin structural alterations in the process of gene silencing. DLC-1-nonexpressing cell lines, SNU-5 and SNU-668, which have the unmethylated alleles, were used to test whether HDAC activity contributes to the loss of DLC-1 mRNA expression. When SNU-5 or SNU-668 cell line was treated with TSA at a concentration of 500 nm, DLC-1 mRNA expression was weakly detected after 6 h and clearly observed after 12 h in both cell lines (Figure 5a). To demonstrate that TSA directly induces DLC-1 mRNA expression by modulating the acetylation status of DLC-1 promoter in two cell lines, a chromatin immunoprecipitation (ChIP) assay was performed using antiacetylated H3 or H4 antibodies. The immunoprecipitated DNA was analysed with PCR using primers that recognize the DLC-1 promoter region. The acetylation of histones H3 and H4 associated with the DLC-1 promoter was observed in both SNU-5 and SNU-668 cell lines by treatment with TSA, whereas that was not present in untreated cell lines (Figure 5b). These results suggest that DLC-1 mRNA expression is modulated by histone deacetylation in SNU-5 and SNU-668 cell lines. To ascertain whether both the activities of DNA methyltransferase (DNMT) and HDAC play a role in the loss of DLC-1 expression more generally, a dual treatment strategy was used in SNU-719 cell line. The treatment of SNU-719 cell line with low doses of 5-AzadC alone weakly induced DLC-1 mRNA expression,

and the combined treatment of TSA with 5-Aza-dC increased this gene expression (Figure 5c). However, treatment with TSA alone had no effect on gene expression.

Discussion DLC-1 was first identified by using a representative difference analysis (RDA), a PCR-based subtractive hybridization technique which was applied to a primary HCC and adjacent noncancerous tissue from the same patient (Yuan et al., 1998). In the original report, the loss of heterozygosity of DLC-1 was detected in seven of 16 HCCs and in 10 of 11 HCC cell lines. A subsequent report confirmed this observation by finding homozygous deletion of DLC-1 in six of 30 human HCCs and two of five HCC cell lines (Ng et al., 2000). However, the loss of DLC-1 mRNA expression was also observed in HCCs without homozygous deletion, suggesting that alternative mechanisms are responsible for inactivating DLC-1 in these tumors. In this study, we observed the loss of DLC-1 mRNA expression without genetic deletion in seven of nine gastric cancer cell lines. MSP and methylation-sensitive Southern blotting showed that five of seven gastric cell lines and 29 of 97 (30%) primary gastric cancer tissues were methylated at the DLC-1 CpG island. In addition, treatment of gastric cancer cell lines that have the methylated alleles with 5-Aza-dC restored DLC-1 mRNA expression, demonstrating that DNA methylation is directly involved in the transcriptional silencing of DLC-1 gene. Recent studies have indicated the existence of a mechanistic linkage between DNA

Figure 5 Effect of TSA on DLC-1 mRNA expression and histone acetylation at the DLC-1 promoter locus in DLC-1-nonexpressing gastric cancer cell lines with the unmethylated alleles. (a) RT–PCR analysis was performed after treating SNU-5 and SNU-668 cell lines with 500 nm TSA for the indicated period. b-actin expression demonstrates the efficiency of the cDNA synthesis. (b) SNU-5 and SNU668 cell lines were cultured with or without TSA (500 nm) for 12 h and treated with formaldehyde to crosslink protein to DNA and the acetylated histones were immunoprecipitated using antibodies specific to the acetylated histone H3 or acetylated histone H4. Immunoprecipitated DNA was analysed by PCR using a pair of primer specific for the DLC-1 promoter region. Negative control reaction using an aliquot precipitated with no antibody (No Ab) was shown. Input denotes 10% of total DNA prior to immunoprecipitation. (c) SNU-719 cell line with the methylated alleles was treated with DMSO, 1 mm 5-Aza-dC, 500 nm TSA, or 1 mm 5-Aza-dC followed by 500 nm TSA, and RT–PCR was performed using specific primers for DLC-1, as described in ‘Materials and methods’. b-actin expression demonstrates the efficiency of cDNA synthesis Oncogene


