A critical re-appraisal of BRCA1 methylation studies in ovarian cancer

A critical re-appraisal of BRCA1 methylation studies in ovarian cancer

Gynecologic Oncology 119 (2010) 376–383 Contents lists available at ScienceDirect Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s ...

287KB Sizes 0 Downloads 20 Views

Gynecologic Oncology 119 (2010) 376–383

Contents lists available at ScienceDirect

Gynecologic Oncology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y g y n o

Review

A critical re-appraisal of BRCA1 methylation studies in ovarian cancer Emir Senturk a, Samantha Cohen b, Peter R. Dottino b, John A. Martignetti a,⁎ a b

Departments of Genetics and Genomic Sciences, Division of Gynecologic Oncology, Mount Sinai School of Medicine, New York, NY 10029, USA Departments of Obstetrics, Gynecology, and Reproductive Science, Division of Gynecologic Oncology, Mount Sinai School of Medicine, New York, NY 10029, USA

a r t i c l e

i n f o

Article history: Received 27 May 2010 Available online 24 August 2010 Keywords: BRCA1 Methylation Epithelial ovarian cancer

a b s t r a c t A central challenge facing gynecologic oncology is achieving personalized care in ovarian cancer treatment. The current ovarian cancer classification scheme distinguishes tumors based on histopathologic subtype, grade, and surgical stage. Recent molecular investigations have highlighted distinguishing genetic features of certain tumors within a given category, and given the rapid pace of technologic advancement combined with plummeting costs for complete genomic sequencing this classification will markedly improve. Clinical studies have begun to explore the influence of currently known distinctions on the natural history of the disease, most recently with particular attention to the BRCA1 status of tumors. Mutations in the BRCA1 gene have long been known to increase a woman's risk of developing ovarian cancer. As has been shown, BRCA1associated ovarian cancers may be associated with characteristic differences in therapeutic response and overall survival, and further defining these subsets may become instrumental in clinical decision-making. Therefore, given the eightfold difference (5–40%) in reported frequency of BRCA1 inactivation by methylation in the pioneering studies in the field, a critical re-appraisal of the literature, techniques, samples used, and interpretations of BRCA1 inactivation is warranted along with a review of the more recent and comprehensive molecular studies. © 2010 Elsevier Inc. All rights reserved.

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . BRCA1 and ovarian cancer . . . . . . . . . . . . . . DNA methylation and transcriptional silencing . . . . A model of methylation-dependent silencing of BRCA1 Studies using patient-derived tumor material . . . Studies using cell lines . . . . . . . . . . . . . . Epigenetic modification of the BRCA1 promoter and its Conclusion . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest statement . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . effects on cis-acting BRCA1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Introduction Epithelial ovarian cancer is the leading cause of death among gynecologic malignancies in the United States [1]. It is estimated that 21,550 women will have been diagnosed and 14,600 women will have died of ovarian cancer in 2009 [2]. Approximately one in 70 women will develop ovarian cancer in her lifetime, and one in 100 will die of this disease. This extremely high fatality-to-case ratio ⁎ Corresponding author. Department of Genetics and Genomic Sciences, Mount Sinai School of Medicine, 1425 Madison Avenue, Box 1498, New York, NY 10029, USA. Fax: +1 212 360 1809. E-mail address: [email protected] (J.A. Martignetti). 0090-8258/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2010.07.026

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . transcription factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

376 377 378 378 378 380 381 381 382 382

is due in part to a lack of effective screening modalities to detect ovarian cancer at an early stage wherein rates of cure exceed 90%. Instead, approximately 75–80% of ovarian carcinoma patients initially present with intraperitoneal spread [1]. The 5-year survival for these patients is poor despite initial response to cytoreductive surgery and adjuvant chemotherapy and it is therefore of enormous importance to find innovative strategies to treat this disease (Table 1). Epithelial ovarian cancer is a clinically and pathologically heterogenous disease comprised of serous, mucinous, clear cell, and endometrioid subtypes. Serous tumors make up more than two thirds of all cases, frequently present at a high grade and stage, and often recur despite initial chemosensitivity to platinum/taxane doublet

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

377

Table 1 Study

Sample set

Method

Region

Results

Mancini et al. [34]

5 ovarian tumors Unpaired normal tissues (lymphocytes, breast, placenta, muscle)

BiSeq

−282 to + 58

Catteau et al. [35]

43 ovarian tumors Unpaired normal tissues (breast, placenta, lymphocytes)

MRED/SB

+ 132

Bianco et al. [36]

20 ovarian tumors

MRED/SB

−86 to + 213

Esteller et al. [37]

31 ovarian tumors 13 sporadic tumors (+) for LOH at BRCA1 locus 18 sporadic tumors (−) for LOH at BRCA1 locus

MSP

−40 to + 46

Baldwin et al. [38]

98 ovarian tumors

MSP

−86 to + 213

2 ovarian tumors (40%) with variable methylation patterns of 16 cytosine residues across CpG motifs in the BRCA1 promoter. Residues not methylated in any normal tissues analyzed. 2 ovarian tumors (5%) with methylation-mediated failure of SmaI endonuclease cleavage at indicated site in the BRCA1 promoter. Negative findings in normal tissues analyzed. 2 ovarian tumors (10%) with methylation mediated failure of CfoI and HpaII endonucleases to cleave at sites within studied region of BRCA1. 4 ovarian tumors (13%) producing visible PCR amplification products with methyl-specific PCR primers exclusively. All 4 ovarian tumors (+) for methylation were also (+) for LOH at the BRCA1 locus. None of the ovarian tumors (−) for LOH at the BRCA1 locus were (+) for methylation. 12 ovarian tumors (15% of tumors not associated with family history; 12% of total sample set) producing visible PCR amplification products with methyl-specific PCR primers exclusively. 5 of 10 methylation-positive tumors (50%) with available genomic DNA were (+) for LOH at the BRCA1 locus.

Wang et al. [39]

81 tumors from pts. with (−) family history 12 normal ovarian specimens 64 ovarian tumors with matching germline DNA from patients with (−) family history

MSP

−40 to + 46

Wilcox et al., 2005 [40]

50 ovarian tumors

MSP

−157 to + 47

Press et al. [41]

49 ovarian tumors

PMR

−147 to + 20

15 ovarian tumors (23%) producing visible PCR amplification products with methyl-specific PCR primers exclusively and demonstrating LOH at the BRCA1 locus. 5 ovarian tumors (7%) with BRCA1 promoter methylation and no LOH at the BRCA1 locus. 8 ovarian tumors (16%) producing visible PCR amplification products with methyl-specific PCR primers exclusively. 9 ovarian tumors (18%) with BRCA1 promoter methylation.

