Abstract

Changes in the epigenetic landscape are widespread in neoplasia, with de novo methylation and histone repressive marks commonly enriched in CpG island associated promoter regions. DNA hypermethylation and histone repression correlate with gene silencing, however, the dynamics of this process are still largely unclear. The tumour suppressor gene p16INK4A is inactivated in association with CpG island methylation during neoplastic progression in a variety of cancers, including breast cancer. Here, we investigated the temporal progression of DNA methylation and histone remodelling in the p16INK4A CpG island in primary human mammary epithelial cell (HMEC) strains during selection, as a model for early breast cancer. Silencing of p16INK4A has been previously shown to be necessary before HMECs can escape from selection. Here, we demonstrate that gene silencing occurs prior to de novo methylation and histone remodelling. An increase in DNA methylation was associated with a rapid loss of both histone H3K27 trimethylation and H3K9 acetylation and a gradual gain of H3K9 dimethylation. Interestingly, we found that regional-specific ‘seeding’ methylation occurs early after post-selection and that the de novo methylation pattern observed in HMECs correlates with the apparent footprint of nucleosomes across the p16INK4A CpG island. Our results demonstrate for the first time that p16INK4A gene silencing is a precursor to epigenetic suppression and that subsequent de novo methylation initially occurs in nucleosome-free regions across the p16INK4A CpG island and this is associated with a dynamic change in histone modifications.

INTRODUCTION

Widespread changes in genomic DNA methylation patterns occur during the transition from a normal cell to a cancer cell, and this is associated with chromatin remodelling and modified gene expression. Chromatin consists of nucleosomes, each containing 147 bp of DNA wrapped around an octamer of core histone proteins, which are separated from each other by ∼50 bp of linker DNA (1). Nucleosome occupancy along DNA promoters plays an important role in transcriptional regulation (2), where both sliding and loss of nucleosomes affect the accessibility of the DNA to transcription factors (reviewed in 3). Malignant cells are characterized by a global reduction in genomic DNA methylation and a localized increase in methylation of CpG island-associated promoter regions (reviewed in 4–7). CpG islands have a frequency of CpG dinucleotides approximately five times greater than the genome, and commonly span the promoter region of ubiquitously expressed ‘housekeeping’ genes, and the 5′ or 3′ regions of many tissue-specific genes (8,9). Typically the CpG islands of transcriptionally active genes in normal cells are unmethylated and associated with permissive histone marks, whereas the CpG islands of transcriptionally inactive genes in cancer cells are densely methylated and associated with repressive histone marks. However, the dynamics of aberrant de novo methylation in CpG island-associated promoters that mediate the transition from the unmethylated, active state to the densely methylated, inactive state remain largely unknown (10,11). It is not easy to address this question in tumour tissue because DNA hypermethylation is often an early event and therefore once the tumour is large enough to detect the aberrant methylation process has already occurred and the hypermethylated genes are already silenced.

In this study, we investigated the temporal change in epigenetic modifications in the CpG island of the p16INK4A gene (also known as CDKN2A and MTS-1). The p16INK4A protein binds and inhibits the activities of cyclin-dependent kinases CDK4 and CDK6 resulting in the hypophosphorylated form of pRb which acts as an inhibitor of cell-cycle progression (reviewed in 12). Many human tumour cell lines and primary tumours have lost expression of wild-type p16INK4A and this has led to the suggestion that the p16INK4A gene encodes a tumour suppressor (13). Indeed, the p16INK4A CpG island promoter is often hypermethylated in many tumours including breast tumours, and this appears to occur early in the oncogenic pathway (14–16). We and others have shown that loss of p16INK4A expression in human mammary epithelial cells (HMECs) is necessary for in vitro lifespan extension (17), and this also correlates with hypermethylation of the CpG island promoter (18–23). HMECs, when cultured in serum-free medium, exhibit two phases of growth (24,25). The first growth phase lasts for several population doublings (PDs), after which growth temporarily ceases (termed selection or M0). Within 2–4 weeks, colonies of small cells with a basal mammary epithelial phenotype (26,27) appear with enhanced growth capacity and these colonies continue to proliferate for another 20–40 PDs before entering a second growth plateau resembling cell crisis, termed agonescence (28). HMECs isolated during the first growth phase are termed pre-selection cells, and those isolated during the second growth phase are termed either post-selection or variant HMECs (vHMECs) (19,22,24). Post-selection HMECs have been shown to share many characteristics of pre-malignant breast-cancer cells, including both genetic and epigenetic lesions (21,23,29,30), and therefore provide an ideal primary cell model to study early epigenetic changes in malignancy.

To investigate the temporal relationship between gene silencing, DNA methylation and chromatin remodelling, we analysed in detail de novo methylation of the p16INK4A CpG island in post-selection HMECs that were not expressing p16INK4A. Using this primary tissue model, we first found that hypermethylation of the p16INK4A CpG island occurs only after the gene is silenced, and secondly we demonstrated that a low level of de novo methylation is associated with a dynamic remodelling of associated chromatin, as early as the first passage following selection. Lastly, we demonstrated that individual post-selection strains share a common pattern of regional-specific initial ‘seeding’ methylation within the p16INK4A CpG island. Using a high-resolution foot-printing technique, known as methylase-based single-promoter analysis assay (MSPA) (31), which exploits the fact that nucleosomes and binding factors restrict M. Sss I CpG methylase from methylating the DNA (32), we found that the ‘seeding’ methylation ‘hot spots’ correlated with the position of nucleosomes in the post-selection HMECs. Our results demonstrate for the first time that p16INK4A gene silencing is a precursor to epigenetic changes in post-selection HMECs and that subsequent de novo methylation occurs primarily in nucleosome-free regions across the p16INK4A CpG island and then progressively spreads to adjacent regions with proliferation.

RESULTS

p16INK4A is inactivated and heterogeneously methylated in post-selection HMECs

To study the underlying epigenetic changes that occur during p16INK4A silencing in HMECs that have undergone selection, we first used quantitative reverse transcription-PCR (qRT–PCR) to show that p16INK4A mRNA expression is inactivated, or down regulated, in the individual post-selection HMECs strains, Bre-12 and Bre-40 (30), and Bre-56, Bre-60 and Bre-80, compared with their isogenic pre-selection HMECs (Fig. 1A). We also confirmed that a loss of p16INK4A mRNA expression correlates with a loss in p16INK4A protein expression in Bre-56 and Bre-80 post-selection cells (Fig. 1B), as previously shown in Bre-40, Bre-60, Bre-70 and Bre-80 post-selection cells (19). Using direct PCR bisulphite sequencing, we reported that p16INK4A silencing in Bre-40, Bre-60, Bre-70 and Bre-80 post-selection HMECs correlates with extensive DNA hypermethylation of the CpG island-associated p16INK4A promoter (19). To better understand the dynamics and initial events leading to CpG island methylation, we have analysed the DNA methylation profiles in more detail by bisulphite clonal sequencing across three neighbouring regions of the p16INK4A CpG island spanning 1035 bp across the start of transcription and first exon (Supplementary Material, Fig. S1). The post-selection HMECs from each strain comprised 10–40 independent colonies that escaped selection and these colonies were pooled prior to analysis to gain a comprehensive view of the DNA methylation patterns. Figure 2 summarizes the bisulphite methylation sequencing data from Bre-38 and Bre-40 pre- and post-selection cells across 71 CpG sites (numbered from –19 CpG to +52 CpG relative to the start of transcription). The actively p16INK4A expressing pre-selection cells (Bre-38 and Bre-40) were essentially unmethylated from CpG sites –19 to +38, even though there was evidence of low-level sporadic CpG methylation noted in Bre-40 (Fig. 2A and C). In the pre-selection cells, the level of methylation was further enriched downstream from the start of transcription at CpG sites +40 to +52 in both Bre-38 and Bre-40 (Fig. 2A and C) correlating to the boundary region of the p16INK4A CpG island (Supplementary Material, Fig. S1). In contrast, in the corresponding Bre-38 and Bre-40 post-selection HMECs [passage 11 (p11) and 10 (p10), respectively] there was extensive DNA methylation across the three regions analysed (Fig. 2B and D) and this occurred in nearly all molecules, suggesting that DNA methylation of p16INK4A is bi-allelic.