3948

methylation and chromatin alteration, mediated by methyl-CpG-binding proteins (i.e. MeCPs and MBDs) that recruit Sin3A/HDAC corepressor complex (Nan et al., 1998; Wade et al., 1998; Dobosy and Selker, 2001; Burgers et al., 2002). Our study performed by a combined treatment of 5-Aza-dC with TSA to a gastric cancer cell line (SNU-719) showed that the combined treatment synergistically increased the level of DLC-1 mRNA expression more than that activated by 5-AzadC alone, which supports a report that showed the essential roles of DNMT and HDAC in the transcriptional silencing of several methylated genes (Cameron et al., 1999). Restoration of histone acetylation by treatment with TSA alone did not induce DLC-1 mRNA expression in this cell line. This result suggests that methylation plays the dominant role over histone deacetylation in DLC-1 silencing in association with DNA methylation. Consistent with our finding, it has been recently reported that TSA alone cannot reactivate genes silenced by DNA methylation, such as DAPK (Satoh et al., 2002), COX-2 (Kikuchi et al., 2002), and Runx3 (Guo et al., 2002). We also demonstrated that TSA can induce the DLC1 mRNA expression and an accumulation of acetylated histone associated with the DLC-1 promoter in the gastric cancer cell lines that have the unmethylated alleles (SNU-5 and -668). Recent published data suggested that Sp1 binding sites are involved in the transcriptional activation of genes such as p21 (Sowa et al., 1997; Huang et al., 2000) and Cox1 (Taniura et al., 2002) in response to HDAC inhibitors. In addition, Doetzlhofer et al. (1999) reported that Sp1 is tightly associated with HDAC and that the two proteins may be part of one complex. An attempt to identify potential transcription factor binding sites using the TFSEARCH program (http://www.cbrc.jp/research/db/TFSEARCH. html) revealed that the promoter region of DLC-1 contains multiple Sp1 binding sites, notably in the region from 280 to 210, which suggests that these Sp1 binding sites might play an important role in the induction of DLC-1 mRNA expression by treatment with TSA in SNU-5 and SNU-668 cell lines. However, further studies using DLC-1 promoter constructs are required to identify the functional roles of the Sp1 binding sites in DLC-1 silencing. Taken together, our experiments suggest that two distinct epigenetic mechanisms for the transcriptional silencing of DLC-1 gene, namely, DNA methylation and histone deacetylation, exist in gastric cancer cells. The precise nature of molecular mechanisms that determine which of the two epigenetic pathways underlying the transcriptional silencing of DLC-1 gene remain to be elucidated. One possible mechanism recently proposed by Robertson (Robertson, 2002; Geiman and Robertson, 2002) is that an altered chromatin environment may predispose DNA methylation. According to his hypothesis, HDAC would first target the region destined to undergo transcriptional silencing via interaction with other proteins. Next, DDM/Lsh, a SNF2 family member, known to alter the chromatin structure by disrupting the histone/DNA contacts (Vignali et al., Oncogene

2000), acts to enable histone methyltransferases (HMTase) to modify the histone tails, specifically on lysine9 of histone H3. Following this chromatin structural alteration and histone methylation, DNMTs can gain access to the DNA sequence that was previously inaccessibly packaged. Finally, HDAC can bind to the methylated DNA, thus causing a ‘tightening’ of the chromatin structure. This model is supported by recent evidence which shows that gene silencing of exogenous retroviral sequences that have been shown to be methylated can occur in the de novo methylase-null cell (dnmt3a/; dnmt3b/) by repressive chromatin structure (Pannell et al., 2000). Therefore, it seems likely that in cases of SNU-5 and SNU-668 cell lines the DLC1 gene silencing is attributed to the histone deacetylation step prior to the DNA methylation step. This may occur because of the disruption of any of the proteins involved in above sequential reactions, including disruption of the chromatin remodeling complex (SNF-2 family proteins and HMTase) or DNMTs. We performed RT–PCR to check DNMTs mRNA in the gastric cancer cell lines. DNMT1, a maintenance methyltransferase and DNMT3a, de novo methyltransferase, were expressed in all gastric cancer cell lines. Interestingly, the expression of DNMT3b, also characterized as a de novo methyltransferase was abolished in SNU-5 and -668 cell lines, but detected in the other gastric cancer cell lines with methylated DLC-1 alleles (data not shown), which suggests that DNMT3b might be a factor that determines whether DLC-1 is silenced by DNA methylation or histone deacetylation in gastric cancer cell lines. Further studies are needed to clarify this hypothesis, including the loss of chromatin remodeling complex in SNU-5 and SNU-668 cell lines. It has been reported that the DLC-1 gene has an 86% homology with rat p122RhoGAP. Transfection of p122RhoGAP, which has a GAP activity specific for RhoA, resulted in the disassembly of actin stress fiber, morphological rounding, and detachment (Sekimata et al., 1999). This imply that DLC-1 may be involved in the regulation of cytoskeletal rearrangement and morphological alteration by inhibiting the activity of Rho protein. Recent studies indicate that Rho proteins regulate signaling pathways that mediate tumor cell invasion and metastasis (Sliva et al., 2000; Tsutsumi et al., 2002). DLC-1 gene was mapped to the 8p21.3–22, a region that is frequently deleted during the process of metastasis of lung cancer (Kurimoto et al., 2001), colorectal cancer (Knosel et al., 2002), and melanoma (Wiltshire et al., 2001). Moreover, a recent study using Gene-chip array technology analysed the different transcriptional profiles of two human mammary carcinoma cell lines, which showed metastatic and nonmetastatic properties, and reported that DLC-1 was downregulated in the metastatic cell line (Euer et al., 2002). Taken together, these studies suggest that the loss of DLC-1 expression may be involved in the process of metastasis. In summary, our findings demonstrate that epigenetic mechanisms, including the methylation of DLC-1 CpG island and histone deacetylation, occur in gastric cancer