Correlation with expression

BRCA1 protein staining absent in 12 of 12 (100%) ovarian tumors with (+) BRCA1 promoter methylation.

BRCA1 protein staining absent in 2 of 9 (22%) of tumors with un-methylatedBRCA1 promoter. BRCA1 protein staining absent in 13 of 15 (87%) ovarian tumors with (+) BRCA1 promoter methylation and LOH at the BRCA1 locus. BRCA1 protein absent or decreased in all 5 ovarian tumors (+) for BRCA1 promoter methylation but without LOH at the BRCA1 locus. 1.3- to 2.5-fold reduction in BRCA1 mRNA expression assessed by RT-PCR.

BRCA1 protein staining absent in 8 of 9 (89%) of ovarian tumors with (+) BRCA1 promoter methylation.

Key: BiSeq—bisulfate sequencing, MRED/SB—methylation-sensitive restriction endonuclease digestion and Southern Blotting, MSP = methylation-specific PCR, PMR—percent methylated reference (see text; ref. [41]).

regimens [3,4]. Non-serous subtypes, on the other hand, may present at earlier stages and have varying degrees of primary chemoresistance and thus carry a less predictable prognosis [4]. Despite these established differences in clinical behavior, all ovarian neoplasms are subject to the same treatment paradigm. A considerable effort to further subclassify ovarian tumors based on their molecular and genetic characteristics is necessary if the benefits of personalized oncologic care are to be extended to patients with ovarian cancer. We are now approaching an era wherein complete sequencing of an ovarian tumor genome will not only be part of a research effort, but also a reality in the clinical setting. Given a number of recent findings, BRCA1 status may soon represent the paradigm for this approach. BRCA1 and ovarian cancer Mutations of BRCA1 have been associated with approximately 50% of all familial breast and ovarian cancers [5]. Germline mutations in

BRCA1 have been reported in 5–15% of unselected ovarian cancer cases in the United States [6]. Women with mutations in the BRCA1 gene have a 39–46% risk for developing ovarian cancer by age 70 [7]. This is in contrast to the general population, in which the average lifetime risk of developing ovarian cancer is 1.5% [8]. Ninety percent of these germline mutations are either known or predicted to be frameshift or nonsense mutations that result in absent or truncated gene products [9]. Beyond mutations, inactivation of tumor suppressor genes can be achieved through different biological mechanisms. A well-established feature of ovarian tumors that develop in BRCA1 carriers is LOH of the remaining wild-type allele [10]. The elevated lifetime risk of malignancy in patients with inherited BRCA1 mutations is therefore a function of combined germline and somatic inactivation of this tumor suppressor, consistent with Knudson's twohit model [11]. In addition to genetic events that have been reported to result in loss of tumor suppressor expression or function, epigenetic means of

378

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

transcriptional silencing can also inactivate tumor suppressors. BRCA1 protein and mRNA levels in ovarian tumors are decreased or absent in as many as 90% of patient cases without evidence of germline BRCA1 mutations or family history of the disease [12]. Most importantly, recent studies of patients with ovarian cancers harboring genetically inactivated or epigenetically silenced BRCA1 have drawn distinct clinical pictures associated with both sets of tumors. Although the median age of ovarian cancer diagnosis is nearly 10 years younger for patients with germline BRCA1 mutations, higher rates of treatment response, longer disease-free intervals, and improved overall survival have been noted in studies comparing this population to patients with sporadic disease [13]. On the other hand, patients whose tumors have silenced BRCA1 expression via promoter methylation have been reported to have significantly shorter disease-free intervals and poorer overall survival compared both to patients with BRCA1 mutations or wild-type BRCA1 patients with ovarian cancer [14]. Differences in experimental methods and, in particular, the region of the BRCA1 promoter investigated across studies have produced a range of reported frequencies of BRCA1 promoter methylation, from as few as 5% of clinical specimens to as many as 40%. This paper reviews the possible models and data supporting the molecular basis of methylation-dependent silencing of BRCA1 in ovarian cancer. Given the increasing interest in DNA methylation inhibitors, their use as adjuvant agents in ovarian cancer treatment protocols, and the promising results of clinical trials employing inhibitors of poly(ADP) ribose polymerase (PARP) in BRCA-deficient tumors [15,16], we believe a critical review of the molecular data suggesting epigenetic inactivation is warranted. DNA methylation and transcriptional silencing Over half of all human and other eukaryote genes, particularly housekeeping genes, do not lie downstream of promoters bearing a TATA box or initiator element. Rather, these genes’ promoter regions are defined by the presence of CpG islands: ~1 kb CG-stretches that lie upstream of the transcriptional start site [17]. The CpG dinucleotide is otherwise underrepresented, comprising less than 1% of the human genome [18]. Roughly 70% of CpG dinucleotides in the human genome that are not within CpG islands are methylated [19]. In contrast, transcriptional start site-associated CpG islands are characteristically unmethylated [17]. Aberrant DNA methylation patterns are more common in human cancer cells than other molecular lesions, including mutations and karyotypic abnormalities [20]. This understanding has led to significant clinical correlation between these novel patterns and their utility as potential markers for screening, disease prognosis, recurrence rates, and overall survival. Kim et al. [21] in a recent review of the colorectal literature, defined how epigenetic dysregulation and hypermethylation may not only be a predictive marker for metastatic and aggressive colorectal cancer phenotypes, but may also serve as a screening tool in the early events of carcinogenesis. Focusing solely on gynecologic malignancies, one recent study isolated 6 novel genes that were frequently methylated in squamous cell carcinoma of the cervix [22]. In addition, these authors demonstrated that methylation rates increased as the severity increased among precancerous dysplastic lesions and that testing for methylation in cervical swabs, along with HPV testing, markedly increased in the sensitivity of cytologic screening. Su et al., demonstrated the associated between abnormal DNA hypermethylation in 6 novel genes and clinical outcome in ovarian cancer [23]. Genomic DNA extracted from tissue and serum was compared across invasive, borderline, and benign ovarian tumors. The methylation rates were significantly higher in those with ovarian cancer, and the presence of methylation had independent effects on recurrence and overall survival, portending a worse prognosis. In addition, the methylation status between the serum and tissue samples was highly concordant,

