DNA Methylation and Breast Carcinogenesis

DNA methylation and breast carcinogenesis

Martin Widschwendter*,1 and Peter A Jones1

1USC/Norris Comprehensive Cancer Center, Keck School of Medicine of the University of Southern California, 1441 Eastlake Avenue, MS 8302L, Los Angeles, California, CA 90089-9181, USA

Knowledge about breast carcinogenesis has accumulated associated with development of breast cancer and lead to during the last decades but has barely been translated a ‘new type of tissue’ (neoplasm) characterized by a into strategies for early detection or prevention of this variety of genetic lesions including gene amplifications, common disease. Changes in DNA methylation have gene deletions, point mutations, loss of heterozygosity, been recognized as one of the most common molecular chromosomal rearrangements, and overall aneuploidy. alterations in human neoplasia and hypermethylation of Besides the above mentioned well known genetic gene-promoter regions is being revealed as one of the alterations, epigenetic alterations are among the most most frequent mechanisms of loss of gene function. The common molecular alterations in human neoplasia heritability of methylation states and the secondary (Baylin and Herman, 2000; Jones, 1996; Jones and nature of the decision to attract or exclude methylation Laird, 1999). Epigenetic changes differ from genetic support the idea that DNA methylation is adapted for a changes mainly in that they occur at a higher frequency specific cellular memory. According to Hanahan and than genetic changes, are reversible upon treatment Weinberg, there are six novel capabilities a cell has to with pharmacological agents and occur at defined acquire to become a cancer cell: limitless replicative regions in a gene. Epigenetics describes a trait that is potential, self-sufficiency in growth signals, insensitivity heritable, yet not based upon a change in primary to growth-inhibitory signals, evasion of programmed cell DNA sequence. DNA methylation is one well known death, sustained angiogenesis and tissue invasion and epigenetic-mechanism and it has become clear in recent metastasis. This review highlights how DNA-methylation years that there is a synergy between genetic and contributes to these features and offers suggestions about epigenetic changes and that Knudson’s two-hit hypoth- how these changes could be prevented, reverted or used esis has to be revised: instead of two possibilities (loss as a ‘tag’ for early detection of breast cancer or, of heterozygosity or homozygous deletion), a third preferably, for detection of premalignant changes. possibility – transcriptional silencing by DNA methy- Oncogene (2002) 21, 5462 – 5482. doi:10.1038/sj.onc. lation of promoters – can disable tumor-suppressor 1205606 genes (Jones and Laird, 1999). Cytosines are methylated in the human genome Keywords: breast cancer; DNA methylation; carcino- mostly when located 5’ to a guanosine. These CpG genesis; prevention; early detection nucleotides have been severely depleted in the verte- brate genome to about 20% of the predicted frequency and most CpG dinucleotides (over 70%) are methylated. However, in small stretches of DNA termed Introduction CpG islands, which are about 500 – 2000 bp in length (Jones and Takai, 2001; Takai and Jones, 2002), the Each year more than 180 000 women in the US are CpG dinucleotide occurs at near the expected diagnosed with breast cancer, the most common cancer frequency and these areas are frequently located in among women in this country. If current breast cancer and around the transcription-start sites of approxi- rates remain constant, a woman born today has a one mately half of human genes. It has been increasingly in 10 chance of developing breast cancer (Eifel et al., recognized over the past 4 – 5 years that the CpG 2001; Howe et al., 2001). Each year, 44 000 women die islands of a large number of genes, which are mostly of breast cancer, making it the second leading cause of unmethylated in normal tissues, are methylated to cancer deaths among American women, after lung varying degrees in human cancers, including breast cancer, and the leading cause of death among women cancer (Yang et al., 2001a). The post-synthetic covalent aged 40 to 55 years. addition of a methyl group to cytosine is mediated by Decades of research have led to a substantial under- the three known active DNA cytosine methyltransfer- standing of the factors involved in the development of ase (DNMT1, 3a and 3b) (Robertson et al., 1999). breast cancer. All these factors cause or are at least When DNA containing a symmetrically methylated CpG dinucleotide is replicated, the result is two double-stranded DNA molecules, each containing a *Correspondence: M Widschwendter; methylated CpG dinucleotide on the parental strand E-mail: [email protected] but containing an unmethylated CpG dinucleotide on DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5463 the newly synthesized strand. The methylated state of breast cancer. Breast cancer is uncommon among the site on the parent molecule is maintained in the women younger than 30 years of age but the incidence daughter molecules when a maintenance methyltrans- increases sharply with age. The rate of increase in ferase recognizes the hemimethylated site and breast cancer incidence continues throughout life but methylates the unmethylated cytosine, restoring the slows somewhat between ages 45 and 50 years. This symmetrically methylated CpG dinucleotide pair. finding strongly suggests the involvement of reproduc- DNMT1 is mainly responsible for maintenance of tive hormones in breast cancer etiology, because non- DNA methylation, whereas DNMT3a and DNMT3b hormone-dependent cancers do not exhibit this change have been shown to methylate hemimethylated and in slope of the incidence curve around the time of unmethylated DNA with equal efficiencies (Okano et menopause (Pike et al., 1993). Several reproductive al., 1999). Overexpression of DNMT1 as well as the factors that alter estrogen status effect risk of breast DNMT3 mRNAs has been reported in human tumors cancer: early age at menarche and late age at (Mizuno et al., 2001; Robertson et al., 1999), but this menopause are associated with increased risk of breast phenomenon is probably only partially responsible for cancer. After menopause, adipose tissue is the major the observed methylation changes. Compared to source of estrogen, and obese postmenopausal women normal tissue, DNA from breast carcinomas is have both higher levels of endogenous estrogen and a generally hypomethylated (Soares et al., 1999); higher risk of breast cancer (Harris et al., 1992; Huang DNMTs appear to be upregulated in tumors when et al., 1997a). Postmenopausal hormone use increases RNA levels are normalized using b-actin or RNA pol the breast cancer risk depending on the duration of use II large subunit, but not when RNA levels are and whether estrogen alone or estrogen in combination normalized with proliferation associated genes, such with progestin is taken (Ross et al., 2000). On the other as histone H4 or PCNA (Eads et al., 1999). In this hand, hormonal manipulations, such as administration study, using colorectal tumors, neither the frequency of antiestrogens (e.g., tamoxifen), are useful in the nor the extent of CpG island hypermethylation in treatment of breast cancer and might reduce breast individual tumors correlated with mRNA expression of cancer incidence in high-risk women (Fisher et al., any of the three DNA methyltransferases. In hepato- 1998). Based on this knowledge a model connecting cellular carcinoma, hypomethylation was also epidemiologic knowledge can be built which incorpo- associated with upregulation of DNMT expression rates recently discovered epigenetic changes in breast (Lin et al., 2001). Therefore, up-regulation of DNMTs cancer. mRNAs may simply be a result of increased cell A woman’s age might be the driving force for the proliferation in cancer. Although it appears that accumulation of mutational load, telomere dysfunction different DNMT enzymes have distinct sequences and increased epigenetic gene silencing. Exposure to targeted for methylation (Hsieh, 1999; Okano et al., growth factors like estrogen increases the likelihood of 1999), this fact does not completely explain the occurrence of these changes in breast epithelial stem significant variety of methylation patterns in tumors. cells as well as the propagation of these changes by These epigenetic markers on DNA can be copied enabling the cells to divide. Growth factors may not after DNA synthesis, resulting in heritable changes in only act directly on epithelial cells but also indirectly chromatin structure. The reciprocal relationship via the stromal microenvironment in which the tumor between the density of methylated cytosine residues cells develop and may profoundly influence many steps and the transcriptional activity of a gene has been of tumor progression. In various experimental tumor widely documented. It should be emphasized, however, models, the microenvironment affects the efficiency of that this inverse correlation has been demonstrated tumor formation, the rate of tumor growth, the extent conclusively only for methylation in the promoter of invasiveness, and the ability of tumor cells to regions and not in the transcribed parts of a gene metastasize. In carcinomas, the influences of the (Jones, 1999). Several tumor-suppressor genes contain microenvironment are mediated, in large part, by CpG islands in their promoters, and many of them paracrine signaling between epithelial tumor cells and show evidence of methylation silencing. After changes neighboring stromal fibroblasts (Elenbaas and Wein- associated with histone deacetylation have occurred berg, 2001). and after these CpG islands have become methylated, the relevant genes become irrevocably silent. A woman’s risk of developing breast cancer is DNA methylation and the establishment of six novel increased if she has a family history of the disease. capabilities towards breast carcinogenesis The largest re-analysis of individual data from 52 epidemiological studies so far completed reported that Recently Hanahan and Weinberg (2000) provided an 12% of women with breast cancer had one affected excellent overview of alterations in cell physiology that relative and 1% had two or more (Collaborative collectively define malignant growth. The six new Group on Hormonal Factors in Breast Cancer, 2001). capabilities, that a cell has to acquire to become Therefore the genesis of most breast cancers cannot malignant are: (1) limitless replicative potential, (2) be explained by heritage. Age as well as duration of self-sufficiency in growth signals, (3) insensitivity to exposure to endogenous or exogenous steroid hormone growth-inhibitory signals, (4) evasion of programmed levels might be one of the best defined risk factors for cell death, (5) sustained angiogenesis and (6) tissue

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5464 invasion and metastasis. Despite the fact that many doublings, before entering a second growth plateau genes mentioned below might have overlapping func- (termed postselection, agonescence or M1). Cells tions in different pathways, we use this classification to emerging from this stage enter a telomere-based highlight how DNA methylation may contribute to crisis-like state (Romanov et al., 2001). p16INK4A, one breast carcinogenesis (Figure 1). of the most commonly inactivated tumor suppressor genes in human cancer (Sherr, 1996), is a cyclin- dependent kinase inhibitor that regulates progression (1) Limitless replicative potential through the G1 phase of the cell cycle by binding and Senescence and genomic integrity are thought to be inhibiting cyclin-dependent kinases 4 and 6 (Jiang et important barriers to the development of malignant al., 1998) thus inhibiting Rb phosphorylation. In early lesions. Human fibroblasts undergo a limited number passage HMECs, p16Ink4A increases as HMECs reach of cell divisions before entering an irreversible arrest, M0, and this is paralleled by a decrease in the level of called senescence. In contrast, human mammary Rb (Foster et al., 1998). HMECs that emerge from the epithelial cells (HMECs) in culture exhibit an initial first plateau lose expression of p16INK4A protein, and growth phase that is followed by a transient growth this correlates with a progressive, region-specific de plateau (termed selection or M0), from which prolif- novo methylation of the p16 CpG island (Brenner et erative cells emerge to undergo further population al., 1998; Foster et al., 1998; Huschtscha et al., 1998;

Figure 1 View of breast carcinogenesis from a DNA methylation standpoint. DNA-methylation of certain genes and therefore irrevocably silencing them, enables some cells to acquire new capabilities needed for tumorigenesis. Prolonged exposure to growth promoting substances like estradiol, which exerts direct effects on epithelial cells or indirect effects via stromal cells, allows cells to propagate heritable changes like DNA-methylation. Cells which accumulated DNA-methylation at various loci as a function of time (age) and as a function of exposure to growth factors (e.g. estrogens), have gained all six novel capabilities needed to be a tumor cell due to the fact that ‘defense strategies’ have irrevocably switched off by DNA-methylation

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5465 Wong et al., 1999). These results raise the question of region or whether Id-1 suppresses p16INK4A expression the mechanism responsible for de novo methylation of in the absence of DNA methylation. this region and whether DNA methylation is respon- The fact that treatment with demethylating agents in sible for p16INK4A gene silencing. vivo as well as in vitro causes a robust re-expression of Recently it has been demonstrated that either Ets1 p16 in large varieties of tumors (Bender et al., 1998) as and Ets2 (through an Ets-binding site) (Ohtani et al., well as in hTERT-immortalized cells (Farwell et al., 2001), or class A basic HLH proteins (Pagliuca et al., 2000) supports the hypothesis that methylation of the 2000) (by activating E-boxes within the p16INK4A p16 promoter may be a very important mechanism for promoter) can activate p16INK4A gene expression and permanent silencing of the p16INK4A gene. that Id1 represses p16 expression by either interfering Wong et al. (1999) have studied the methylation with basic HLH proteins or Ets factors (Alani et al., status of seven CpG sites in the p16 CpG island on 2001; Ohtani et al., 2001). Id genes can immortalize individual DNA molecules by sequencing PCR clones various types of cells, and Id1 overexpression causes of bisulfite-treated genomic DNA. They found that this more aggressive phenotypes of breast cancer and island was initially methylated at a subset of sites in regulates steroid-hormone-responsive growth of breast three discrete regions in association with p16 transcrip- cancer cells (referenced in Norton, 2000). Although tional repression and escape from M0 growth arrest. there is a striking chronological association between Methylation gradually increased in density with Id-1 regulation of cellular senescence through tran- continued passage and spread to sites in adjacent scriptional repression of p16INK4A (Alani et al., 2001) regions. Noteworthy, the E-box motif of the p16INK4A and p16INK4A promoter methylation (Foster et al., promoter lies about one nucleosome distance upstream 1998; Wong et al., 1999), it is unclear whether Id-1 of the first region, whereas regions 1, 2 and 3 are also directly or indirectly causes methylation at this specific separated by approximately one nucleosome distance.