Figure 1.

Suppression of p16INK4A expression in post-selection HMECs. (A) mRNA levels of p16INK4A in Bre-12, Bre-40, Bre-56, Bre-60 and Bre-80 post-selection HMECs were determined by quantitative RT–PCR. After normalizing expression to 18S rRNA, the fold change in expression levels was made relative to pre-selection HMECs. All donor post-selection strains show reduced levels of p16INK4A mRNA. (B) Western blot showing p16INK4A protein expression in Bre-56 pre-selection cells, and p16INK4A silencing in Bre-56 and Bre-80 post-selection cells. Actin levels were used as a loading control.

Figure 1.

Suppression of p16INK4A expression in post-selection HMECs. (A) mRNA levels of p16INK4A in Bre-12, Bre-40, Bre-56, Bre-60 and Bre-80 post-selection HMECs were determined by quantitative RT–PCR. After normalizing expression to 18S rRNA, the fold change in expression levels was made relative to pre-selection HMECs. All donor post-selection strains show reduced levels of p16INK4A mRNA. (B) Western blot showing p16INK4A protein expression in Bre-56 pre-selection cells, and p16INK4A silencing in Bre-56 and Bre-80 post-selection cells. Actin levels were used as a loading control.

Figure 2.

p16INK4A silencing in post-selection cells is accompanied by DNA hypermethylation. Bisulphite methylation clonal sequencing analysis was performed across three neighbouring regions of the p16INK4A CpG island associated promoter region. Amplicon locations: region I, region II and region III. The start of transcription, as indicated by Genbank accession number p16INK4A (AB060808), is indicated by a black arrow. CpG sites are numbered relative to the start of transcription. (A) Bre-38 pre-selection cells [passage 5 (p5)]. (B) Bre-38 post-selection cells (p11). (C) Bre-40 pre-selection cells (p5). (D) Bre-40 post-selection cells (p10). White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation in post-selection HMECs.

Figure 2.

p16INK4A silencing in post-selection cells is accompanied by DNA hypermethylation. Bisulphite methylation clonal sequencing analysis was performed across three neighbouring regions of the p16INK4A CpG island associated promoter region. Amplicon locations: region I, region II and region III. The start of transcription, as indicated by Genbank accession number p16INK4A (AB060808), is indicated by a black arrow. CpG sites are numbered relative to the start of transcription. (A) Bre-38 pre-selection cells [passage 5 (p5)]. (B) Bre-38 post-selection cells (p11). (C) Bre-40 pre-selection cells (p5). (D) Bre-40 post-selection cells (p10). White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation in post-selection HMECs.

Even though the DNA methylation was generally extensive, there was also considerable heterogeneity noted, with some molecules having no DNA methylation or methylation at only a few of the CpG sites, despite p16INK4A expression being suppressed in these cells. The DNA methylation was also not uniformly distributed between the CpG sites. Blocks of adjacent CpG sites appeared to be more resistant to methylation, whereas other CpG sites appeared to be predominately methylated in ‘hot spots’ (Fig. 2B and D). Four such ‘hot spots’ were found to occur in similar locations in Bre-38 and Bre-40 post-selection HMECs across regions I and II. To ascertain whether the same CpG sites were also ‘hot spots’ for the de novo methylation in different post-selection HMEC strains, we further analysed region II in other post-selection HMEC strains, Bre-60 (p7) and Bre-70 (p9) (Fig. 3). Focal ‘hot spots’ of methylation of adjacent CpG sites were also observed for these post-selection HMECs, even though the location of the specific sites were slightly different from strain to strain.

Figure 3.

DNA hypermethylation of p16INK4A occurs at focal ‘hot spot’ regions. Bisulphite methylation clonal sequencing analysis was performed at region II of the p16INK4A CpG island-associated promoter, revealing that there is a distinct pattern of DNA methylation. CpG sites are numbered relative to the start of transcription. (A) Bre-38 post-selection cells (p11). (B) Bre-40 post-selection cells (p10). (C) Bre-60 post-selection cells (p7). (D) Bre-70 post-selection cells (p9). White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation.

Figure 3.

DNA hypermethylation of p16INK4A occurs at focal ‘hot spot’ regions. Bisulphite methylation clonal sequencing analysis was performed at region II of the p16INK4A CpG island-associated promoter, revealing that there is a distinct pattern of DNA methylation. CpG sites are numbered relative to the start of transcription. (A) Bre-38 post-selection cells (p11). (B) Bre-40 post-selection cells (p10). (C) Bre-60 post-selection cells (p7). (D) Bre-70 post-selection cells (p9). White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation.

p16INK4A silencing in post-selection HMECs occurs prior to DNA hypermethylation

We previously showed that p16INK4A is inactivated and methylated in post-selection HMECs (19), but what was not clear at the time was if DNA methylation was causing p16INK4A silencing or if p16INK4A silencing was promoting DNA methylation. The fact that the methylation pattern in all the different post-selection HMEC strains was heterogeneous, and a number of molecules remained unmethylated or had only minimal methylation in the early passages after selection (Figs 2B,D and Fig. 3), led us to ask if p16INK4A silencing preceded DNA methylation. To address this question, we used laser capture microscopy to isolate individual post-selection HMECs that had just emerged from selection and stained negative for p16INK4A expression (Fig. 4A–C). At this early time point, each post-selection colony had only expanded to approximately 30 cells or 4–5 PDs. We performed bisulphite sequencing on pools of 20 individually laser-captured cells that were negative for p16INK4A expression from a colony at the time of selection (Fig. 4B), and compared the methylation profile to 20 pooled individually laser-captured pre-selection or senescent cells, from the same dish, that were positive for p16INK4A expression (Fig. 4A, Supplementary Material, Fig. S2). Figure 4D summarizes the methylation results and shows that there was little or no methylation evident in either the p16INK4A silent HMECs at the time of selection, or the surrounding p16INK4A expressing pre-selection cells (Fig. 4D). The silencing of the p16INK4A gene therefore appears to occur independently and prior to subsequent de novo methylation.