signifying its potential as a screening marker. Furthermore, identification of these types of epigenetic changes within a tumor has potential therapeutic implications, as discussed below. DNA methyltransferases (DNMTs) catalyze the covalent attachment of methyl groups to 5' cytosine residues in CpG dinucleotides [24]. It had originally been proposed that three different DNMTs in mammals fulfill separate methylation-dependent functions, despite their common underlying catalytic properties. DNMT1 was observed to be localized to replication foci during S phase and was thus believed to be responsible for the transfer of methylation-based epigenetic information during DNA replication [20]. Thus, DNMT1 alone was thought to transmit both physiologic and aberrant methylation patterns to successive generations of cells in embryonic development and cancer, respectively. DNMT3a and DNMT3b, on the other hand, were thought to be exclusively responsible for de novo methylation [25]. More recently, in vitro studies have suggested that all three DNMTs may cooperate to carry out both de novo and maintenance methylation. Inappropriate repressor-mediated targeting of DNMTs to specific promoters, deregulated DNMT expression, and mutations in genes encoding DNMTs are all thought to be causal [26]. Myc, for example, which is often aberrantly expressed in ovarian cancer, has been shown to act in concert with DNMT3a to silence expression of the cyclin-dependent kinase inhibitor p21 [27]. Although replication studies are lacking, Baldwin et al. [28] showed that Myc escapes TGFβdirected downregulation in 5 primary ovarian cancer cell lines compared to 5 human ovarian surface epithelial lines. Additionally, Chen et al. [29] performed immunohistochemistry on a series of 28 ovarian tumors and 12 normal ovaries and found ~ 66% of tumors to overexpress Myc relative to normal tissue. Using an in vitro system, direct binding of Miz-1-associated Myc to DNMT3a at the p21 promoter was required to both site-specifically target DNMT3a and produce significant p21 transcriptional downregulation in U2OS human osteosarcoma cells. Collectively, these findings suggest that Myc may co-opt a cell's DNA methylation machinery to produce an oncogenic program of transcriptional dysregulation. Transcriptional silencing of several tumor suppressor genes, including BRCA1, can result from aberrant patterns of DNA methylation affecting these genes' promoter regions. Transcriptional repression of genes downstream of CpG islands in the presence of methylated cytosine residues is believed to result from at least two mechanisms. The presence of methylated cytosines can interfere directly with the binding of sequence-specific transcription factors that would otherwise promote gene expression. DNA methylation can also indirectly effect transcriptional repression through chromatin remodeling. The involvement of methyl-CpG binding proteins, histone deacetylases, and histone methyltransferases in this process has been extensively studied and is reviewed elsewhere [30]. A model of methylation-dependent silencing of BRCA1 expression Studies using patient-derived tumor material The BRCA1 gene on chromosome 17q21.31 lies approximately 200 bp downstream of a bidirectional promoter, which is shared with BRCA1's 5' neighbor NBR2 [31] (Fig. 1). A 5' CpG island spanning ~2.7 kb, identified by sequencing and analysis of the BRCA1 locus, encompasses the promoter region and the first two BRCA1 exons [32]. Within this region, a sub-region spanning positions −202 to +20 relative to the BRCA1 exon 1a transcriptional start site (hereafter referred to as the transcription start site) has been identified as an essential positive regulatory element [33]. Deletion of this region results in complete loss of BRCA1 promoter activity in fibroblasts and MCF7 breast cancer cells [33]. Although a number of clinical studies directed at assessing BRCA1 promoter methylation in tumor samples have been conducted, none have defined a “threshold” of methylation required for BRCA1 inactivation. This complicates attempts at

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

379

Fig. 1. The BRCA1 gene locus. The ~ 800 bp region demarcated by the asterisk encompasses the area of the BRCA1 promoter-associated CpG island normally devoid of methylation in normal tissue but found to be hypermethylated in a subset ovarian tumors. The CpG island extends into coding regions of the BRCA1 gene. Also shown are the binding sites of two transcription factors, which have been identified and that may become disrupted in the setting of abnormal methylation patterns. The different promoter regions analyzed and reported on by different groups are shown. (A) Mancini et al. (1998), (B) Bianco et al. (2000) and Baldwin et al. (2000), (C) Catteau et al. (1999), and (D) Esteller et al. (2000) and Wang et al. (2003), (E) Wilcox et al. (2005), and (F) Press et. Al (2008).

functional correlation and evaluation between studies. In large part this is a consequence of a lack of correlative studies defining BRCA1 mRNA or protein expression in most published reports of patientderived material. Additionally, the methods employed across studies to assess methylated CpGs vary widely and include methylationsensitive PCR, methylation-specific PCR, and bilsulfite sequencing. In one of the studies reporting the highest frequency of methylation, Mancini et al. [34] demonstrated BRCA1 promoter methylation via bisulfite sequencing of extracted DNA in 2 of 5 (40%) ovarian tumor samples. They reported variable methylation patterns of 16 cytosine residues within CpG motifs across a region 282 bp upstream and 58 bp downstream of the BRCA1 transcription start site (Arrow A, Fig. 1). None of these CpG dinucleotides were methylated in samples of human lymphocytes and placenta control samples from individuals that were also analyzed. In a larger set of 43 ovarian tumors, DNA from tumors was digested using methylation-sensitive restriction endonucleases and Southern blotting was performed to analyze the resulting banding patterns across a 3.8-kb region of the BRCA1 gene that included the ~ 2.7-kb CpG island [35]. High molecular weight bands corresponding to undigested, methylated DNA were observed in only 2 tumors (~5%), suggesting that methylation is not a high frequency event. The SmaI endonuclease, which was used in these studies, would not allow for interrogation of the region upstream of the first BRCA1 exon that Mancini et al. had explored. In addition, it could be argued that the use of methylation-sensitive restriction endonucleases failed to provide a quantitative assessment of methylation. Given the different regions analyzed and different techniques employed in assessing BRCA1 promoter methylation status, these two studies, suggesting a range