Figure 2 Preventive, therapeutic and diagnostic possibilities using DNA methylation as a target. Changes in the chromatin structure of genes not activated might serve as a substrate for DNA-methylation. Exposure to carcinogens can not be handled properly because genes needed for this purpose cannot be expressed any more. Accumulation of these genes irrevocably silenced may lead to cancer. Green numbers: Mechanisms of prevention or reversal of DNA methylation (most of them have not been studied extensively in breast cancer). (1) Challenging breast epithelial stem cells early in life (analogous to pregnancy at young age or to a pregnancy complicated by pre-eclampsia) might prevent methylation of important cell-death pathways and thereby giving these cells the chance to recall efficient patterns of genes to adequately handle exposure to carcinogens. (2) HDAC inhibitors – applied systemically or retrograde intraductally – might prove to be efficient to prevent DNA methylation and hence irreversible silencing. (3) After DNA methylation of certain genes has already been established, demethylating agents might revert this process. (4) An established cancer might be treated with demethylating agents alone or in combination with biological response modifiers as outlined recently (Widsch- wendter and Jones, 2002). Red numbers: Diagnostic possibilities in breast cancer using DNA methylation. (5) Examining DNA- methylation of certain genes in breast nipple fluid might be used for early detection of premalignant lesions or non-invasive cancer. (6) Chemopreventive strategies might be monitored by DNA methylation (surrogate endpoint biomarker) in breast nipple fluid. (7) Evidence for systemic disease might be gained due to analysis of DNA methylation in patients’ blood and might serve as a predictive marker. Relapse of disease might be detected early by means of studying DNA methylation in patients’ blood and therapies might be monitored using DNA methylation as a tumor marker. It has to be stressed that for different purposes ((5) to (7)), different combinations of various genes might prove to be useful

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5466 Table 1 Genes for which direct or indirect evidence exists for involvement of methylation in breast carcinogenesis Gene Alternate gene name Function Methylation

APAF1 Apoptosis Protease- Activation of procaspase-9, results in initiation of a indirect Activating Factor 1 cascade involving the downstream executioners caspase-3, -6, and -7?apoptosis APC Adenomatous Polyposis of Cell adhesion, signal transduction, stabilization of direct the Colon the cytoskeleton, regulation of cell cycle and apoptosis BCSG1 Breast Cancer Specific Increases motility and invasiveness direct Gene 1; Synuclein-g BRCA1 Breast Cancer type 1 Involved in DNA repair, recombination, direct checkpoint control of the cell cycle and transcription. Interacts with p53, STAT-factors, SRBC, etc. Caspase-8 Apoptosis-related cysteine apoptosis indirect protease CCND2 Cyclin D2 Cell cycle regulation direct DAPK Death-associated protein Mediator of interferon-g induced apoptosis direct kinase 1 E-Cad E-Cadherin Epithelial cell – cell adhesion, suppresses invasion direct and metastasis ER Estrogen receptor a and b Regulation of cell proliferation, predictor of direct endocrine therapy FHIT Fragile histidine triad gene Controls proliferation and apoptosis, tumor direct suppressor GPC3 Glypican 3 apoptosis direct GSTP1 Glutathione S-transferase Carcinogen detoxification direct P1 H-Cad H-Cadherin Cell-cell adhesion direct HIN1 High in normal 1 Putative cytokine, inhibits cell growth direct HOXA5 Homeo box A5 Upregulates p53, apoptosis direct hTERT Telomere reverse synthesizes the telomere ends of linear direct transcriptase chromosomes, implicated in human cell immortalization IF-regulated Interferon regulated genes Interleukin 6, ICAM 1, Superoxide dismutase and indirect genes Elafin are regulated by Interferons and mediate tumorsuppressive functions; involved in senescence mac25 Insulin-like growth factor Cell cycle regulation, apoptosis, involved in indirect binding-related protein 1 senescence Maspin Protease inhibitor 5 Inhibitor of angiogenesis, reduces cells’ ability to direct induce tumors and metastasize NES1 Kallikrein 10 Inhibition of anchorage-independent growth and direct tumor formation Nm23-H1 Metastasis inhibition factor Metastasis suppressor activity direct NM23 NOEY2 Ras homolog gene family Suppresses clonogenic growth; regulation of cyclin direct member I D1 and p21 p16 Cyclin-dependent kinase Cell cycle regulation, involved in senescence direct inhibitor 2A p21 Cyclin-dependent kinase Cell cycle regulation indirect inhibitor 1A p53 Transformation-related Apoptosis, cell cycle regulation, inhibition of direct protein 53 growth and invasion p73 P53 related protein p73 Inhibitor of angiogenesis, apoptosis indirect PR Progesterone receptor Growth regulation direct Prostasin Protease serine 8 Suppresion of invasion direct RAR-b Retinoic acid receptor b Apoptosis, involved in senescence, inhibition of direct proliferation RASSF1A Ras Association domain Reduces colony formation, suppresses anchorage- direct family protein 1 independent growth, and inhibits tumor formation, apoptosis RFC Reduced folate carrier Cellular uptake of methotrexate direct RIZ1 Retinoblastoma protein- Tumor suppressor direct binding zinc finger protein SOCS1 Suppressor of cytokine Suppresses growth rate and anchorage-independent direct signaling 1 growth, induction of apoptosis, regulation of STAT activation Continued

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5467 Table 1 (Continued ) Gene Alternate gene name Function Methylation SRBC Serum deprivation Interaction with BRCA1 direct response factor (sdr)- related gene product that binds to c-kinase STAT1 Signal transducer and Growth regulation, induction of apoptosis, indirect activator of transcription involved in senescence SYK Spleen tyrosine kinase Inhibits tumor growth and metastasis direct TGFbRII Transforming growth Cell cycle regulation direct factor b receptor II THBS1 Thrombospondin 1 Inhibition of angiogenesis and invasion direct TIMP3 Tissue inhibitor of Suppresses tumor growth, angiogenesis, invasion direct metalloproteinase-3 and metastasis TMS1 Target of methylation apoptosis direct induced silencing 1 TWIST TWIST Inhibits oncogene- and p53-dependent cell death direct ZAC Pleomorphic adenoma Induction of apoptosis and cell cycle regulation direct gene-like 14-3-3s Stratifin Cell cycle regulation direct

The overall maintenance methylating activity in or repressors such as WT1 which interact with the fibroblasts is greatly decreased during cellular senes- hTERT gene regulatory region to modulate telomerase cence but is strikingly elevated in immortalized cells activity in aging cells and tumorigenesis. hTERT is (Lopatina et al., 2002). Detailed analyses demonstrated methylated in some breast cancer cell lines (Devereux that the major maintenance methyltransferase, et al., 1999), but no specific methylation patterns at DNMT1, declined steadily in activity with cellular certain CpG sites or regions of the hTERT promoter senescence and immortalization. However, DNMT3B, emerged that correlated with expression in all of the which has significant de novo methylating activity, diverse cell lines examined (Devereux et al., 1999). As increased markedly in activity in aging and immorta- outlined in the Introduction, exposure to estrogen is a lized cells (Lopatina et al., 2002). These studies major risk factor for breast cancer. It has been shown indicated that reduced genome-wide methylation in that estrogen is also able to activate telomerase via aging cells (Wilson and Jones, 1983) may be attributed direct and indirect effects on hTERT promoter (Kyo et to attenuated DNMT1 activity but that regional or al., 1999) in a breast cancer cell line. gene-localized hypermethylation in aging and immor- Besides p16INK4A, the retinoblastoma protein-inter- talized cells may be linked to increased de novo acting zinc finger gene (RIZ1), which is a tumor methylation by other DNMTs. In summary, in suppressor gene and a member of a nuclear histone/ HMECs during escape from M0, specific regions like protein methyltransferase superfamily, can affect Rb. the p16 CpG island might serve as a target for DNMTs RIZ1 inactivation is commonly found in many types of responsible for de novo methylation. In senescing cells human cancers and occurs through loss of mRNA Ets transcription factors or bHLHs protect this region. expression, frameshift mutation, chromosomal dele- Inhibitors of cellular senescence like Id-1 may prevent tion, and missense mutation. It has been shown that binding of protective transcription factors, and thereby loss of RIZ1 mRNA in human cancers is associated enable spreading of methylation. It is intriguing to with DNA methylation of its promoter CpG island speculate whether de novo DNMTs directly or (Du et al., 2001). Methylation of the RIZ1 promoter indirectly bind to ‘unprotected’ E-box motifs initiating was found in 44% of breast cancer specimens and de novo methylation in regions spaced by one strongly correlated with lost or decreased RIZ1 mRNA nucleosome distance. expression. Treatment with the methylation inhibitor 5- Beside inactivation of p16INK4A, telomerase activity aza-2’-deoxycytidine (5-aza-CdR) activated RIZ1 is required to immortalize human mammary epithelial mRNA expression in cancer cells (Du et al., 2001). cells (Kiyono et al., 1998). Although telomerase, which Additional evidence has suggested that RIZ1 protein maintains the integrity of chromosome ends, is down- may be a specific effector of estrogen action downregulated as cells differentiate leading to attrition of stream of the hormone-receptor interaction, chromosomal termini and ultimate replicative senes- presumably involved in proliferation control (Abbon- cence, it is up-regulated in most cancer cells which danza et al., 2000). show no net loss of average telomere length. The D-type cyclins (cyclins D1, D2, and D3) are involved mRNA level of the catalytic component of telomerase, in regulation of transition from G1 to S during the cell hTERT, is the major determinant of telomerase cycle. Cyclin D2 is unique among the three D-type activity but little is known about control of hTERT cyclins, as its expression is significantly upregulated transcription. Recently Tollefsbol and Andrews (2001) under conditions of growth arrest in phenotypically have proposed mechanisms whereby cytosine methyla- normal human fibroblasts. Ectopic overexpression of tion may alter the binding of activators such as c-Myc cyclin D2 effectively blocked cell cycle progression

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5468 (Meyyappan et al., 1998). Hypermethylation of the epithileum contains an enzyme, aldehyde dehydrogen- CpG island in the cyclin D2 promoter was detected by ase 6, that is capable of synthesizing retinoic acid from methylation-specific PCR in nearly half of the breast retinal and retinol. Expression of this enzyme is lost in cancers and was associated with silencing of cyclin D2 breast cancer, explaining the defect of retinoic acid gene expression. Promoter hypermethylation was also biosynthesis in these cells. detected in ductal carcinoma in situ, suggesting that Together, this evidence might explain the chemopre- loss of cyclin D2 expression is an early event in ventive potentials of retinoids: exposure to retinoic acid tumorigenesis (Evron et al., 2001a). in cells that have lost their own ability to synthesize A variety of other genes are known to be induced this ligand might activate the RAR-bRARE. Beside upon senescence in addition to p16. Retinoids are tumor suppressive effects mediated by RAR-b (Lotan important cellular dietary factors that regulate differ- et al., 1995), this scenario prevents chromatin conden- entiation and cellular growth and serve as ligands for sation at this region which would cause DNA specific nuclear receptors, the retinoic acid receptors methylation and irreversible suppression of the RAR-b. (RARs) (Minucci and Pelicci, 1999). Ligand-activated Very recently, Di Croce et al. (2002) report an receptors regulate gene transcription through target alternative mechanism of RAR-b methylation. In acute retinoic acid-responsive elements (RAREs) found in promyelocytic leukemias, PML-RAR acts as an promoter regions. Transcription of the RAR-b2 gene is oncogene and induces hypermethylation of the RAR- induced by retinoic acid and RAR-b is able to repress bRARE by recruiting DNA methyltransferases to this AP1, a condition which might be incompatible with promoter. Retinoic acid treatment induces promoter tumor progression (reviewed in Altucci and Grone- demethylation, gene re-expression, and reversion of the meyer, 2001). In P19 embryonal carcinoma cells, transformed phenotype. retinoic acid markedly increases nuclease sensitivity Another gene associated with senescence is mac25 within the promoter, including a site near the RARE (IGFBP-rP1), Insulin-like Growth Factor Binding to which the nuclear receptor retinoid X receptor Protein-related Protein-1). This gene was selected by (RXR)-RAR heterodimer binds. These changes became differential display of mRNA in a search for genes undetectable upon removal of RA, which coincides overexpressed in senescent human mammary epithelial with the extinction of transcription (Bhattacharyya et cells (Swisshelm et al., 1995). Mac25 accumulates in al., 1997). In the absence of agonists, corepressor senescent normal mammary epithelial cells but is complexes bind to the RAR-RXR heterodimer. This absent or low in breast cancer cell lines and tissue complex attracts histone deacetylases to an area (Landberg et al., 2001; Swisshelm et al., 1995). In surrounding RAREs to remove acetyl groups from prostate cancer it was demonstrated that IGFBP-rP1/ nucleosomal histones, resulting in chromatin condensa- mac25 alters cell cycle kinetics by delaying G1 tion and gene silencing (reviewed in Altucci and progression and was associated with induction of Gronemeyer, 2001). This transcriptionally silent chro- apoptosis (Sprenger et al., 2002). In the breast, strong matin might serve as a de novo methylation target IGFBP-rP1/mac25 staining was observed in luminal (Bird, 2002). Analyses with receptor-selective ligands epithelial cells of normal lobules and ducts, in apocrine and an antagonist showed that increase in restriction cells of cysts and fibroadenomas. Moderate to weak site accessibility correlates with transcriptional activa- protein expression was found in hyperplastic and DCIS tion, which parallels the RA-induced in vivo footprint cells, but no specific staining was detected in invasive of the promoter. It has been demonstrated that most carcinoma cells (Burger et al., 1998). Low IGFBP-rP1/ tumor cells show a loss of RAR-b expression, but that mac25 was associated with high cyclin E protein RAR-a and -g as well as retinoid X receptor b were content, retinoblastoma protein (pRb) inactivation, variably expressed in both normal and cancerous low bcl-2 protein, poorly differentiated tumors and breast tissue and cells (Widschwendter et al., 1995, higher stage. There was a significantly impaired 1997; Xu et al., 1997). RAR-b gene expression is prognosis for patients with low IGFBP-rP1/mac25 induced both by retinoic acid and by fenretinide in protein tumors (Landberg et al., 2001). Treatment with normal cells, but tumor cells fail to respond to either. retinoids increases IGFBP-rP1/mac25 in normal In contrast, RAR-b expression increases with serial mammary epithelial cells but not in tumor cells. It passage in senescing cells (Lee et al., 1995; Swisshelm et was proposed that IGFBP-rP1/mac25 expression may al., 1994). RAR-b promoter methylation has been be regulated by RAR-b and represent one of the demonstrated to be an important mechanism causing downstream genes on the RAR-b senescence pathway suppression of the RAR-b2 message (Arapshian et al., (Swisshelm et al., 1995). Although there is no direct 2000; Bovenzi et al., 1999; Sirchia et al., 2000; evidence that methylation causes silencing of IGFBP- Widschwendter et al., 2000) in breast cancer. No rP1/mac25 in breast cancer, restriction landmark RAR-b2 methylation was observed in normal breast genomic scanning for methylation has demonstrated tissue nor in mortal or immortal human mammary that IGFBP-rP1/mac25 methylation and subsequent epithelial cells (Sirchia et al, 2000). Alterations of the suppression of this gene was shown to be involved in chromatin structure mediated by low intracellular murine SV40T/t antigen-induced hepatocarcinogenesis levels of retinoic acid might precede DNA methylation (Komatsu et al., 2000). at the RAR-b2 promoter (Sirchia et al., 2000). Indeed, Using a cDNA microarray approach, Perou et al. Rexer et al. (2001) have found that normal breast (1999) identified a cluster of interferon (IFN)-regulated