3949

cells, and that is associated with the transcriptional silencing of the DLC-1 gene, suggesting that these processes may play important roles in gastric carcinogenesis.

Materials and methods Tumor samples and cell lines

GGA GTG-30 (antisense) and methylated reactions (nt. from 53 to þ 119), 50 -CCC AAC GAA AAA ACC CGA CTA ACG-30 (sense) and 50 -TTT AAA GAT CGA AAC GAG GGA GCG-30 (antisense). All amplifications used a hot start AmpliTaq DNA polymerase (Perkin-Elmer, Foster City, CA, USA) and the following conditions: 951C for 12 min, and 35 cycles of 951C for 30 s, 551C for 30 s, and 721C for 30 s, followed by final extension at 721C for 10 min. Each set of PCR reaction products were loaded onto 2.5% agarose gels, ethidium bromide-stained, and visualized under UV light.

Tumors and corresponding normal tissues were obtained from 97 patients with primary gastric carcinoma, during surgery at the Seoul National University Hospital, Seoul, Korea. After surgical removal, the tissues were frozen immediately in liquid nitrogen and stored until required. Nine gastric cancer cell lines (SNU-1, -5, -16, -484, -601, -620, -638, -668, and -719) were obtained from the Korean Cell Line Bank (Seoul) and one hepatocellular carcinoma cell line (HepG2) used as a positive control was obtained from the American Type Culture Collection (Rockville, MD, USA). Cell lines were maintained in RPMI-1640 supplemented with 10% fetal bovine serum and gentamicin (10 mg/ml) at 371C in a humidified 5% CO2 atmosphere.

Sodium bisulfite-treated genomic DNAs were amplified using the following primers (nt. from þ 20 to þ 310): 50 -GTT TTT AGT TAG GAT ATG GT-30 (sense) and 50 -CTT CTT TCT ACA CAT CAA ACA -30 (antisense). The conditions of the amplification were as follows: 951C for 12 min, and 35 cycles of 951C for 30 s, 521C for 30 s, and 721C for 30 s, followed by final extension at 721C for 10 min. PCR reaction products were gelpurified and cloned into pEZ-T vector (RNA, Inc., Seoul, South Korea). The inserted PCR fragments of individual clones were sequenced using an ABI PRISM 377 DNA sequencer (Perkin-Elmer).

Northern blot analysis

In vitro methylation assay

mRNA was isolated as previously described (Song et al., 2001). Blots were hybridized with a random-labeled 1160-bp HindIII cDNA fragment of DLC-1 (GenBank Accession no. NM_006094) and a b-actin cDNA probe.