between 5% and 40%, are difficult to compare. Promoter methylation was also absent in unpaired control samples of normal breast tissue, lymphoid cells, and chorio-placental cells, consistent with Mancini et al.'s findings. Bianco et al. [36] also applied methylation-sensitive endonuclease digestion to demonstrate BRCA1 promoter methylation in 2 of 20 (10%) sporadic tumors. The endonucleases employed by this group allowed them to assay a region spanning 86 bp upstream and 213 bp region downstream of the BRCA1 transcriptional start site (Arrow B, Fig. 1). Similarly, Esteller et al. [37] found the BRCA1 promoter to be methylated in 4 of 31 (~ 13%) ovarian tumors using methylationspecific PCR (MSP). These studies investigated the BRCA1 promoter 40 bp upstream and 46 bp downstream of the transcriptional start site. Interestingly, all tumors demonstrating BRCA1 promoter methylation also harbored LOH at the BRCA1 locus (Arrow C, Fig. 1). Nine additional tumors demonstrating LOH at the BRCA1 locus were found to be unmethylated. This finding suggests methylation at yet other sites in the BRCA1 promoter not assessed in this particular study or alternative mechanisms of BRCA1 inactivation. Again, as with the earlier studies discussed, correlative assessments of BRCA1 expression were not reported. An additional and potentially important caveat to the interpretation of all these studies is the failure to account for possible germline and/or tumor-specific BRCA1 mutations. None of these reports delineated either criteria for the exclusion of familial cancers by way of questionnaire or BRCA1 mutational analysis. In contrast, Baldwin et al. [38] analyzed a larger set of ovarian tumors and normal ovaries by MSP. The latter series of experiments allowed interrogation of the −86 to + 213 region relative to the BRCA1 transcriptional start

380

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

site, consistent with Bianco et al.'s study (Fig. 1). Twelve of 81 (~15%) ovarian tumors in patients without a family history of ovarian cancer had evidence of BRCA1 promoter methylation. None of the 12 normal ovaries had evidence of methylation. Tumor and genomic DNA from patients with BRCA1 promoter methylation were screened for the three key founder mutations in the Ashkenazi Jewish population: BRCA1 185delAG, 5382insC, and BRCA2 6174delT. None of the tumors that demonstrated BRCA1 promoter methylation had concurrent BRCA1 mutations. Baldwin et al. [38] also performed immunohistochemistry on paired paraffin-embedded tumor tissues. BRCA1 expression was detected in the nuclei of adjacent stromal cells but absent in all 12 of the tumors, suggesting BRCA1 inactivation through promoter methylation. Only 5 of these 12 methylated samples exhibited LOH at the BRCA1 locus. While discordant findings between these studies may be a product of biases inherent in population samplings, they suggest the possibility of alternative sites of inactivating BRCA1 methylation not detected in either group's assay. Three recent clinical studies in ovarian tumors have offered more direct evidence of downregulated BRCA1 expression in the setting of BRCA1 promoter methylation. In the first, 64 ovarian cancers with matching germline DNA were assessed for methylation in the −40 to + 46 region of the BRCA1 promoter by MSP (Arrow D, Fig. 1) [39]. Fifteen (23%) of these tumors demonstrated both LOH at the BRCA1 locus and BRCA1 promoter methylation. Immunohistochemical staining revealed loss of BRCA1 protein expression (b10% positive cells) in 13 (87%) of these 15 tumors. Notably, all of the ovarian tumors in the set that demonstrated BRCA1 promoter methylation in the absence of LOH (5/64; 7%) had decreased BRCA1 expression, ranging from b10% to 30% positive cells by IHC. In the second of these studies, 8 of 50 (16%) tumors demonstrated BRCA1 promoter methylation in the −157 to + 47 region, also assessed by MSP (Arrow E, Fig. 1) [40]. RT-PCR showed 1.3- to 2.5-fold reduction in BRCA1 mRNA in the promoter-methylated tumors compared to non-methylated samples. In addition, the authors describe the heterogeneous nature of hypermethylated tumors in terms of histologic subtype, stage and grade. In particular, 20% of stage I tumors displayed BRCA1 promoter methylation, including 2 that represented stage IA disease. The authors therefore speculated that the ability to detect these epigenetic changes in early stage tumors could offer insight into the development of early detection tools. Finally, a study by Press et al. [41] of 49 of ovarian tumors found 18 tumors (37%) to have loss of BRCA1 expression, with 8 of these (16%) attributed to methylation-directed means. When the analysis was limited to only high grade serous ovarian carcinomas, these figures rose to 42% and 24%, respectively. The investigators assessed the mutation status of BRCA1, LOH status of the BRCA1 locus, and methylation of the BRCA1 promoter across the −147 to −20 region (Arrow F, Fig. 1). To quantify the extent of BRCA1 promoter methylation, real-time PCR amplification of bilsulfite-modified DNA was used to generate fluorescent signals commensurate with the presence of methylated CpG dinucleotides in this region. Fluorescence readout from BRCA1 promoter assessments were divided by fluorescent readout from the same reaction directed at the MYOD1 promoter in each sample, which lacked CpG dinucleotides. The resulting ratio was divided by the BRCA1:MYOD1 ratio of bisulfite-treated, Jurkat genomic DNA (positive control) and multiplied by 100 to generate a percentage of methylated reference (PMR) score. Samples bearing a PMR of 4 or greater were considered methylated, and those with a PMR of less than 4, unmethylated. Using qRT-PCR, tumor-specific BRCA1 relative expression levels were divided by the average relative expression level of all tumors to produce expression ratios. Tumors with BRCA1 expression ratios of less than 0.7 were considered to have reduced BRCA1 expression. Additionally, tumors were assessed for BRCA1 protein expression by immunohistochemistry, and those tumors with less than 1% BRCA1 staining were considered negative