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5469 genes (STAT-1, etc.) highly expressed in normal human et al., 2001). The sis-inducible element (SIE)-1, a STAT mammary epithelial cells under three circumstances: responsive element located upstream of the p21WAF1 addition of interferon, senescence and confluence. The CpG island was shown to be completely methylated in latter two suggest that there may be circumstances that rhabdomyosarcomas. Using electrophoretic mobility activate expression of these genes other than the shift assays methylation within SIE-1 significantly presence of interferon. They also showed that this inhibited binding of activated STAT1 and abrogated entire cluster of IFN-regulated genes was highly STAT-mediated transcription activation in response to expressed in some of the tumors, moderately expressed IFN-g. However demethylation at SIE-1 reactivated in others and apparently silent in others (Perou et al., p21WAF1 expression and restored the responsiveness to 1999). A recent paper (Liang et al., 2002) used high IFN-g (Chen et al., 2000). Another protein shown to density oligonucleotide gene expression microarrays to interact with BRCA1 is SRBC (serum deprivation examine the effects of 5-aza-CdR treatment on a response factor (sdr)-related gene product that binds to human bladder tumor cell line. Treatment with this c-kinase), which was isolated in a yeast two-hybrid demethylating agent showed 60 genes to be induced screening with BRCA1 as the probe (Xu et al., 2001). more than fourfold. Half of these genes belonged to Although the function of SRBC is not yet clear, it has the interferon signaling pathway and some also play been shown that suppression of this gene is associated important roles in breast carcinogenesis. Alterations of with hypermethylation of CpG dinucleotides in its IL6 expression are associated with pathogenesis in promoter region (Xu et al., 2001). breast cancer (Basolo et al., 1993) and the repression of In addition to an active telomerase and presence of IL6 is associated with hypermethylation (Armenante et H-ras, SV40 large T antigen is known to be necessary al., 1999). Expression of the intercellular adhesion to transform human mammary epithelial cells into molecule-1 (ICAM-1) is associated with good prognosis human breast cancer cells (Elenbaas et al., 2001). p53 and might serve as a suppressor of tumor progression inhibition – achieved by SV40 large T antigen – was (Ogawa et al., 1998). Despite the lack of direct proof, it required for the transformation of these cells. Inactiva- was suggested that methylation of CpG islands may tion of p53 by mutations is found in 20% of breast play a role in the downregulation of ICAM-1 (Arnold cancers (Pharoah et al., 1999). Hypermethylation in the et al., 2001). Elafin is an elastase inhibitor, constitu- p53 promoter region is an alternative pathway to tively expressed in normal mammary epithelial cells, tumorigenesis where there is no p53 gene mutation but downregulated in most tumor cell lines (Zhang et (Kang et al., 2001). About 60% of tumors showed al., 1995). Manganese-containing superoxide dismutase neither methylation nor p53 mutation (Kang et al., functions as a tumor suppressor gene in breast cancer 2001). DNA damage leads to stabilization of p53, (Li et al., 2001a) and it is silenced by DNA which is required for maintenance of the G2 arrest methylation of the 5’-CpG island (Huang et al., 1997b). through the transactivation of the p21 and 14-3-3s The interferon-g response has been postulated to be genes. 14-3-3s is required to sequester cdc2-cyclinB1 part of the endogenous tumor surveillance system. At complexes in the cytoplasm, whereas p21 may prevent the cellular level, IFN-g mediates activation of an any cdc2-cyclin B1 that enters the nucleus from antiviral state and causes cell growth arrest in the G1 becoming activated (Chan et al., 1999). Hypermethyla- phase of the cell cycle. The biological effects of IFN-g tion of 14-3-3s was detected in more than 90% of are mediated through a heterodimeric transmembrane breast cancers and was associated with lack of gene receptor which is capable of activating the Janus kinase expression (Ferguson et al., 2000). Interestingly, beside (JAK)-STAT pathway, leading to tyrosine phosphor- invasive breast cancers, methylation of 14-3-3s was ylation of the STAT1a protein (Stark et al., 1998). detected in 83% of ductal carcinoma in situ and 38% STAT-1 plays an important role in growth arrest, in of atypical hyperplasias, but none of the hyperplasias promoting apoptosis and is implicated as a tumor without atypia showed hypermethylation (Umbricht et suppressor. Beside IFN-g, induction and activation of al., 2001). Interestingly, patients with breast cancer STAT-1 is mediated by retinoic acid via the RAR-b showed 14-3-3s hypermethylation in normal adjacent signaling pathways in breast cancer cells (Shang et al., breast epithelium as well, whereas epithelium from 1999). Inhibition of DNMTs stimulates the expression normal breasts contained unmethylated 14-3-3s, irre- of STAT1, 2 and 3 in colon tumor cells (Karpf et al., spective of age (Umbricht et al., 2001). 1999). It has been demonstrated that STAT1 acts in The polycyclic aromatic hydrocarbon 7,12-dimethyl- concert with the BRCA1 tumor suppressor to differ- benz anthracene (DMBA) is a very potent inducer of entially activate transcription of a subset of IFN-g mammary tumors in mice (Mehta, 2000). DMBA target genes (Ouchi et al., 2000). Induction of the induces NF-kB and increases c-myc levels in HMECs cyclin-dependent kinase inhibitor p21WAF1 was syner- very early, occurring prior to malignant transformation gistically activated by BRCA1, whereas the IRF-1 gene (Kim et al., 2000). c-myc may be one of the was unaffected. BRCA1 promoter hypermethylation transcription factors necessary for induction of telo- has been described to be present in about 13% of merase (Tollefsbol and Andrews, 2001) as outlined unselected primary breast carcinomas (Esteller et al., above. Glutathione S-transferases (GSTs) are a super- 2000) and these tumors may mimick the gene family of genes, responsible for the detoxification of expression profiles found in breast cancers from xenobiotics. GSTP1 null mice demonstrated an patients with BRCA1 germline mutations (Hedenfalk increased skin tumorigenesis after topical application

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5470 of DMBA (Henderson et al., 1998). MSP-based studies increased breast cancer risk nor do the chemopreven- demonstrated that GSTP1 promoter methylation is tive properties of antiestrogens fit into this model. It is associated with gene inactivation in about 30% of therefore intriguing to speculate whether dysregulation primary breast carcinomas (Esteller et al., 1998). The of ER-a and PR expression are involved in breast GSTP1 CpG island is hypermethylated in ER positive, carcinogenesis. If expression of ER-a is necessary in the GSTP1 non-expressing cell lines but is undermethy- normal breast epithelium to protect the cells from lated in ER negative, GSTP1 expressing cell lines aberrantly expressing growth factor receptors, suppres- (Jhaveri and Morrow, 1998). sion of ER-a might undermine this protective mechanism. DNA methylation of the promoter region and the first exon appear to play a role in inactivating (2) Self-sufficiency in growth signals the ER-a and PR (Lapidus et al., 1996, 1998; Breast cancer can be divided into hormone responsive Ottaviano et al., 1994) and treating cells with and non-responsive tumors. Hormone (or estrogen) demethylating agents and/or HDAC inhibitors can responsive tumors can be controlled by the use of anti- restore ER-a expression (Yang et al., 2000, 2001b). estrogens. A hormone independent tumor has acquired There is a discrepancy concerning whether methylation capabilities which enables it to grow in the absence of of the ER-a CpG island is associated with the estrogen estrogen. receptor status in vivo (Falette et al., 1990; Hori et al., Estradiol is obligatory for normal development of 1999; Iwase et al., 1999; Kay et al., 1998; Lapidus et the female breast as well as for induction and al., 1998). So far it has not been possible to clarify progression of mammary carcinoma. Most of the whether DNA methylation of genes coding for steroid effects of estradiol are mediated by estrogen receptors receptors is an initiating or an early event in breast (ERs). An excellent review by Nilsson et al. (2001) has carcinogenesis or whether it occurs later. highlighted the roles of ERs in the breast: There are The functions of the recently discovered second ER, two distinct and functional ERs, called ERa and ERb the ERb, in the breast remain to be defined but from (Enmark et al., 1997; Green et al., 1986; Kuiper et al., what we have learned about its activities in in vitro 1996). During pubertal growth and during the estrous systems, this estrogen receptor may have a protective cycle, the majority of proliferating cells both in role in the breast (Gustafsson and Warner, 2000). terminal end buds and ducts are ERa negative (Clarke Studies in human and rodent mammary glands as well et al., 1997a,b; Zeps et al., 1998). Induction of the as in human breast cancer biopsies revealed that ERb progesterone receptor (PR) by estradiol does occur in is by far the more abundant of the two ERs. Studies of ERa containing cells, and this induction occurs at the mammary glands of ERb knockout mice revealed much lower plasma levels of estradiol than are required abnormal epithelial growth, overexpression of Ki67 for epithelial cell proliferation (Clarke et al., 1997b). and severe cystic gland disease as mice age (Gustafsson These observations have led to the concept (Wiesen et and Warner, 2000). In prostate cancer it has been al., 1999) of two distinct types of responses to estradiol shown that expression of ERb can be silenced by DNA in the breast: (1) an indirect action in the mammary methylation (Nojima et al., 2001). Our own prelimin- epithelium which occurs via ER-containing stromal ary, unpublished data has demonstrated that ERb cells and (2) a direct effect on ERa containing cells that methylation is a common event in breast cancer as occurs at low estradiol concentrations and results in well. induction of PR and differentiation of the epithelium. The gene for the PR encodes two isoforms, PR-A Approximately 60% of proliferating cells in the and PR-B, which differ in both their N-terminal mammary gland contain neither ERa nor ERb (Saji et sequences and biological activities. The PR-B transcript al., 2000). This observation raises the question of why is preferentially induced by ER while the PR-A is not. ERa containing breast epithelial cells do not divide but Since ligand-bound ER is a major transcriptional ER-a containing breast cancer cells divide in response activator of PR-B gene expression, the presence of to estradiol. This question is still unanswered but there PR is indicative of functional ER. PR gene methylation is evidence that one of the changes in breast cancer has been demonstrated in about 40% of PR-negative involves induction of other growth factor receptors breast tumors and several PR-negative breast cancer (Ethier, 1995; Lippman and Dickson, 1990). ERa may cell lines (Lapidus et al., 1996). suppress the expression of certain growth factor Signal transducers and activators of transcription receptors in normal mammary epithelium. Upon (STATs) are transcription factors activated in response estradiol withdrawal, as occurs during menopause, to cytokines and growth factors. Constitutively active growth factor receptors are expressed in ERa positive STAT3 has been shown to mediate oncogenic cells. Once this has occurred, ER-a can be activated transformation in cultured cells and induce tumor not only by estradiol but also by growth factor- formation in mice. A number of tumor-derived cell stimulated tyrosine kinases (El Ashry et al., 1997; lines as well as samples from human cancer have been Marsh et al., 1999; Pietras et al., 1995) and the normal reported to constitutively express active STAT3 protein regulation of cell growth is lost. This scenario does (Garcia et al., 2001). Inhibition of the STAT3 signaling neither explain the fact that prolonged exposure to pathway using the Janus Kinase-selective inhibitor, estrogens (early menarche, late menopause, usage of AG490, and a dominant negative STAT3 (STAT3b) hormone replacement therapy) is associated with significantly suppresses the growth of ovarian and

