Purpose: Genetic alterations were previously identified in normal epithelia adjacent to invasive cancers. The aim of this study was to determine DNA methylation in histologically normal tissues from multiple geographic zones adjacent to primary breast tumors.

Experimental Design: First, methylation status of a 4-kb region of RASSF1A promoter was interrogated using oligonucleotide-based microarray in 144 samples (primary tumors, 47; adjacent normals, 69; reduction mammoplasty tissues, 28). Second, allelic imbalance (AI)/loss of heterozygosity (LOH) surrounding RASSF1A promoter were analyzed in 30 samples (tumors, 8; adjacent normals, 22). Third, global methylation screening of 49 samples (tumors, 12; adjacent normals, 25; reduction mammoplasty, 12) was done by differential methylation hybridization. Real-time quantitative methylation-specific PCR was used to validate the microarray findings.

Results: DNA methylation in the core RASSF1A promoter was low in reduction mammoplasty tissues (P = 0.0001) when compared with primary tumors. The adjacent normals had an intermediate level of methylation. The regions surrounding the core were highly methylated in all sample types. Microsatellite markers showed AI/LOH in tumors and some of the adjacent normals. Concurrent AI/LOH and DNA methylation in RASSF1A promoter occurred in two of six tumors. Global methylation screening uncovered genes more methylated in adjacent normals than in reduction mammoplasty tissues. The methylation status of four genes was confirmed by quantitative methylation-specific PCR.

Conclusions: Our findings suggest a field of methylation changes extending as far as 4 cm from primary tumors. These frequent alterations may explain why normal tissues are at risk for local recurrence and are useful in disease prognostication.

Historical data indicate that when lumpectomies were done as definitive therapy without adjuvant radiotherapy, the recurrence rates for cancer in the ipsilateral breast could be as high as 40% despite apparent negative pathologic margins (1). Introduction of adjuvant radiotherapy, however, reduces the rate of local recurrence to ≤10%, making breast conservation a comparable alternative to modified radical mastectomy (2). The natural history of breast cancer with its localized patterns of recurrence strongly suggests that the residual “normal” breast is in fact at risk for harboring occult, precancerous cells. To date, the degree to which the histologically normal breast is affected by premalignant changes has not been determined, nor has there been a quantitative and geographic means to assess and map this risk. In addition, the geographic extent of premalignant change throughout the breast is not known.

Previous studies have confirmed that genetic alterations exist in histologically normal breast tissues immediately adjacent to invasive cancers (3). Additional studies have suggested a preponderance of alterations in the upper outer quadrant of the breast where the majority of cancers occur (4). These normal epithelial cells frequently harbor loss of heterozygosity (LOH; refs. 3, 57), microsatellite and chromosomal instability (8), and gene mutations (9, 10). In addition to genetic changes, altered DNA methylation has recently been implicated in preneoplastic lesions associated with various tumor types (1113). This epigenetic alteration, by adding a methyl group to the cytosine residues of CpG dinucleotides, is frequently observed in 1- to 2-kb CpG islands located at the promoter/first exon of genes (14). Promoter hypermethylation contributes to the reorganization of chromatin structure and plays a role in the transcriptional inactivation in cancer (15). Epigenetic dysregulation is shown to be responsible for the silencing of genes that encode cellular proteins responsible for growth suppression, genomic stability and repair, apoptosis, and cell cycle regulation in many tumor types (15).

We have recently reported that increased methylation density within promoter CpG islands can be linked to tumor progression (11, 16). The prevalence of this epigenetic event occurring in histologically normal epithelium surrounding the tumor sites remains to be determined. Because the density of DNA methylation can be progressively accumulated during neoplastic transformation from premalignant lesions to invasive cancers (12, 1719), we have hypothesized that frequent promoter hypermethylation of tumor suppressor loci in normal breast tissue is a harbinger for local recurrence.

The goal of this study is to identify epigenetic biomarkers frequently hypermethylated in normal breast tissue immediately adjacent to, and up to 4 cm from, tumor sites. We evaluated four zones of normal breast tissue in the ipsilateral and contralateral breasts of patients undergoing unilateral or bilateral mastectomy for invasive breast cancer. The previously developed methylation-specific oligonucleotide (MSO; refs. 20, 21) microarray was used to comprehensively map a 4-kb promoter/first exon region of RASSF1A in a large series of breast tumors, adjacent normal tissues, and breast samples of unrelated healthy individuals. The promoter/first exon of this gene is densely methylated in primary breast tumors (11), but our present study further shows that this event may be initiated via methylation spreading from outer flanks to the promoter/first exon core in normal adjacent tissues. The increased presence of epigenetic alterations in adjacent tissues may signify a greater risk of local recurrence. We therefore extended the study to identify novel loci frequently hypermethylated in normal epithelia adjacent to cancer sites by a second microarray technique, differential methylation hybridization (22, 23). This microarray study suggests that a more extensive field of epigenetically altered breast epithelia is present in breast cancer patients than previously thought. The identified epigenetic markers are useful for predicting future local recurrence.