Figure 4.

p16INK4 silencing in HMECs occurs prior to DNA hypermethylation. Laser capture dissection techniques were used to isolate single Bre-70 pre- and post-selection cells that were positive or negative for p16INK4A expression. (A) Bre-70 pre-selection cells expressing p16INK4A (positive staining) and small post-selection cells silenced for p16INK4A. (B) Selection of p16INK4A silenced cells for dissection. (C) Cells after dissection, showing specific isolation of p16INK4A silenced cells. (D) Bisulphite methylation clonal sequencing analysis of p16INK4A from single Bre-70 post-selection (p16INK4A silenced) and pre-selection cells (p16INK4A expressed) isolated by laser capture microscopy. Newly emerging post-selection HMECs that were negative for p16INK4A did not exhibit methylation. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site.

Figure 4.

p16INK4 silencing in HMECs occurs prior to DNA hypermethylation. Laser capture dissection techniques were used to isolate single Bre-70 pre- and post-selection cells that were positive or negative for p16INK4A expression. (A) Bre-70 pre-selection cells expressing p16INK4A (positive staining) and small post-selection cells silenced for p16INK4A. (B) Selection of p16INK4A silenced cells for dissection. (C) Cells after dissection, showing specific isolation of p16INK4A silenced cells. (D) Bisulphite methylation clonal sequencing analysis of p16INK4A from single Bre-70 post-selection (p16INK4A silenced) and pre-selection cells (p16INK4A expressed) isolated by laser capture microscopy. Newly emerging post-selection HMECs that were negative for p16INK4A did not exhibit methylation. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site.

p16INK4A hypermethylation originates from focal CpG sites and progressively spreads with proliferation

We performed a detailed analysis on post-selection cells derived from a single HMEC colony to ascertain if the heterogeneity in DNA methylation observed in post-selection HMECs was due to the fact that we had pooled individual colonies prior to passaging or if the heterogeneity reflected the inherent stochastic nature of de novo and or maintenance methylation. We isolated HMECs from a single Bre-80 colony at selection and cultured the cells for an additional 39 passages and analysed methylation at passages 1, 4, 10, 23, 26 and 39. Figure 5 shows the progressive expansion of DNA methylation we observed across region II of the p16INK4A CpG island. In the pre-selection Bre-80 cells, region II was essentially unmethylated with only a few single methylated CpG sites; however, by passage 1 after selection, ‘hot spots’ of methylation were already observed in some of the molecules. With increasing passage number, the density of methylation also progressively increased, and by passage 39 all molecules were hypermethylated in nearly 50% of CpG sites, with two discrete ‘hot spots’ flanking a region of CpG sites that was more resistant to methylation. Indeed, the overall methylation pattern generated from the single Bre-80 clone was remarkably similar to the patterns observed from the pooled post-selection colonies in each of different HMEC strains (for example Bre-60 and Bre-70) (Fig. 5B), suggesting a directive process.

Figure 5.

DNA hypermethylation of p16INK4A in a single Bre-80 colony increases and expands with successive passaging. (A) Bisulphite methylation clonal sequencing analysis of p16INK4A in a single Bre-80 colony at selection (top panel). A single Bre-80 colony was cultured until the end of its lifespan, and the DNA methylation status of p16INK4A determined at passage numbers 1, 4, 10, 23, 26 and 39 (relative to selection). Density of methylation of the CpG sites increases with increasing passage number. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation. (B) Quantitation of bisulphite clonal sequencing data for Bre-60 (p7), Bre-70 (p9) and Bre-80 (p23) post-selection cells, highlighting similar methylation patterns.

Figure 5.

DNA hypermethylation of p16INK4A in a single Bre-80 colony increases and expands with successive passaging. (A) Bisulphite methylation clonal sequencing analysis of p16INK4A in a single Bre-80 colony at selection (top panel). A single Bre-80 colony was cultured until the end of its lifespan, and the DNA methylation status of p16INK4A determined at passage numbers 1, 4, 10, 23, 26 and 39 (relative to selection). Density of methylation of the CpG sites increases with increasing passage number. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation. (B) Quantitation of bisulphite clonal sequencing data for Bre-60 (p7), Bre-70 (p9) and Bre-80 (p23) post-selection cells, highlighting similar methylation patterns.

Nucleosome occupancy of the p16INK4A CpG island mimics the post-selection HMEC methylation profile

The regularity of methylation ‘hot spots’ that we observed across the p16INK4A CpG island, in each of the different post-selection HMEC strains, prompted us to address if this signature correlates with the position of nucleosome repeat units. We used the new high-resolution foot-printing technique, known as MSPA (31), to determine the position of the nucleosomes across the three neighbouring regions of the p16INK4A CpG island promoter. The advantage of this method is that it provides an in vitro map of regions that are protected from the DNA methyltransferase enzyme. Nuclei were prepared from Bre-38 pre-selection HMECs and analysed for accessibility to M. Sss I CpG methylase. M. Sss I CpG methylase treatment of control naked genomic DNA was also performed as a control for the extent of enzyme activity. Figure 6A confirms that p16INK4A was essentially unmethylated in Bre-38 pre-selection HMECs and after M. Sss I treatment the control genomic DNA became extensively methylated (Fig. 6B). In contrast, when the Bre-38 pre-selection nuclei were treated with M. Sss I, the p16INK4A DNA molecules were methylated in three distinct patterns (Fig. 6C). Class I clones, exhibited two focal regions of methylation flanking a central region of protection from M. Sss I treatment, whereas class II and III clones appeared to be protected from methylation at one or other end of the p16INK4A amplicons examined. When the methylation pattern of the clones were analysed collectively from regions I, II and III, three discrete regions of ∼150 bp in length are shown to be more protected from M. Sss I treatment (Fig. 6D), consistent with the presence of three nucleosomes. However, in the region spanning the start of transcription, an extended region of in vitro M. Sss I methylation was observed, suggesting the loss of a nucleosome in the actively expressing Bre-38 pre-selection HMECs (Fig. 6D). Interestingly, the regions that were more susceptible to in vitro M. Sss I methylation of Bre-38 nuclei, correspond in general to the wave signature pattern of methylation observed in vivo in the Bre-38 post-selection HMECs, as summarized in Figure 6E. Conversely, the regions that were protected from M. Sss I methylation in vitro (Fig. 6D) correspond to the regions protected in vivo in the Bre-38 post-selection HMECs (Fig. 6E). In both cases, the region spanning the start of transcription was more susceptible to methylation, but in contrast to the in vitro M. Sss I methylation data, the in vivo methylation profile observed in the Bre-38 post-selection HMECs where p16INK4A expression was inactivated was more discrete, consistent with a gain of a nucleosome near the start of transcription (Fig. 6D,E). Supplementary Material, Figure S3 shows a comparative analysis of in vitro MSPA pre-selection Bre-38 data with the in vivo Bre-38 post-selection p16INK4A methylation patterns using lowess curves that were generated and smoothing applied over 10 data points. A clear correlation can be observed between the apparent nucleosome footprint generated using the MSPA technique and the DNA methylation wave signature pattern observed in the post-selection HMECs. However, after the start of transcription, the phasing of the methylation footprint, reflecting nucleosome positioning, appeared to be more compact in the post-selection cells, which is consistent with the inactive state of p16INK4A in these cells and de novo methylation in the nucleosome-free linker regions.