for expression. Samples with epigenetic BRCA1 loss were defined as having a PMR of 4 or greater and low relative BRCA1 expression by qRT-PCR or negative BRCA1 expression by IHC. In total, 9 of 49 ovarian tumors (18%) had evidence of BRCA1 promoter methylation by this technique, of which 8 (89%; 16% of total) were negative for BRCA1 protein expression by immunohistochemistry. Notably, methylationmediated silencing of BRCA1 protein expression was limited only to high grade serous or undifferentiated ovarian tumor histotypes. None of the low-grade serous, endometrioid, or clear cell histotypes exhibited evidence of loss of BRCA1 protein expression, BRCA1 mutation, or BRCA1 promoter methylation. Beyond subclassification by BRCA1 status into those tumors that did or did not have loss and by which mechanism (epigenetic or mutation) that loss occurred, the authors further examined the possibility of a shared complex genetic background or relationship in BRCA1 negative tumors. It should be stated that these high grade tumors were otherwise indivisible based on histologic morphology. Indeed, correlations between the PI3K/AKT and p53 pathways were identified depending on whether or not BRCA1 was epigenetically or mutationally lost. The strongest correlation to a distinct oncogenic signaling pathway was with PI3K/AKT. Specifically, 7 of 8 high grade serous ovarian tumors classified as having lost BRCA1 expression through epigenetic inactivation were found to have an increased copy number of the PIK3CA gene locus (which encodes the p110A catalytic subunit of phosphatidylinositol 3-kinase (PI3K)). None of the BRCA1 mutation positive cases had PIK3CA locus amplification. Conversely, 8 of 9 high grade serous ovarian tumors that had lost BRCA1 expression through mutation and LOH had decreased PTEN copy number. PIK3CA copy number variation was found to produce increased PIK3CA and decreased PTEN relative expression levels by qRT-PCR, respectively. These distinct molecular abnormalities were found to be mutually exclusive. Of particular note, these findings strongly suggest that while the PI3K/PTEN/AKT pathway is commonly dysfunctional in ovarian cancers that have lost BRCA1, the mechanism of this dysfunction is perhaps dependent on the more precise nature of BRCA1 loss. In addition, immunohistochemical analysis of high-grade serous or undifferentiated ovarian tumors in this study further identified a “p53+/p21−” phenotype, wherein high levels of p53 expression, defined as N50% of nuclei positivity, was observed to coincide with low levels of p21 expression. This phenotype was noted in 8 of 9 tumors (89%) with epigenetic loss of BRCA1 but only 4 of 9 (44%) of tumors with loss of BRCA1 consequent to germline or somatic mutation. While these findings will need to be validated in additional sample sets and in larger numbers, the suggestion that epigenetic and mutational loss of BRCA1 result in distinct molecular abnormalities may have therapeutic relevance in the future. These recent efforts to characterize the frequency and distribution of BRCA1 methylation and to define the molecular consequences of methylation-mediated loss of BRCA1 protein expression in ovarian cancer exemplify the nature of the work that remains to be done in the field. Future studies should expand on these preliminary data made available over the past decade by combining reliable measures of BRCA1 promoter methylation and BRCA1 protein expression. Data relating BRCA1 methylation status to patient treatment response and survival will also be of great clinical interest, as this may define and clinically distinct subset of patients for whom a separate diagnostic and therapeutic approach may eventually exist. Studies using cell lines While studies examining patient-derived clinical specimens have not offered quantitative or mechanistic insight into methylationdependent BRCA1 transcriptional downregulation, in vitro studies of BRCA1 promoter methylation have been more informative with regard to molecular mechanisms of BRCA1 inactivation. In this regard,

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

it is also important to note that the BRCA1 promoter region analyzed in these in vitro studies (extending up to 2 kb upstream of the transcription start site) is in general much larger than those investigated in clinical specimens. Potentially tempering these positive strengths, a caveat is that only a few of these studies have been performed in ovarian-derived cell lines, and thus cell type dependent contexts must be kept in mind. Rice et al. demonstrated significantly reduced BRCA1 mRNA expression by RT-PCR in five primary and one immortalized breast cancer cell line compared to normal human mammary epithelial cells [42]. Bisulfite sequencing-directed interrogation of the −567 to +255 region revealed one of these primary lines to be methylated at 18 of 30 CpG dinucleotides. BRCA1 expression in this cell line (UACC3199) was among the lowest in those assayed. In later studies, the group was able to enrich for acetyl-histone H3- and H4-associated DNA via chromatin immunoprecipitation using antibodies directed at these two histones [43]. Subsequent PCR of the BRCA1 promoter DNA from −232 to + 198 in the UACC3199 line revealed no amplification product compared to that of normal human mammary cells, indicating that BRCA1 promoter DNA is likely not associated with acetylated histones in these cancer cells. These findings also suggested that H3 and H4 are likely hypoacetylated in the setting of BRCA1 promoter methylation, consistent with the condensed chromatin structure and reduced protein accessibility of the promoter that would be expected in this setting. To explore the means by which BRCA1 promoter-associated histones become hypoacetylated, Magdinier et al. [44] proposed that transcriptional silencing of BRCA1 is dependent on the binding of MeCP2, a methyl-CpG binding protein that can recruit HDACs to the methylated BRCA1 promoter. MepC2 and a vector containing the −1714 to +43 region of the BRCA1 CpG island fused to the luciferase gene were cotransfected into Bosc23 human embryonic kidney cells. In the absence of MeCP2, luciferase activity dropped by 2- to 3-fold after HpaII or HhaI methylase treatment, respectively In the presence of MeCP2, luciferase activity was inhibited approximately 50- and 90-fold, respectively. The physiologic significance of these findings has not been defined. Furthermore, the differences in luciferase inhibition seen with HpaII and HhaI methylase treatment correspond to a greater number of methylated sites seen with HhaI over HpaII methylase. These data suggest nonspecific binding of MeCP2 to methylcytosines in the BRCA1 promoter, as opposed to a sequence-specific interaction. More recently, Auriole et al. [45] reported that Methyl-CpG Binding Domain Protein 2 binding to the methylated BRCA1 CpG island across the −1928 to −1004 region mediated BRCA1 silencing. The specific binding of MBD2 was determined by quantitative ChIP using antibodies directed against MBD2 in HeLa cells and PCR primers directed at four subsegments of this region. MeCP2 was unable to extract comparable bound fractions of the same region in the setting of siRNA-mediated knockdown of MBD2. MBD2 therefore might be responsible for coordinating the transcriptional repression directed by methylation of the BRCA1 promoter. Although these binding studies appear convincing, studies examining BRCA1 expression in MBD2 knockdown versus control cells will be necessary to fully characterize the role of MBD2. Epigenetic modification of the BRCA1 promoter and its effects on cis-acting BRCA1 transcription factors In vitro studies examining epigenetic BRCA1 downregulation have nominated two distinct transcription motifs that may be involved. In their study of clinical specimens, Mancini et al. [34] identified a CRE motif located 155–179 bp upstream of the BRCA1 transcriptional start site. To investigate this motif's involvement in transcriptional repression, they performed gel shift assays corresponding to the BRCA1/CRE binding site in MCF7 breast cancer-derived cell extracts. BRCA1 binding was inhibited in the presence of methylated