Tissue collection, histology, and DNA extraction. Tissues were collected in accordance with the protocols approved by the Institutional Review Boards of the University of South Florida and the Ohio State University. Breast tissues from patients that underwent mastectomy at various stages of breast carcinoma were collected. Mastectomy specimens were marked for tumor and its four surrounding zones (1, 2, 3, and 4 cm away from the grossly visible tumor boundary). Samples from each zone were collected and frozen in liquid nitrogen. In bilateral and prophylactic mastectomy cases, normal breast tissues were taken from the four quadrants of the breast. Tissues were also obtained from healthy individuals undergoing breast reduction mammoplasty. Samples were embedded in Tissue-Tek optimum cutting temperature compound and 5-μm sections were cut and mounted on Mercedes Platinum StarFrost Adhesive slides. The slides were stained using a standard H&E protocol and the tissue boundaries were marked. Using the marked slide as a “map,” tissues were then microdissected. Adipose tissues were trimmed and the tumor and normal tissues were separated and stored in liquid nitrogen. Histologic examination of all tissue sections and microdissection of samples were conducted in close collaboration with a pathologist to ensure consistency in the clinical diagnoses. To ensure that the DNA methylation observed in the tumor-associated normals was contributed mainly by the epithelial components, the marked sections of H&E-stained slides from the adjacent normal sections were further quantified for infiltrating inflammatory cells by careful microscopic examination.

DNA was isolated from breast tissues using the Puregene DNA Purification method (Gentra Systems, Inc., Minneapolis, MN). Briefly, tissues were ground in liquid nitrogen, resuspended in 10 mL of lysis buffer, and incubated for 20 minutes at room temperature. RNase was added (50 μL of 10 mg/mL) and incubated at 37° C for 30 minutes. Proteinase K (100 μL of 10 mg/mL) was added and incubated at 55°C for 1 hour. Then, samples were added with 3 mL of protein precipitation solution and mixed well for 3 minutes. The samples were centrifuged at 11,500 × g for 15 minutes at 4°C and the supernatants were collected. Genomic DNA was precipitated by adding 10 mL of isopropanol and mixing gently. DNA was pelleted by centrifuging at 10,000 × g for 10 minutes. The pellet was washed with 75% ethanol and resuspended in 100 μL of deionized water. The amount of DNA was quantified by measuring A260. Due to limited amount of material available from microdissected tissues, different sets of samples were used for some of the analyses described below.

MSO microarray. Preparation of MSOs was carried out as described (11, 21). These oligonucleotides were used to interrogate bisulfite-treated DNA covering the various regions of RASSF1A. Of note, the oligonucleotides used in the promoter/first exon region are the same as those previously used. Each oligonucleotide was suspended in microspotting solution (Telechem, Sunnyvale, CA) and printed in triplicate as microdots on the superaldehyde-coated glass slides (Telechem) using a GMS 417 microarray. Slide processing was done following the protocol of the manufacturer. For DNA target preparation, bisulfite-treated DNA was amplified by PCR from regions 1 to 5 located in the RASSF1A promoter/first exon region and its flanking area. The primer sequences and PCR conditions are available on request. Purified PCR products were labeled at the 3′ terminus with Cy5-dCTP (Amersham Pharmacia, Piscataway, NJ) by terminal transferase (New England Biolabs, Beverly, MA). These labeled products were resuspended in the Unihybridization Solution (Telechem), denatured at 95°C for 5 minutes, and applied to MSO slides. Hybridization was conducted in a moist chamber at 50°C for 4 to 5 hours. Slides were rinsed and washed once at 50°C with 2× SSC-0.2% SDS for a total of 15 minutes, followed by washing once with 2× SSC at room temperature for 5 minutes and drying by centrifugation at 500 rpm for 5 minutes. These microarray slides were scanned with a GenePix 4000A scanner (Axon Instruments, Union City, CA). The fluorescent images were analyzed with GenePix Pro 6.0 software. Signal intensities of individual spots were obtained and exported to Excel spreadsheets for further analysis. MSO data for each probe pair were first normalized according to their standard curve. The normalized signals were used to derive the intensity ratio of M/(M + U) (M, methylated allele; U, unmethylated allele) for each probe set. Statistical analyses were conducted using SigmaStat 2.0 software (Jandel Scientific, San Rafael, CA).

