Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Sep 29.
Published in final edited form as: Oncogene. 2006 Apr 10;25(39):5446–5461. doi: 10.1038/sj.onc.1209533

Histone modifications silence the GATA transcription factor genes in ovarian cancer

C Caslini 1,3, CD Capo-chichi 1, IH Roland 1, E Nicolas 2, AT Yeung 2, X-X Xu 1
PMCID: PMC7523731  NIHMSID: NIHMS1626363  PMID: 16607277

Abstract

Altered expression of GATA factors was found and proposed as the underlying mechanism for dedifferentiation in ovarian carcinogenesis. In particular, GATA6 is lost or excluded from the nucleus in 85% of ovarian tumors and GATA4 expression is absent in majority of ovarian cancer cell lines. Here, we evaluated their DNA and histone epigenetic modifications in five ovarian epithelial and carcinoma cell lines (human ‘immortalized’ ovarian surface epithelium (HIO)-117, HIO-114, A2780, SKOV3 and ES2). GATA4 and GATA6 gene silencing was found to correlate with hypoacetylation of histones H3 and H4 and loss of histone H3/lysine K4 tri-methylation at their promoters in all lines. Conversely, histone H3/lysine K9 di-methylation and HP1γ association were not observed, excluding reorganization of GATA genes into heterochromatic structures. The histone deacetylase inhibitor trichostatin A, but not the DNA methylation inhibitor 5′-aza-2′-deoxycytidine, re-established the expression of GATA4 and/or GATA6 in A2780 and HIO-114 cells, correlating with increased histone H3 and H4 acetylation, histone H3 lysine K4 methylation and DNase I sensitivity at the promoters. Therefore, altered histone modification of the promoter loci is one mechanism responsible for the silencing of GATA transcription factors and the subsequent loss of a target gene, the tumor suppressor Disabled-2, in ovarian carcinogenesis.

Keywords: ovarian epithelial cells, ovarian carcinomas, chromatin, transcription repression, GATA transcription factors, Disabled-2 (DAB2)

Introduction

The GATA transcription factors play an essential role in lineage determination during development and function in maintenance of cell differentiation in adult tissues (Molkentin, 2000). Loss of GATA4, GATA5 and GATA6 factors has been implicated in ovarian, lung and gastric cancer development (Bai et al., 2000; Lassus et al., 2001; Akiyama et al., 2003; Capo-chichi et al., 2003; Guo et al., 2004). In addition, downregulation of the candidate tumor suppressor gene Disabled-2 (DAB2 for the gene and Dab2 for the protein), a GATA6 transcriptional target, has been observed in several cancers, including ovarian carcinomas (Mok et al., 1998; Fazili et al., 1999; Sheng et al., 2000). The loss of function of GATA factors leads to the loss of epithelial-specific differentiation markers such as Dab2, laminin and collagen IV, and it was suggested that the inappropriate expression of GATA factors causes the loss of the gene expression pattern responsible for epithelial characteristics (i.e., epithelial dedifferentiation) and subsequent transformation of ovarian surface epithelial cells (Capo-chichi et al., 2003).

The mechanism triggering the loss of function of GATA factors in ovarian cancer is not clear. Although deletions of chromosome regions at both GATA4 and GATA5 loci have been found in different types of cancer, they are not frequently observed in ovarian cancer (Bova et al., 1993; Fujiwara et al., 1993; Nemer et al., 1999; Lassus et al., 2001). Hence, other mechanisms must be considered to explain the frequent loss of function observed for the GATA transcription factors in ovarian cancer. Delocalization of GATA proteins from the nucleus to the cytoplasm, observed in a fraction of ovarian tumors, suggests that alteration of nucleocytoplasmic trafficking is one of these mechanisms (Capo-chichi et al., 2003). Alternatively, epigenetic modifications at the GATA gene promoters should be considered to explain the transcriptional silencing of these genes in the absence of genetic alterations.

Epigenetic modulation of chromatin through DNA methylation, histone modification or RNA-associated interference is thought to play a major role in the development of different types of diseases, especially cancer (Jones and Laird, 1999; Herman and Baylin, 2003; Egger et al., 2004; Hake et al., 2004). The mechanism for the epigenetic silencing of gene expression in cancer is an actively developing area of study. Aberrant methylation of CpG islands around promoter regions is generally believed to be the cause of gene silencing (Jones and Laird, 1999; Herman and Baylin, 2003; Egger et al., 2004). Methylation of the CpG islands recruits histone-modifying enzymes that covalently modify histone tails (Baylin and Bestor, 2002). Also, histone modification and gene silencing can occur before DNA methylation, such as in transgenes (Mutskov and Felsenfeld, 2004) and in the case of p16INK4a (Bachman et al., 2003). The histone modification subsequently leads to changes in chromatin structure and results in the alteration of transcriptional activity of the locus (Gregory et al., 2001; Rice and Allis, 2001; Berger, 2002). The association between various histone modifications and transcriptional activity is known as the ‘histone code’ (Jenuwein and Allis, 2001; Rice and Allis, 2001; Hake et al., 2004). Multiple acetylations at both histone H3 and H4 subunits are associated with active transcription and lack of histone acetylation (hypoacetylation) correlates with transcription silencing (Rice and Allis, 2001). Increased levels of acetylation at the ε-amino group of the histone H3 and H4 N-terminal lysine residues are achieved by the enzymatic activity of histone acetyltransferases (HATs) and are found in more decondensed chromatin regions, which are more accessible to many transcription factors. The removal of acetyl groups by histone deacetylases (HDACs) can lead to chromatin condensation of the unmodified histone and result in repression of gene transcription. Additionally, the unmodified histone lysine residues can be mono-, di- or tri-methylated by histone methyltransferases (HMTs) to further strengthen the epigenetic mark of the transcriptional status. Histone H3 di- or tri-methylation at lysine 4 also associates with active transcription (Rice and Allis, 2001; Santos-Rosa et al., 2002; Schneider et al., 2004). In contrast, histone H3 di- and tri-methylation at lysine 9 are associated with transcription silencing. These modifications recruit the chromodomain-containing proteins HP1α, β, and γ to the locus and leads to the assembling of heterochromatin, a transcriptional inactive state (Lachner et al., 2001). The histone modifications and associated proteins in a chromosome locus can be determined by the chromatin immunoprecipitation (ChIP) assay (Spencer et al., 2003; Umlauf et al., 2004).

To investigate further the causative factors of the loss of GATA factors in ovarian cancer, we examined the potential involvement of epigenetic mechanisms for GATA4 and GATA6 genes silencing in ovarian cancer cell lines.

Results

Close correlation between GATA6 and Dab2 loss with neoplastic morphological transformation of ovarian surface epithelia

Previously, we found that GATA4 and GATA6 are strongly expressed in ovarian surface epithelial cells in normal ovarian tissues, and either their expression or nuclear localization is lost in around 85% of ovarian tumors (Capo-chichi et al., 2003). A transcription activator/target relationship was established between GATA6 and Dab2 in cultured cells (Morrisey et al., 2000; Capo-chichi et al., 2003). To further verify the association between the expression and loss of GATA6 and Dab2 in tumor tissues, we examined ‘epithelial transition zones’, which are contiguous epithelia linking morphologically normal monolayer cells with neoplastic multilayer tumor cells. In the portion of ovarian tumors confined to ovaries (but not the tumors found in omentum), even in high-grade cancer, transitional epithelia can often be found, and we have observed a close correlation between the loss of Dab2 expression and the epithelial morphological transformation (Sheng et al., 2000; Yang et al., 2002a).

In 16 cases of epithelial transition zones analysed, consistently we found in all cases that loss of Dab2 expression associates with loss of GATA6 expression and correlates with morphological transformation, as shown in two examples (Figure 1). In the first ovarian carcinoma shown, GATA6 is expressed in the nucleus of monolayered ovarian surface epithelial cells (Figure 1a and b, arrow), and its expression is lost in the multilayered region of the ovarian epithelium (Figure 1a and b, double arrows). The loss of GATA6 correlates with the loss of the cytoplasmic staining of Dab2 (Figure 1c and d). Another example shows that the loss of GATA6 expression (Figure 1e and g) correlates with the loss of Dab2 (Figure 1f), gain of CA125 (an ovarian tumor-specific antigen) (Figure 1h) and morphological transformation. In the monolayered region of the epithelium, though most of the cells are positive for nuclear GATA6 staining, some cells (red arrowhead) lacking GATA6 are present (Figure 1g) but are still Dab2-positive (Figure 1f), suggesting that the loss of GATA6 precedes the loss of Dab2. Thus, these observations re-enforce the idea that loss of GATA6 leads to loss of its transcription target Dab2, and loss of Dab2 may be the cause of epithelial morphological transformation.