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5471 breast cancer cell lines harboring constitutively active RASSF1A is methylated and silenced in about 62% of STAT3. This also induced significant apoptosis in primary breast cancers irrespective of grade including ovarian and breast cancer cell lines expressing high grade I tumors, indicating that RASSF1A methylation levels of constitutively active STAT3 but had a less might be an early event during breast cancer pathogen- profound effect on normal cells lacking constitutively esis (Dammann et al., 2001). Interestingly, RASSF1A active STAT3 (Burke et al., 2001). It has been well methylation was found in 7.5% of normal tissues. documented that either GH or its downstream effector SYK is a protein tyrosine kinase that is widely IGF-I stimulates primate mammary epithelial prolif- expressed in hematopoietic cells. It has been shown eration in vivo (Ng et al., 1997) and that these factors that SYK is commonly expressed in normal human at least in part act by activating STAT3 (Gronowski et breast tissue, benign breast lesions and low-tumorigenic al., 1995). STAT3 activity is regulated by a negative breast cancer cell lines. However, SYK is expressed at feedback loop in which activated STAT3 stimulates low or undetectable levels in invasive breast carcinoma transcription of various genes including Suppressor of tissue and cell lines. Transfection of wild-type SYK Cytokine Signaling (SOCS) 1. SOCS1 suppresses the into a SYK-negative breast cancer cell line markedly JAK/STAT pathway by inhibiting JAK2 activity inhibited its tumor growth and metastasis formation in (Kishimoto and Kikutani, 2001). In mice, SOCS1 athymic mice. Conversely, overexpression of a kinase- plays a role as a negative regulator of prolactin deficient SYK in a SYK-positive breast cancer cell line signaling and suggesting that SOCS1 is required for significantly increased its tumorigenicity and growth. the prevention of lactation prior to parturition (Linde- Suppression of tumor growth by the reintroduction of man et al., 2001). Recently it was demonstrated that SYK appeared to be the result of aberrant mitosis and SOCS1 was aberrantly methylated and silenced in cytokinesis (Coopman et al., 2000). SYK 5’ CpG human primary hepatocellular carcinoma (Yoshikawa hypermethylation has been detected in six out of 20 et al., 2001). The restoration of SOCS1 suppressed breast cancer cell lines, and the aberrant methylation both growth rate and anchorage-independent growth status was strongly associated with loss of SYK gene of cells in which SOCS1 was methylation-silenced and expression and could be reverted upon treatment of JAK2 was constitutively activated. This growth cells with a methylation inhibitor. SYK was also suppression was caused by apoptosis and was hypermethylated in 32% of unselected breast tumors, reproduced by AG490, a specific chemical JAK2 whereas all of the matched neighboring normal breast inhibitor that reversed the constitutive phosphorylation tissues were methylation free (Yuan et al., 2001). of STAT3 in SOCS1 inactivated cells. Our own unpublished data demonstrate SOCS1 methylation in (3) Insensitivity to growth-inhibitory signals a considerable percentage of breast cancers. The Ras superfamily of GTPases act as important Transforming growth factor b (TGFb) acts in a regulatory switches to coordinate extracellular stimuli number of ways (most still elusive) to prevent the with activation of intracellular signaling pathways and phosphorylation that inactivates pRb. In this fashion, appropriate biological responses. In breast cancer, the TGFb blocks advances through the G1 phase of the signaling pathways involving these GTPases may be cell cycle (Hanahan and Weinberg, 2000). The pRb upregulated due to increased coupling to growth factor signaling circuit, as governed by TGFb, can be receptors or other tyrosine kinases commonly over- disrupted in a variety of different ways: Suppression expressed in this disease, increased expression of of type II TGFb receptor by overexpressing a regulators, the Ras protein itself, or downstream dominant-negative mutant II TGFb receptor enhanced effectors. Functional studies utilizing both in vitro tumorigenesis in the mammary gland in response to and in vivo models demonstrated that Ras signaling can carcinogen (Bottinger et al., 1997). Breast cancer cell regulate a variety of endpoints relevant to breast lines that express the estrogen receptor are refractory cancer progression, including anchorage dependent to TGF-b effects, whereas estrogen receptor-negative and independent growth, tumorigenesis, steroid sensi- cells are often TGFb sensitive (Arteaga et al., 1988). tivity and invasion (Malaney and Daly, 2001). Loss or undetectable expression of type II receptor has Although activated Ras proteins are usually associated been reported to contribute to TGFb resistance in with driving growth and transformation, they may also ER+ breast cancer cells (Kalkhoven et al., 1995; Sun et induce senescence, apoptosis, and terminal differentia- al., 1994). Treatment of ER positive breast cancer cells tion. The subversion of these anti-neoplastic effects with the DNA methyltransferase inhibitor 5-aza-CdR during Ras-dependent tumor development may be as leads to accumulation of type II TGFb receptor important as the acquisition of the pro-neoplastic transcripts and protein (Ammanamanchi et al., 1998). effects. It has been shown that RASSF1 (RAS Our unpublished data demonstrate type II TGFb association family 1 gene) C binds Ras in a GTP- receptor methylation in a subset of breast cancer dependent manner, both in vivo and directly in vitro. specimens. Moreover, activated Ras enhances, and dominant Using serial analysis of gene expression (SAGE) to negative Ras inhibits, the cell death induced by compare the gene expression profiles of normal and transient transfection of RASSF1C (Vos et al., 2000). ductal carcinoma in situ (DCIS) mammary epithelial Recent data support a role of RASSF1A in the cells, HIN-1 (high in normal-1) was identified (Krop et proapoptotic pathway (Khokhlatchev et al., 2002). al., 2001). HIN-1 expression is significantly down-

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5472 regulated in 94% of human breast carcinomas and in transport-mediated MTX resistance in breast cancer cell 95% of preinvasive lesions, such as ductal and lobular lines (Moscow et al., 1995). Transfection of MDA-MB- carcinoma in situ. This decrease in HIN-1 expression is 231 breast cancer cells with RFC cDNA restored accompanied by hypermethylation of its promoter in methotrexate uptake and increased methotrexate sensi- the majority of breast cancer cell lines (490%) and tivity by approximately 50-fold. A CpG island in the primary tumors (74%). HIN-1 is a putative cytokine promoter region of RFC was found to be methylated in with no significant homology to known proteins. MDA-MB-231 cells, but was unmethylated in RFC Reintroduction of HIN-1 into breast cancer cells expressing, methotrexate-sensitive MCF-7 breast cancer inhibits cell growth. These results indicate that HIN-1 cells. Treatment of MDA-MB-231 cells with 5-aza-CdR is a candidate tumor suppressor gene that is inactivated restored RFC expression (Worm et al., 2001). at high frequency in the earliest stages of breast tumorigenesis (Krop et al., 2001). (4) Evasion of programmed cell death Using differential display PCR, a gene called NOEY2 (ARHI) has been identified (Yu et al., 1999). p53 has emerged as one of the leaders among the This gene has high homology to ras and rap and is multitude of players involved in apoptosis due to its expressed consistently in normal ovarian and breast ability to selective induce apoptosis in stressed or epithelial cells but not in ovarian and breast cancers abnormal cells, thereby protecting the organism from (Yu et al., 1999). Expression of NOEY2 through cancer development (reviewed in Vousden, 2000). transfection suppressed clonogenic growth of breast Therefore p53 function is thought to be controlled by and ovarian cancer cells. Growth suppression was several mechanisms, one of the most effective being associated with down-regulation of the cyclin D1 regulation of protein stability. MDM2 is an E3 ligase promoter activity and induction of p21WAF1/CIP1.In that targets both p53 and itself for ubiquitination. an effort to identify mechanisms leading to NOEY2 MDM2 is a transcriptional target of p53, creating a silencing in cancer it was demonstrate that this gene is negative feedback loop where p53 activates expression expressed monoallelically and is imprinted maternally. of MDM2, which keeps p53 levels low during normal Loss of heterozygosity of the gene was detected in 41% conditions. DNA damage, oncogene activation, telo- of ovarian and breast cancers. In most of the cancer mere erosion and hypoxia inhibit MDM2 and thereby samples with loss of heterozygosity, the nonimprinted stabilize p53. Oncogenes like myc and ras can induce functional allele was deleted. Southern blot analysis stabilization of p53 by enlisting the activity of ARF, a demonstrated hypermethylation in two out of eight protein that functions by binding directly to MDM2, breast cancer cell lines (Yu et al., 1999). inhibiting the ubiquitination of p53 and allowing Another gene which is expressed in normal mammary accumulation of p53 in the nucleus. ARF expression epithelial cells but dramatically decreased in breast can be directly regulated by transcription factors cancer cell lines is the normal epithelial cell-specific-1 DMP1 and E2F1 or by DAP kinase, or downregulated (NES1)/kallikrein 10 (Dhar et al., 2001). Although the by factors such as Twist and Bmi1. Beside control of function of this gene has not yet been clarified, stable p53 degradation, HOXA5 regulates synthesis of p53 by expression of NES1 in the NES1-negative MDA-MB- binding and activating consensus HOX binding sites in 231 breast cancer cell line suppressed the oncogenicity as the p53 promoter (Raman et al., 2000). Both major revealed by inhibition of the anchorage-independent apoptotic pathways – via death receptor signaling or growth and tumor formation in nude mice (Goyal et al., mitochondrial perturbations, including cytochrome c 1998). Recently a strong correlation between exon 3 release, activation of APAF-1 and subsequently hypermethylation and loss of NES1 mRNA expression activating caspases – can be mediated by p53. in a panel of breast cancer cell lines and in primary Inactivation of p53 by mutations is not that common tumors has been described (Li et al., 2001b). Treatment in human breast cancer and is seen in about 20% of of NES1-nonexpressing cells with a demethylating agent tumors (Pharoah et al., 1999). p53 mRNA levels are led to reexpression of NES1, suggesting an important 5 – 10-fold lower in breast cancer cell lines than in role of hypermethylation in the loss of NES1 expression normal breast epithelium (Raman et al., 2000). Breast (Li et al., 2001b). cancer cell lines and breast cancer specimens display a Beside physiological growth inhibitors, a variety of coordinate loss of p53 and HOXA5 mRNA and pharmacological substances also mediate growth protein expression. The HOXA5 promoter region was suppression. For instance methotrexate (MTX) is part methylated in 16 out of 20 p53-negative breast tumor of combination chemotherapy regimens used in the specimens but unmethylated in human mammary treatment of various malignancies, including acute epithelial cells with a finite lifespan as well as in lymphoblastic leukemia, non-Hodgkin’s lymphoma, immortalized HMECs. DNMT inhibitors restored osteosarcoma, and breast cancer (Bertino, 1993). The expression of HOXA5 (Raman et al., 2000). major route for cellular uptake of MTX involves the Some genes involved in p53 protein stability have reduced folate carrier (RFC), a bidirectional anion also been shown to be regulated by DNA-methylation. transporter with high affinity for reduced folates and DAP kinase has been reported to be methylated in 7% antifolates but low affinity for folic acid (Sirotnak and of breast cancers (Esteller et al., 2001). Twist was Tolner, 1999). Transport studies in in vitro models, have shown to be methylated in 21 out of 50 invasive and shown that down-regulation of RFC activity results in four out of 14 in situ breast cancers (Evron et al.,

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5473 2001b). Expression of downstream members of the p53 main mediators of angiogenesis. Each binds to pathway like caspase-8 (Fulda et al., 2001) or Apaf1 transmembrane tyrosine kinase receptors displayed by (Soengas et al., 2001) have been shown to be directly or endothelial cells. A counterbalancing between positive indirectly altered by DNA methylation. and negative signals encourage or block angiogenesis. TMS1 (target of methylation-induced silencing) THBS1 (Thrombospondin-1) binds to CD36, a belongs to a family of apoptotic signaling molecules transmembrane receptor on endothelial cells coupled that contain a CARD (caspase recruitment domain). to intracellular Src-like tyrosine kinases (Bull et al., TMS1 is likely to function as an adaptor protein, 1994) and is a prototypical angiogenesis inhibitor. acting in the initiation phase of an apoptotic pathway THBS1 has been shown to be positively regulated by by coupling death receptors at the cell surface, or the p53 tumor suppressor protein and consequently, intrinsic death signals, to the activation of the caspase loss of p53 function (as outlined above) leads to cascade. THBS1 decrease, liberating endothelial cells from TMS1 is aberrantly methylated and silenced in inhibitory effects (Dameron et al., 1994). To ascertain human breast cancer cells and aberrant methylation the participation of the THBS1 in tumor progression, of TMS1 was evident in 40% (11 out of 27) of primary mammary tumor-prone mice that either lack or breast tumors analysed. Ectopic expression of TMS1 specifically overexpress THBS1 in the mammary gland induced apoptosis in human embryonic kidney cells have been generated (Rodriguez-Manzaneque et al., and inhibited the survival of human breast cancer cells 2001). Tumor burden and vasculature were signifi- (Conway et al., 2000). cantly increased in THBS1-deficient animals, and ZAC, a new zinc finger protein was named according capillaries within the tumor appeared distended and to its functional properties, namely induction of sinusoidal. In contrast, THBS1 overexpressors showed apoptosis and control of cell cycle progression. ZAC delayed tumor growth or lacked frank tumor develop- is expressed in normal mammary gland and maps to ment. Absence of THBS1 resulted in an increased 6q24-q25, a recognized hot spot for deletion on 6q in association of VEGF with its receptor and higher levels breast cancer. A survey of eight breast cancer cell lines of active matrix metalloproteinase-9 (MMP9), a showed either a deeply reduced or complete loss of molecule previously shown to facilitate both angiogen- ZAC expression. Treatment of three of these cell lines esis and tumor invasion. Enzymatic activation of with the methylation-interfering agent 5-aza-CdR proMMP9 was suppressed by THBS1 in vitro (Rodri- induced ZAC re-expression. In addition, Northern blot guez-Manzaneque et al., 2001). Besides regulation due and RNase protection assay analysis of ZAC expres- to p53, de novo methylation may serve as a potential sion in 23 unselected primary breast tumors indicated a way to inactivate THBS1 expression in human reduced expression in several samples (Bilanges et al., neoplasms (Li et al., 1999). 1999). p73 inhibits expression of VEGF (Salimath et al, Another gene, Glypican 3 (GPC3), a membrane- 2000) and DNA methylation causes its suppression in bound heparan sulfate proteoglycan that is mutated in lymphoid leukemia cells (Liu et al., 2001). In breast the Simpson-Golabi-Behmel syndrome, was also shown cancer p73 methylation has not been studied properly. to induce apoptosis in breast cancer (Gonzalez et al., Maspin, a unique member of the serpin family, is a 1998). GPC3 expression is also silenced in human secreted protein whose downregulation is associated breast cancer, and this silencing is due, at least in part, with the development of breast cancers (Zou et al., to hypermethylation of the GPC3 promoter. Although 1994). Maspin is an effective inhibitor of angiogenesis. the exact function of the gene is not yet known, it was It was shown to act directly on cultured endothelial demonstrated that ectopic expression of GPC3 inhib- cells to stop their migration towards bFGF and VEGF ited growth in eight out of 10 breast cancer cell lines and to limit mitogenesis and tube formation. It blocked (Xiang et al., 2001). neovascularization in the rat cornea pocket model FHIT, a diadenosine hydrolase may be involved in (Zhang et al., 2000). HMECs expressed maspin mRNA growth control pathways of the cell. Studies on and displayed a completely non-methylated maspin protein-protein interactions, cell lines, including gene promoter with an open chromatin structure. In tumorigenicity tests, and knockout mice suggest that contrast, seven of nine breast cancer cell lines had no the FHIT protein is involved in cell proliferation and detectable maspin expression and six of these seven apoptosis, and might act as a tumor suppressor. In maspin-negative breast cancer cell lines also displayed several different cancers, including breast cancer, an aberrant pattern of cytosine methylation of the alterations in the FHIT gene have been detected in maspin promoter. Moreover, maspin gene expression high frequency (Ingvarsson, 2001). 5’ CpG island was re-activated in MCF-7 cells by treatment with a methylation of the FHIT gene has been detected DNA demethylating agent (Domann et al., 2000). frequently in breast cancer and correlated with loss of gene expression (Zochbauer-Muller et al., 2001). (6) Tissue invasion and metastasis Invasion and metastasis involves changes in the (5) Sustained angiogenesis physical coupling of cells to their microenvironment VEGF (vascular endothelial growth factor) and FGF1/ and activation of extracellular proteases. Epithelial 2 (acidic and basic fibroblast growth factors) are the cells maintain contact with their neighbors through