Allelic imbalance/LOH analysis. Microsatellite markers for allelic imbalance (AI)/LOH analysis were selected from Genethon and Marshfield genetic maps according to their physical location on the chromosome 3p and analyzed as previously described (24). Twenty-nine polymorphic microsatellite markers were selected to span the entire short arm of chromosome 3 from 1 to 87.5 Mb (telomere to centromere). The average intermarker distance is 2.98 Mb. However, a denser coverage was implemented for 3p21.3, in which RASSF1A lies, using six microsatellite markers between 47 and 54 Mb with an average intermarker distance of 1.17 Mb.

After standard PCR, the amplicons were separated by capillary electrophoresis and the signal was detected with an Applied Biosystems semiautomated DNA sequencer 3700 (Applied Biosystems, Perkin-Elmer Corp., Norwalk, CT). The results were analyzed by automated fluorescence detection using the GeneScan collection and analysis software (GeneScan, Applied Biosystems, Perkin-Elmer). Scoring of AI/LOH was done by inspection of the GeneScan analysis output. Normal breast tissue from the contralateral breast or tissue of greatest distance to the tumor (which corresponds to normal breast tissue obtained at least 4 cm away from the tumor margin) was used as the normal control tissue (corresponding germ line). A ratio of peak heights of alleles between normal tissue and tumor DNA of 0.67 was employed to define AI/LOH as described by others and us previously (2527). Data were also analyzed by a more stringent cutoff of 0.5 to ensure the AI/LOH findings.

Differential methylation hybridization using CpG island microarray. Differential methylation hybridization was done according to the detailed protocol published by Yan et al. (16), with a slight modification of using McrBC (New England Biolabs), a methylaion-cutting enzyme, to interrogate DNA samples. Briefly, DNA isolated from various tissue types was digested with MseI, ligated to linkers, and half of each sample was digested with McrBC and the other half was mock digested without the enzyme and the cofactor GTP. The digested and mock-digested, linker-ligated DNAs were used as templates for PCR amplification. The amplified products were tagged with fluorescent dyes using the indirect labeling method. The mock-digested samples were coupled to Cy5 and the digested samples were coupled to Cy3. The two halves of each sample were combined and cohybridized to the CpG 12K array from the University Health Network Microarray Centre (Toronto, Ontario, Canada). The identity of the arrayed CpG island clones can be accessed from an online CGI Library Browser8

and a detailed description of this array was recently presented by Heisler et al. (28). The dye-coupled DNA was hybridized and washed as previously described (22). The washed slides were scanned with a GenePix 4000A scanner (Axon Instruments) and the acquired microarray images were analyzed with GenePix 6.0 software. The microarray data will be deposited in GEO9 for public access.

Microarray data analysis. The quality of differential methylation hybridization experiments was assessed with ArrayQuality (an R package, part of the Bioconductor open sources software).10

Data from good quality experiments were background corrected and normalized using the mitochondrial clones as described by Nouzova et al. (29), as well as clones without the methylation-cutting restriction sites. Targets differentially methylated between breast tumors, histologically normal tissues adjacent to tumors, and uninvolved normal breast tissues were selected using the Prediction Analysis for Microarray. By using the “nearest shrunken centroid method” as described by Tibshirani et al. (30), a list of significant genes of which the methylation status characterized each tissue type was generated. A subset of these genes is listed in Supplementary Table S1.

Bisulfite conversion and real-time quantitative methylation-specific PCR. Approximately 0.5 μg of genomic DNA was bisulfite modified with EZ DNA Methylation Kit (Zymo Research, Orange, CA). The modified DNA was subjected to real-time quantitative methylation-specific PCR to determine the methylation levels of studied genes using SYBR Green–based detection technology (Applied Biosystems, Foster City, CA). To normalize the amount of methylated DNA among test samples, total input DNA was gauged by a separate PCR amplification using primers devoid of CpG dinucleotides targeting a region of COL2A1. Primer sequences for test genes and COL2A1, as well as PCR conditions, are available on request. The amount of methylated DNA was determined by the threshold cycle number for each sample against a standard curve generated by SssI-treated DNA (Chemicon, Temecula, CA; 30-0.1 ng; R2 > 0.99) and expressed as the ratio of the methylated DNA to COL2A1 DNA in the same sample.

Statistical analyses. Statistical analyses (other than those described previously in the microarray method sections) were done using the SigmaPlot software (version 9.0). The hypermethylation differences among sample types were determined by t statistics. The difference was considered significant when P < 0.05.