381

oligonucleotide. Additionally, increasing concentrations of methylated oligonucleotide failed to compete for binding to proteins that had already bound its unmethylated counterpart. Nevertheless, evidence for the precise role of CREB in directing BRCA1 transcription is sparse. Atlas et al. generated luciferase reporter plasmids under the control of the promoter region located 169 to 179 bp upstream of the BRCA1 transcription start site and assayed luciferase activity in transiently transfected MCF7 and T-47D breast cancer cells and in HepG2 liver cells [46]. The introduction of a mutation in the promoter sequence to inhibit CREB binding (CpG → ApT at −172/−173) resulted in a 20–40% reduction in luciferase activity. Unfortunately, while the authors reported a lack of cAMP-directed upregulation of BRCA1 transcription by forskolin treatment, they did not provide any data nor offer any experimental evidence to support this statement. In addition, while these findings suggest that CREB may be a constitutively active element in directing BRCA1 expression, it is important to note that the promoter region employed in these studies was 13 bp shorter than the CRE region described by Mancini et al. [34]. In IOSE-29 and IOSE-80pc immortalized ovarian surface epithelial and OVCAR-3 and SkOV-3 ovarian cancer cells, Graves et al. [47] also used BRCA1 promoter-driven luciferase reporter constructs to investigate the function of the previously identified CRE motif. They focused on the −207 to +27 region. Introduction of a CpG → ApT mutation at −172/−173 resulted in an 80% decrease in luciferase activity. To further interrogate the role of the CRE motif, oligonucleotide probes directed against the −184 to −155 bp region of the BRCA1 promoter were mixed with nuclear extracts from IOSE-80pc and OVCAR-3 cells. Gel mobility shift assays performed with antibodies against Fra-2 and c-Jun revealed binding of both of these AP-1 transcription factor components to this region of the BRCA1 promoter. Co-transfection of c-Jun and Fra-2 produced a 150–400% increase in luciferase activity. These effects were not recapitulated when the CRE-binding mutant luciferase reporter was used. Together these findings suggest that AP-1 may be playing a role in BRCA1 expression and that positive regulatory interaction the BRCA1 promoter with these AP-1 components may be lost in the setting of methylated cytosines critical for their binding. It remains to be shown whether disruption of this interaction plays a role in tumor formation, however, and in vivo correlates to these initial mechanistic insights looking more directly at BRCA1 transcription are necessary. Furthermore, these investigators did not address the decrease in luciferase activity seen with reporter constructs under the control of a larger BRCA1 promoter region spanning positions −970 to + 27. A less-explored site of potential transcriptional control of BRCA1 that may also succumb to epigenetic modification is a RIBS (EcoRI bandshift) element at position −202 to −182. The RIBS element contains three potential ets transcription factor binding sites, the central one of which has been shown to be recognized by the GABPα/ β ets family member [48]. Interestingly, Graves et al. were able to demonstrate the importance of the RIBS element by generating a luciferase promoter under the control of a truncated BRCA1 promoter region encompassing position −182 to +27. The measured luciferase activity in the ovarian cancer cell lines they transfected with this truncated promoter construct was comparable to that seen with promoterless vector controls. As with this groups' other findings, the pitfalls of luciferase assays must be addressed by in vivo studies correlating BRCA1 promoter methylation to BRCA1 expression. It remains to be seen whether BRCA1 promoter methylation produces an oncogenic degree of BRCA1 transcriptional downregulation. Conclusion In pioneering early studies, aberrant methylation of the BRCA1 promoter was reported to range from 5% to 40% in sporadic ovarian

382

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383

cancers. Even with this broad estimate, an understanding of the biological significance of this finding was, and remains unclear since the effect of methylation on BRCA1 expression levels and downstream consequences has not always been clearly defined. In addition, the overlap between somatic and germline BRCA1 mutations, LOH status, and epigenetic silencing of BRCA1 from sample sets has not been well established, making estimation of the frequency of BRCA1 epigenetic inactivation in ovarian cancer difficult. Methylation-dependent mechanisms disrupting the binding and function of transcription factors that would otherwise regulate BRCA1 expression have been elucidated in vitro, and recent work has shown BRCA1 promoter hypermethylation to produce downregulation of BRCA1 at both transcript and protein levels both in vitro and in clinical specimens. Based upon these more recent and encompassing studies, epigenetic inactivation of BRCA1–demonstration of DNA promoter methylation with concurrent loss of BRCA1 expression–may occur in up to 20% of late-stage ovarian tumors. Now, the consequences of BRCA1 downregulation produced by this epigenetic mechanism will need to be thoroughly investigated because tumors with functional loss secondary to germline or somatic BRCA1 mutations may have distinct molecular abnormalities from those with methylation-mediated silencing [41]. Therapeutic trials with novel agents may therefore need to distinguish between patients whose tumors are BRCA1 mutation positive and those with epigenetic loss of BRCA1. With the advent of more powerful and cost-efficient technologies to completely sequence genomes, an exciting front has been opened into cancer biology and therapeutics, and patients with ovarian cancer will undoubtedly benefit from continued investigation. At present, methylation-mediated differences in the biological and clinical aspects of cancer have been explored in only a few disease-specific settings. Preclinical and clinical studies of hypomethylating agents in cancer have been well-executed in myelodysplastic syndromes, for which two hypomethylating agents–5-azacytidine and 5-aza-2'deoxycytidine–are currently FDA-approved [49]. Recent work with breast cancer cells has also suggested a role for hypomethylating agents in solid tumors when administered concurrently with histone deacetylase inhibitors [50]. Similarly, recent published studies support the synergistic effects of these agents in the treatment of lung cancer, not only in animal models, but in clinical trials [51–53]. To date, there have been 3 clinical trials evaluating the combination of hypomethylating agents with chemotherapy in patients with ovarian cancer [15]. Fang et al. [54], in a phase I/II trial of decitabine in combination with carboplatin, performed methylation analysis of 5 genes, including BRCA1. Demethylation of BRCA1 and biologic activity of this drug combination was demonstrated while using a safe and effective dosing schedule. Given the possibility of treatment with hypomethylating agents and PARP inhibitors, it will be important to address basic biological and clinical questions about the behavior of ovarian cancers that arise consequent to genetic versus epigenetic lesions in BRCA1. In addition, a more critical evaluation of the literature regarding BRCA1 promoter methylation and ovarian cancer is warranted. In moving forward, we would recommend the adoption of reliable measures of BRCA1 methylation coupled with direct assessment of BRCA1 expression in well-curated clinical samples. Preliminary data suggest that BRCAmethylated ovarian tumors are of predominantly high-grade serous histotype, similar to their BRCA1-mutated counterparts. Studies defining samples within their set by histotype, however, have only been published recently and their implications warrant validation. Only when critical regions and effects of methylation on expression are defined can we more accurately classify these tumors and correctly stratify patients for targeted protocols and treatments. Important clinical questions remain to be addressed by these studies. Are the natural histories of ovarian cancers harboring germline BRCA1 mutation or that have succumbed to BRCA1 epigenetic silencing similar? If so, can both cancers be subject to the