Figure 1.

Figure 1

Open in a new tab

Loss of GATA6 associated with loss of Disabled-2 (Dab2) and morphological transformation of ovarian surface epithelia. Adjacent sections of ovarian tumors containing contiguous links between monolayer epithelial and neoplastic cells were stained to detect GATA6, Dab2, and CA125 expression. (a–d) An example of ovarian carcinoma stained with GATA6 (a, b) and Dab2 (c, d) in adjacent sections. The areas indicated by ‘*’ are shown in a higher magnification in the right panels. GATA6 is positive in ovarian surface epithelium found in the benign portion of an ovarian tumor (arrow) and is lost in neoplastic lesions (double arrow). (e–h) In another example of ovarian carcinoma, in monolayer areas (arrow), both GATA6 (e, and area indicated by ‘*’ shown in a higher magnification in g), and Dab2 (f) are positive, and CA125 (h) is negative. The loss of GATA-6 correlates with the loss of Dab2 and the gain of CA125. Red arrowhead indicates GATA-6-negative cells.

Expressions of GATA transcription factors are frequently suppressed in ovarian cancer cell lines and their rapid induction by inhibition of histone deacetylases

To determine the mechanism for the loss of GATA factors in ovarian cancer cells, we chose a panel of representative ovarian epithelial and cancer cells with various status of GATA factors and Dab2 expression for analysis (Capo-chichi et al., 2003). These five cell lines are the human ‘immortalized’ ovarian surface epithelium (HIO) cells HIO-107 (GATA4 is positive, GATA6 and Dab2 are weak), HIO-114 (GATA4 is negative, GATA6 and Dab2 are positive), and tumor cells A2780 (GATA4, GATA6 and Dab2 all are negative), SKOV-3 (GATA4 is negative, GATA6 is positive, Dab2 is weak) and ES2 (GATA4 is negative, GATA6 and Dab2 are positive).

As it is known that GATA transcription factors can be silenced by epigenetic alterations (Akiyama et al., 2003; Guo et al., 2004), we examined the effects of the HDACs inhibitor trichostatin A (TSA) and DNA methylation inhibitor 5′-aza-2′-deoxycytidine (5′-Aza) on the re-expression of GATA4, GATA6 and Dab2 in tumor cell lines by both semiquantitative reverse transcription (RT)–polymerase chain reaction (PCR) assay (Figure 2a) and quantitative real-time RT–PCR (Figure 2b). GATA4 expression was restored robustly in HIO-114 and A2780 cells after exposure to TSA for 20h, but no expression of GATA4 was observed in either SKOV-3 or ES2 cells. The expression of both GATA6 and DAB2 genes was also restored in A2780 cells after TSA treatment. Thus, treatment of ovarian cancer cell lines with HDACs inhibitors can re-establish the expression of GATA4, GATA6 and DAB2 genes in some cell lines (HIO-114 and A2780), but the regulation of GATA factors expression is more complex in other cell lines (ES2 and SKOV3). In ES2 and SKOV3 cells, neither inhibition of histone deacetylation nor DNA methylation, separately or in combination, were sufficient to restore GATA4 expression (Figure 2b).

Figure 2.

Figure 2

Open in a new tab

Expression of GATA factors and Disabled-2 (Dab2) in ovarian surface epithelial and cancer cells, and effect of trichostatin A (TSA). A panel of ovarian epithelial and cancer cells were analysed for mRNA expression. (a) Cells were treated without (M) or with TSA (T) for 20h. RNA was isolated and used for semiquantitative reverse transcription (RT)–polymerase chain reaction (PCR) analysis. The figure shown is representative of six independent experiments. (b) Cells were treated with or without TSA, 5′-aza-2′-deoxycytidine (5′-Aza), or both. RNA was prepared for analysis using real-time, quantitative RT–PCR. An arbitrary unit of 40 was given to the amount detected in the cDNA preparation generated with 40ng of RNA of the reference sample (A2780+TSA for GATA4, ES2 for POLR2F and GATA6). The error bars represent the standard deviation. The levels of POL2RF expression indicating the total RNA and the efficiency of the reverse transcription are comparable in all samples. (c) Time-course analysis of re-expression of GATA4, GATA6 and Dab2 by semiquantitative RT–PCR following histone deacetylase inhibition. (d) During the same time course, histone proteins were isolated by acid extraction and their post-translational modification analysed by Western blotting using specific antibodies to Ac H3, Ac H4, Me2 H3K4 and Me3 H3K4. Coomassie staining of the gel indicates equal amount of total protein loaded per lane.

To test whether the re-expression of GATA4 or GATA6 temporally correlated with the inhibition of HDACs, HIO-114 and A2780 cells were treated with TSA and the time course of GATA gene expression was analysed by semiquantitative RT–PCR (Figure 2c). The change in genome-wide histone acetylation and methylation patterns was determined by Western blot analysis (Figure 2d). Within 4h of TSA addition, GATA4 mRNA was induced in HIO-114 and both GATA4 and GATA6 mRNAs in A2780 cells. In A2780 cells, the increase of DAB2 mRNA was detected at 4h, closely following the induction of GATA6. Trichostatin A treatment increased global accumulation of acetylated lysine 9 and 14 of histone H3 and lysine 5, 8, 12 and 16 of histone H4 in both cell lines within 1h (Figure 2d). Levels of di- and tri-methylation of lysine 4 of histone H3 also increased following TSA treatment.

Thus, the re-expression of GATA4, GATA6 and Dab2 is caused by the global alteration in the pattern of post-translational modification of lysines of histones H3 and H4 following the inhibition of HDACs. The closely correlated time course is consistent with Dab2 expression as a consequence of GATA6 expression, which is induced by the changes in histone modification of the locus following TSA treatment in A2780 cells.

Lack of CpG island methylation at the GATA4 and GATA6 promoters in A2780 ovarian cancer cell lines

We next wanted to confirm the DNA methylation status of GATA4 and GATA6 promoters. The 5′ sequence and promoter regions of each gene were obtained from previous reports and identified by database searches on the National Center for Biotechnology Information. Both GATA4 and GATA6 genes contain CpG-rich regions at the 5′ of exon 1 (Figure 3a and b), which was subjected to methylation analysis by bisulfite treatment/sequencing and methylation-specific PCR (MSP) approaches. The GATA6 promoter in A2780 cells was found largely unmethylated by both techniques (Figure 3c and d). Using the MSP assay, both GATA4 and GATA6 promoters were found unmethylated in SKOV-3, A2780, HIO-114 and HIO-107 cells. In ES2 cells, the GATA6 promoter was found unmethylated (Figure 3c). However, we have not been able to amplify the GATA4 promoter by two primer pairs for either methylated or unmethylated sequences (Table 3), suggesting possible mutation or polymorphic sequences of the GATA4 gene in ES2 cells.

Figure 3.

Figure 3

Open in a new tab

GATA gene promoter methylation status in ovarian surface epithelial and cancer cells. Promoter regions of GATA4 (a) and GATA6 (b) are illustrated and the location of methylation-specific primers (MSP) is indicated. The frequency of CpG sequence is illustrated by plotting the number of CpG sites per 100bp along the 5′ sequences of the genes. (c) The methylation status of the GATA4 and GATA6 promoter regions was analysed by methylation-specific polymerase chain reaction (PCR) showing PCR products specific, indicative of either methylated ‘M’ or unmethylated ‘U’ genomic loci. (d) DNA from A2780 cells was analysed by bisulfite sequencing at a region (BS) of the GATA6 promoter. Sequences from six clones of PCR products are shown with the methylation status of the CpG sites indicated. The results are representative of five independent experiments with similar findings.

Table 3.