Figure 6.

MSPA of p16INK4A in Bre-38 pre-selection cells. Bisulphite methylation clonal sequencing analysis of p16INK4A in M. Sss I treated Bre-38 pre-selection cells was performed to ascertain whether nucleosomes influence p16INK4A methylation patterns in post-selection HMECs. (A) Untreated Bre-38 pre-selection cells. (B) M. Sss I treated control genomic DNA. (C) M. Sss I treated Bre-38 pre-selection nuclei. Methylation patterns were divided into three distinct classes; class I (top), class II (middle) and class III (bottom). (D) Quantitation of in vitro Bre-38 M. Sss I MSPA clonal data. (E) Quantitation of in vivo Bre-38 post-selection clonal methylation data from Fig. 2B. The methylation footprints at the single molecule level reveal ∼150 bp protected regions consistent with nucleosome units across each of the three regions. Proposed nucleosome units (∼150 bp) are indicated by green ovals, while nucleosome devoid regions are indicated by dotted white ovals. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation in Bre-38 post-selection cells (as shown in Fig. 2B).

Figure 6.

MSPA of p16INK4A in Bre-38 pre-selection cells. Bisulphite methylation clonal sequencing analysis of p16INK4A in M. Sss I treated Bre-38 pre-selection cells was performed to ascertain whether nucleosomes influence p16INK4A methylation patterns in post-selection HMECs. (A) Untreated Bre-38 pre-selection cells. (B) M. Sss I treated control genomic DNA. (C) M. Sss I treated Bre-38 pre-selection nuclei. Methylation patterns were divided into three distinct classes; class I (top), class II (middle) and class III (bottom). (D) Quantitation of in vitro Bre-38 M. Sss I MSPA clonal data. (E) Quantitation of in vivo Bre-38 post-selection clonal methylation data from Fig. 2B. The methylation footprints at the single molecule level reveal ∼150 bp protected regions consistent with nucleosome units across each of the three regions. Proposed nucleosome units (∼150 bp) are indicated by green ovals, while nucleosome devoid regions are indicated by dotted white ovals. CpG sites are numbered relative to the start of transcription. White circle, unmethylated CpG site; black circle, methylated CpG site. Red bars indicate focal ‘hot spots’ of DNA methylation in Bre-38 post-selection cells (as shown in Fig. 2B).

To determine if a similar nucleosome footprint was observed in breast-cancer cells, we performed MSPA on MDAMB453 cells, a breast-cancer cell line in which p16INK4A is also expressed and unmethylated. First, as for Bre-38 cells, regions I and II spanning the p16INK4A CpG island are essentially unmethylated in the expressing MDAMB453 cells (Supplementary Material, Fig. S4A). Secondly, when MDAMB453 nuclei were treated with M. Sss I, we also observed distinct focal regions of in vitro methylation in similar focal regions to the pre-selection Bre-38 M. Sss I treated nuclei (Supplementary Material, Fig. S4B). Thirdly, the M. Sss I methylation clonal pattern found in the MDAMB453 cells also mimics the in vivo wave pattern of methylation observed for the post-selection cells Bre-38 (Supplementary Material, Fig. S4C). The combined data supports the hypothesis that the nucleosome footprint dictates the de novo methylation signature pattern across the p16INK4A CpG island in both normal breast cells and breast-cancer cells.

p16INK4A epigenetic silencing in post-selection HMECs is associated with dynamic chromatin remodelling

To determine the relationship between p16INK4A silencing in post-selection HMECs, DNA methylation and chromatin remodelling, we performed chromatin immunoprecipitation (ChIP) assays with acetylated H3K9 (H3K9Ac), dimethylated H3K9 (H3K9Me2) and trimethylated H3K27 (H3K27Me3) antibodies to compare histone modifications in pre- and post-selection cells. p16INK4A-associated chromatin isolated from actively expressing pre-selection HMECs was enriched for active H3K9Ac marks and rapid deacetylation was observed in the post-selection cells at the first passage after selection which remained deacetylated with further passaging (Fig. 7A). In contrast, p16INK4A associated chromatin from pre-selection HMECs was depleted in the repressive H3K9Me2 mark, but after selection slowly accumulated histone methylation marks over several passages (Fig. 7B). Interestingly, p16INK4A-associated chromatin was also found to be enriched in the polycomb mark H3K27Me3 in pre-selection cells, and this was rapidly depleted in post-selection HMECs at the first passage after selection (Fig. 7C). Similar results were observed using different HMEC strains (Supplementary Material, Fig. S5). Together these data show that p16INK4A silencing at selection promotes a progressive increase in de novo CpG methylation in post-selection HMECs (Fig. 7D) and a parallel deacetylation of H3K9 and loss of trimethylated H3K27, whereas dimethylation of H3K9 occurs subsequently and gradually increases with DNA methylation accumulation.

Figure 7.

Interplay between accumulation of repressive chromatin and de novo CpG methylation associated with p16INK4A epigenetic silencing. Chromatin from pre- and post-selection HMECs was immunoprecipitated with (A) acetylated H3K9 (H3K9Ac), (B) dimethylated H3K9 (H3K9Me2) and (C) trimethylated H3K27 (H3K27Me3) antibodies. The amount of immunoprecipitated DNA (relative binding) was quantified by real-time PCR and was calculated as a ratio of immunoprecipitated DNA to the total amount of input. (D) Bisulphite clonal sequencing data for Bre-80 (as described in Fig. 5) was quantitated as average CpG methylation at each passage. A rapid deacetylation of H3K9 was associated with a loss of H3K27me3 and gain of DNA methylation followed by a slower gain of H3K9 methylation.

Figure 7.

Interplay between accumulation of repressive chromatin and de novo CpG methylation associated with p16INK4A epigenetic silencing. Chromatin from pre- and post-selection HMECs was immunoprecipitated with (A) acetylated H3K9 (H3K9Ac), (B) dimethylated H3K9 (H3K9Me2) and (C) trimethylated H3K27 (H3K27Me3) antibodies. The amount of immunoprecipitated DNA (relative binding) was quantified by real-time PCR and was calculated as a ratio of immunoprecipitated DNA to the total amount of input. (D) Bisulphite clonal sequencing data for Bre-80 (as described in Fig. 5) was quantitated as average CpG methylation at each passage. A rapid deacetylation of H3K9 was associated with a loss of H3K27me3 and gain of DNA methylation followed by a slower gain of H3K9 methylation.

DISCUSSION

The results presented in this paper provide for the first time a comprehensive analysis of the temporal steps involved in p16INK4A silencing, epigenetic reprogramming and nucleosome positioning in a model for early malignancy of breast cancer. HMECs isolated from disease-free breast tissue provide a uniquely informative system to study early events in breast tumourigenesis (reviewed in 33). Post-selection or variant HMECs exhibit many preneoplastic characteristics, including transcriptional silencing of the p16INK4A tumour suppressor gene and overexpression of cyclo-oxygenase 2 (Cox-2) (18–23) as well as epigenetic deregulation of the TGF-β pathway (30). Epigenetic deregulation of p16INK4A also occurs commonly in pre-malignant lesions and rare foci of morphologically normal epithelial cells exhibiting p16INK4A methylation have been identified in vivo in disease-free breast tissue (22). It has been postulated that these foci are cancer precursors, which can promote malignancy with additional epigenetic and or genetic changes (29).