Loss of methylation demarcation in the RASSF1A promoter/first exon in primary breast tumors. We conducted methylation mapping on a 4-kb region (2 kb upstream and downstream from the transcription start site) of the RASSF1A gene in breast tumors using MSO microarray (Fig. 1A). A panel of 88 oligonucleotides (17- to 25-mer) were designed and arrayed on microscopic slides for microarray hybridization (Fig. 1B). Each pair was used to evaluate the methylation status of one to three CpG sites in a stretch of DNA (20). Standardization curves, generated from a series of SssI-treated control samples (0%, 33%, 66%, and 100% methylation), were used to calculate the degrees of methylation in the interrogating sequences. Of note, the oligonucleotides used to interrogate the methylation status of RASSF1A promoter/first exon were the same used in a previous study. As such, the sensitivity and the effectiveness of this subset of oligonucleotides in quantifying methylation status had been verified by COBRA analysis and bisulfite sequencing in breast tumors, normal breast tissues, and breast cancer cell lines. Figure 1C shows a detailed methylation mapping of the 4-kb region in 23 invasive ductal carcinomas and 17 control breast tissues of healthy individuals. The outer flanking regions (regions 1 and 5) are heavily methylated (>80% methylation) in tumor and control samples. Approaching the immediate flanking regions (regions 2 and 4) of the RASSF1A promoter/first exon core, the overall levels (∼40% on average) of methylation in these zones is decreased relative to those of regions 1 and 5. This trend becomes significant (P = 0.0001) in the RASSF1A core (region 3), whereas high-level methylation (40-90%) is present in 74% (16 of 23 samples) of primary tumors analyzed and low-level methylation (∼20%) is present in control samples. These MSO results show that the sharp demarcation in methylation level along the border of the RASSF1A core and its outer flanks is lost in the majority of primary tumors.

Fig. 1.

MSO microarray analysis of DNA samples obtained from primary breast tumors and normal tissues from breast reduction mammoplasty surgery. A, genomic map of the RASSF1A promoter/first exon and the nearby regions. Vertical lines, positions of CpG sites in the genomic sequence. Position “0” in exon 1 marks the transcription start site. Gray boxes, repeat sequences in this region. Blue bars, genomic locations for MSO hybridization targets in regions 1 to 5. B, schematic representations of MSO microarray. Briefly, oligonucleotides capable of distinguishing the methylated (M) and unmethylated (U) alleles in the five interrogated regions flanking the RASSF1A promoter/first exon are arrayed onto the superaldehyde slide. Probes prepared from bisulfite-converted DNA were labeled with fluorescent dye and hybridized onto the slide. Fluorescence signal intensities from the M and the U oligonucleotides of each interrogating site are used to calculate the level of methylation in CG dinucleotide(s) included in these oligonucleotides. C, methylation status of regions flanking RASSF1A promoter/first exon in breast tumors and uninvolved breast normal tissues. Methylation level is expressed as %M [%M = M/(M + U) × 100, and the resultant value adjusted by its respective standard curve] and is depicted by the gradient blue scale (right). Each row represents %M data from one sample. Each column represents %M data from one set of oligonucleotides. White boxes with dotted line strike through indicate data not acquired. Genomic location of regions 1 to 5 is depicted in (A). Region 3 contains oligonucleotides that interrogate both the promoter and the first exon region of RASSF1A.

Fig. 1.

MSO microarray analysis of DNA samples obtained from primary breast tumors and normal tissues from breast reduction mammoplasty surgery. A, genomic map of the RASSF1A promoter/first exon and the nearby regions. Vertical lines, positions of CpG sites in the genomic sequence. Position “0” in exon 1 marks the transcription start site. Gray boxes, repeat sequences in this region. Blue bars, genomic locations for MSO hybridization targets in regions 1 to 5. B, schematic representations of MSO microarray. Briefly, oligonucleotides capable of distinguishing the methylated (M) and unmethylated (U) alleles in the five interrogated regions flanking the RASSF1A promoter/first exon are arrayed onto the superaldehyde slide. Probes prepared from bisulfite-converted DNA were labeled with fluorescent dye and hybridized onto the slide. Fluorescence signal intensities from the M and the U oligonucleotides of each interrogating site are used to calculate the level of methylation in CG dinucleotide(s) included in these oligonucleotides. C, methylation status of regions flanking RASSF1A promoter/first exon in breast tumors and uninvolved breast normal tissues. Methylation level is expressed as %M [%M = M/(M + U) × 100, and the resultant value adjusted by its respective standard curve] and is depicted by the gradient blue scale (right). Each row represents %M data from one sample. Each column represents %M data from one set of oligonucleotides. White boxes with dotted line strike through indicate data not acquired. Genomic location of regions 1 to 5 is depicted in (A). Region 3 contains oligonucleotides that interrogate both the promoter and the first exon region of RASSF1A.