same therapeutic interventions with the same expected clinical outcomes? As described, preliminary evidence, including genetic studies in colorectal and ovarian carcinoma, suggests that epigenetic dysregulation and hypermethylation may have prognostic significance for more aggressive disease [21,23]. It is also known that BRCA1 germline-associated cancers may have more favorable clinical outcomes [13]. The clinical consequences of epigenetic dysregulation are only beginning to be explored. How patient outcomes with BRCA1 epigenetic silencing versus BRCA germline mutations compared with other BRCA somatic mutations such as defects in homologous recombination, remains to be seen [16]. It has become clear, however, that therapeutic modalities may be quite different depending on the epigenetic/genetic BRCA-associated defects identified. The impaired DNA damage responses unique to BRCA-deficient tumors have already exposed a potential Achilles' heel in ovarian cancer, evidenced by the emergence of drugs that exploit their biologic deficits [16]. Nevertheless, molecular differences in ovarian tumors with genetically- and epigenetically-inactivated BRCA1 status have already been observed [41]. Whether these drugs show similar efficacy in tumors that have achieved BRCA1 deficiency by epigenetic means, or whether these tumors indeed represent a clinically distinct class, remains to be seen. As results become available from the international groups focused on sequencing hundreds, if not thousands of late-stage cancer genomes over the next several years–including The Cancer Genome Atlas (TCGA in the USA), Cancer Genome Project (United Kingdom) and the International Cancer Genome Consortium (ICGC)–the frequency of these events and the associated oncogenic pathways which are dysregulated will be known. Ultimately, the classification of ovarian tumors based on BRCA1 status and further subclassification based on the molecular mechanism of inactivation represents an exciting development in both understanding the molecular etiology of this disease and a possible tool for more effective clinical decisionmaking regarding its treatment. Conflict of interest statement The authors declare that there are no conflicts of interest.

References [1] Fleming GF, Ronnett BM, et al. In: Barakat RR, Markman M, Randall ME, editors. Principles and Practice of Gynecologic Oncology. 5th ed. Lippincott Williams & Wilkins; 2009. [2] SEER cancer statistics review 1975-2006 [homepage on the Internet]. Bethesda, MD: National Cancer Institute; 2008 [cited 9/1/2009]. [3] Soslow RA. Histologic subtypes of ovarian carcinoma: An overview. Int J Gynecol Pathol 2008;27(2):161–74. [4] Cannistra SA. Cancer of the ovary. N Engl J Med 2004;351(24):2519–29. [5] Easton DF, Ford D, Bishop DT. Breast and ovarian cancer incidence in BRCA1mutation carriers. Breast cancer linkage consortium. Am J Hum Genet 1995;56(1): 265–71. [6] Ramus SJ, Gayther SA. The contribution of BRCA1 and BRCA2 to ovarian cancer. Mol Oncol 2009;3(2):138–50. [7] Lancaster JM, Powell CB, Kauff ND, et al. Society of gynecologic oncologists education committee statement on risk assessment for inherited gynecologic cancer predispositions. Gynecol Oncol 2007;107(2):159–62. [8] Berek JS, Hacker NF. Practical gynecologic oncology; 2004. [9] Xu CF, Solomon E. Mutations of the BRCA1 gene in human cancer. Semin Cancer Biol 1996;7(1):33–40. [10] Smith SA, Easton DF, Evans DG, Ponder BA. Allele losses in the region 17q12-21 in familial breast and ovarian cancer involve the wild-type chromosome. Nat Genet 1992;2(2):128–31. [11] Knudson Jr AG. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci U S A 1971;68(4):820–3. [12] Russell PA, Pharoah PD, De Foy K, et al. Frequent loss of BRCA1 mRNA and protein expression in sporadic ovarian cancers. Int J Cancer 2000;87(3):317–21. [13] Cass I, Baldwin RL, Varkey T, Moslehi R, Narod SA, Karlan BY. Improved survival in women with BRCA-associated ovarian carcinoma. Cancer 2003;97(9):2187–95. [14] Chiang JW, Karlan BY, Cass L, Baldwin RL. BRCA1 promoter methylation predicts adverse ovarian cancer prognosis. Gynecol Oncol 2006;101(3):403–10. [15] Matei DE, Nephew KP. Epigenetic therapies for chemoresensitization of epithelial ovarian cancer. Gynecol Oncol 2009. [16] Fong PC, Boss DS, Yap TA, et al. Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med 2009;361(2):123–34. [17] Illingworth RS, Bird AP. CpG islands—“a rough guide”. FEBS Lett 2009;583(11): 1713–20.