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5474 adherens junctions. Cadherins traverse the membrane, Demethylation coupled with histone deacetylase associating with cadherins on adjacent cells. E- inhibition resulted in reactivated expression of the Cadherin (E-Cad), suppresses tumor cell invasion and prostasin mRNA in MDA-MB-231 and MDA-MB- metastasis in experimental tumor models. Decreased E- 435s cells (Chen and Chai, 2002). Tissue inhibitor of Cad expression is common in poorly differentiated, metalloproteinase-3 (TIMP-3) antagonizes matrix advanced-stage carcinomas. The 5’ CpG island of E- metalloproteinase activity and can suppress tumor Cad is densely methylated in E-Cad-negative breast growth, angiogenesis, invasion, and metastasis (Uria carcinoma cell lines and primary breast carcinoma et al., 1994). The TIMP-3 promoter is methylated in tissue but is unmethylated in normal breast tissue. The about 30% of human breast cancer cell lines and treatment with 5-aza-CdR, partially restored E-Cad primary breast tumors and TIMP-3 expression was RNA and protein levels in E-Cad-negative breast restored after 5-aza-CdR-mediated demethylation of carcinoma cell lines (Graff et al., 2000). the TIMP-3 proximal promoter region (Bachman et al., Expression of a second cadherin, H-cadherin 1999). (CDH13), was shown to be significantly reduced in The nm23 gene family exhibits metastasis suppressor human breast carcinoma cell lines and breast cancer activity in breast cancer in vivo (Leone et al., 1993) and specimens. Introduction of CDH13 cDNA markedly low nm23-H1 expression was a significant predictor of diminished tumor cell growth and resulted in a poor survival in univariate and multivariate analyses significant change from invasive morphology to a normal (Heimann et al., 1998). Two CpG islands are present in cell-like morphology in the Matrigel outgrowth assay the nm23-H1 promoter and bisulfite sequencing of (Lee, 1996). Methylation of CDH13 has frequently been these CpG islands in a panel of cell lines and in 20 observed in primary breast tumors (18 out of 55, 33%) infiltrating ductal carcinomas revealed that one island and cell lines (seven out of 20, 35%). Gene expression exhibited infrequent differential methylation. was restored in methylated cell lines tested after Treatment with 5-aza-CdR, increased the nm23-H1 treatment with 5-aza-CdR (Toyooka et al., 2001). expression of five of 11 human breast carcinoma cell On the cytoplasmic face, b-catenin connects to the lines in vitro, including three of three metastatically cadherin tail and associates with a-catenin, which in competent lines. Increased nm23-H1 expression was turn binds to actin. Accumulation of b-Catenin occurs accompanied by a reduction in motility in vitro, with in the absence of APC (adenomatous polyposis coli) minimal effect on proliferation (Hartsough et al., and might lead to activation of c-myc and cyclin D1 2001). (reviewed in Fearnhead et al., 2001). In addition, APC Overexpression of Breast Cancer Specific Gene 1 is involved in signal transduction, stabilization of the (BCSG1) in breast cancer cells leads to a significant cytoskeleton and regulation of cell cycle and apoptosis increase in motility and invasiveness in vitro and a (Fearnhead et al., 2001). The APC promoter 1A was profound augmentation of metastasis in vivo (Jia et al., methylated in 34 out of 77 breast cancer tumors and 1999). BCSG1 is not expressed in normal breast tissues cell lines (44%). There was complete concordance but highly expressed in advanced infiltrating breast between promoter methylation and silencing of its carcinomas. A recent publication demonstrated that transcript in cell lines. Demethylation with 5-aza-CdR BCSG1 might be regulated by methylation and that treatment restored transcript 1A expression in all eight treatment with demethylating agents activated BCSG1 methylated cell lines tested (Virmani et al., 2001). Jin et transcription (Lu et al., 2001). It has not yet been al. (2001) found methylation of the APC promoter demonstrated whether treatment with demethylating CpG island in 18 out of 50 (36%) primary breast agents causes more aggressive cancers. cancers and in none of 21 non-cancerous breast tissue samples. Prostasin is a serine protease, which decreases Timing of DNA methylation in breast carcinogenesis invasiveness in vitro (Chen et al., 2001). Prostasin mRNA and protein were shown to be expressed in We propose that DNA methylation and silencing of normal HMECs, the poorly invasive breast carcinoma certain genes (Figure 1) might support clonal selection cell line MCF-7 and the nonmetastatic breast carcino- due to growth advantages. However, there are certain ma cell line MDA-MB-453, but absent in highly genes which do not support this theory in that there is invasive and metastatic breast carcinoma cell lines no obvious selection advantage to their inactivation. MDA-MB-231 and MDA-MB-435s. Enforced reex- p16 inactivation is mandatory for immortalization of pression of prostasin in MDA-MB-231 and MDA-MB- HMECs (Kiyono et al., 1998) and p16 silencing in 435s reduced the in vitro invasiveness of either cell line these cells is mediated by DNA methylation (Brenner by 50%. Examination of the prostasin gene promoter et al., 1998; Foster et al., 1998; Huschtscha et al., and first exon revealed a GC-enriched region that 1998). But this specific p16 promoter region is very contains transcription regulatory elements. The promo- rarely methylated in breast cancer (Lehmann et al., ter and exon 1 region of the prostasin gene was 2002). Unexpectedly, high p16INK4A mRNA expression investigated for DNA methylation in HMEC and the was associated with high tumor grade (P=0.006), 54 carcinoma cell lines. The results revealed a methylation axillary lymph node involvement (P=0.004), ER pattern that correlates with prostasin expression in negativity (P=0.0001), and increased risk of relapse these cells. (P=0.006) (Hui et al., 2000). These conflicting results

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5475 imply that p16 methylation might be needed to DNA methylation is an important mechanism overcome senescence as an initiating event in breast contributing to epigenetic mechanisms (Bird, 2002; carcinogenesis. Once immortalization has been reached, Jones and Laird, 1999). Epigenetics can explain these cells may establish a microenvironment enabling mitotically and/or meiotically heritable changes in gene other cells – without p16 methylation – to progress function that cannot be explained by changes in DNA. towards cancer. Passive demethylation later on in The heritability of methylation states and the second- breast carcinogenesis by inhibiting DNMT1 binding ary nature of the decision to invite or exclude to the p16 promoter may serve as an alternative methylation support the idea that DNA methylation explanation for this phenomenon. is adapted for a specific cellular memory function in Expression of genes like COX-2 (Ristimaki et al., development (Bird, 2002). 2002), BCSG1 (Lu et al., 2001) or Gelsolin (Thor et al., By studying the tyrosine aminotransferase (Tat) gene 2001) are associated with poor prognostic parameters in fetal rat hepatocytes as well as rat hepatoma cells, a in breast cancer but on the other hand are frequently mechanism for memorizing hormone action has been methylated (unpublished data; Lu et al., 2001; described (Thomassin et al., 2001). Glucocorticoids Mielnicki et al., 1999). These genes probably become regulate DNA demethylation within a key enhancer of methylated ‘by chance’ during carcinogenesis by the rat liver-specific Tat gene and additional DNA- mechanisms (probably gene silencing) responsible for associated factors are subsequently recruited. Demethy- the methylation of the majority of the other genes. lation persists after the wave of expression has Carcinogenesis is a highly complex process in which subsided, and reinduction of the silent gene by further interactions between epithelial and stromal cells hormone treatment is significantly stronger as a result. (Krtolica et al., 2001), paracrine, endocrine, as well This system provides a model for a DNA methylation- as autocrine mechanisms are all involved in this mediated memory of the first hormone induction process (Elenbaas and Weinberg, 2001; Krtolica et (Kress et al., 2001). al., 2001; Moinfar et al, 2000). Most of the data supporting the ‘memory-function’ of DNA methylation pattern stem from embryogenesis. So far there is indirect evidence that this mechanism holds true for carcinogenesis. As outlined above there DNA methylation, memory and breast carcinogenesis are genes according to certain pathways, which are switched on during senescence of HMECs (e.g. p16, Several lines of evidence suggest that DNA methyla- RAR-b2, STAT-1 and other interferon regulated genes, tion does not intervene to silence active promoters, but mac25, etc.). DNA methylation and permanent silen- effects genes that are already silent (reviewed in Bird, cing is frequently observed within these genes after 2002). The signal for this putative gene silencing- HMECs have become immortal and in breast cancer related de novo methylation is unknown, but the cell lines and in breast cancer specimens. possibility that chromatin states recruit the DNA On the other hand, evidence is now accumulating methylation machinery to a particular DNA sequence that some of these methylation changes may initiate in is attractive (Selker, 1990). The acetylation and subpopulations of normal cells as a function of age and methylation state of nucleosomal histones is tightly progressively increase during carcinogenesis. Age- correlated with transcriptional activity (Jenuwein and related methylation appears to be widespread and is Allis, 2001). In Neurospora, an intimate link between one of the earliest changes marking the risk for histone methylation and DNA methylation has been neoplasia (Ahuja et al., 1998). Age-related methylation described (Tamaru and Selker, 2001). Methylation of involves at least 50% of the genes which are already silenced genes may occur to silence them hypermethylated in colon cancer, and it was proposed irrevocably. X inactivation and silencing of transpo- that such age-related methylation may partly account sable elements are two examples (Bird, 2002; Heard et for the fact that most cancers occur as a function of al., 2001). On the other hand, transcriptionally active increased age (Ahuja and Issa, 2000). Although DNA chromatin might protect against DNA methylation at methylation has not been studied in terms of age in CpG islands. Promoter activity early in development non-neoplastic breast tissue, an age related increase in may create a methylation-free CpG island; in other ERa gene methylation was demonstrated in the right words, unmethylated CpG islands might be footprints atrium of the heart (Post et al., 1999) as well as in the of embryonic promoter activity (Bird, 2002). Indeed, a colon (Ahuja and Issa, 2000). DNA methylation was CpG-island promoter whose product RNA is not also used as a tag to characterize stem cells that expected to occur in the early embryo (a-globin, 68k maintain human colon crypts (Yatabe et al., 2001). neurofilament) is nevertheless expressed, whereas tran- Thereby methylation was shown to increase with aging scripts from a CpG-deficient promoter (b-globin, opsin, but varied between crypts and was mosaic within single casein) are not detected (Daniels et al., 1997; Macleod crypts. et al., 1998). Recent studies in human prostate cancer Premalignant breast lesions are thought to arise cells support this hypothesis: a combination of prior primarily from stem cells in normal terminal duct gene silencing and random ‘seeds’ of methylation lobular units (TDLUs) (Rudland, 1993). In normal trigger hypermethylation of the GSTP1 gene (Song et TDLUs a low rate of proliferation has been reported al., 2002). averaging only about 2% (Ferguson and Anderson,

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5476 1981; Going et al, 1988; Kamel et al., 1989; Longacre evidenced by increased expression of mdm2 and p21 and Bartow, 1986; Meyer, 1977). In premenopausal (CIP1/WAF1). Importantly, exposure to perphenazine women, the rate fluctuates and is approximately twice (a compound that induces mammary gland differentia- as high in the luteal as in the follicular phase of the tion but does not confer protection) does not induce menstrual cycle (Ferguson and Anderson, 1981; Going p53 expression, indicating that p53 is not a marker of et al, 1988; Longacre and Bartow, 1986; Meyer, 1977), differentiation. The proliferative block and induction of whereas in postmenopausal women, the proliferation p53 are operative in both rats and mice, which rate is somewhat lower and relatively stable (Lelle et supports the generality of the proposed hypothesis al., 1987; Meyer and Connor, 1982). (Sivaraman et al., 2001). Using an analogous approach, We hypothesize that DNA methylation contributes RbAp46 has also been shown to be persistently to silence specific genes in some stem cells in TDLUs upregulated after E/P-treatment (Ginger et al., 2001). due to their lack of expression (Bird, 2002). These This protein interacts with BRCA1, interacts directly methylation changes may not be detectable in young with histones H3 and H4 and is a component of individuals due to the large background of unmethy- multisubunit complexes that are involved in histone lated CpG islands, but may accumulate as a function deacetylation, histone acetylation, nucleosome disrup- of age. In vitro some HMECs – after having tion and nucleosome assembly (reviewed in Ginger et methylated and therefore irrevocably suppressed p16 al., 2001). Using another model system, Boulanger and – escape senescence (Foster et al., 1998), most likely Smith (2001) were able to demonstrate that premature due to a selection benefit. stem cell senescence may also reduce mammary cancer Methylation of various genes, not needed in risk. quiescent breast stem cells may also accumulate in These results may suggest that during development vivo. Stimuli like exposure to endogenous and and differentiation, early inductive processes that exogenous estrogens might induce proliferation of influence cell fate at a later stage may leave marks at these cells. Due to methylation and permanent distinct gene loci that are maintained through several silencing of genes needed for physiologic turnover rounds of mitosis. DNA methylation is part of this and senescence, these cells are prone to become epigenetic memory that restricts or permits differential immortal. Breast cancer might then be a manifestation expression of genes in descendant cells. This can result of the above mentioned six essential alterations in cell in establishment of a cell-type-specific DNA methyla- physiology – in part mediated by DNA methylation. tion pattern that may restrict the ability to transcribe genes needed for apoptosis, senescence and other physiological processes in some cells. These cells may Prevention of breast cancer be predestined to immortalization and de-differentiation, and become precursors of neoplastic cells. In Prevention of breast cancer can be achieved by a better other words, events that switch on pathways (and understanding of the etiological factors contributing to therefore prevent methylation of genes involved) in the development of the disease. There is strong normal cells needed for physiological turnover, might evidence that women who experience a full-term prevent these pathways from becoming irrevocably pregnancy early in their lives have a significantly switched off by methylation. reduced risk for developing breast cancer (Chodosh et Beside early age at first pregnancy, there are two al., 1999; MacMahon et al., 1970). This was other pregnancy-associated breast cancer risk reducing recapitulated in rat models that demonstrated that factors. Evidence suggests that immune, hormonal, or early full-term pregnancy confers resistance to chemical genetic mechanisms that induce hypertension or carcinogen-induced mammary tumorigenesis. This preeclampsia during pregnancy reduce the risk of protection can be also mimicked with the hormones breast cancer both in the mother (Polednak and estrogen (E) and progesterone (P) or with human Janerich, 1983; Thompson et al., 1989; Troisi et al., chorionic gonadotropin (Chodosh et al., 1999; Sivara- 1998) and in the daughter (Ekbom et al., 1992, 1997). man et al., 2001; and references therein) given either Secondly, independent reductions in the rate of before or after carcinogen challenge to induce a subsequent breast cancer were associated with increases refractory state. An excellent study by Sivaraman et in blood pressure between the second and the third al. (2001), described p53 to be a potential mediator of trimesters, infarctions in the maternal floor of the pregnancy- and hormone-induced resistance to placenta, low placental weight, and small placental mammary carcinogenesis. A striking increase in p53 diameter (Cohn et al., 2001). and down-stream effectors in the rat mammary gland Decreased perfusion of the placenta which is a was observed after E/P-treatment compared with prerequisite for development of preeclampsia and is untreated age matched virgin rats. These changes in associated with small placentas and maternal floor turn dictate the proliferative response to carcinogen infarction of the placenta, causes generation of reactive challenge and include a block in carcinogen-induced oxygen species. Markers of oxidative stress have been increase in mammary epithelial cell proliferation and detected in the blood of women with preeclampsia for an increased and sustained expression of nuclear p53 in over 40 years (Dekker and Sibai, 2001; Roberts and the hormone-treated mammary gland. This hormone- Cooper, 2001; and references therein). Reactive oxygen induced nuclear p53 is transcriptionally active as species have been demonstrated to induce p53 and