Close modal

Promoter hypermethylation of RASSF1A is frequently observed in normal tissues adjacent to breast cancers. To study DNA methylation in gene promoters as a marker of field defect surrounding primary breast cancer, we had to ensure that the microdissected sections in these adjacent normals did not contain many infiltrating inflammatory cells. The H&E-stained slides marked originally as maps for microdissection purposes were carefully evaluated under a microscope. Adjacent normals had an average of 2.1% inflammatory cell infiltration (range, 0-20%) and primary tumors had an average of 4% infiltration (range, 0-10%). Thus, our samples had negligible contamination of inflammatory components.

The MSO analysis was extended to determine the methylation status of RASSF1A in normal tissues adjacent to breast cancers. In this regard, samples of four successive zones, each 1 to 4 cm (N1, N2, N3, and N4) from the tumor site, were obtained in the ipsilateral breast (Fig. 2A). Unremarkable breast ducts of these samples were identified and confirmed by histologic examination for microdissection (Fig. 2B). In two cases (4974 and 4768), four quadrants (Q1, Q2, Q3, and Q4) from the contralateral breast where no tumor was visible were also procured for the analysis. MSO results show that primary tumors from the six sets of samples exhibit hypermethylation in region 3 (i.e., the RASSF1A promoter/first exon core; Fig. 2C). We also observed extensive methylation of this region in adjacent breast ducts (n = 29), but the overall level of hypermethylation was lower than those of primary tumor sites (Fig. 3A). This methylation is significantly higher (1.75-fold) compared with the methylation levels of the aforementioned normal breast samples of unrelated healthy individuals (P = 0.01; see Fig. 3A). In MSO profiling, the zone delineating differential methylation between the RASSF1A core and its neighboring regions becomes less distinct in these adjacent samples.

Fig. 2.

MSO microarray analysis of DNA samples obtained from primary breast tumors and histologically normal adjacent tissues. A, breast tissues from patients undergoing mastectomy were collected. Mastectomy specimens were marked for tumor (T) and its surrounding zones N1 (1 cm), N2 (2 cm), N3 (3 cm), and N4 (4 cm) from the grossly visible tumor boundary. In bilateral and prophylactic mastectomy cases, normal breast tissues were taken from the four quadrants of the breast: Q1, Q2, Q3, and Q4. B, examples of H&E sections of tissues taken from tumor sites and adjacent area. Breast tissues were embedded in Tissue-Tek optimum cutting temperature compound and 5-μm sections were cut and mounted on Mercedes Platinum StarFrost Adhesive slides. The slides were stained using a standard H&E protocol and tissue boundaries marked. Images were captured at ×400 magnification. C, methylation status of regions flanking the RASSF1A promoter/first exon in breast tumors and adjacent histologically normal breast epithelia. Methylation level is expressed as %M [%M = M/(M + U) × 100, and the resultant value adjusted by its respective standard curve] and is depicted by the gradient blue scale (right). Each block represents %M data from one patient with methylation status of the primary tumor (T) indicated on top followed by information from histologically normal epithelia from mastectomy specimens (N) and bilateral prophylactic mastectomy normal epithelia (Q). Each column represents %M data from one set of oligonucleotides. White boxes with dotted line strike through indicate data not acquired. Genomic location of regions 1 to 5 is depicted in Fig. 1A. Region 3 contains oligonucleotides that interrogate both the promoter and the first exon region of RASSF1A. R, right breast; L, left breast.

Fig. 2.

MSO microarray analysis of DNA samples obtained from primary breast tumors and histologically normal adjacent tissues. A, breast tissues from patients undergoing mastectomy were collected. Mastectomy specimens were marked for tumor (T) and its surrounding zones N1 (1 cm), N2 (2 cm), N3 (3 cm), and N4 (4 cm) from the grossly visible tumor boundary. In bilateral and prophylactic mastectomy cases, normal breast tissues were taken from the four quadrants of the breast: Q1, Q2, Q3, and Q4. B, examples of H&E sections of tissues taken from tumor sites and adjacent area. Breast tissues were embedded in Tissue-Tek optimum cutting temperature compound and 5-μm sections were cut and mounted on Mercedes Platinum StarFrost Adhesive slides. The slides were stained using a standard H&E protocol and tissue boundaries marked. Images were captured at ×400 magnification. C, methylation status of regions flanking the RASSF1A promoter/first exon in breast tumors and adjacent histologically normal breast epithelia. Methylation level is expressed as %M [%M = M/(M + U) × 100, and the resultant value adjusted by its respective standard curve] and is depicted by the gradient blue scale (right). Each block represents %M data from one patient with methylation status of the primary tumor (T) indicated on top followed by information from histologically normal epithelia from mastectomy specimens (N) and bilateral prophylactic mastectomy normal epithelia (Q). Each column represents %M data from one set of oligonucleotides. White boxes with dotted line strike through indicate data not acquired. Genomic location of regions 1 to 5 is depicted in Fig. 1A. Region 3 contains oligonucleotides that interrogate both the promoter and the first exon region of RASSF1A. R, right breast; L, left breast.