One of the biggest challenges in dissecting the processes involved in epigenetic reprogramming in cancer is that the changes occur early in oncogenesis and are consequently difficult to study in clinical samples. We therefore used HMECs grown in serum-free conditions to study the early epigenetic changes that are associated with p16INK4A silencing. We and others have previously shown that p16INK4A silencing in post-selection cells that arise under these culture conditions is associated with DNA hypermethylation of the CpG island promoter and this occurs just as the first post-selection cells emerge from selection (18–23). We now demonstrate that DNA hypermethylation of p16INK4A CpG island occurs only after the gene is silent and that the initial ‘seeds’ of de novo methylation accumulate in the accessible regions of the chromosome and this is associated with a dynamic remodelling of the chromatin.

Over the years, there has been constant debate over the mechanism of epigenetic silencing of tumour suppressor genes in cancer (11,34,35). Central to this debate is the question of whether DNA methylation and histone modifications are a consequence or cause of gene silencing? A favoured dogma is that DNA methylation causes gene silencing and was demonstrated in early in vitro studies (36–38). However, more recently the hypothesis that DNA methylation is a consequence of gene silencing is gaining support. We have previously reported that transcriptional silencing of a genetically manipulated GSTP1 gene in LNCaP cells precedes DNA methylation (39), and the low level ‘seeding’ methylation observed in normal cells promotes H3K9 deacetylation and H3K9 methylation. In contrast, Bachman et al. 2003 (35) reported that silencing of p16INK4A and H3K9 methylation preceded DNA methylation also using an experimentally contrived situation. The advantage of the HMEC system is that it allows us to address the interplay between gene inactivation and epigenetic changes at the earliest time points after p16INK4A silencing in an in vivo setting. Using laser capture dissection of single cells clearly demonstrated that de novo methylation occurred post gene silencing, and analysis of a single colony showed that methylation was progressive rather than being a single aberrant event that encompasses the entire island, supporting previous findings (10,19). Even though the methylation of individual CpG sites was stochastic, it was clear that there was an overall similar pattern between different HMEC strains. Using the new single-molecule footprinting technique developed to visualize nucleosome occupancy at high resolution (31), we were able to demonstrate that the signature wave pattern of methylation observed in the post-selection cells was similar to the observed nucleosome footprint across the p16INK4A CpG island. Our results suggest that chromatin accessibility is dictating the initial access to the DNA methyltransferase enzyme, thereby only permitting aberrant de novo methylation to limited regions within the p16INK4A CpG island which then spreads progressively with PD. Our data also suggests that the nucleosomes are quite mobile in the actively expressing cells, and not precisely positioned along the p16INK4A promoter. This is not unexpected, as it is well known that nucleosomes are not a simple static unit; rather, they are a dynamic element that can slide along the DNA, thus permitting access or constraints to transcriptional machinery (3). Indeed, the transitory nature of nucleosomes was emphasized in a recent MSPA analysis of the GRP78 promoter during endoplasmic reticulum (ER) stress (40). Similar to our finding, silent promoters have generally been shown to be enriched for nucleosomes relative to their active counterparts (41), and nucleosome depletion at CpG islands has been described for several epigenetically regulated human gene promoters (31,42). For example, the methylated and silent MLH1 promoter was found to be occupied by three nucleosomes in RKO cells, which were evicted upon demethylation and activation of the promoter by 5-aza-2'-deoxycytidine (42).

The final implication of our work relates to chromatin remodelling that occurred in parallel with the de novo methylation. As summarized in Fig. 8, we found the p16INK4A CpG island in pre-selection cells was associated with the lack of a nucleosome at the start of transcription and bivalent histones consisting of the active H3K9Ac mark and the repressive polycomb EZH2-associated H3K27Me3 mark. After p16INK4A inactivation, the nucleosomes are remodelled with the gain of a nucleosome across the start of transcription, and a rapid deacetylation of H3K9 in concert with the rapid removal of the H3K27Me3 polycomb mark. In contrast, dimethylation of H3K9 accumulates more gradually in parallel with the accumulation of DNA methylation, resulting in consolidation of p16INK4A gene silencing. Our data supports recent findings that many genes including those involved in cell-fate determination, stem-cell renewal, cell growth and cell division are marked by polycomb in the normal cells, but are susceptible to aberrant DNA methylation in cancer cells (43,44). What remains unclear is the mechanism responsible for rapid change from the bivalent histone mode to one associated with DNA methylation and histone deacetylation and methylation. Our studies demonstrate that gene silencing in pre-malignancy is key to tipping the balance from an epigenetically plastic condition to one that becomes repressively locked.

Figure 8.

Summary of epigenetic silencing of p16INK4A in post-selection HMECs. In pre-selection HMECs, the p16INK4A transcription start site (TSS) is devoid of a nucleosome. The CpG island-associated promoter region is marked by DNA methylation ‘seeds’, there is active gene transcription and the chromatin is in a bivalent state as it is marked by both active (H3K9Ac) and repressive (H3K27Me3) histone modifications. As a consequence of stochastic gene silencing, p16INK4A undergoes epigenetic deregulation in post-selection HMECs through a DNA methylation-associated mechanism. In early post-selection cells, there is nucleosome gain across the TSS, which is accompanied by loss of H3K27Me3, deacetylation of H3K9, and evidence of DNA methylation expansion from ‘seeds’ within nucleosome linker regions. In late passage post-selection HMECs, there is consolidation of gene silencing through an enrichment of H3K9Me and extensive spreading and accumulation of DNA methylation.

Figure 8.

Summary of epigenetic silencing of p16INK4A in post-selection HMECs. In pre-selection HMECs, the p16INK4A transcription start site (TSS) is devoid of a nucleosome. The CpG island-associated promoter region is marked by DNA methylation ‘seeds’, there is active gene transcription and the chromatin is in a bivalent state as it is marked by both active (H3K9Ac) and repressive (H3K27Me3) histone modifications. As a consequence of stochastic gene silencing, p16INK4A undergoes epigenetic deregulation in post-selection HMECs through a DNA methylation-associated mechanism. In early post-selection cells, there is nucleosome gain across the TSS, which is accompanied by loss of H3K27Me3, deacetylation of H3K9, and evidence of DNA methylation expansion from ‘seeds’ within nucleosome linker regions. In late passage post-selection HMECs, there is consolidation of gene silencing through an enrichment of H3K9Me and extensive spreading and accumulation of DNA methylation.

MATERIALS AND METHODS

Cells and cell culture

Breast tissue removed from reduction mammoplasties was obtained with institutional Ethics Committee approval and informed donor consent. HMEC cultures were prepared from normal breast tissue for donor strains Bre-12, Bre-38, Bre-40, Bre-56, Bre-60, Bre-70 and Bre-80 as previously described (19,30). The pre-selection and post-selection HMECs were subcultured according to the protocol described in (19,24).