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5477 senescence (Chen et al., 1998; Chen, 2000; Kinscherf et breast epithelial stem cells for a short period of time al., 1998; Krtolica et al., 2001; Sugano et al., 1995). (analogous to pregnancy at early age or pregnancies It is tempting to speculate that in early life short with placental pathologies) might prevent cells from term exposure to situations causing a challenge of p53 permanently switching off genes which would be dependent pathways or other senescence-inducing needed later in life to resist carcinogenic triggers. pathways might prevent genes involved, from becoming Classes of agents currently undergoing evaluation in irrevocably silenced by methylation. In other words clinical prevention trials or those for which testing is HMECs might ‘recall’ the challenge their predecessors planned in the near future include new selective have experienced decades before and are therefore able estrogen receptor modulators, aromatase inactivators/ to adequately respond to a carcinogenic stimulus. inhibitors, gonadotrophin-releasing hormone agonists, monoterpenes, isoflavones, retinoids, rexinoids, vitamin D derivatives, and inhibitors of tyrosine kinase, DNA methylation as a tool for early detection of breast cyclooxygenase-2, and polyamine synthesis (reviewed cancer in Fabian, 2001). Most of these substances have shown to be directly or indirectly involved in the above It is questionable whether screening for breast cancer mentioned six novel capabilities needed for develop- by mammography reduces mortality (Miettinen et al., ment of breast cancer. Whether they will have an 2002; Olsen and Gotzsche, 2001). An alternative DNA- impact on ‘preventing DNA methylation’ remains to be based approach for early detection of breast cancer elucidated. might be promising since DNA extracted from Once a gene has been silenced by histone deacetyla- patient’s plasma, serum or other body fluids could be tion (HDACs), the use of HDAC inhibitors, such as easily amplified by PCR technology and is therefore suberoylanilide hydroxamic acid (SAHA) might be a potentially more sensitive than conventional tests. promising way to reinduce expression. In breast cancer RNA is not as useful as a detection marker because cell lines SAHA causes the inhibition of proliferation, of its inherent instability. Detection of promoter CpG accumulation of cells in a dose-dependent manner in island hypermethylation offers several advantages G(1) then G(2)-M phase of the cell cycle, and compared to other DNA alterations in cancer induction of differentiation (Munster et al., 2001). (reviewed in Widschwendter and Jones, 2002). Methy- SAHA and other HDAC inhibitors are currently in lated DNA can be detected with a very high degree of Phase I clinical trials. specificity, even in the presence of a vast excess of Finally, genes that have already been methylated can unmethylated DNA. MethyLight technology for be demethylated and reexpressed as outlined above. instance can detect a single hypermethylated allele Demethylating chemopreventive strategies have been against a background of 10 000 unmethylated alleles shown in vivo to be effective (Laird et al., 1995; (Eads et al., 2000). Methylated DNA from patients Ramchandani et al., 1997). 5-aza-CdR has already with manifest breast cancer has been detected in blood been tested in a phase II study in elderly patients with (Silva et al., 1999a,b) as well as in ductal lavage fluid high-risk myelodysplastic syndrome (Wijermans et al., (Evron et al., 2001b). Detection of DNA methylation 2000). However systemic treatment with this drug in patients’ blood might prove to be useful as a commonly causes myelosuppression. It remains to be predictive marker at the moment of primary diagnosis tested whether intraductal, retrograde application of or as a marker for early detection of relapse of disease. various substances (possibly combined with systemic As the aim is to detect cancer early or even non- treatment) might be another chemopreventive invasive or premalignant lesions, fluid collected from approach. the breasts may be a very promising substrate. It will DNA-methylation is a very effective strategy to take a long time to realize the aim to reduce the burden irrevocably switching off certain genes. Cells which of diseases within the population due to screening have accumulated DNA-methylation of promoter based on these first results. As we have reached Phase regions of genes which would be needed to adequately II (according to suggestions by Sullivan et al. (2001)) respond to carcinogenic stimuli, are prone to become there are many questions to answer before a retro- tumor cells. Future research might prove whether this spective longitudinal study (Phase III) will be knowledge translates into strategies for early detection conducted: which genes should be analysed? Which or even prevention of breast cancer. method should be used? For instance, a gene which is methylated with high prevalence in aggressive cancer might be unmethylated in a non-invasive cancer and vice versa. Acknowledgements We thank Mihaela Velicescu and Dan Weisenberger for critical reading of this manuscript. This work was DNA methylation as a target for new preventive and supported by a Grant from the ‘Fonds zur Fo¨ rderung der therapeutic approaches wissenschaftlichen Forschung’ project #J2024 and from ‘Jubila¨ umsfonds der O¨ sterreichischen Nationalbank’ and The accumulated knowledge offers promising new NIH Grants 1R01 CA 83867-01 and 1R01 CA 82422-01 preventive and therapeutic strategies. Challenging from the National Cancer Institute.

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5478 References

Abbondanza C, Medici N, Nigro V, Rossi V, Gallo L, Piluso Chodosh LA, D’Cruz CM, Gardner HP, Ha SI, Marquis ST, G, Belsito A, Roscigno A, Bontempo P, Puca AA, Rajan JV, Stairs DB, Wang JY and Wang M. (1999). Molinari AM, Moncharmont B and Puca GA. (2000). Cancer Res., 59, 1765 – 1771s. Proc. Natl. Acad. Sci. USA, 97, 3130 – 3135. Clarke RB, Howell A and Anderson E. (1997a). Breast Ahuja N and Issa JP. (2000). Histol. Histopathol., 15, 835 – Cancer Res. Treat., 45, 121 – 133. 842. Clarke RB, Howell A, Potten CS and Anderson E. (1997b). Ahuja N, Li Q, Mohan AL, Baylin SB and Issa JP. (1998). Cancer Res., 57, 4987 – 4991. Cancer Res., 58, 5489 – 5494. Cohn BA, Cirillo PM, Christianson RE, van den Berg BJ and Alani RM, Young AZ and Shifﬂett CB. (2001). Proc. Natl. Siiteri PK. (2001). J. Natl. Cancer Inst., 93, 1133 – 1140. Acad. Sci. USA, 98, 7812 – 7816. Collaborative Group on Hormonal Factors in Breast Altucci L and Gronemeyer H. (2001). Nat. Rev. Cancer, 11, Cancer. (2001). Lancet, 358, 1389 – 1399. 181 – 193. Conway KE, McConnell BB, Bowring CE, Donald CD, Ammanamanchi S, Kim SJ, Sun LZ and Brattain MG. Warren ST and Vertino PM. (2000). Cancer Res., 60, (1998). J. Biol. Chem., 273, 16527 – 16534. 6236 – 6242. Arapshian A, Kuppumbatti YS and Lopez R. (2000). Coopman PJ, Do MT, Barth M, Bowden ET, Hayes AJ, Oncogene, 19, 4066 – 4070. Basyuk E, Blancato JK, Vezza PR, McLeskey SW, Armenante F, Merola M, Furia A and Palmieri M. (1999). Mangeat PH and Mueller SC. (2000). Nature, 406, 742 – Biochem. Biophys. Res. Commun., 258, 644 – 647. 747. Arnold JM, Cummings M, Purdie D and Chenevix-Trench Dameron KM, Volpert OV, Tainsky MA and Bouck N. G. (2001). Br. J. Cancer, 85, 1351 – 1358. (1994). Science, 265, 1582 – 1584. Arteaga CL, Tandon AK, Von Hoff DD and Osborne CK. Dammann R, Yang G and Pfeifer GP. (2001). Cancer Res., (1988). Cancer Res., 48, 3898 – 3904. 61, 3105 – 3109. BachmanKE,HermanJG,CornPG,MerloA,CostelloJF, Daniels R, Lowell S, Bolton V and Monk M. (1997). Hum. Cavenee WK, Baylin SB and Graff JR. (1999). Cancer Reprod., 12, 2251 – 2256. Res., 59, 798 – 802. Dekker G and Sibai B. (2001). Lancet, 357, 209 – 215. Basolo F, Conaldi PG, Fiore L, Calvo S and Toniolo A. Devereux TR, Horikawa I, Anna CH, Annab LA, Afshari (1993). Int. J. Cancer, 55, 926 – 930. CA and Barrett JC. (1999). Cancer Res., 59, 6087 – 6090. Baylin SB and Herman JG. (2000). Trends Genet., 16, 168 – Dhar S, Bhargava R, Yunes M, Li B, Goyal J, Naber SP, 174. Wazer DE and Band V. (2001). Clin. Cancer Res., 7, 3393 – Bender CM, Pao MM and Jones PA. (1998). Cancer Res., 58, 3398. 95 – 101. Di Croce L, Raker VA, Corsaro M, Fazi F, Fanelli M, Bertino JR. (1993). J. Clin. Oncol., 11, 5 – 14. FarettaM,FuksF,LoCF,KouzaridesT,NerviC, Bhattacharyya N, Dey A, Minucci S, Zimmer A, John S, Minucci S and Pelicci PG. (2002). Science, 295, 1079 – Hager G and Ozato K. (1997). Mol. Cell Biol., 17, 6481 – 1082. 6490. Domann FE, Rice JC, Hendrix MJ and Futscher BW. (2000). Bilanges B, Varrault A, Basyuk E, Rodriguez C, Mazumdar Int. J. Cancer, 85, 805 – 810. A, Pantaloni C, Bockaert J, Theillet C, Spengler D and Du Y, Carling T, Fang W, Piao Z, Sheu JC and Huang S. Journot L. (1999). Oncogene, 18, 3979 – 3988. (2001). Cancer Res., 61, 8094 – 8099. Bird A. (2002). Genes Dev., 16, 6 – 21. EadsCA,DanenbergKD,KawakamiK,SaltzLB,BlakeC, Bottinger EP, Jakubczak JL, Haines DC, Bagnall K and Shibata D, Danenberg PV and Laird PW. (2000). Nucleic Wakeﬁeld LM. (1997). Cancer Res., 57, 5564 – 5570. Acids Res., 28, E32. Boulanger CA and Smith GH. (2001). Oncogene, 20, 2264 – Eads CA, Danenberg KD, Kawakami K, Saltz LB, 2272. Danenberg PV and Laird PW. (1999). Cancer Res., 59, Bovenzi V, Le NL, Cote S, Sinnett D, Momparler LF and 2302 – 2306. Momparler RL. (1999). Anticancer Drugs, 10, 471 – 476. EifelP,AxelsonJA,CostaJ,CrowleyJ,CurranJrJW, Brenner AJ, Stampfer MR and Aldaz CM. (1998). Oncogene, DeshlerA,FultonS,HendricksCB,KemenyM, 17, 199 – 205. Kornblith AB, Louis TA, Markman M, Mayer R and Bull HA, Brickell PM and Dowd PM. (1994). FEBS Lett., Roter D. (2001). J. Natl. Cancer Inst., 93, 979 – 989. 351, 41 – 44. Ekbom A, Hsieh CC, Lipworth L, Adami HQ and Burger AM, Zhang X, Li H, Ostrowski JL, Beatty B, Trichopoulos D. (1997). J. Natl. Cancer Inst., 89, 71 – 76. Venanzoni M, Papas T and Seth A. (1998). Oncogene, 16, Ekbom A, Trichopoulos D, Adami HO, Hsieh CC and Lan 2459 – 2467. SJ. (1992). Lancet, 340, 1015 – 1018. Burke WM, Jin X, Lin HJ, Huang M, Liu R, Reynolds RK El Ashry D, Miller DL, Kharbanda S, Lippman ME and and Lin J. (2001). Oncogene, 20, 7925 – 7934. Kern FG. (1997). Oncogene, 15, 423 – 435. Chan TA, Hermeking H, Lengauer C, Kinzler KW and Elenbaas B, Spirio L, Koerner F, Fleming MD, Zimonjic Vogelstein B. (1999). Nature, 401, 616 – 620. DB, Donaher JL, Popescu NC, Hahn WC and Weinberg Chen B, He L, Savell VH, Jenkins JJ and Parham DM. RA. (2001). Genes Dev., 15, 50 – 65. (2000). Cancer Res., 60, 3290 – 3298. Elenbaas B and Weinberg RA. (2001). Exp. Cell Res., 264, Chen LM and Chai KX. (2002). Int. J. Cancer, 97, 323 – 329. 169 – 184. Chen LM, Hodge GB, Guarda LA, Welch JL, Greenberg Enmark E, Pelto-Huikko M, Grandien K, Lagercrantz S, NM and Chai KX. (2001). Prostate, 48, 93 – 103. Lagercrantz J, Fried G, Nordenskjold M and Gustafsson Chen QM. (2000). Ann. NY Acad. Sci., 908, 111 – 125. JA. (1997). J. Clin. Endocrinol. Metab., 82, 4258 – 4265. Chen QM, Bartholomew JC, Campisi J, Acosta M, Reagan Esteller M, Corn PG, Baylin SB and Herman JG. (2001). JD and Ames BN. (1998). Biochem J., 332, 43 – 50. Cancer Res., 61, 3225 – 3229.