Close modal
Fig. 3.

Methylation status of regions flanking RASSF1A promoter/first exon. Genomic location of regions 1 to 5 is depicted in Fig. 1A. A, the average probe set %M of each sample within the three sample types is presented for each studied regions. Gray line, average %M within samples from each group. The probe sets of regions 2 to 4 are distributed across the RASSF1A promoter/first exon and %M from these probe sets showed the greatest separations between the different sample types as indicated by statistical comparison (*, P < 0.05; **, P < 0.01). Probe sets covering the regions flanking this CpG island are nondiscriminative. B, the %M for individual sample set of tumor (T)/mastectomy normal tissue (N)/bilateral prophylactic mastectomy (Q) for region 3 (promoter and exon) are shown. The %M was evaluated across samples obtained from each individual patient. Samples from patients 4810 and 4817 clearly show a stepwise decrease in %M as samples moved away from the gross margin of the primary tumor (*, P < 0.05; **, P < 0.01).

Fig. 3.

Methylation status of regions flanking RASSF1A promoter/first exon. Genomic location of regions 1 to 5 is depicted in Fig. 1A. A, the average probe set %M of each sample within the three sample types is presented for each studied regions. Gray line, average %M within samples from each group. The probe sets of regions 2 to 4 are distributed across the RASSF1A promoter/first exon and %M from these probe sets showed the greatest separations between the different sample types as indicated by statistical comparison (*, P < 0.05; **, P < 0.01). Probe sets covering the regions flanking this CpG island are nondiscriminative. B, the %M for individual sample set of tumor (T)/mastectomy normal tissue (N)/bilateral prophylactic mastectomy (Q) for region 3 (promoter and exon) are shown. The %M was evaluated across samples obtained from each individual patient. Samples from patients 4810 and 4817 clearly show a stepwise decrease in %M as samples moved away from the gross margin of the primary tumor (*, P < 0.05; **, P < 0.01).

Close modal

We further observed a gradient pattern in 4810 and 4817 sample sets in which more methylation tends to occur in adjacent tissues closest to the cancer site (1 cm or N1) than those 2 to 4 cm away from the sites (Figs. 2C and 3B). This gradient is not obvious in the adjacent samples from the remaining patients, suggesting an absence of well-defined epigenetic margins for the other four tumors even at 4 cm away from the tumor site. Interestingly, extensive methylation of region 3 also occurred in the contralateral breast of patients 4974 and 4768.

AI/LOH of chromosome 3p is observed in normal tissues adjacent to breast cancers. The RASSF1A gene is located on chromosome 3p, a region previously reported to exhibit genetic abnormalities in normal tissues adjacent to breast cancers (8, 31, 32). We therefore determined whether AI/LOH occurred in our studied panel and whether both genetic and epigenetic alterations could be concurrently detected in adjacent tissues. A total of 29 microsatellite markers spanning the entire 3p region were genotyped in 8 breast tumors in conjunction with 22 corresponding adjacent normal tissues (Fig. 4). Thirteen of these 29 markers showed AI/LOH in these tumors. Specifically, one tumor, 4822, showed an obvious loss of the entire 3p region, whereas five of the remaining seven tumors showed AI/LOH in several locations, including D3S1581 (telomeric to RASSF1A), D3S3118 (centromeric to RASSF1A), and D3S663 (3p21.31). When the corresponding adjacent tissues were examined among the 13 positive markers, in 5 instances AI/LOH of the same loci were found in one or more zones near the tumor sites. Of note, we found 11 additional instances of AI/LOH in normal tissues surrounding tumors in which no AI/LOH was detected in these markers. Also of note, the AI/LOH finding using the 0.67 cutoff (less stringent) was very similar to those identified by the 0.5 cutoff (more stringent), providing confidence to these findings.

Fig. 4.