A 701-bp fragment corresponding to nucleotides 428 to þ 273 in the DLC-1 promoter region was amplified using normal human genomic DNA as a template with following primers: 50 -ACA CCT CCG CCA AGT AAA TG-30 (sense) and 50 -AAG CGC TGT GGG GAA GTC G-30 (antisense) and cloned into the pEZ-T vector (RNA, Inc.). The reporter construct was prepared by excising an insert from the T vector with NcoI, and subcloning this between the NcoI sites of the pGL3-basic vector (Promega, Madison, WI, USA). The sequences and orientations of the inserts were determined by direct sequencing. This reporter construct was treated overnight with three U of SssI (CpG) methylase (New England Biolabs, Beverly, MA, USA) per microgram of vector DNA in the presence (methylated) or absence (unmethylated) of 1 mm S-adenosylmethionine. Complete methylation of the vector DNA was confirmed by digestion with HpaII. Equal amounts of methylated and unmethylated vectors were introduced into the SNU-484 and SNU-601 cell lines using lipofectin (Life Technologies, Inc., Gaithersburg, MD, USA) with a bgalactosidase expression vector (pSV b-gal). Luciferase activity was measured 48 h after transfection using a TR717 Microplate Luminometer (Tropix, Inc., Bedford, MA, USA) and a Bioluminescent Reporter Gene Assay System (Tropix, Inc.) according to the manufacturer’s instructions. Relative luciferase activity was calculated and normalized versus b-galactosidase activity.

Southern blot analysis For Southern blotting, genomic DNA prepared using a DNA extraction kit (IntronBiotechnology, Seoul, South Korea) was digested with HindIII and separated by electrophoresis in 0.7% agarose gel. The 1160-bp DLC-1 cDNA and a b-actin cDNA probes were used. For the methylation study, genomic DNAs were digested overnight with EcoRI and then further digested with the methylation-sensitive restriction enzymes, BssHII or NaeI. DNA fragments were separated on a 1% agarose gel and transferred to a positively charged nylon membrane (Schleicher & Schuell, Keene, NH, USA). Blots were hybridized overnight with a random prime-labeled 448-bp DLC-1 probe (138 to þ 310 with respect to the putative transcription initiation site; GenBank Accession no. NM_006094 and AF404867) in 65% formamide/6 SSC/1% SDS at 421C. Putative transcription initiation site of the DLC-1 gene was determined by the longest 50 UTR (untranslated region) of the DLC-1 cDNA (GenBank Accession no. NM_006094). Sodium bisulfite treatment and MSP Genomic DNAs (1 mg) were treated with sodium bisulfite at 501C for 18–20 h using a CpGenome amplification kit (Intergen, Purchase, NY, USA). Sodium bisulfite-modified DNA was subject to MSP using primer sets that distinguish between unmethylated (U) and methylated (M) DNA. Primers designed on the basis of the DLC-1 promoter sequence information (GenBank Accession no. AF404867) contained six CpG sites. To discriminate between unmethylated and methylated products, primers were designed to generate different PCR products that varied in size by 6-bp (178 versus 172-bp). The unmethylated and methylated primer sequences were as follows: unmethylated reactions (nt. from 56 to þ 122), 50 -AAA CCC AAC AAA AAA ACC CAA CTA ACA-30 (sense) and 50 -TTT TTT AAA GAT TGA AAT GAG

Bisulfite sequencing

Drug treatment and RT–PCR Cells were seeded at a density of 1 106 cells/100 mm dish and treated 24 h later with varying concentrations of 5-Aza-dC for 4 days or 500 nm TSA for varying period. For the synergistic study, 1 mm 5-Aza-dC was present in culture for 4 days, and/or 500 nm TSA was added for the last 12 h. At the end of the treatment, the medium was removed and the RNA was isolated as described above. cDNA was synthesized using MMLV reverse transcriptase (Invitrogen, Carsbad, CA, USA) with random hexamers following the manufacturer’s instructions and amplified with primers specific for the DLC-1 gene (nt. from þ 313 to þ 574): 50 -GGA CAC CAT GAT CCT AAC AC-30 (sense) and 50 -CTC ATC CTC GTC TGA ATC Oncogene


3950 GT-30 (antisense). PCR reactions were performed as follows, 951C for 5 min, and 35 cycles of 951C for 30 s, 551C for 30 s, and 721C for 30 s, followed by incubation at 721C for 10 min. b-actin was amplified, at the same time, to estimate the efficiency of cDNA synthesis as described previously (Jong et al., 1999). PCR products were resolved on 1% agarose gels. ChIP assay Chip was performed using antibodies specific for acetylated histone H3 and H4, as described previously (Richon et al., 2000). Briefly, cells were plated at a density of 5 106 cells/15cm dish and incubated overnight. The next day, the cells were cultured in the presence of 500 nm TSA for 12 h. Formaldehyde was added directly to the culture media to a final concentration of 1% and incubated at 721C for 10 min. The cells were suspended in SDS lysis buffer (1% SDS/10 mm EDTA/50 mm Tris-HCl, pH 8.1), and the lysates were sonicated to shear the DNA into lengths of 200–1000 bp. An aliquot of the chromatin preparation was used as an internal control for the amount of