E. Senturk et al. / Gynecologic Oncology 119 (2010) 376–383 [18] Lander ES, Linton LM, Birren B, et al. Initial sequencing and analysis of the human genome. Nature 2001;409(6822):860–8921. [19] Bird AP. The relationship of DNA methylation to cancer. Cancer Surv 1996;28: 87–101. [20] Esteller M. Aberrant DNA methylation as a cancer-inducing mechanism. Annu Rev Pharmacol Toxicol 2005;45:629–56. [21] Kim MS, Lee J, Sidransky D. DNA methylation markers in colorectal cancer. Cancer Metastasis Rev 2010;29(1):181–206. [22] Lai HC, Lin YW, Huang TH, et al. Identification of novel DNA methylation markers in cervical cancer. Int J Cancer 2008;123(1):161–7. [23] Su HY, Lai HC, Lin YW, Chou YC, Liu CY, Yu MH. An epigenetic marker panel for screening and prognostic prediction of ovarian cancer. Int J Cancer 2009;124(2): 387–93. [24] Garinis GA, Patrinos GP, Spanakis NE, Menounos PG. DNA hypermethylation: when tumour suppressor genes go silent. Hum Genet 2002;111(2):115–27. [25] Okano M, Bell DW, Haber DA, Li E. DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 1999;99 (3):247–57. [26] Klose RJ, Bird AP. Genomic DNA methylation: the mark and its mediators. Trends Biochem Sci 2006;31(2):89–97. [27] Brenner C, Deplus R, Didelot C, et al. Myc represses transcription through recruitment of DNA methyltransferase corepressor. EMBO J 2005;24(2): 336–46. [28] Baldwin RL, Tran H, Karlan BY. Loss of c-myc repression coincides with ovarian cancer resistance to transforming growth factor beta growth arrest independent of transforming growth factor beta/Smad signaling. Cancer Res 2003;63(6): 1413–9. [29] Chen CH, Shen J, Lee WJ, Chow SN. Overexpression of cyclin D1 and c-myc gene products in human primary epithelial ovarian cancer. Int J Gynecol Cancer 2005;15(5):878–83. [30] Miranda TB, Jones PA. DNA methylation: the nuts and bolts of repression. J Cell Physiol 2007;213(2):384–90. [31] Xu CF, Brown MA, Chambers JA, Griffiths B, Nicolai H, Solomon E. Distinct transcription start sites generate two forms of BRCA1 mRNA. Hum Mol Genet 1995;4(12):2259–64. [32] Smith TM, Lee MK, Szabo CI, et al. Complete genomic sequence and analysis of 117 kb of human DNA containing the gene BRCA1. Genome Res 1996;6(11):1029–49. [33] Thakur S, Croce CM. Positive regulation of the BRCA1 promoter. J Biol Chem 1999;274(13):8837–43. [34] Mancini DN, Rodenhiser DI, Ainsworth PJ, et al. CpG methylation within the 5' regulatory region of the BRCA1 gene is tumor specific and includes a putative CREB binding site. Oncogene 1998;16(9):1161–9. [35] Catteau A, Harris WH, Xu CF, Solomon E. Methylation of the BRCA1 promoter region in sporadic breast and ovarian cancer: correlation with disease characteristics. Oncogene 1999;18(11):1957–65. [36] Bianco T, Chenevix-Trench G, Walsh DC, Cooper JE, Dobrovic A. Tumour-specific distribution of BRCA1 promoter region methylation supports a pathogenetic role in breast and ovarian cancer. Carcinogenesis 2000;21(2):147–51.

383

[37] Esteller M, Silva JM, Dominguez G, et al. Promoter hypermethylation and BRCA1 inactivation in sporadic breast and ovarian tumors. J Natl Cancer Inst 2000;92(7): 564–9. [38] Baldwin RL, Nemeth E, Tran H, et al. BRCA1 promoter region hypermethylation in ovarian carcinoma: a population-based study. Cancer Res 2000;60(19):5329–33. [39] Wang C, Horiuchi A, Imai T, et al. Expression of BRCA1 protein in benign, borderline, and malignant epithelial ovarian neoplasms and its relationship to methylation and allelic loss of the BRCA1 gene. J Pathol 2004;202(2):215–23. [40] Wilcox CB, Baysal BE, Gallion HH, Strange MA, DeLoia JA. High-resolution methylation analysis of the BRCA1 promoter in ovarian tumors. Cancer Genet Cytogenet 2005;159(2):114–22. [41] Press JZ, De Luca A, Boyd N, et al. Ovarian carcinomas with genetic and epigenetic BRCA1 loss have distinct molecular abnormalities. BMC Cancer 2008;8:17. [42] Rice JC, Massey-Brown KS, Futscher BW. Aberrant methylation of the BRCA1 CpG island promoter is associated with decreased BRCA1 mRNA in sporadic breast cancer cells. Oncogene 1998;17(14):1807–12. [43] Rice JC, Futscher BW. Transcriptional repression of BRCA1 by aberrant cytosine methylation, histone hypoacetylation and chromatin condensation of the BRCA1 promoter. Nucleic Acids Res 2000;28(17):3233–9. [44] Magdinier F, Billard LM, Wittmann G, et al. Regional methylation of the 5' end CpG island of BRCA1 is associated with reduced gene expression in human somatic cells. FASEB J 2000;14(11):1585–94. [45] Auriol E, Billard LM, Magdinier F, Dante R. Specific binding of the methyl binding domain protein 2 at the BRCA1-NBR2 locus. Nucleic Acids Res 2005;33(13):4243–54. [46] Atlas E, Stramwasser M, Mueller CR. A CREB site in the BRCA1 proximal promoter acts as a constitutive transcriptional element. Oncogene 2001;20(48):7110–4. [47] Graves ML, Zhou L, MacDonald G, Mueller CR, Roskelley CD. Regulation of the BRCA1 promoter in ovarian surface epithelial cells and ovarian carcinoma cells. FEBS Lett 2007;581(9):1825–33. [48] Atlas E, Stramwasser M, Whiskin K, Mueller CR. GA-binding protein alpha/beta is a critical regulator of the BRCA1 promoter. Oncogene 2000;19(15):1933–40. [49] Atallah E, Garcia-Manero G. Use of hypomethylating agents in myelodysplastic syndromes. Clin Adv Hematol Oncol 2007;5(7):544–52. [50] Billam M, Sobolewski MD, Davidson NE. Effects of a novel DNA methyltransferase inhibitor zebularine on human breast cancer cells. Breast Cancer Res Treat 2009. [51] Gray SG, Al-Sarraf N, Baird AM, Cathcart MC, McGovern E, O'Byrne KJ. Regulation of EP receptors in non-small cell lung cancer by epigenetic modifications. Eur J Cancer 2009;45(17):3087–97. [52] Schrump DS, Nguyen DM. Targeting the epigenome for the treatment and prevention of lung cancer. Semin Oncol 2005;32(5):488–502. [53] Momparler RL, Bouffard DY, Momparler LF, Dionne J, Belanger K, Ayoub J. Pilot phase I-II study on 5-aza-2'-deoxycytidine (decitabine) in patients with metastatic lung cancer. Anticancer Drugs 1997;8(4):358–68. [54] Fang F, Balch C, Schilder J, et al. Clinical and pharmacodynamic activity of decitabine in vivo for patients with recurrent, platinum-resistant, epithelial ovarian cancer. Proceedings of the 100th annual meeting of the American association for cancer research; 2009 Apri 18–22; Denver, CO. Philadelpha (PA): AACR; 2009.