AI/LOH analysis. Microsatellite markers flanking RASSF1A and along the short arm of chromosome 3 were selected from Genethon and Marshfield genetic maps and locations harboring TSG are marked on top. The distance spanned by the markers in megabases (telomere to centromere) is shown immediately below. AI/LOH was evaluated in eight sets of primary tumors (T) and adjacent histologically normal tissue from mastectomy from the same subject and sample information are displayed on the left. Black squares, loci of AI/LOH identified with a 0.5 cutoff. Gray squares, loci of AI/LOH identified with a 0.67 cutoff. Clear squares, informative loci without AI/LOH. Shaded squares, uninformative loci. Sample sets identified by a star signify those also evaluated by the MSO microarray analysis.

Fig. 4.

AI/LOH analysis. Microsatellite markers flanking RASSF1A and along the short arm of chromosome 3 were selected from Genethon and Marshfield genetic maps and locations harboring TSG are marked on top. The distance spanned by the markers in megabases (telomere to centromere) is shown immediately below. AI/LOH was evaluated in eight sets of primary tumors (T) and adjacent histologically normal tissue from mastectomy from the same subject and sample information are displayed on the left. Black squares, loci of AI/LOH identified with a 0.5 cutoff. Gray squares, loci of AI/LOH identified with a 0.67 cutoff. Clear squares, informative loci without AI/LOH. Shaded squares, uninformative loci. Sample sets identified by a star signify those also evaluated by the MSO microarray analysis.

Close modal

We further compared the frequency of genetic versus epigenetic events in the RASSF1A region (3p21.3-31.1) in six tumors sets (4768, 4810, 4879, 4974, 5037, and 4817). Promoter hypermethylation of this gene was detected in adjacent tissues of these tumors (see Fig. 2C). AI/LOH around the RASSF1A region (D3S1581 and D3S3118) was only detected in adjacent tissues of 4810 and 4974 tumors, suggesting that both genetic and epigenetic alterations corroborated to silence this gene in these two samples. Adjacent normal tissues of the remaining four patients showed no detectable AI/LOH. Although this study was conducted in limited sample sets, the initial results seem to indicate that promoter hypermethylation of RASSF1A is more frequent than AI/LOH of the same region in normal tissues adjacent to breast cancers.

Global profiling identifies novel loci frequently hypermethylated in normal tissues adjacent to breast cancers. Next, we used differential methylation hybridization to globally screen CpG islands for methylation alterations in a separate set of paired tumor/adjacent normal samples (12 invasive ductal carcinomas and 25 adjacent normal tissues). An additional 12 breast samples obtained from healthy individuals were used as controls. Amplicons were prerestricted with McrBC, an enzyme known to cut methylated [5′..(G/A)C(N55-103)(G/A)mC] sequences (29, 33). Therefore, methylated DNAs were restricted away and could not be amplified by linker-mediated PCR whereas unmethylated fragments remained uncut and thus amplified. The McrBC-cut (test) and uncut (control) amplicons from each sample were labeled with Cy3 and Cy5, respectively, and cohybridized to a microarray panel of 12,000 CpG island tags previously used to identify differentially methylated loci in solid tumors (16, 22). CpG tags that hybridized predominately with the control amplicon, but not with the test amplicon, appear as red spots. These are methylated loci, which are restricted and therefore absent in the test DNA, but present in the control DNA. The methylation data are normalized using mitochondrial clones and clones without McrBC restriction sites. Subsequently, to analyze CGI loci that best characterized differences between the tumor, adjacent normals, and the reduction normal mammoplasty samples, the shrunken centroid analysis was conducted in the microarray data set, using Prediction Analysis for Microarray software in R. Tenfold cross-validation was used to assess loci commonly methylated in tumor and adjacent groups, but not in the normal breast. Relative to the overall predicted centroids as shown in Fig. 5A as horizontal units (representing the log ratios of average DNA methylation in the analyzed samples), a positive value indicates that more methylation occurs in a locus whereas a negative value indicates less or no methylation. Collectively, this analysis identifies a subset of putative loci preferentially hypermethylated in tumor and adjacent groups (Fig. 5A; a subset of these loci is presented in Supplementary Table S1). The shrunken centroid map also reveals higher levels of methylation in tumor group than in adjacent group. Initial methylation assessment of nine loci by nonquantitative methylation-specific PCR further substantiates these microarray findings (Fig. 5B).

Fig. 5.

Global methylation status as determined by differential methylation hybridization microarray. A, Prediction Analysis of Microarray plot output depicting the predicted centroids, shown as horizontal units, representing the log ratios of average DNA methylation generated from the three sample types. CpG island loci from breast reduction mammoplasty normal tissues are projected to the left indicating a hypomethylated status (see Table S1 for gene identities). These loci from primary breast tumors are projected to the right indicating a hypermethylated status. The shrunken centroid analysis selects the histologically normal tissue from breast mastectomy from the same patient to be neutral in methylation status. Microarray data will be deposited in GEO8 for public access. B, validation of methylation status of genes identified by the shrunken centroid analysis to be differentially methylated among the three sample types. Bisulfite-converted genomic DNA from the three pooled sample types and in vitro methylated DNA (by SssI methylase as a positive control) were subjected to methylation-specific PCR. The PCR band intensities of nine loci are shown and their gene designations are indicated on the right.