Primer information

Analysis Primer name Primer 5′–3′ sequence Size (bp)
Semiquantitative RT–PCR Dab2 F4: ACATCTTTGCTCCTCCCGTCTC —
Dab2 R4: GCTGGTTCCAACCTGAAACAGC 303
Quantitative RT–PCR GATA-6 F1: TTGTGGACTCTACATGAAACTCCA —
GATA-6 R1: TTATGTTCTTAGGTTTTCGTTTCCTG 95
POLR2F F: TGCCATGAAGGAACTCAAGG —
POLR2F R: TCATAGCTCCCATCTGGCAG 71
POLR2F P: CCCCATCATCATTCGCCGTTACC —
Semiquantitative and quantitative ChIP/PCR GATA4 F1: CCGAGAGAGGACTTGAATGGG —
GATA4 R1: CTGCTTTTTTGGGGGTGTTTC 327
GATA4 F3: ACAGGAGATGGGAAGTGTCGC —
GATA4 R3: GGTGACCTCTTGGGCTCAACTC 289
GATA6 F2: CATTTCCAGTCCCTTTTGCCC —
GATA6 R2: TTCCACATCAGTCGTGTCCGAG 259
GATA6 F14: TGCTTTCTTGCTGGCACTTCAC —
GATA6 R14: GCTGAGACAAACACTATGGTGGG 308
GATA6 F23: GGGACAGGGATTCTTTTGTTGG —
GATA6 R23: TGATTCACCAGAGGTCTCAAGCC 250
Bisulfite sequencing GATA6 F1: TAATGGATGAGGGGAGGGTTG —
GATA6 F1.2: TGGATGAGGGGAGGGTTGGAG 570
GATA6 R1: CAAAACCCCCCACCCCTAC 567
Methylation-specific PCR GATA4 M F: TTTGGGTATTATAGCGAATTTAATC —
GATA4 M R: AAACGCAAACTACGAAACTATACG 154
GATA4 U F: TTGGGTATTATAGTGAATTTAATTGA —
GATA4 U R: AACACAAACTACAAAACTATACAAA 152
GATA-4 M F2: ATTATAGCGAATTTAATCGATTTTC —
GATA-4 M R2: GACTACACCTCCGCTAAACG 162
GATA-4 U F2: TATAGTGAATTTAATTGATTTTTGG —
GATA-4 U R2: CCAACTACACCTCCACTAAACAC 162
GATA6 M F: TAAGGTTTGGAGCGTTTTTTTC —
GATA6 M R: AAAACGCCTAACCTAACTAACGAC 175
GATA6 U F: TTTAAGGTTTGGAGTGTTTTTTTTG —
GATA6 U R: CAAAACACCTAACCTAACTAACAAC 178
Open in a new tab

Abbreviations: ChIP, chromatin immunoprecipitation; RT–PCR, reverse transcription–polymerase chain reaction.

These results are consistent with the finding that the DNA methyltransferase inhibitor, 5′-Aza, had no effect on re-expression of GATA genes or synergistic or enhancing activity with HDACs inhibitor TSA in these several ovarian epithelial and cancer cells analysed.

Histone modifications at the GATA4 and GATA6 promoters in ovarian cancer cell lines

As the re-expression of GATA4 and GATA6 factors observed following treatment of ovarian cancer cell lines with TSA temporally correlates with their genome-wide modification of histones, we performed ChIP (Spencer et al., 2003; Umlauf et al., 2004) to verify the histone modification of promoter regions of GATA4 and GATA6 genes. A panel of antibodies against various modified forms of histones, including multi-acetylated histone H3 or H4, histone H3 di- or tri-methylated at lysine 4 and histone H3 di-methylated at lysine 9, was used. Antibodies to HP1γ were used to determine whether the gene loci are associated with heterochromatin. Antibodies against GATA4 and GATA6 were also used in the ChIP assay to detect possible auto- or inter-regulation of the GATA factors. The input signal represents 2% of chromatin input in the ChIP assay, and was used as a positive control for comparison.

The GATA4-positive HIO-107 cells exhibit hyperacetylation of both histones H3 and H4 at the GATA4 promoter, with higher acetylation occurring more proximal to the untranslated exon 1 of GATA4 (Figure 4a–c). GATA4 and GATA6 transcription factors were not found associated in proximity to exon 1. Consistent with the absence of GATA4 expression, both semiquantitative analysis (Figure 4b and c) and quantification by real-time PCR (Table 1) of the ChIP assays showed GATA4 promoter hypoacetylation in HIO-114, A2780, ES2 and SKOV3 cells.

Figure 4.

Figure 4

Open in a new tab

Chromatin histone modification at GATA4 and GATA6 promoters in ovarian surface epithelial and cancer cells. (a) The GATA4 gene 5′ region is illustrated and two regions 5′ of exon 1 amplified by F1/R1 and F3/R3 primer pairs are indicated. (b) The chromatin immunoprecipitation (ChIP) assay was performed using antibodies to GATA4, GATA6, Ac H3, Ac H4, Me2 H3K4, Me3 H3K4, Me2 H3K9 and HP1γ. Isolated DNA was used for amplification of GATA4 promoter loci using F1/R1 and/or F3/R3 primer pairs. The bright field images of the polymerase chain reaction (PCR) products are shown. (c) The ChIP assay was quantitatively determined by ImageJ program, and the results are presented as the percentage of the input signal generated with 2% of the chromatin that was immunoprecipitated by the specific antibodies in each cell lines. (d) The GATA6 gene 5′ region is illustrated and three regions amplified by F14/R14, F23/R23 and F2/R2 primer pairs are indicated. (e) Chromatin immunoprecipitation assay was performed and the isolated DNA was used for amplification of GATA6 promoter locus using the three primer pairs. (f) The PCR products of the ChIP assay were quantitatively determined by ImageJ program. (g) The relative value of a ChIP assay quantitatively determined by real-time PCR is presented. The error bars represent the standard deviation of the assay. The scheme here and data in Table 1 represent two independent sets of experiments. ‘*’, Chromatin immunoprecipitation assay was not performed in human ‘immortalized’ ovarian surface epithelium (HIO)-107 and HIO-114 using antibodies to Me3 H3K4. ‘In’, for PCR of total input chromatin; ‘no’, for no antibodies; and ‘H2O’, for no DNA/chromatin. Four independent ChIP experiments were carried out to verify these representative results.

Table 1.

Real-time PCR in ChIP assay to determine histone modification of GATA4 and GATA6 promoter loci in HIO and ovarian cancer cells

Antibodies A2780 SKOV3 ES2 HIO-107 HIO-114
GATA4
 No. of Ab <4 <4 <4 — —
 Ac H3 5.8 24.7 <4 ND ND
 Ac H4 9.6 5.0 6.3 ND ND
 Me2 H3(K4) 136.0 34.4 24.3 ND ND
 Me3 H3(K4) 4.7 <4 <4 ND ND
 Me2 H3(K9) <4 <4 <4 ND ND
 HP1γ <4 <4 <4 ND ND
GATA6
 No. of Ab <4 <4 <4 <4 4.4
 Ac H3 10.9 283.5 49.9 29.85 79.75
 Ac H4 11.1 90.8 73.7 31.7 55.9
 Me2 H3(K4) 36.9 683.9 386.0 353.6 395
 Me3 H3(K4) 4.9 67.8 22.8 ND ND
 Me2 H3(K9) <4 <4 <4 <4 <4
 HP1γ <4 <4 <4 <4 <4
Open in a new tab

Abbreviations: Ab, antibodies; ChIP, chromatin immunoprecipitation; HIO, human ‘immortalized’ ovarian surface epithelium cells; ND, not determined; <4, background level. The deviations of the real-time PCR results are normally < 10% of the value and are not shown here. The number indicates the relative abundance of DNA copies of the GATA4 or GATA6 loci associated with the immunoprecipitated proteins.

Histone H3 methylation at lysine 9 contributes to the repression of gene transcription by creating a binding site for the chromatin organization modifier (chromo) domain of the heterochromatin assembly proteins HP1α, β and γ. Neither histone H3/lysine 9 di-methylation nor HP1γ association was found at the GATA4 promoter of any cell line analysed, excluding a mechanism of silencing through heterochromatic reorganization of the GATA4 locus (Figure 4b, Table 1). Compared with the other four cell lines, the GATA4-positive HIO-107 cells exhibit higher acetylation of both histone H3 and H4 at the GATA4 locus. Thus, histone H3 and H4 hypoacetylation of the gene locus correlates with and is likely the causative factor for the loss of GATA4 expression in these cells.

The ChIP analysis at the GATA6 promoters by semiquantitative PCR (Figure 4d–f) or quantitative PCR (Figure 4g and Table 1) also showed a strong correlation among the patterns of histone acetylation and methylation and the expression status. In ES2 and SKOV3 cell lines, high levels of histone H3 and H4 acetylation and histone H3/lysine 4 tri-methylation correlated with GATA6 gene expression. The contrast is seen in A2780 cells (Figure 4d–g). Nevertheless, no di-methylation at histone H3/lysine 9 or HP1γ association was observed at GATA6 promoter, excluding the reorganization of the locus into heterochromatic structures.