DNA (input). Immunoprecipitation was performed using antiacetylated histone H3 or H4 antibodies (Upstate Biotechnologies, Lake Placid, NY, USA) for 16 h at 41C. Immunocomplexes were collected by incubation with protein A-agarose for 2 h at 41C, washed five times for 5 min, and eluted twice with 250 ml of elution buffer (0.1 m NaHCO3/1% SDS). RNase A (0.03 mg/ml) and NaCl (0.3 m) were added, and crosslinks were reversed by incubation for 4–5 h at 651C. Samples were digested with Proteinase K (0.24 mg/ml) for 2 h at 451C. DNA was recovered by phenol/chloroform extraction and ethanol precipitation, and analysed by PCR using following primers (nt. from -243 to -120): 50 -AGA GGA GAG GCG GGG CCT-30 (sense) and 50 -CTT AGC GAC GGG CTG TTC TCC-30 (antisense). Acknowledgements This work was supported in part by grants from the Ministry of Science & Technology of Korea through the National Research Laboratory Program for Cancer Epigenetics and by 2002 BK21 Project for Medicine, Dentistry, and Pharmacy.

References Baffa R, Santoro R, Bullrich F, Mandes B, Ishii H and Croce CM. (2000). Clin. Cancer Res., 6, 1372–1377. Burgers WA, Fuks F and Kouzarides T. (2002). Trends Genet., 18, 275–277. Cameron EE, Bachman KE, Myohanen S, Herman JG and Baylin SB. (1999). Nat. Genet., 21, 103–107. Crespo P, Schuebel KE, Ostrom AA, Gutkind JS and Bustelo XR. (1997). Nature, 385, 169–172. Dobosy JR and Selker EU. (2001). Cell Mol. Life Sci., 58, 721– 727. Doetzlhofer A, Rotheneder H, Lagger G, Koranda M, Kurtev V, Brosch G, Wintersberger E and Seiser C. (1999). Mol. Cell. Biol., 19, 5504–5511. Emi M, Fujiwara Y, Nakajima T, Tsuchiya E, Tsuda H, Hirohashi S, Maeda Y, Tsuruta K, Miyaki M and Nakamura Y. (1992). Cancer Res., 52, 5368–5372. Euer N, Schwirzke M, Evtimova V, Burtscher H, Jarsch M, Tarin D and Weidle UH. (2002). Anticancer Res., 22, 733– 740. Fujiwara Y, Ohata H, Emi M, Okui K, Koyama K, Tsuchiya E, Nakajima T, Monden M, Mori T, Kurimasa A, Oshimura M and Nakmura Y. (1994). Genes Chromosomes Cancer, 10, 7–14. Geiman TM and Robertson KD. (2002). J. Cell Biochem., 87, 117–125. Guo WH, Weng LQ, Ito K, Chen LF, Nakanishi H, Tatematsu M and Ito Y. (2002). Oncogene, 28, 8351–8355. Hart MJ, Eva A, Evans T, Aaronson SA and Cerione RA. (1991). Nature, 354, 311–314. Herman JG, Graff JR, Myohanen S, Nelkin BD and Baylin SB. (1996a). Proc. Natl. Acad. Sci. USA, 93, 9821–9826. Herman JG, Jen J, Merlo A and Baylin SB. (1996b). Cancer Res., 56, 722–727. Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, Markowitz S, Willson JK, Hamilton SR, Kinzler KW, Kane MF, Kolodner RD, Vogelstein B, Kunkel TA and Baylin SB. (1998). Proc. Natl. Acad. Sci. USA, 95, 6870–6875. Huang L, Sowa Y, Sakai T and Pardee AB. (2000). Oncogene, 19, 5712–5719. Jones PA and Taylor SM. (1980). Cell, 20, 85–93. Oncogene