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5479 Esteller M, Corn PG, Urena JM, Gabrielson E, Baylin SB Hedenfalk I, Duggan D, Chen Y, Radmacher M, Bittner M, and Herman JG. (1998). Cancer Res., 58, 4515 – 4518. Simon R, Meltzer P, Gusterson B, Esteller M, Kallioniemi Esteller M, Silva JM, Dominguez G, Bonilla F, Matias-Guiu OP, Wilfond B, Borg A and Trent J. (2001). N. Engl. J. X, Lerma E, Bussaglia E, Prat J, Harkes IC, Repasky EA, Med., 344, 539 – 548. Gabrielson E, Schutte M, Baylin SB and Herman JG. Heimann R, Ferguson DJ and Hellman S. (1998). Cancer (2000). J. Natl. Cancer Inst., 92, 564 – 569. Res., 58, 2766 – 2771. Ethier SP. (1995). J. Natl. Cancer Inst., 87, 964 – 973. Henderson CJ, Smith AG, Ure J, Brown K, Bacon EJ and Evron E, Dooley WC, Umbricht CB, Rosenthal D, Sacchi N, Wolf CR. (1998). Proc. Natl. Acad. Sci. USA, 95, 5275 – Gabrielson E, Soito AB, Hung DT, Ljung B, Davidson NE 5280. and Sukumar S. (2001b). Lancet, 357, 1335 – 1336. Hori M, Iwasaki M, Yoshimi F, Asato Y and Itabashi M. Evron E, Umbricht CB, Korz D, Raman V, Loeb DM, (1999). Breast Cancer, 6, 79 – 86. Niranjan B, Buluwela L, Weitzman SA, Marks J and Howe HL, Wingo PA, Thun MJ, Ries LA, Rosenberg HM, Sukumar S. (2001a). Cancer Res., 61, 2782 – 2787. Feigal EG and Edwards BK. (2001). J. Natl. Cancer Inst., Fabian CJ. (2001). Breast Cancer Res., 3, 99 – 103. 93, 824 – 842. Falette NS, Fuqua SA, Chamness GC, Cheah MS, Greene Hsieh CL. (1999). Mol. Cell Biol., 19, 8211 – 8218. GL and McGuire WL. (1990). Cancer Res., 50, 3974 – Huang Y, Peng J, Oberley LW and Domann FE. (1997b). 3978. Free Radic. Biol. Med., 23, 314 – 320. Farwell DG, Shera KA, Koop JI, Bonnet GA, Matthews CP, Huang Z, Hankinson SE, Colditz GA, Stampfer MJ, Hunter Reuther GW, Coltrera MD, McDougall JK and Klingel- DJ, Manson JE, Hennekens CH, Rosner B, Speizer FE and hutz AJ. (2000). Am. J. Pathol., 156, 1537 – 1547. Willett WC. (1997a). JAMA, 278, 1407 – 1411. Fearnhead NS, Britton MP and Bodmer WF. (2001). Hum. Hui R, Macmillan RD, Kenny FS, Musgrove EA, Blamey Mol. Genet., 10, 721 – 733. RW, Nicholson RI, Robertson JF and Sutherland RL. Ferguson AT, Evron E, Umbricht CB, Pandita TK, Chan (2000). Clin. Cancer Res., 6, 2777 – 2787. TA, Hermeking H, Marks JR, Lambers AR, Futreal PA, Huschtscha LI, Noble JR, Neumann AA, Moy EL, Barry P, Stampfer MR and Sukumar S. (2000). Proc. Natl. Acad. Melki JR, Clark SJ and Reddel RR. (1998). Cancer Res., Sci. USA, 97, 6049 – 6054. 58, 3508 – 3512. Ferguson DJ and Anderson TJ. (1981). Br.J.Cancer,44, Ingvarsson S. (2001). Semin. Cancer Biol., 11, 361 – 366. 177 – 181. Iwase H, Omoto Y, Iwata H, Toyama T, Hara Y, Ando Y, Fisher B, Costantino JP, Wickerham DL, Redmond CK, Ito Y, Fujii Y and Kobayashi S. (1999). Br. J. Cancer, 80, Kavanah M, Cronin WM, Vogel V, Robidoux A, Dimitrov 1982 – 1986. N, Atkins J, Daly M, Wieand S, Tan-Chiu E, Ford L and Jenuwein T and Allis CD. (2001). Science, 293, 1074 – 1080. Wolmark N. (1998). J. Natl. Cancer Inst., 90, 1371 – 1388. Jhaveri MS and Morrow CS. (1998). Gene, 210, 1–7. Foster SA, Wong DJ, Barrett MT and Galloway DA. (1998). Jia T, Liu YE, Liu J and Shi YE. (1999). Cancer Res., 59, Mol. Cell Biol., 18, 1793 – 1801. 742 – 747. FuldaS,KuferMU,MeyerE,vanValenF,Dockhorn- Jiang H, Chou HS and Zhu L. (1998). Mol. Cell Biol., 18, Dworniczak B and Debatin KM. (2001). Oncogene, 20, 5284 – 5290. 5865 – 5877. JinZ,TamuraG,TsuchiyaT,SakataK,KashiwabaM, Garcia R, Bowman TL, Niu G, Yu H, Minton S, Muro- Osakabe M and Motoyama T. (2001). Br.J.Cancer,85, Cacho CA, Cox CE, Falcone R, Fairclough R, Parsons S, 69 – 73. Laudano A, Gazit A, Levitzki A, Kraker A and Jove R. Jones PA. (1996). Cancer Res., 56, 2463 – 2467. (2001). Oncogene, 20, 2499 – 2513. Jones PA. (1999). Trends Genet., 15, 34 – 37. Ginger MR, Gonzalez-Rimbau MF, Gay JP and Rosen JM. Jones PA and Laird PW. (1999). Nat. Genet., 21, 163 – 167. (2001). Mol. Endocrinol., 15, 1993 – 2009. Jones PA and Takai D. (2001). Science, 293, 1068 – 1070. Going JJ, Anderson TJ, Battersby S and MacIntyre CC. Kalkhoven E, Roelen BA, de Winter JP, Mummery CL, (1988). Am. J. Pathol., 130, 193 – 204. Eijnden-van Raaij AJ, van der Saag PT and van der BB. Gonzalez AD, Kaya M, Shi W, Song H, Testa JR, Penn LZ (1995). Cell Growth Differ., 6, 1151 – 1161. and Filmus J. (1998). J. Cell Biol., 141, 1407 – 1414. Kamel OW, Franklin WA, Ringus JC and Meyer JS. (1989). GoyalJ,SmithKM,CowanJM,WazerDE,LeeSWand Am.J.Pathol.,134, 107 – 113. Band V. (1998). Cancer Res., 58, 4782 – 4786. KangJH,KimSJ,NohDY,ParkIA,ChoeKJ,YooOJand Graff JR, Gabrielson E, Fujii H, Baylin SB and Herman JG. Kang HS. (2001). Lab. Invest., 81, 573 – 579. (2000). J. Biol. Chem., 275, 2727 – 2732. Karpf AR, Peterson PW, Rawlins JT, Dalley BK, Yang Q, Green S, Walter P, Kumar V, Krust A, Bornert JM, Argos P Albertsen H and Jones DA. (1999). Proc. Natl. Acad. Sci. and Chambon P. (1986). Nature, 320, 134 – 139. USA, 96, 14007 – 14012. Gronowski AM, Zhong Z, Wen Z, Thomas MJ, Darnell Jr Kay PH, Hahnel R, Gunn H, Harmon D and Song J. (1998). JE and Rotwein P. (1995). Mol. Endocrinol., 9, 171 – 177. Anticancer Res., 18, 1709 – 1712. Gustafsson JA and Warner M. (2000). J. Steroid Biochem. Khokhlatchev A, Rhabizadeh S, Xavier R, Nedwidek M, Mol. Biol., 74, 245 – 248. Chen T, Zhang XF, Seed B and Avruch J. (2002). Curr. Hanahan D and Weinberg RA. (2000). Cell, 100, 57 – 70. Biol., 12, 253 – 265. Harris JR, Lippman ME, Veronesi U and Willett W. (1992). Kim DW, Sovak MA, Zanieski G, Nonet G, Romieu- N. Engl. J. Med., 327, 319 – 328. Mourez R, Lau AW, Hafer LJ, Yaswen P, Stampfer M, Hartsough MT, Clare SE, Mair M, Elkahloun AG, Sgroi D, Rogers AE, Russo J and Sonenshein GE. (2000). Osborne CK, Clark G and Steeg PS. (2001). Cancer Res., Carcinogenesis, 21, 871 – 879. 61, 2320 – 2327. Kinscherf R, Claus R, Wagner M, Gehrke C, Kamencic H, Heard E, Rougeulle C, Arnaud D, Avner P, Allis CD and Hou D, Nauen O, Schmiedt W, Kovacs G, Pill J, Metz J Spector DL. (2001). Cell, 107, 727 – 738. and Deigner HP. (1998). FASEB J., 12, 461 – 467.

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5480 Kishimoto T and Kikutani H. (2001). Nat. Genet., 28, 4–5. Macleod D, Ali RR and Bird A. (1998). Mol. Cell Biol., 18, Kiyono T, Foster SA, Koop JI, McDougall JK, Galloway 4433 – 4443. DA and Klingelhutz AJ. (1998). Nature, 396, 84 – 88. MacMahon B, Cole P, Lin TM, Lowe CR, Mirra AP, Komatsu S, Okazaki Y, Tateno M, Kawai J, Konno H, Ravnihar B, Salber EJ, Valaoras VG and Yuasa S. (1970). Kusakabe M, Yoshiki A, Muramatsu M, Held WA and Bull. World Health Organ., 43, 209 – 221. Hayashizaki Y. (2000). Biochem. Biophys. Res. Commun., Malaney S and Daly RJ. (2001). J. Mammary Gland Biol. 267, 109 – 117. Neoplasia, 6, 101 – 113. Kress C, Thomassin H and Grange T. (2001). FEBS Lett., Marsh SK, Bansal GS, Zammit C, Barnard R, Coope R, 494, 135 – 140. Roberts-Clarke D, Gomm JJ, Coombes RC and Johnston KropIE,SgroiD,PorterDA,LunettaKL,LeVangieR,Seth CL. (1999). Oncogene, 18, 1053 – 1060. P, Kaelin CM, Rhei E, Bosenberg M, Schnitt S, Marks JR, Mehta RG. (2000). Eur. J. Cancer, 36, 1275 – 1282. Pagon Z, Belina D, Razumovic J and Polyak J. (2001). Meyer JS. (1977). Hum. Pathol., 8, 67 – 81. Proc. Natl. Acad. Sci. USA, 98, 9796 – 9801. Meyer JS and Connor RE. (1982). Cancer, 50, 746 – 751. Krtolica A, Parrinello S, Lockett S, Desprez PY and Campisi Meyyappan M, Wong H, Hull C and Riabowol KT. (1998). J. (2001). Proc. Natl. Acad. Sci. USA, 98, 12072 – 12077. Mol. Cell Biol., 18, 3163 – 3172. Kuiper GG, Enmark E, Pelto-Huikko M, Nilsson S and Mielnicki LM, Ying AM, Head KL, Asch HL and Asch BB. Gustafsson JA. (1996). Proc. Natl. Acad. Sci. USA, 93, (1999). Exp. Cell Res., 249, 161 – 176. 5925 – 5930. Miettinen OS, Henschke CI, Pasmantier MW, Smith JP, Kyo S, Takakura M, Kanaya T, Zhuo W, Fujimoto K, Libby DM and Yankelevitz DF. (2002). Lancet, 359, 404 – Nishio Y, Orimo A and Inoue M. (1999). Cancer Res., 59, 405. 5917 – 5921. Minucci S and Pelicci PG. (1999). Semin. Cell Dev. Biol., 10, Laird PW, Jackson-Grusby L, Fazeli A, Dickinson SL, Jung 215 – 225. WE, Li E, Weinberg RA and Jaenisch R. (1995). Cell, 81, Mizuno S, Chijiwa T, Okamura T, Akashi K, Fukumaki Y, 197 – 205. Niho Y and Sasaki H. (2001). Blood, 97, 1172 – 1179. Landberg G, Ostlund H, Nielsen NH, Roos G, Emdin S, Moinfar F, Man YG, Arnould L, Bratthauer GL, Ratschek Burger AM and Seth A. (2001). Oncogene, 20, 3497 – 3505. M and Tavassoli FA. (2000). Cancer Res., 60, 2562 – 2566. Lapidus RG, Ferguson AT, Ottaviano YL, Parl FF, Smith MoscowJA,GongM,HeR,SgagiasMK,DixonKH, HS, Weitzman SA, Baylin SB, Issa JPJ and Davidson NE. Anzick SL, Meltzer PS and Cowan KH. (1995). Cancer (1996). Clin. Cancer Res., 2, 805 – 810. Res., 55, 3790 – 3794. LapidusRG,NassSJ,ButashKA,ParlFF,WeitzmanSA, Munster PN, Troso-Sandoval T, Rosen N, Rifkind R, Marks Graff JG, Herman JG and Davidson NE. (1998). Cancer PA and Richon VM. (2001). Cancer Res., 61, 8492 – 8497. Res., 58, 2515 – 2519. Ng ST, Zhou J, Adesanya OO, Wang J, LeRoith D and Lee SW. (1996). Nat. Med., 2, 776 – 782. Bondy CA. (1997). Nat. Med., 3, 1141 – 1144. Lee X, Si SP, Tsou HC and Peacocke M. (1995). Exp. Cell Nilsson S, Makela S, Treuter E, Tujague M, Thomsen J, Res., 218, 296 – 304. Andersson G, Enmark E, Pettersson K, Warner M and Lehmann U, Langer F, Feist H, Glockner S, Hasemeier B Gustafsson JA. (2001). Physiol. Rev., 81, 1535 – 1565. and Kreipe H. (2002). Am.J.Pathol.,160, 605 – 612. Nojima D, Li LC, Dharia A, Perinchery G, Ribeiro-Filho L, Lelle RJ, Heidenreich W, Stauch G, Wecke I and Gerdes J. Yen TS and Dahiya R. (2001). Cancer, 92, 2076 – 2083. (1987). J. Cancer Res. Clin. Oncol., 113, 73 – 77. Norton JD. (2000). J. Cell Sci., 113, 3897 – 3905. Leone A, Flatow U, VanHoutte K and Steeg PS. (1993). Ogawa Y, Hirakawa K, Nakata B, Fujihara T, Sawada T, Oncogene, 8, 2325 – 2333. Kato Y, Yoshikawa K and Sowa M. (1998). Clin. Cancer LiB,GoyalJ,DharS,DimriG,EvronE,SukumarS,Wazer Res., 4, 31 – 36. DE and Band V. (2001b). Cancer Res., 61, 8014 – 8021. Ohtani N, Zebedee Z, Huot TJ, Stinson JA, Sugimoto M, Li Q, Ahuja N, Burger PC and Issa JP. (1999). Oncogene, 18, Ohashi Y, Sharrocks AD, Peters G and Hara E. (2001). 3284 – 3289. Nature, 409, 1067 – 1070. Li Z, Khaletskiy A, Wang J, Wong JY, Oberley LW and Li Okano M, Bell DW, Haber DA and Li E. (1999). Cell, 99, JJ. (2001a). Free Radic. Biol. Med., 30, 260 – 267. 247 – 257. Liang G, Gonzales FA, Jones PA, Orntoft TF and Thykajaer Olsen O and Gotzsche PC. (2001). Lancet, 358, 1340 – 1342. T. (2002). Cancer Res., 62, 961 – 966. Ottaviano YL, Issa JP, Parl FF, Smith HS, Baylin SB and LinCH,HsiehSY,SheenIS,LeeWC,ChenTC,ShyuWC Davidson NE. (1994). Cancer Res., 54, 2552 – 2555. and Liaw YF. (2001). Cancer Res., 61, 4238 – 4243. Ouchi T, Lee SW, Ouchi M, Aaronson SA and Horvath CM. Lindeman GJ, Wittlin S, Lada H, Naylor MJ, Santamaria M, (2000). Proc. Natl. Acad. Sci. USA, 97, 5208 – 5213. Zhang JG, Starr R, Hilton DJ, Alexander WS, Ormandy Pagliuca A, Gallo P, De Luca P and Lania L. (2000). Cancer CJ and Visvader J. (2001). Genes Dev., 15, 1631 – 1636. Res., 60, 1376 – 1382. Lippman ME and Dickson RB. (1990). Prog. Clin. Biol. Res., PerouCM,JeffreySS,vandeRM,ReesCA,EisenMB,Ross 354A, 147 – 178. DT, Pergamenschikov A, Williams CF, Zhu SX, Lee JC, LiuM,TaketaniT,LiR,TakitaJ,TakiT,YangHW, Lashkari D, Shalon D, Brown PO and Botstein D. (1999). Kawaguchi H, Ida K, Matsuo Y and Hayashi Y. (2001). Proc. Natl. Acad. Sci. USA, 96, 9212 – 9217. Leuk. Res., 25, 441 – 447. Pharoah PD, Day NE and Caldas C. (1999). Br.J.Cancer, Longacre TA and Bartow SA. (1986). Am. J. Surg. Pathol., 80, 1968 – 1973. 10, 382 – 393. Pietras RJ, Arboleda J, Reese DM, Wongvipat N, Pegram Lopatina N, Haskell JF, Andrews LG, Poole JC, Saldanha S MD, Ramos L, Gorman CM, Parker MG, Sliwkowski MX and Tollefsbol T. (2002). J. Cell Biochem., 84, 324 – 334. and Slamon DJ. (1995). Oncogene, 10, 2435 – 2446. Lotan R, Xu XC, Lippman SM, Ro JY, Lee JS, Lee JJ and Pike MC, Spicer DV, Dahmoush L and Press MF. (1993). Hong WK. (1995). N.Engl.J.Med.,332, 1405 – 1410. Epidemiol. Rev., 15, 17 – 35. LuA,GuptaA,LiC,AhlbornTE,MaY,ShiEYandLiuJ. Polednak AP and Janerich DT. (1983). J. Reprod. Med., 28, (2001). Oncogene, 20, 5173 – 5185. 314 – 318.