Inhibition of histone deacetylation re-establishes the transcription-associated histone modifications at the GATA4 and GATA6 promoters

We performed the ChIP assay on GATA4 and GATA6 promoter loci to determine whether the patterns of chromatin modification are altered upon TSA treatment. Trichostatin A increased the acetylation of histone H3 and H4 at both GATA4 and GATA6 promoter loci, determined by both semiquantitative PCR (Figure 5a) and real-time quantitative PCR (Figure 5b, and Table 2) in A2780 cells. Additionally, an elevation of the di- and tri-methylation of histone H3/lysine 4 occurred at the promoter region of GATA4 and GATA6 genes, which presumably is the consequence of histone hyperacetylation caused by the addition of TSA.

Figure 5.

Figure 5

Open in a new tab

Alteration of histone modification at GATA4 and GATA6 promoters by trichostatin A (TSA) in ovarian cancer cells. Cells were incubated with or without TSA for 20h. Chromatin immunoprecipitation (ChIP) was used for the analysis of GATA4 (primer pair F3/R3) and GATA6 (primer pair F2/R2) promoter regions. (a) Chromatin immunoprecipitation assay of A2780 cells. The semiquantitative polymerase chain reaction (PCR) result is shown in bright field images. (b) The ChIP assay was determined by quantitative real-time PCR for GATA4 and GATA6 promoters. The error bars represent the standard deviation of results in the real-time reverse transcription (RT)–PCR experiments. The quantitative PCR results for the entire set of experiments are summarized in Table 2. (c) Gels of the semiquantitative PCR ChIP result are shown for GATA4 promoter. (d) The ChIP was quantified by real-time PCR for GATA4 promoter locus (primer pair F3/R3) comparing the effects of TSA in A2780, ES2 and SKOV-3 cells. Two independent experiments showed similar results.

Table 2.

Real-time PCR in ChIP assay to determine histone modification of GATA4 and GATA6 promoter loci in ovarian cancer cells with or without TSA treatment

Gene GATA4
 GATA6
Antibodies A2780 A2780+TSA SKOV3 SKOV3+TSA ES2 ES2+TSA A2780 A2780+TSA
No. of Ab 7.2 <4 <4 <4 <4 <4 <4 <4
Ac H3 28.3 87.3 19.0 26.8 9.3 16.7 10.8 41.1
Ac H4 62.5 224.1 52.7 73.5 20.9 46.1 11.7 76.5
Me2 H3(K4) 200.6 368.6 104.7 103.3 56.6 48.6 188.7 380.7
Me3 H3(K4) 9.3 73.2 4.3 13.9 4.1 4.4 7.4 93.9
Me2 H3(K9) <4 <4 <4 <4 <4 <4 <4 <4
HP1γ <4 <4 <4 <4 <4 <4 <4 <4
Open in a new tab

Abbreviations: Ab, antibodies; ChIP, chromatin immunoprecipitation; TSA, trichostatin A. See Table 1 for explanation.

At the GATA6 locus in A2780 cells, TSA increased Me2 H3K4 level from 188.7 to 380.7, as determined by real-time PCR (Table 2), compared to 683.9 and 386.0 in the GATA6-positive SKOV-3 and ES2 cells, respectively (Table 1). The re-expression of GATA6 by TSA treatment in A2780 cells also correlates well with the increase in Me3 H3K4, from 7.4 to 93.9 (Table 2), compared to 67.8 and 22.8 in SKOV3 and ES2 cells (Table 1), respectively. In A2780 cells, TSA treatment caused a very similar magnitude in the changes of histone modification of the GATA4 promoter locus as those of the GATA6 locus (Figure 5a and b).

An unexpected finding is that GATA4 expression was not restored in ES2 or SKOV-3 cells following TSA treatment alone or in combination with 5′-Aza to inhibit DNA methylation (Figure 2). Consistently, we did not observe significant changes in histone modification of the GATA4 locus in ES2 and SKOV3 cells following TSA treatment (Figure 5c and d, Table 2). Thus, inhibition of histone deacetylation is insufficient to alter the transcription inactive pattern of histone modification and restore GATA4 gene expression in ES2 and SKOV3 cells.

Inhibition of histone deacetylation increases the DNase I sensitivity of GATA4 and GATA6 promoters

To test whether the inhibition of HDACs was able to remodel the chromatin structure at the GATA4 and GATA6 genes into a transcriptionally permissive configuration, we determined the effects of TSA on the DNase I sensitivity of GATA4 and GATA6 promoters.

We designed a strategy to examine a 3.8kb XmnI restriction fragment for GATA4 gene (Figure 6a) and a 5.4kb BamH1 restriction fragment for GATA6 gene (Figure 6c) for their sensitivity to DNase I digestion. The DNase I sensitivity of these fragments may reflect the accessibility of the DNA to the hydrolytic enzyme and thus the conformation of the promoters. Probes located at the 5′ end of the fragments were used to detect the presence and sizes of the DNA fragments by Southern blot.

Figure 6.

Figure 6

Open in a new tab

DNase I hypersensitivity of GATA4 and GATA6 promoters in ovarian cancer cells and effect of trichostatin A (TSA). (a) Schematic illustration of the GATA4 promoter region indicating the restriction sites of the XmnI (X) restriction enzyme used for complete digestion and location of the probe (1260bp) used for Southern blot hybridization. (b) DNase I hypersensitivity assay of the GATA4 promoter in ES2 and A2780 cells cultured without or with TSA. The parental band indicated (arrow) is the XmnI fragment including exon 1 of the GATA4 gene. (c) Schematic illustration of the GATA6 promoter region indicating the restriction sites of the BamH1 (b), and location of the probe (985bp) used for Southern blot hybridization. (d) DNase I hypersensitivity assay of the GATA6 promoter in ES2 and A2780 cells treated without or with TSA. Parental band (arrow) and DNase I hypersensitivity bands (*) are indicated. The experiment was repeated and a similar conclusion was reached.

DNase I sensitivity of both GATA4 and GATA6 genes was assayed by digesting isolated nuclei from ES2 and A2780 cells with an increasing amount of DNase I (Figure 6). Trichostatin A had no significant impact on the sensitivity of the GATA4 promoter fragment to DNase I in ES2 cells (Figure 6b). This is consistent with the absence of GATA4 expression in ES2 cells and the lack of effect of TSA. Conversely, the GATA4 promoter fragment in A2780 cells is more sensitive to DNase I digestion and TSA treatment increased its DNase I sensitivity.

The GATA6 promoter fragment of the GATA6-positive ES2 cells is highly sensitive to DNase digestion, though TSA treatment only slightly further increased this sensitivity (Figure 6d). The promoter of the GATA6-negative A2780 cells is less sensitive to DNase I digestion: at 160U/ml of DNase I, the GATA6 promoter fragment was only partially digested (Figure 6d). Treatment with TSA increased the digestion of the A2780 GATA6 promoter to the extent similar to that of ES2 cells.

Thus, DNase I sensitivity of the GATA4 and GATA6 promoters correlates well with histone modification and gene expression. These results suggest that, in A2780 cells, TSA treatment leads to a more open chromatin structure at both the GATA4 and GATA6 promoters, associated with induction of their expression. In ES2 cells the GATA6 gene is actively transcribed and therefore the promoter is highly accessible to DNase I digestion. Trichostatin A treatment appears insufficient to open the GATA4 promoter chromatin conformation in ES2 cells, and thus GATA4 expression is not inducible by TSA in the cells.

Global changes in histone modification and histone deacetylase activity in ovarian cancer cells

Prompted by the recent report that general loss of histone H4 acetylation at Lys16 is a hallmark of human cancer (Fraga et al., 2005), we examined the panel of ovarian cell lines for global histone modification. We found that A2780 is unique, in that the total histone H4 acetylation is reduced, but histone H3 acetylation or trimethylation are comparable to the other cell lines (Figure 7a). The cellular HDAC activities in the panel of cells are comparable (Figure 7b), which cannot be attributed to the global loss of histone H4 acetylation in A2780 cells and also do not explain the histone hypoacetylation and silencing of GATA gene loci in ovarian cancer cells.

Figure 7.