Quantitative real-time reverse transcription PCR

RNA was extracted from Bre-12, Bre-40, Bre-56, Bre-60 and Bre-80 pre- and post-selection HMECs using Trizol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. cDNA was reverse transcribed from 1 µg of total RNA using SuperScript III RNase H Reverse Transcriptase (Invitrogen) according to the manufacturer's protocol. The reaction was primed with 150 ng of random primers (Boehringer-Mannheim, Castle Hill, NSW, Australia). The reverse transcription reaction was diluted 1:10 with sterile H2O before addition to the reverse transcription PCR. p16INK4A expression primer sequences are described in Supplementary Material, Table S1. p16INK4A expression was quantitated using the 7900HT Applied Biosystems Sequence Detection System as described (30).

Western blotting

Western blot analysis using Bre-56 pre- [passage 4 (p4)] and post-selection (p12), and Bre-80 post-selection (p52) cells was performed essentially as described in (45). p16INK4A was detected with an anti-p16 (Ab-1) mouse monoclonal antibody (DSC-50.1/H4: Calbiochem/Merck KGaA, Darmstadt, Germany).

DNA isolation

DNA was isolated from less than 1 × 106 pre- and post-selection cells using simple lysis buffer (2 µg tRNA, 280 ng/μl proteinase K, 1% SDS) (46), or from 1–3 × 106 cells using either Puregene DNA isolation kit (Gentra Systems, Inc., Minneapolis, USA) according to the manufacturer's instructions, or with Trizol Reagent with the following modifications to the manufacturer's protocol: 300 µl of 100% ethanol and 20 µg of tRNA was added to the reserved organic phase, the samples inverted and incubated at room temperature for 3 min. Samples were centrifuged at 12 000 g for 15 min at 4°C. The supernatant was carefully removed, the DNA pellet resuspended in 18 µl of simple lysis buffer and the sample incubated at 55°C overnight prior to bisulphite conversion.

Laser capture microdissection

HMECs were stained for p16INK4A expression with a Envision DAKO kit as previously described (19). Single pre-selection Bre-70 cells that stained p16INK4A positive and post-selection Bre-70 cells that were p16INK4A negative, were isolated using the PALM Robot Microbeam laser microdissection system (P.A.L.M GmbH, Bernried, Germany) (47). Twenty cells for each cell type were captured in the tube cap in duplicate for two separate experiments and placed in 18 µl lysis buffer (100 mm Tris–HCl pH 8.0, 3% SDS, 50 mm EDTA, 200 µg/ml Proteinase K).

DNA methylation studies

Bisulphite genomic sequencing was used to analyse the methylation status of three neighbouring regions of p16INK4A in pre- and post-selection HMECs (Supplementary Material, Fig. S1). The bisulphite reaction was carried out on extracted DNA for 16 h at 55°C on up to 2 µg of digested DNA, under conditions described previously (48,49). Laser-captured cells were incubated in 18 µl of DNA lysis buffer for 30 mins at 37°C prior to bisulphite treatment for 4 h, as described (49). After bisulphite conversion, the DNA was ethanol precipitated, dried, resuspended in 10–50 µl H2O and stored at −20°C. Triplicate PCR amplifications were performed for p16INK4A and pooled. The primer sequences and location of the p16INK4A amplicons in relation to the CpG island and start of transcription are summarized in Supplementary Material, Table S1 and Supplementary Material, Figure S1. The methylation status of p16INK4A was determined by bisulphite clonal sequencing of the pooled PCR products, as described (50) to ensure representative clonal analysis. Example DNA sequence traces are shown in Supplementary Material, Fig. S6.

Methylase-based single-promoter analysis assays

Nucleosome positioning assays were performed by a high-resolution, MSPA, essentially as described in (31,42) and with the following modifications. Actively growing Bre-38 pre-selection HMECs, or MDAMB453 breast cancer cells, were trypsinized and washed twice with cold PBS. Cells were resuspended in 1 ml of cold RSB buffer (10 mm Tris–HCl; pH 7.4, 10 mm NaCl and 3 mm MgCl2) per 10 × 106 cells and incubated on ice for 10 min. One microlitre of 10% NP-40 detergent per 10 × 106 cells was added and the cells were homogenized at 4°C with the tight pestle in a glass dounce homogenizer for at least 15 strokes. Nuclei were washed with 1 ml of cold RSB buffer and resuspended in 74.25 µl of 1× M. Sss I buffer+sucrose per 1 × 106 nuclei. 1 × 106 nuclei were treated with 60 units of M. Sss I (New England BioLabs, Inc., Beverly, MA, USA) in a final volume of 150 µl for 15 min at 37°C. M. Sss I treated purified control gDNA (6 µg) was used as a positive control. Reactions were stopped by the addition of an equal volume of stop solution (20 mm Tris–HCl; pH 7.9, 600 mm NaCl, 1% SDS and 10 mm EDTA) and proteinase K treatment for 48 h at 37°C. DNA was purified by phenol chloroform extraction, ethanol precipitated and resuspended in water. One microgram of sheared M. Sss I treated nuclei or control gDNA was bisulphite converted for 16 h at 55°C, under conditions previously described (48). After bisulphite conversion, the DNA was ethanol precipitated, dried, resuspended in 50 µl water and stored at −20°C. Bisulphite genomic sequencing was used to analyse the methylation status of individual molecules modified by M. Sss I across the three neighbouring regions of the p16INK4A CpG island associated promoter region described in Supplementary Material, Figure S1. Lowess curves were generated and smoothing applied over 10 data points in order to highlight similarities between the methylation patterns seen for the in vitro MSPA data with the in vivo Bre-38 post-selection methylation data.

Chromatin immunoprecipitation assays

Chromatin immunoprecipitation assays were carried out according to the manufacturer's instructions (Upstate/Millipore, Temecula, CA, USA) using pooled Bre-60 and Bre-70 pre-selection cedllsd, Bre-80 post-selection ceslls, Bre-12 pre- and post-selection cells, Bre-40 pre- and post-selection cells and Bre-38 pre- and post-selection cells, essentially as described (30). The eluted complexes were immunoprecipitated with antibodies specific for acetylated H3K9 (H3K9Ac, Upstate/Millipore), dimethylated H3K9 (H3K9Me2, Upstate/Millipore) and trimethylated H3K27 (H3K27Me3, Upstate/Millipore) and DNA yield was measured by quantitative real-time PCR as described (30). Chromatin immunoprecipitation amplification primers are described in Supplementary Material, Table S1.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work is supported by Cancer Institute NSW (CI NSW), National Breast Cancer Foundation (NBCF) and National Health and Medical Research Council (NH&MRC) project grants; Dora Lush Biomedical Postgraduate Scholarship from the NH&MRC, CI NSW Research Scholar Award and NBCF Excellence Award (R.A.H.); and NH&MRC fellowships (S.J.C. and R.R.R.).

ACKNOWLEDGEMENTS

We would like to thank Marcel Coolen for critical reading of the manuscript and help with figures, Christopher Molloy and Jane Noble for experimental assistance and Professors Thea Tlsty and Peter Jones for technical advice.

Conflict of Interest statement. None declared.