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5481 Post WS, Goldschmidt-Clermont PJ, Wilhide CC, Heldman SwisshelmK,RyanK,LeeX,TsouHC,PeacockeMand AW, Sussman MS, Ouyang P, Milliken EE and Issa JP. Sager R. (1994). Cell Growth Differ., 5, 133 – 141. (1999). Cardiovasc. Res., 43, 985 – 991. Swisshelm K, Ryan K, Tsuchiya K and Sager R. (1995). Raman V, Martensen SA, Reisman D, Evron E, Odenwald Proc. Natl. Acad. Sci. USA, 92, 4472 – 4476. WF, Jaffee E, Marks J and Sukumar S. (2000). Nature, Takai D and Jones PA. (2002). Proc. Natl. Acad. Sci. USA, 405, 974 – 978. 99, 3740 – 3745. Ramchandani S, MacLeod AR, Pinard M, von Hofe E and Tamaru H and Selker EU. (2001). Nature, 414, 277 – 283. Szyf M. (1997). Proc.Natl.Acad.Sci.USA,94, 684 – 689. Thomassin H, Flavin M, Espinas ML and Grange T. (2001). Rexer BN, Zheng WL and Ong DE. (2001). Cancer Res., 61, EMBO J., 20, 1974 – 1983. 7065 – 7070. Thompson WD, Jacobson HI, Negrini B and Janerich DT. Ristimaki A, Sivula A, Lundin J, Lundin M, Salminen T, (1989). J. Natl. Cancer Inst., 81, 1571 – 1574. Haglund C, Joensuu H and Isola J. (2002). Cancer Res., 62, Thor AD, Edgerton SM, Liu S, Moore DH and Kwiatkowski 632 – 635. DJ. (2001). Clin. Cancer Res., 7, 2415 – 2424. Roberts JM and Cooper DW. (2001). Lancet, 357, 53 – 56. Tollefsbol TO and Andrews LG. (2001). Med. Hypotheses, Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J, 56, 630 – 637. Gonzales FA and Jones PA. (1999). Nucleic Acids Res., 27, Toyooka KO, Toyooka S, Virmani AK, Sathyanarayana 2291 – 2298. UG, Euhus DM, Gilcrease M, Minna JD and Gazdar AF. Rodriguez-Manzaneque JC, Lane TF, Ortega MA, Hynes (2001). Cancer Res., 61, 4556 – 4560. RO, Lawler J and Iruela-Arispe ML. (2001). Proc. Natl. Troisi R, Weiss HA, Hoover RN, Potischman N, Swanson Acad. Sci. USA, 98, 12485 – 12490. CA, Brogan DR, Coates RJ, Gammon MD, Malone KE, Romanov SR, Kozakiewicz BK, Holst CR, Stampfer MR, Daling JR and Brinton LA. (1998). Epidemiology, 9, 641 – Haupt LM and Tisty TD. (2001). Nature, 409, 633 – 637. 647. Ross RK, Paganini-Hill A, Wan PC and Pike MC. (2000). J. Umbricht CB, Evron E, Gabrielson E, Ferguson A, Marks J Natl. Cancer Inst., 92, 328 – 332. and Sukumar S. (2001). Oncogene, 20, 3348 – 3353. Rudland PS. (1993). Histol. Histopathol., 8, 385 – 404. Uria JA, Ferrando AA, Velasco G, Freije JM and Lopez- Saji S, Jensen EV, Nilsson S, Rylander T, Warner M and Otin C. (1994). Cancer Res., 54, 2091 – 2094. Gustafsson JA. (2000). Proc. Natl. Acad. Sci. USA, 97, Virmani AK, Rathi A, Sathyanarayana UG, Padar A, 337 – 342. Huang CX, Cunnigham HT, Farinas AJ, Milchgrub S, Salimath B, Marme D and Finkenzeller G. (2000). Oncogene, Euhus DM, Gilcrease M, Herman J, Minna JD and 19, 3470 – 3476. Gazdar AF. (2001). Clin. Cancer Res., 7, 1998 – 2004. Selker EU. (1990). Trends Biochem. Sci., 15, 103 – 107. Vos MD, Ellis CA, Bell A, Birrer MJ and Clark GJ. (2000). J. Shang Y, Baumrucker CR and Green MH. (1999). Oncogene, Biol. Chem., 275, 35669 – 35672. 18, 6725 – 6732. Vousden KH. (2000). Cell, 103, 691 – 694. Sherr CJ. (1996). Science, 274, 1672 – 1677. Widschwendter M, Berger J, Daxenbichler G, Muller- Silva JM, Dominguez G, Garcia JM, Gonzalez R, Villanueva Holzner E, Widschwendter A, Mayr A, Marth C and MJ, Navarro F, Provencio M, San Martin S, Espana P and Zeimet AG. (1997). Cancer Res., 57, 4158 – 4161. Bonilla F. (1999a). Cancer Res., 59, 3521 – 3526. Widschwendter M, Berger J, Hermann M, Muller HM, Silva JM, Dominguez G, Villanueva MJ, Gonzalez R, Garcia Amberger A, Zeschnigk M, Widschwendter A, Abendstein JM, Corbacho C, Provencio M, Espana P and Bonilla F. B, Zeimet AG, Daxenbichler G and Marth C. (2000). J. (1999b). Br.J.Cancer,80, 1262 – 1264. Natl. Cancer Inst., 92, 826 – 832. Sirchia SM, Ferguson AT, Sironi E, Subramanyan S, Widschwendter M, Daxenbichler G, Dapunt O and Marth C. Orlandi R, Sukumar S and Sacchi N. (2000). Oncogene, (1995). Cancer Res., 55, 2135 – 2139. 19, 1556 – 1563. Widschwendter M and Jones PA. (2002). Clin. Cancer Res., Sirotnak FM and Tolner B. (1999). Annu. Rev. Nutr., 19, 91 – 8, 17 – 21. 122. Wiesen JF, Young P, Werb Z and Cunha GR. (1999). Sivaraman L, Conneely OM, Medina D and O’Malley BW. Development, 126, 335 – 344. (2001). Proc. Natl. Acad. Sci. USA, 98, 12379 – 12384. Wijermans P, Lubbert M, Verhoef G, Bosly A, Ravoet C, Soares J, Pinto AE, Cunha CV, Andre S, Barao I, Sousa JM Andre M and Ferrant A. (2000). J. Clin. Oncol., 18, 956 – and Cravo M. (1999). Cancer, 85, 112 – 118. 962. Soengas MS, Capodieci P, Polsky D, Mora J, Esteller M, Wilson VL and Jones PA. (1983). Science, 220, 1055 – 1057. Opitz-Araya X, McCombie R, Herman JG, Gerald WL, Wong DJ, Foster SA, Galloway DA and Reid BJ. (1999). Lazebnik YA, Cordon-Cardo C and Lowe SW. (2001). Mol. Cell Biol., 19, 5642 – 5651. Nature, 409, 207 – 211. Worm J, Kirkin AF, Dzhandzhugazyan KN and Guldberg P. Song JZ, Stirzaker C, Harrison J, Melki JR and Clark SJ. (2001). J. Biol. Chem., 276, 39990 – 40000. (2002). Oncogene, 21, 1048 – 1061. Xiang YY, Ladeda V and Filmus J. (2001). Oncogene, 20, Sprenger CC, Vail ME, Evans K, Simurdak J and Plymate 7408 – 7412. SR. (2002). Oncogene, 21, 140 – 147. Xu XC, Sneige N, Liu X, Nandagiri R, Lee JJ, Lukmanji F, Stark GR, Kerr IM, Williams BR, Silverman RH and Hortobagyi G, Lippman SM, Dhingra K and Lotan R. Schreiber RD. (1998). Annu. Rev. Biochem., 67, 227 – 264. (1997). Cancer Res., 57, 4992 – 4996. Sugano T, Nitta M, Ohmori H and Yamaizumi M. (1995). Xu XL, Wu LC, Du F, Davis A, Peyton M, Tomizawa Y, Jpn. J. Cancer Res., 86, 415 – 418. Maitra A, Tomlinson G, Gazdar AF, Weissman BE, Sullivan PM, Etzioni R, Feng Z, Potter JD, Thompson ML, Bowcock AM, Baer R and Minna JD. (2001). Cancer Res., Thornquist M, Winget M and Yasui Y. (2001). J. Natl. 61, 7943 – 7949. Cancer Inst., 93, 1054 – 1061. Yang X, Ferguson AT, Nass SJ, Phillips DL, Butash KA, Sun L, Wu G, Willson JK, Zborowska E, Yang J, Wang SM, Herman JG and Davidson NE. (2000). Cancer Rajkarunanayake I, Wang J, Gentry LE, Wang XF and Res., 60, 6890 – 6894. Brattain MG. (1994). J. Biol. Chem., 269, 26449 – 26455.

Oncogene DNA methylation and breast carcinogenesis M Widschwendter and PA Jones 5482 Yang X, Phillips DL, Ferguson AT, Nelson WG, Herman JG Zeps N, Bentel JM, Papadimitriou JM, D’Antuono MF and and Davidson NE. (2001b). Cancer Res., 61, 7025 – 7029. Dawkins HJ. (1998). Differentiation, 62, 221 – 226. Yang X, Yan L and Davidson NE. (2001a). Endocr. Relat. Zhang M, Volpert O, Shi YH and Bouck N. (2000). Nat. Cancer, 8, 115 – 127. Med., 6, 196 – 199. Yatabe Y, Tavare S and Shibata D. (2001). Proc. Natl. Acad. Zhang M, Zou Z, Maass N and Sager R. (1995). Cancer Res., Sci. USA, 98, 10839 – 10844. 55, 2537 – 2541. Yoshikawa H, Matsubara K, Qian GS, Jackson P, Groop- Zochbauer-Muller S, Fong KM, Maitra A, Lam S, Geradts man JD, Manning JE, Harris CC and Herman JG. (2001). J, Ashfaq R, Virmani AK, Milchgrub S, Gazdar AF and Nat. Genet., 28, 29 – 35. Minna JD. (2001). Cancer Res., 61, 3581 – 3585. Yu Y, Xu F, Peng H, Fang X, Zhao S, Li Y, Cuevas B, Kuo Zou Z, Anisowicz A, Hendrix MJ, Thor A, Neveu M, Sheng WL, Gray JW, Siciliano M, Mills GB and Bast RC. (1999). S, Raﬁdi K, Seftor E and Sager R. (1994). Science, 263, Proc. Natl. Acad. Sci. USA, 96, 214 – 219. 526 – 529. Yuan Y, Mendez R, Sahin A and Dai JL. (2001). Cancer Res., 61, 5558 – 5561.

Oncogene