Figure 7

Open in a new tab

Comparison of the global histone modification in cell lines. (a) Cells were treated without or with trichostatin A (TSA) for 20h. Histone proteins were isolated by acid-extraction and their post-translational modification was analysed by Western blotting using specific antibodies to Ac H3, Ac H4 and Me3 H3K4. This experiment was repeated and confirmed. (b) Cells were treated with or without TSA for 3h, preparations of cell total extract, cytosolic or nuclear fractions were determined for histone deacetylase (HDAC) activity as represented by the fluorescent units. The results were obtained from a triplicate assay and the error bars represent the standard deviation.

Factors associated with active and silent GATA4 and GATA6 loci in ovarian cancer cells

To explore further the epigenetic determining factors, we examined a panel of candidate proteins for their differential association with either active or silent GATA4 and GATA6 promoter loci by ChIP. Comparing ES2 and A2780 with or without TSA, we found no significant difference in the association of methylated DNA-binding proteins to GATA promoters (MeCP1, MeCP2, and MBD2). HDAC1 and HDAC2 were found to be weakly associated with both silent and active loci.

However, Bmi1, a member of the Polycomb group genes functioning in chromatin remodeling and transcription repression (Park et al., 2004; Valk-Lingbeek et al., 2004), did associate with the inactive GATA4 and GATA6 promoter loci in A2780 cells, but not the active GATA6 promoter loci in ES2 cells (Figure 8). Additionally, the association of Bmi1 exhibited a 2–3-fold reduction upon treating A2780 cells with TSA (Figure 8b), correlating with an increase in gene expression. Thus, Bmi1 may be involved in the binding and silencing of GATA loci in A2780 cells, which will be investigated in future studies.

Figure 8.

Figure 8

Open in a new tab

Comparison of GATA4 and GATA6 promoter-associated proteins in A2780 and ES2 cells and effect of trichostatin A (TSA). (a) ES2 and A2780 cells were treated without (M) or with (T) 100ng/ml of TSA for 4h. GATA4 and GATA6 promoter regions were analysed by chromatin immunoprecipitation using primers F3/R3 and F2/R2, respectively, and a panel of antibodies. The result shown represents two independent experiments. (b) The result was quantitatively determined by ImageJ program and presented as the percentage of the input signal generated with 2% of the immunoprecipitated chromatin.

Discussion

In the present study, we determined that a causative mechanism for the loss of GATA4 and GATA6 in ovarian carcinomas is the alteration of chromatin conformation, including the hypoacetylation of histone H3 and H4 and the associated reduction in di- and tri-methylation of histone H3 lysine 4. In certain cells (HIO-114 and A2780), inhibition of HDACs is sufficient to restore GATA4 and/or GATA6 expression, resulting in the re-expression of Dab2, a target of the GATA6 transcription factor. Specifically, we observed histone H3 and H4 hypoacetylation and loss of histone H3/lysine 4 tri-methylation at the promoter of silenced GATA4 and GATA6 genes. An increase in HDAC or a decrease in HAT activity specifically associated with GATA4 or GATA6 gene loci, but not a change in their global activities, may account for the promoter histone H3 and H4 hypoacetylation. High levels of histone H3/lysine 4 di-methylation, along with the absence of histone H3/lysine 9 di-methylation or HP1γ association of the loci, suggest that the genes exhibit euchromatic, but not heterochromatic, structures at these loci. We also demonstrated that the patterns of histone modification correlate with chromatin conformation, as probed by the sensitivity to DNase I digestion. Lastly, we identify Bmi1, a Polycomb group gene previously linked to the stem cell characteristics in cancer (Park et al., 2004; Valk-Lingbeek et al., 2004), as a candidate for the binding and silencing of GATA loci in A2780 cells.

A recent study describes that DNA methylation of their promoter is responsible for the transcriptional silencing of GATA4 and GATA5 in lung, colorectal and gastric cancer cell lines (Akiyama et al., 2003; Guo et al., 2004). In that study, the silenced GATA4 and GATA5 genes were re-activated by treatment with 5′-Aza in every colorectal cancer and gastric cancer cell line tested, whereas treatment of the gastric cancer cell line AZ521 with the HDACs inhibitor, TSA, failed to reactivate GATA4 and GATA5 gene expression (Akiyama et al., 2003). Conversely, the GATA6 promoter was found unmethylated and the gene expressed in all of the gastrointestinal cancer cell lines analysed, though chromatin structure was not examined.

In comparison, we found that the epigenetic mechanisms for silencing of GATA genes are somewhat different in ovarian cancer cells. In the ovarian epithelial and cancer cell line HIO-114 and A2780, inhibition of HDACs by TSA is sufficient to restore the acetylation of histone and re-expression of GATA4 and/or GATA6 genes. Treatment with 5′-Aza individually or in combination with TSA did not restore or enhance the expression of the genes, which is consistent with the finding that both the GATA4 and GATA6 promoters are unmethylated in these cells. These data indicate that both histone modifications and DNA methylation can act independently or cooperatively to silence the GATA transcription factors and their target genes in different types of cancers. The reason for different mechanisms in silencing of GATA factors in cancer development is currently unknown and may rely on the tissue and/or organ specificity of certain chromatin-modifying enzymes or protein effectors that read the epigenetic modifications (Hake et al., 2004). In A2780 cells, inhibition of histone deacetylation not only led to increase in histone H3 and H4 acetylation, but also resulted in histone H3K4 methylation. It is likely that histone hyperacetylation is a signal to recruit HMT enzyme to the loci. In ES2 and SKOV3 cells, however, both inhibition of histone deacetylation and DNA methylation are not sufficient to induce expression of GATA4. Likely, histone acetylation by HATs is inactive at the locus despite the inhibition of HDACs, or histone H3K4 methyltransferase is not recruited to further modify the locus to stabilize the acetylation of the histone and the maintenance of an active chromatin conformation in these two cell lines. Consistently, the GATA4 promoter fragment was found to be insensitive to DNase I digestion and TSA treatment did not increase the sensitivity in ES2 cells.

Adding to the cases of GATA factor disregulation, a recent report (Wakana et al., 2005) shows that GATA4 and GATA5 promoters are methylated in three and two, respectively, out of eight ovarian cancer cell lines examined. Thus, the mechanism for the silencing of GATA factors appears to vary among cell lines.

Although loss of GATA4 expression was observed in a fraction (14%) of ovarian carcinomas (Capo-chichi et al., 2003), GATA4 expression is lost in most cultured ovarian epithelial cell lines, either non-tumorigenic or cancerous. Re-expression of GATA4 suppresses cell growth (Capo-chichi et al., 2003), suggesting that the loss of GATA4 may enhance tumor malignancy and augment growth and adaptation to culture conditions. GATA6 is inactivated by either loss of expression or mislocalization in most (82%) ovarian tumors and cancer cells (Capo-chichi et al., 2003). In many ovarian cancer cells other than A2780, GATA6 mRNA can often be detected by RT–PCR, but the level is low, as judged by Northern blotting (Capo-chichi et al., 2003; Wakana et al., 2005).

Several GATA6 transcriptional targets, including Dab2 and collagen IV, have been suggested from the analysis of GATA6 knockout embryonic stem cells (Morrisey et al., 2000). The consequential loss of Dab2 and laminin following GATA6 suppression by siRNA was demonstrated in ovarian epithelial cells (Capo-chichi et al., 2003). Based on the idea that GATA6 determines cell lineage differentiation in development, the loss of GATA6 functions and associated transcription targets such as the epithelial-specific markers Dab2, collagen IV and laminin was speculated to be the underlying mechanism for epithelial de-differentiation in ovarian cancer (Capo-chichi et al., 2003). Dab2 is considered to be a tumor suppressor and the inactivation of DAB2 is an early step in ovarian tumorigenicity (Fazili et al., 1999). Dab2 is thought to have a role in the structural organization of an epithelium, and loss of Dab2 leads to disorganized proliferation in cancer (Sheng et al., 2000). Additionally, Dab2 knockout in mice leads to early embryonic lethality, owing to the disorganization of the primitive endoderm layer, an epithelial cell type of early embryos (Yang et al., 2002b). This phenotype of Dab2 loss on the structure of the extraembryonic endodermal epithelium reminisces the close correlation between Dab2 loss and morphological neoplastic transformation of ovarian surface epithelia. Thus, the knockout study substantiates the role of Dab2 in epithelial organization, that underlies its mechanism in tumor suppression.