REFERENCES

1
Kornberg
R.D.
Chromatin structure: a repeating unit of histones and DNA
Science
 , 
1974
, vol. 
184
 (pg. 
868
-
871
)
2
Schones
D.E.
Cui
K.
Cuddapah
S.
Roh
T.Y.
Barski
A.
Wang
Z.
Wei
G.
Zhao
K.
Dynamic regulation of nucleosome positioning in the human genome
Cell
 , 
2008
, vol. 
132
 (pg. 
887
-
898
)
3
Li
B.
Carey
M.
Workman
J.L.
The role of chromatin during transcription
Cell
 , 
2007
, vol. 
128
 (pg. 
707
-
719
)
4
Baylin
S.B.
Esteller
M.
Rountree
M.R.
Bachman
K.E.
Schuebel
K.
Herman
J.G.
Aberrant patterns of DNA methylation, chromatin formation and gene expression in cancer
Hum. Mol. Genet.
 , 
2001
, vol. 
10
 (pg. 
687
-
692
)
5
Baylin
S.B.
DNA methylation and gene silencing in cancer
Nat. Clin. Pract. Oncol.
 , 
2005
, vol. 
2
 
Suppl 1.
(pg. 
S4
-
S11
)
6
Jones
P.A.
Overview of cancer epigenetics
Semin. Hematol.
 , 
2005
, vol. 
42
 (pg. 
S3
-
S8
)
7
Gal-Yam
E.N.
Saito
Y.
Egger
G.
Jones
P.A.
Cancer epigenetics: modifications, screening, and therapy
Annu. Rev. Med.
 , 
2008
, vol. 
59
 (pg. 
267
-
280
)
8
Gardiner-Garden
M.
Frommer
M.
CpG islands in vertebrate genomes
J. Mol. Biol.
 , 
1987
, vol. 
196
 (pg. 
261
-
282
)
9
Bird
A.P.
CpG-rich islands and the function of DNA methylation
Nature
 , 
1986
, vol. 
321
 (pg. 
209
-
213
)
10
Wong
D.J.
Foster
S.A.
Galloway
D.A.
Reid
B.J.
Progressive region-specific de novo methylation of the p16 CpG island in primary human mammary epithelial cell strains during escape from M(0) growth arrest
Mol. Cell. Biol.
 , 
1999
, vol. 
19
 (pg. 
5642
-
5651
)
11
Clark
S.J.
Melki
J.
DNA methylation and gene silencing in cancer: which is the guilty party?
Oncogene
 , 
2002
, vol. 
21
 (pg. 
5380
-
5387
)
12
Sherr
C.J.
The ink4a/arf network in tumour suppression
Nat. Rev. Mol. Cell. Biol.
 , 
2001
, vol. 
2
 (pg. 
731
-
737
)
13
Rocco
J.W.
Sidransky
D.
p16(MTS-1/CDKN2/INK4a) in cancer progression
Exp. Cell. Res.
 , 
2001
, vol. 
264
 (pg. 
42
-
55
)
14
Herman
J.G.
p16(INK4): involvement early and often in gastrointestinal malignancies
Gastroenterology
 , 
1999
, vol. 
116
 (pg. 
483
-
485
)
15
Belinsky
S.A.
Nikula
K.J.
Palmisano
W.A.
Michels
R.
Saccomanno
G.
Gabrielson
E.
Baylin
S.B.
Herman
J.G.
Aberrant methylation of p16(INK4a) is an early event in lung cancer and a potential biomarker for early diagnosis
Proc. Natl Acad. Sci. USA
 , 
1998
, vol. 
95
 (pg. 
11891
-
11896
)
16
Nuovo
G.J.
Plaia
T.W.
Belinsky
S.A.
Baylin
S.B.
Herman
J.G.
In situ detection of the hypermethylation-induced inactivation of the p16 gene as an early event in oncogenesis
Proc. Natl Acad. Sci. USA
 , 
1999
, vol. 
96
 (pg. 
12754
-
12759
)
17
Noble
J.R.
Rogan
E.M.
Neumann
A.A.
Maclean
K.
Bryan
T.M.
Reddel
R.R.
Association of extended in vitro proliferative potential with loss of p16INK4 expression
Oncogene
 , 
1996
, vol. 
13
 (pg. 
1259
-
1268
)
18
Foster
S.A.
Wong
D.J.
Barrett
M.T.
Galloway
D.A.
Inactivation of p16 in human mammary epithelial cells by CpG island methylation
Mol. Cell. Biol.
 , 
1998
, vol. 
18
 (pg. 
1793
-
1801
)
19
Huschtscha
L.I.
Noble
J.R.
Neumann
A.A.
Moy
E.L.
Barry
P.
Melki
J.R.
Clark
S.J.
Reddel
R.R.
Loss of p16INK4 expression by methylation is associated with lifespan extension of human mammary epithelial cells
Cancer Res.
 , 
1998
, vol. 
58
 (pg. 
3508
-
3512
)
20
Brenner
A.J.
Stampfer
M.R.
Aldaz
C.M.
Increased p16 expression with first senescence arrest in human mammary epithelial cells and extended growth capacity with p16 inactivation
Oncogene
 , 
1998
, vol. 
17
 (pg. 
199
-
205
)
21
Romanov
S.R.
Kozakiewicz
B.K.
Holst
C.R.
Stampfer
M.R.
Haupt
L.M.
Tlsty
T.D.
Normal human mammary epithelial cells spontaneously escape senescence and acquire genomic changes
Nature
 , 
2001
, vol. 
409
 (pg. 
633
-
637
)
22
Holst
C.R.
Nuovo
G.J.
Esteller
M.
Chew
K.
Baylin
S.B.
Herman
J.G.
Tlsty
T.D.
Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia
Cancer Res.
 , 
2003
, vol. 
63
 (pg. 
1596
-
1601
)
23
Crawford
Y.G.
Gauthier
M.L.
Joubel
A.
Mantei
K.
Kozakiewicz
K.
Afshari
C.A.
Tlsty
T.D.
Histologically normal human mammary epithelia with silenced p16(INK4a) overexpress COX-2, promoting a premalignant program
Cancer Cell
 , 
2004
, vol. 
5
 (pg. 
263
-
273
)
24
Hammond
S.L.
Ham
R.G.
Stampfer
M.R.
Serum-free growth of human mammary epithelial cells: rapid clonal growth in defined medium and extended serial passage with pituitary extract
Proc. Natl Acad. Sci. USA
 , 
1984
, vol. 
81
 (pg. 
5435
-
5439
)
25
Stampfer
M.R.
Bartley
J.C.
Induction of transformation and continuous cell lines from normal human mammary epithelial cells after exposure to benzo[a]pyrene
Proc. Natl Acad. Sci. USA
 , 
1985
, vol. 
82
 (pg. 
2394
-
2398
)
26
Sandhu
C.
Donovan
J.
Bhattacharya
N.
Stampfer
M.
Worland
P.
Slingerland
J.
Reduction of Cdc25A contributes to cyclin E1-Cdk2 inhibition at senescence in human mammary epithelial cells
Oncogene
 , 
2000
, vol. 
19
 (pg. 
5314
-
5323
)
27
Stampfer
M.R.
Yaswen
P.
Casto
B.C.
Schuler
C.F.
Transformation of Human Epithelial Cells: Moleculare and Oncogene Mechanisms
 , 
1992
Boca Raton
CRC Press Inc.
28
Tlsty
T.D.
Romanov
S.R.
Kozakiewicz
B.K.
Holst
C.R.
Haupt
L.M.
Crawford
Y.G.
Loss of chromosomal integrity in human mammary epithelial cells subsequent to escape from senescence
J. Mammary Gland Biol. Neoplasia
 , 
2001
, vol. 
6
 (pg. 
235
-
243
)
29
Tlsty
T.D.
Crawford
Y.G.
Holst
C.R.
Fordyce
C.A.
Zhang
J.
McDermott
K.
Kozakiewicz
K.
Gauthier
M.L.
Genetic and epigenetic changes in mammary epithelial cells may mimic early events in carcinogenesis
J. Mammary Gland Biol. Neoplasia
 , 
2004
, vol. 
9
 (pg. 
263
-
274
)
30
Hinshelwood
R.A.
Huschtscha
L.I.
Melki
J.
Stirzaker
C.
Abdipranoto
A.
Vissel
B.
Ravasi
T.
Wells
C.A.
Hume
D.A.
Reddel
R.R.
, et al.  . 
Concordant epigenetic silencing of transforming growth factor-beta signaling pathway genes occurs early in breast carcinogenesis
Cancer Res.
 , 
2007
, vol. 
67
 (pg. 
11517
-
11527
)
31
Fatemi
M.
Pao
M.M.
Jeong
S.
Gal-Yam
E.N.
Egger
G.
Weisenberger
D.J.
Jones
P.A.
Footprinting of mammalian promoters: use of a CpG DNA methyltransferase revealing nucleosome positions at a single molecule level
Nucleic Acids Res.
 , 
2005
, vol. 
33
 pg. 
e176
 