Jong HS, Park YI, Kim S, Sohn JH, Kang SH, Song SH, Bang YJ and Kim NK. (1999). Cancer (Philadelphia), 86, 559– 565. Kang SH, Bang YJ, Im YH, Yang HK, Lee DA, Lee HY, Lee HS, Kim NK and Kim S-J. (1999). Oncogene, 18, 7280–7286. Kikuchi T, Itoh F, Toyota M, Suzuki H, Yamamoto H, Fujita M, Hosokawa M and Imai K. (2002). Int. J. Cancer, 97, 272–277. Knosel T, Petersen S, Schwabe H, Schluns K, Stein U, Schlag PM, Dietel M and Petersen I. (2002). Virchows Arch., 440, 187–194. Kurimoto F, Gemma A, Hosoya Y, Seike M, Takenaka K, Uematsu K, Yoshimura A, Shibuya M and Kudoh S. (2001). Int. J. Mol. Med., 8, 89–93. Morii N, Kumagai N, Nur-E-Kamal MS, Narumiya S and Maruta H. (1993). J. Biol. Chem., 268, 27160–27163. Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Eisenman RN and Bird A. (1998). Nature, 393, 386–389. Ng IO, Liang ZD, Cao L and Lee TK. (2000). Cancer Res., 60, 6581–6584. Pannell D, Osborne CS, Yao S, Sukonnik T, Pasceri P, Karaiskakis A, Okano M, Li E, Lipshitz HD and Ellis J. (2000). EMBO J., 19, 5884–5894. Qiu RG, Chen J, McCormick F and Symons M. (1995). Proc. Natl. Acad. Sci. USA, 92, 11781–11785. Richon VM, Sandhoff TW, Rifkind RA and Marks PA. (2000). Proc. Natl. Acad. Sci. USA, 97, 10014–10019. Ridley AJ and Hall A. (1992). Cell, 70, 389–399. Robertson KD. (2002). Oncogene, 21, 5361–5379. Satoh A, Toyota M, Itoh F, Kikuchi T, Obata T, Sasaki Y, Suzuki H, Yawata A, Kusano M, Fujita M, Hosokawa M, Yanagihara K, Tokino T and Imai K. (2002). Br. J. Cancer, 86, 1817–1823. Sekimata M, Kabuyama Y, Emori Y and Homma Y. (1999). J. Biol. Chem., 274, 17757–17762. Sliva D, Mason R, Xiao H and English D. (2000). Biochem. Biophys. Res. Commun., 268, 471–479. Song SH, Jong HS, Choi HH, Inoue H, Tanabe T, Kim NK and Bang YJ. (2001). Cancer Res., 61, 4628–4635.


3951 Song SH, Jong HS, Choi HH, Kang SH, Ryu MH, Kim NK, Kim WH and Bang YJ. (2000). Int. J. Cancer, 87, 236–240. Sowa Y, Orita T, Minamikawa S, Nakano K, Mizuno T, Nomura H and Sakai T. (1997). Biochem. Biophys. Res. Commun., 241, 142–150. Stirzaker C, Millar DS, Paul CL, Warnecke PM, Harrison J, Vincent PC, Frommer M and Clark SJ. (1997). Cancer Res., 57, 2229–2237. Takakura M, Kyo S, Sowa Y, Wang Z, Yatabe N, Maida Y, Tanaka M and Inoue M. (2001). Nucleic Acids Res., 29, 3006–3011. Taniura S, Kamitani H, Watanabe T and Eling TE. (2002). J. Biol. Chem., 277, 16823–16830. Tsutsumi S, Gupta SK, Hogan V, Collard JG and Raz A. (2002). Cancer Res., 62, 4484–4490. Van Lint C, Emiliani S and Verdin E. (1996). Gene Expression, 5, 245–253.

Vignali M, Hassan AH, Neely KE and Workman JL. (2000). Mol. Cell Biol., 20, 1899–1910. Wade PA, Jones PL, Vermaak D, Veenstra GJ, Imhof A, Sera T, Tse C, Ge H, Shi YB, Hansen JC and Wolffe AP. (1998). Cold Spring Harb. Symp. Quant. Biol., 63, 435–445. Wiltshire RN, Dennis TR, Sondak VK and Meltzer PS. (2001). Trent. JM. Cancer Genet. Cytogenet., 131, 97–103. Wistuba II, Behrens C, Virmani AK, Milchgrub S, Syed S, Lam S, Mackay B, Minna JD and Gazdar AF. (1999). Cancer Res., 59, 1973–1979. Yaremko ML, Recant WM and Westbrook CA. (1995). Genes Chromosomes Cancer, 13, 186–191. Yuan BZ, Miller MJ, Keck CL, Zimonjic DB, Thorgeirsson SS and Popescu NC. (1998). Cancer Res., 58, 2196–2199. Zhang Y and Dufau ML. (2002). J. Biol. Chem., 277, 33431– 33438.

Oncogene