The association between the loss of GATA6, Dab2 and the neoplastic morphological transformation of ovarian surface epithelia reported in the current study may provide some clues to the early events in cancer development. A scenario can be postulated that impairment in executing epigenetic inherent markers such as the histone acetylation status of GATA6 promoter leads to the loss of GATA6 expression in some cells (such as the cells indicated by arrowhead in Figure 1g). The absence of GATA6 leads to the loss of expression of its transcription targets including Dab2, and loss of Dab2 subsequently allows cells with oncogenic potential (containing mutations that allow proliferation and survival) to escape the constraint of epithelial structure and undergo neoplastic growth. Such a scenario is consistent with observations that mutations such as R-Raf (Ho et al., 2004) or p53 (Zheng et al., 1993, 1995) are already present in monolayer, morphologically benign ovarian surface epithelial cells that are contiguous with neoplastic lesions. Thus, we consider that the events in morphological transformation, the loss of GATA6 and the consequential loss of Dab2, may represent a necessary step in carcinogenesis in collaboration with the activation of oncogenes and inactivation of tumor suppressor genes.

This study suggests that loss of GATA6 due to alteration of chromatin modification leads to the loss of Dab2, which allows the oncogenic competent cells to escape the constraint of epithelial organization and further results in morphological neoplastic transformation of ovarian surface epithelial cells. The finding presents an example of the causative mechanism of chromatin alteration in carcinogenesis.

Materials and methods

Cell cultures and treatments

The HIO cell lines HIO-107 and HIO-114 were established by the laboratory of Dr Andrew Godwin (Fox Chase Cancer Center, Philadelphia, PA, USA) and were described previously (Capo-chichi et al., 2003). Ovarian cancer cell lines A2780, SKOV-3 and ES2 were obtained from the American Type Culture Collection. The HIO cell lines were cultured in a 1:1 mixture of Media 199 and MCDB-105 media supplemented with 4% fetal calf serum (FCS), 3mM L-glutamine, 100U/ml penicillin/streptomycin and 0.2IU/ml pork insulin (Eli Lilly, Indianapolis, IN, USA). The ovarian cancer cell lines were routinely maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS, 3mM L-glutamine, 100U/ml penicillin/streptomycin (DMEM-F). Cells were seeded at low density 48h before treatment with 100ng/ml of TSA (Sigma, St. Louis, MO, USA) or 5 μM 5′-aza-2′ deoxycytidine (5′-Aza, Sigma). For combined treatment, 5′-Aza was added first and TSA was added 20h before cell harvesting.

Northern blot analysis

Total RNA isolated from 100mm plates of 80% confluent cells was used for Northern blot analysis as described previously (Capo-chichi et al., 2003). Probes for human GATA6 were a generous gift from Dr David Wilson (Washington University, St Louis, MO, USA) and from Dr Kenneth Walsh (St Elizabeth’s Medical Center, Boston, MA, USA). The probes for human DAB2 and GATA4 were reported previously (Capo-chichi et al., 2003).

Semiquantitative reverse transcription–PCR

Total RNA was purified using the RNAqueous total RNA isolation system (Ambion, Austin, TX, USA) and traces of DNA were removed using DNA-free™ (Ambion). For RT–PCR, 0.4 μM of total RNA was reverse-transcribed according to the Superscript first-strand cDNA synthesis system (Invitrogen, Carsbad, CA, USA) and 2 μl of RT reaction amplified by Platinum Taq DNA polymerase (Invitrogen) in a 50 μl reaction mixture. We used the previously described primers and conditions for the amplification of GATA4, GATA5 and GATA6 (Bai et al., 2000). DAB2 was amplified using the primers hDAB2 F4 and hDAB2 R4 (Table 3). Amplification by 30 cycles was used for the semiquantitative PCR expression analysis of all these genes. After electrophoresis in 2% agarose with ethidium bromide the bands were visualized with an Alpha Imager 2200 Documentation & Analysis System (Alpha Innotech Corporation, San Leandro, CA, USA). The intensities of the signals were found to correlate with expression detected by Northern blot.

Chromatin immunoprecipitation and PCR analysis

Chromatin immunoprecipitation (Spencer et al., 2003; Umlauf et al., 2004) was carried out using the ChIP assay kit (Upstate Biotechnology Laboratories, Lake Placid, NY, USA) essentially as described by the manufacturer. Briefly, 2×106 cells were seeded approximately 24–48h before harvest. The cells were crosslinked for 10min at room temperature by addition of formaldehyde to a final concentration (f.c.) of 1%. The reaction was stopped by adding glycine to an f.c. of 0.125M. Cells were resuspended in sodium dodecyl sulfate (SDS) lysis buffer (50mM Tris-Cl, pH 8.0; 10mM ethylenediaminetetraacetic acid (EDTA); 1% SDS; 1mM phenylmethylsulfonyl fluoride (PMSF); complete protease inhibitor cocktail tablets (Roche, Indianapolis, IN, USA)), and the chromatin was sonicated 10 times for 10s each using a 2mm tip and 60% power to generate fragments averaging 1–2kb. GammaBind Plus sepharose beads (Amersham Biosciences) were incubated with the chromatin for 1h at 4°C. The precleared chromatin was incubated overnight at 4°C with 4 μg of the following primary antibodies: rabbit anti-acetyl histone H3 (06–599, Upstate Biotech, Inc. (UBI, Charlottesville, VA, USA)); rabbit anti-acetyl histone H4 (06–866, UBI); rabbit anti-dimethyl histone H3 (K4) (07–030, UBI); rabbit anti-trimethyl histone H3 (K4) (ab8580, Abcam, Cambridge, MA, USA); rabbit anti-dimethyl histone H3 (K9) (07–521, UBI; or ab7312, Abcam); mouse anti-HP1γ (05–690, UBI); rabbit anti-GATA4 (H112, Santa Cruz); rabbit anti-GATA6 (H92 or C20, Santa Cruz); mouse anti-RNA Pol II (ab5408, Abcam); rabbit anti-HDAC1 (Santa Cruz Inc, Santa Cruz, CA); rabbit anti-HDAC2 (Santa Cruz); rabbit anti-p66 (MeCP1) (07–365, UBI); rabbit anti-MeCP2 (07–013, UBI); rabbit anti-MBD2/3 (07–199, UBI); and mouse anti-Bmi-1 (UBI, 05–637).

Immune complexes were collected on GammaBind Plus sepharose beads for 1–2h at 4°C. Following washes, the immune complex was eluted from the beads in 1% SDS and 0.1M NaHCO3. Crosslinking was reversed by incubation in 0.2M NaCl for 5h at 75°C. The resulting DNA was purified by proteinase K treatment, phenol/chloroform/isoamyl alcohol extraction, ethanol precipitation and resuspended in water. Chromatin immunoprecipitation DNA was detected by semiquantitative PCR using GATA4 and GATA6 promoter-specific primers (Table 3) for 30 cycles. After electrophoresis in 1.5% agarose with ethidium bromide, the images were recorded and converted to bright field format, and the bands were quantified as a percentage of the input signal (2% of immunoprecipitated chromatin) by densitometric analysis using the ImageJ software from the National Institutes of Health. Centromeric heterochromatin, a known target of H3/K9 methylation and HP1 binding, was used as a positive control for ChIP assay with anti-HP1γ. Each ChIP experiment was repeated at least three times and representative results are shown.

Histone protein extraction and Western blot analysis

A2780 and HIO114 cell nuclei were extracted in 10mM Tris-Cl, pH 7.5, 10mM NaCl, 3mM MgCl2, 1% Nonidet (NP-40), 100 μM PMSF, and complete protease inhibitor cocktail tablets (Roche). Histone proteins were extracted by resuspension of the nuclear pellet in 0.5N H2SO4 with 10% glycerol and incubation overnight at 4°C. The acid-extracted proteins (5–10 μg/lane) were fractionated on 10% NuPAGE Novex Bis–Tris gel (Invitrogen) and transferred to nitrocellulose membrane with pore size 0.2 μm (Pierce, Rockford, IL, USA). The membranes were blocked with 5% milk in 1× phosphate buffered saline (PBS) with 0.05% Tween-20 (PBS-T) and immunoprobed with antibodies to acetyl histone H3 (1:6000), acetyl histone H4 (1:3000), dimethyl histone H3 (K4) (1:5000) and trimethyl histone H3 (K4) (1:5000). A donkey anti-rabbit horseradish peroxidase-conjugated IgG (Amersham, Piscataway, NJ, USA, 1:15000) was used as secondary antibody and chemoluminescent signals were detected with the SuperSignal West Dura Extended Duration Substrate (Pierce).