32
Kladde
M.P.
Simpson
R.T.
Chromatin structure mapping in vivo using methyltransferases
Methods Enzymol.
 , 
1996
, vol. 
274
 (pg. 
214
-
233
)
33
Hinshelwood
R.A.
Clark
S.J.
Breast cancer epigenetics: normal human mammary epithelial cells as a model system
J. Mol. Med
 , 
2008
34
Baylin
S.
Bestor
T.H.
Altered methylation patterns in cancer cell genomes: cause or consequence?
Cancer Cell
 , 
2002
, vol. 
1
 (pg. 
299
-
305
)
35
Bachman
K.E.
Park
B.H.
Rhee
I.
Rajagopalan
H.
Herman
J.G.
Baylin
S.B.
Kinzler
K.W.
Vogelstein
B.
Histone modifications and silencing prior to DNA methylation of a tumor suppressor gene
Cancer Cell
 , 
2003
, vol. 
3
 (pg. 
89
-
95
)
36
Stein
R.
Razin
A.
Cedar
H.
In vitro methylation of the hamster adenine phosphoribosyltransferase gene inhibits its expression in mouse L cells
Proc. Natl Acad. Sci. USA
 , 
1982
, vol. 
79
 (pg. 
3418
-
3422
)
37
Busslinger
M.
Hurst
J.
Flavell
R.A.
DNA methylation and the regulation of globin gene expression
Cell
 , 
1983
, vol. 
34
 (pg. 
197
-
206
)
38
Yisraeli
J.
Frank
D.
Razin
A.
Cedar
H.
Effect of in vitro DNA methylation on beta-globin gene expression
Proc. Natl Acad. Sci. USA
 , 
1988
, vol. 
85
 (pg. 
4638
-
4642
)
39
Stirzaker
C.
Song
J.Z.
Davidson
B.
Clark
S.J.
Transcriptional gene silencing promotes DNA hypermethylation through a sequential change in chromatin modifications in cancer cells
Cancer Res.
 , 
2004
, vol. 
64
 (pg. 
3871
-
3877
)
40
Gal-Yam
E.N.
Jeong
S.
Tanay
A.
Egger
G.
Lee
A.S.
Jones
P.A.
Constitutive nucleosome depletion and ordered factor assembly at the GRP78 promoter revealed by single molecule footprinting
PLoS. Genet.
 , 
2006
, vol. 
2
 pg. 
e160
 
41
Mito
Y.
Henikoff
J.G.
Henikoff
S.
Genome-scale profiling of histone H3.3 replacement patterns
Nat. Genet.
 , 
2005
, vol. 
37
 (pg. 
1090
-
1097
)
42
Lin
J.C.
Jeong
S.
Liang
G.
Takai
D.
Fatemi
M.
Tsai
Y.C.
Egger
G.
Gal-Yam
E.N.
Jones
P.A.
Role of nucleosomal occupancy in the epigenetic silencing of the MLH1 CpG island
Cancer Cell
 , 
2007
, vol. 
12
 (pg. 
432
-
444
)
43
Ohm
J.E.
McGarvey
K.M.
Yu
X.
Cheng
L.
Schuebel
K.E.
Cope
L.
Mohammad
H.P.
Chen
W.
Daniel
V.C.
Yu
W.
, et al.  . 
A stem cell-like chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing
Nat. Genet.
 , 
2007
, vol. 
39
 (pg. 
237
-
242
)
44
Widschwendter
M.
Fiegl
H.
Egle
D.
Mueller-Holzner
E.
Spizzo
G.
Marth
C.
Weisenberger
D.J.
Campan
M.
Young
J.
Jacobs
I.
, et al.  . 
Epigenetic stem cell signature in cancer
Nat. Genet.
 , 
2007
, vol. 
39
 (pg. 
157
-
158
)
45
Huschtscha
L.I.
Neumann
A.A.
Noble
J.R.
Reddel
R.R.
Effects of simian virus 40 T-antigens on normal human mammary epithelial cells reveal evidence for spontaneous alterations in addition to loss of p16(INK4a) expression
Exp. Cell. Res.
 , 
2001
, vol. 
265
 (pg. 
125
-
134
)
46
Millar
D.S.
Warnecke
P.M.
Melki
J.R.
Clark
S.J.
Methylation sequencing from limiting DNA: embryonic, fixed, and microdissected cells
Methods
 , 
2002
, vol. 
27
 (pg. 
108
-
113
)
47
Micke
P.
Ostman
A.
Lundeberg
J.
Ponten
F.
Laser-assisted cell microdissection using the PALM system
Methods Mol. Biol.
 , 
2005
, vol. 
293
 (pg. 
151
-
166
)
48
Clark
S.J.
Harrison
J.
Paul
C.L.
Frommer
M.
High sensitivity mapping of methylated cytosines
Nucleic Acids Res.
 , 
1994
, vol. 
22
 (pg. 
2990
-
2997
)
49
Clark
S.J.
Statham
A.
Stirzaker
C.
Molloy
P.L.
Frommer
M.
DNA methylation: bisulphite modification and analysis
Nat. Protoc.
 , 
2006
, vol. 
1
 (pg. 
2353
-
2364
)
50
Frigola
J.
Song
J.
Stirzaker
C.
Hinshelwood
R.A.
Peinado
M.A.
Clark
S.J.
Epigenetic remodeling in colorectal cancer results in coordinate gene suppression across an entire chromosome band
Nat. Genet.
 , 
2006
, vol. 
38
 (pg. 
540
-
549
)