Quantitative PCR analysis of gene expression and chromatin immunoprecipitation products

Real-time PCR was used to quantify mRNA levels following reverse transcription and to determine the relative amount of genomic DNA following ChIP assay. To determine gene expression, contaminating DNA from the RNA preparations was removed using TURBO DNA-free™ (Ambion). RNA was quantified using the Agilent 2100 BioAnalyzer in combination with a RNA 6000 Nano LabChip. RNA was reverse-transcribed (RT) using MessageSensor RT kit™ (Ambion). Random decamers and anchored oligo-dT were used for priming. For each sample, the RT step was performed with 40, 10 and 1ng RNA.

Taqman assays were used to measure the expression of GATA4 and POLR2F (locus ID:5435). The Taqman set for GATA4 was purchased from Applied Biosystems (Hs00171403_m1) and the Taqman set for POLR2F was designed using the PrimerExpress software (Applied Biosystems, Foster City, CA, USA). The 5′ and 3′ ends of the probes were labeled with the reporter dye FAM (Glenn Research, Sterling, VA, USA) and the quencher dye Black Hole Quencher 1 (BHQ1) (Applied Biosystems), respectively. The Taqman universal master mix (Applied Biosystems) was used for PCR. SYBR Green assays with Quantitect SYBR Green PCR kit (Qiagen, Valencia, CA, USA) were used for real-time PCR for the measurement of GATA6 expression.

Relative quantification of each gene transcript in the different samples was performed using a standard curve generated with a reference sample (A2780þTSA for GATA4, A2780 for GATA-5, ES2 for GATA6). Arbitrary units of 40, 10 and 1 were given to the amounts detected in the cDNA preparations generated with 40, 10 or 1ng RNA of the reference sample, respectively. Threshold cycle (Ct) measurements were converted to amounts using the standard curve. The amounts below 0.5 were considered below detection levels. The amounts above detection level were corrected for input of cDNA in the PCR reaction and the average was calculated.

For quantitative ChIP/PCR analysis, we used SYBR Green assays with Quantitect SYBR. The sequences of the primers are given in Table 3. For each sample, two PCR reactions were performed (0.5 and 0.125 μl of immunoprecipitation (IP)). Relative quantification in each IP sample was performed using a 3-points fourfold dilution curve generated with a 2% solution of input DNA. Threshold cycle measurements were converted to percentage of material present in the input using the standard curve. The amounts <5% were considered below detection levels. The amounts above detection level were corrected for input of material in the PCR reaction and averaged. In all experiments, the real-time PCR was performed in triplicate using the same preparation of ChIP DNA or cDNA from the reversed transcription reaction. The deviations of the PCR results are typically <10% of the value.

Methylation-specific polymerase chain reaction and bisulfite genomic sequencing

DNA was prepared from HIO and ovarian cancer cell lines according to the DNAzol reagent procedure (Invitrogen) and subjected to sodium bisulfite modification (Frommer et al., 1992; Herman et al., 1996). Briefly, 1–2 μg of genomic DNA was denatured in 0.2M NaOH and incubated for 16h at 50°C with 33 μl of 10mM hydroquinone and 525 μl of 3M sodium bisulfite in a final volume of 613.7 μl. The sodium bisulfite-modified genomic DNA was then purified according to the Wizard DNA Clean-Up system (Promega, Madison, WI, USA), ethanol precipitated and amplified by PCR in a 50 μl Platinum Taq DNA polymerase mixture. Methylation-specific PCR analysis of sodium bisulfite-modified genomic DNA was performed as described previously (Herman et al., 1996). The primers (Table 3) used for the specific amplification of DNA originally methylated (M) or unmethylated (U) were designed using the Methprimer program (Li and Dahiya, 2002). For bisulfite-genomic sequencing, a nested PCR was performed by further amplifying a 1:1000 dilution of the first PCR reaction with an internal primer (Table 3). The PCR products were cloned into the pGEM-T easy vector (Promega) according to the manufacturer’s protocol. Plasmid DNA obtained from single-colony bacterial clones were subjected to automated sequencing analysis using T7 and SP6 primers.

DNase I hypersensitivity assay

Nuclei of A2780 and ES2 cells were prepared as described (Berk and Sharp, 1977; Jackson and Felsenfeld, 1985; Gong et al., 1996) with minor modifications. Typically, 5×107 cells were resuspended in 6ml ice-cold homogenization buffer (50mM Tris-Cl, pH 7.4, 1mM EDTA, 0.1mM ethyleneglycol tetraacetate, 15mM NaCl, 50mM KCl, 0.15mM spermine, 0.5mM spermidine, 0.2% NP-40, 5% (wt/vol) sucrose) and incubated for 3min on ice. The extent of cell lysis (>90%) was observed microscopically by staining an aliquot with Trypan blue. The cell suspension was centrifuged (20min, 2300r.p.m., 4°C) through a 3.5ml cushion of homogenization buffer containing 10% sucrose. Nuclei were resuspended in 2–3ml of 8.5% sucrose washing buffer (10mM Tris-Cl, pH 7.4, 15mM NaCl, 50mM Kcl, 0.15mM spermine, 0.5mM spermidine and 8.5% (w/v) sucrose) plus 3mM CaCl2 and 0.5ml aliquots were digested with 0, 20, 40, 80, 160 or 320U/ml of DNase I (Worthington Biochemical Corp., Lakewood, NJ) at 25°C for 5min. The reactions were stopped by bringing the samples to 12.5 μM EDTA, 1% SDS, 0.2mg/ml Proteinase K and 50 μM NaCl, and then incubated overnight at 55°C. The DNA was purified by phenol/chloroform/isoamyl alcohol extraction, ethanol precipitation, and resuspended in 10mM Tris, pH 8.5. The DNA was then digested to completion with restriction enzymes, gel fractionated by electrophoresis in 1% agarose and analysed by Southern blotting. Probes for Southern blots were generated by PCR, and 32P-labeled with Prime-It II random primer labeling kit (Stratagene, La Jolla, CA, USA) to a specific activity of 1–2×109 dpm/μg of DNA.

Histone deacetylase assay

The HDAC Fluorimetric Assay (BIOMOL, Plymouth Meeting, PA, USA) was used to measure the HDAC activity in cell or nuclear extracts. Cells were incubated with the cell-permeable Fluor the Lys™ (FdeL) Substrate for cellular HDACs (class I–III). Following treatment with a developer agent of whole cell lysate, nuclear or cytosol fractions, the deacetylated substrate produces a fluorophore (Ex. 360nm; Em. 460nm). The arbitrary fluorescence units (AFU) represent relative HDAC activity.

Epithelial transition zones from ovarian tumor tissues

Hematoxylin and eosin-stained slides of about 380 archived ovarian tumor tissues from Fox Chase Cancer Center Tumor Bank were examined carefully for the presence of epithelial transitions: contiguous epithelia linking monolayer ovarian epithelial cells with neoplastic lesions. The tumors are of various grades and stages (mainly III and IV) and histological subtype (around 70% serous). These tumor tissues are of ovarian origin and contain non-cancerous ovarian remnants and are not from metastases or implants to the omentum or peritoneum. The use of these human tissues in our research has been examined and approved by the Institutional Review Board Committee, and safety and ethical guidelines were followed in using the human tumor tissues according to institutional requirements.

Immunohistochemistry

The immunostaining of ovarian tissues and tumors was performed as described in our previous publications (Fazili et al., 1999; Capo-chichi et al., 2003). Briefly, sections were first deparaffinized, rehydrated and subjected to antigen retrieval treatment. For antigen retrieval, the slides in a holder in citrate buffer (pH 6) were boiled for 5min in an electronic steamer, followed by cooling for 10min before immunostaining. All tissues were then exposed to 3% hydrogen peroxide for 5min, primary antibodies for 25min, secondary antibodies for 20min, diaminobenzidine as chromogen for 5min and hematoxylin as a counterstain for 1min. These incubations were performed at room temperature in a box lined with wet paper to provide moisture; between each incubation, sections were washed several times with Tris-buffered saline (TBS) buffer. The primary antibodies were from commercial sources: monoclonal mouse anti-Dab2 IgG was purchased from Transduction Lab, Lexington, KY, USA; and anti-GATA4 and anti-GATA6 rabbit antisera were from Santa Cruz. The specificity of these antibodies has been documented in our previous investigations (Fazili et al., 1999; Capo-chichi et al., 2003).

Acknowledgements

We appreciate Dr Elizabeth Smith for reading and commenting during the process of preparing the paper. We acknowledge the assistance by the Histopathology Facility, the DNA Sequencing Facility, the Fannie E Rippel Biochemistry and Biotechnology Facility, and the Cell Culture Facility of Fox Chase Cancer Center. Drs Kathy Qi Cai, Paul Cairns and Andrew Godwin are greatly appreciated for their intellectual and technical advice in performing these experiments. We thank Malgorzata Rula, Lisa Vanderveer and Jennifer Smedberg for their technical assistance, and Ms Patricia Bateman for her excellent secretarial support. This work was supported by grants R01 CA79716 and R01 CA75389 to XX Xu from NCI, NIH, funds from Ovarian Cancer SPORE P50 CA83638 (RF Ozols, PI), and the Core Grant #CA006927. The work was also supported by an appropriation from the Commonwealth of Pennsylvania.

References

  1. Akiyama Y, Watkins N, Suzuki H, Jair KW, van Engeland M, Esteller M et al. (2003). Mol Cell Biol 23: 8429–8439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bachman KE, Park BH, Rhee I, Rajagopalan H, Herman JG, Baylin SB et al. (2003). Cancer Cell 3: 89–95. [DOI] [PubMed] [Google Scholar]
  3. Bai Y, Akiyama Y, Nagasaki H, Yagi OK, Kikuchi Y, Saito N et al. (2000). Mol Carcinog 28: 184–188. [DOI] [PubMed] [Google Scholar]
  4. Baylin S, Bestor TH. (2002). Cancer Cell 1: 299–305. [DOI] [PubMed] [Google Scholar]
  5. Berger SL. (2002). Curr Opin Genet Dev 12: 142–148. [DOI] [PubMed] [Google Scholar]
  6. Berk AJ, Sharp PA. (1977). Cell 12: 721–732. [DOI] [PubMed] [Google Scholar]
  7. Bova GS, Carter BS, Bussemakers MJ, Emi M, Fujiwara Y, Kyprianou N et al. (1993). Cancer Res 53: 3869–3873. [PubMed] [Google Scholar]
  8. Capo-chichi CD, Roland IH, Vanderveer L, Bao R, Yamagata T, Hirai H et al. (2003). Cancer Res 63: 4967–4977. [PubMed] [Google Scholar]
  9. Egger G, Liang G, Aparicio A, Jones PA. (2004). Nature 429: 457–463. [DOI] [PubMed] [Google Scholar]
  10. Fazili Z, Sun W, Mittelstaedt S, Cohen C, Xu XX. (1999). Oncogene 18: 3104–3113. [DOI] [PubMed] [Google Scholar]
  11. Fraga MF, Ballestar E, Villar-Garea A, Boix-Chornet M, Espada J, Schotta G et al. (2005). Nat Genet 37: 391–400. [DOI] [PubMed] [Google Scholar]
  12. Frommer M, McDonald LE, Millar DS, Collis CM, Watt F, Grigg GW et al. (1992). Proc Natl Acad Sci USA 89: 1827–1831. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Fujiwara Y, Emi M, Ohata H, Kato Y, Nakajima T, Mori T et al. (1993). Cancer Res 53: 1172–1174. [PubMed] [Google Scholar]
  14. Gong QH, McDowell JC, Dean A. (1996). Mol Cell Biol 16: 6055–6064. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gregory P, Wagner DK, Horz W. (2001). Exp Cell Res 265: 195–202. [DOI] [PubMed] [Google Scholar]
  16. Guo M, Akiyama Y, House MG, Hooker CM, Heath E, Gabrielson E et al. (2004). Clin Cancer Res 10: 7917–7924. [DOI] [PubMed] [Google Scholar]
  17. Hake SB, Xiao A, Allis CD. (2004). Br J Cancer 90: 761–769. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Herman JG, Baylin SB. (2003). N Engl J Med 349: 2042–2054. [DOI] [PubMed] [Google Scholar]
  19. Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB. (1996). Proc Natl Acad Sci USA 93: 9821–9826. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Ho CL, Kurman RJ, Dehari R, Wang TL, Shih Ie M. (2004). Cancer Res 64: 6915–6918. [DOI] [PubMed] [Google Scholar]
  21. Jackson PD, Felsenfeld G. (1985). Proc Natl Acad Sci USA 82: 2296–2300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jenuwein T, Allis CD. (2001). Science 293: 1074–1080. [DOI] [PubMed] [Google Scholar]
  23. Jones PA, Laird PW. (1999). Nat Genet 21: 163–167. [DOI] [PubMed] [Google Scholar]
  24. Lachner M, O’Carroll D, Rea S, Mechtler K, Jenuwein T. (2001). Nature 410: 116–120. [DOI] [PubMed] [Google Scholar]
  25. Lassus H, Laitinen MP, Anttonen M, Heikinheimo M, Aaltonen LA, Ritvos O et al. (2001). Lab Invest 81: 517–526. [DOI] [PubMed] [Google Scholar]
  26. Li LC, Dahiya R. (2002). Bioinformatics 18: 1427–1431. [DOI] [PubMed] [Google Scholar]
  27. Mok SC, Chan WY, Wong KK, Cheung KK, Lau CC, Ng SW et al. (1998). Oncogene 16: 2381–2387. [DOI] [PubMed] [Google Scholar]
  28. Molkentin JD. (2000). J Biol Chem 275: 38949–38952. [DOI] [PubMed] [Google Scholar]
  29. Morrisey EE, Musco S, Chen MY, Lu MM, Leiden JM, Parmacek MS. (2000). J Biol Chem 275: 19949–19954. [DOI] [PubMed] [Google Scholar]
  30. Mutskov V, Felsenfeld G. (2004). EMBO J 23: 138–149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Nemer G, Qureshi ST, Malo D, Nemer M. (1999). Mamm Genome 10: 993–999. [DOI] [PubMed] [Google Scholar]
  32. Park IK, Morrison SJ, Clarke MF. (2004). J Clin Invest 113: 175–179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rice JC, Allis CD. (2001). Curr Opin Cell Biol 13: 263–273. [DOI] [PubMed] [Google Scholar]
  34. Santos-Rosa H, Schneider R, Bannister AJ, Sherriff J, Bernstein BE, Emre NC et al. (2002). Nature 419: 407–411. [DOI] [PubMed] [Google Scholar]
  35. Schneider R, Bannister AJ, Myers FA, Thorne AW, Crane-Robinson C, Kouzarides T. (2004). Nat Cell Biol 6: 73–77. [DOI] [PubMed] [Google Scholar]
  36. Sheng Z, Sun W, Smith E, Cohen C, Xu XX. (2000). Oncogene 19: 4847–4854. [DOI] [PubMed] [Google Scholar]
  37. Spencer VA, Sun JM, Li L, Davie JR. (2003). Methods 31: 67–75. [DOI] [PubMed] [Google Scholar]
  38. Umlauf D, Goto Y, Feil R. (2004). Methods Mol Biol 287: 99–120. [DOI] [PubMed] [Google Scholar]
  39. Valk-Lingbeek ME, Bruggeman SW, van Lohuizen M. (2004). Stem Cells Cancer Polycomb Connect Cell 118: 409–418. [DOI] [PubMed] [Google Scholar]
  40. Wakana K, Akiyama Y, Aso T, Yuasa Y. (2005). Cancer Lett (Epub ahead of print). [DOI] [PubMed] [Google Scholar]
  41. Yang DH, Smith ER, Cohen C, Wu H, Patriotis C, Godwin AK et al. (2002a). Cancer 94: 2380–2392. [DOI] [PubMed] [Google Scholar]
  42. Yang DH, Smith ER, Roland IH, Sheng Z, He J, Martin WD et al. (2002b). Dev Biol 251: 27–44. [DOI] [PubMed] [Google Scholar]
  43. Zheng J, Benedict WF, Xu HJ, Hu SX, Kim TM, Velicescu M et al. (1995). J Natl Cancer Inst 87: 1146–1153. [DOI] [PubMed] [Google Scholar]
  44. Zheng J, Wan M, Zweizig S, Velicescu M, Yu MC, Dubeau L. (1993). Cancer Res 53: 4138–4142. [PubMed] [Google Scholar]

RESOURCES