Fig. 5.

Global methylation status as determined by differential methylation hybridization microarray. A, Prediction Analysis of Microarray plot output depicting the predicted centroids, shown as horizontal units, representing the log ratios of average DNA methylation generated from the three sample types. CpG island loci from breast reduction mammoplasty normal tissues are projected to the left indicating a hypomethylated status (see Table S1 for gene identities). These loci from primary breast tumors are projected to the right indicating a hypermethylated status. The shrunken centroid analysis selects the histologically normal tissue from breast mastectomy from the same patient to be neutral in methylation status. Microarray data will be deposited in GEO8 for public access. B, validation of methylation status of genes identified by the shrunken centroid analysis to be differentially methylated among the three sample types. Bisulfite-converted genomic DNA from the three pooled sample types and in vitro methylated DNA (by SssI methylase as a positive control) were subjected to methylation-specific PCR. The PCR band intensities of nine loci are shown and their gene designations are indicated on the right.

Close modal

Putative loci, including RASSF1A, were selected for detailed analysis using real-time quantitative methylation-specific PCR (Fig. 6). Methylation-specific PCR primers for ZNF582 showed a single distinct product in the gel-based experiment, which was found to produce two PCR product peaks in a subset of the samples in quantitative methylation-specific PCR experiments. Modifying the annealing temperature and the PCR reaction mix did not resolve this issue. As such, other gene loci were tested in its place. Consistent with the MSO finding, the overall methylation level of the RASSF1A promoter/first exon is significantly higher in primary tumors than in normal controls (P = 0.05). Hypermethylation of this gene can be seen in adjacent normal tissues of 6320, 4974, and 4993. Promoter hypermethylation of three newly identified loci, CYP26A1, KCNAB1, and SNCA, is detected in 47%, 41%, and 38% of breast tumors, respectively; no or low methylation is detected in normal breast samples. This independent confirmation of global methylation profiles of breast tumors and their adjacent normal tissues uncovers a remarkable finding. We notice in these four studied genes that ∼70% of the time, the genes are methylated in the adjacent normal tissues whenever they are found to be hypermethylated in the primary tumors. In some instances, the levels of methylation in adjacent tissues (N1 and N4 of 6320 and N3 of 6316 for CYP26A1) are higher than those of their corresponding tumors. Treatment of the two breast cancer cell lines with the demethylating drug 5′-aza-2′-deoxycytidine results in a decrease in methylation and an increase in expression in these four genes (see Supplementary data).

Fig. 6.

Validation of differential methylation status of promoter CpG island loci by SYBR-based quantitative methylation-specific PCR. Standard methylation-specific PCR primers detect the methylated alleles in each sample types for a specific test locus and the resultant threshold cycle numbers are adjusted against a standard curve generated by in vitro methylated DNA. Primers representing bisulfite-converted DNA but devoid of CGs quantify the input bisulfite-converted DNA. The percent relative methylation is the ratio of the adjusted methylated alleles to the adjusted COL2A1 DNA found in the same sample. The relative methylation data for RASSF1A and three promoter CpG islands, CYP26A1, KCNAB1, and SNCA, selected from methylation microarray analysis are shown. Sample designations (see Fig. 2A) are shown below the x axis.

Fig. 6.

Validation of differential methylation status of promoter CpG island loci by SYBR-based quantitative methylation-specific PCR. Standard methylation-specific PCR primers detect the methylated alleles in each sample types for a specific test locus and the resultant threshold cycle numbers are adjusted against a standard curve generated by in vitro methylated DNA. Primers representing bisulfite-converted DNA but devoid of CGs quantify the input bisulfite-converted DNA. The percent relative methylation is the ratio of the adjusted methylated alleles to the adjusted COL2A1 DNA found in the same sample. The relative methylation data for RASSF1A and three promoter CpG islands, CYP26A1, KCNAB1, and SNCA, selected from methylation microarray analysis are shown. Sample designations (see Fig. 2A) are shown below the x axis.

Close modal

Convincing evidence indicates that histone modifications precede promoter hypermethylation to initiate transcriptional silencing of tumor suppressor genes (15, 3437). Dysregulation of signal transduction in neoplastic cells can lead to transcriptional silencing of downstream target genes (38). As a result, repressive histone modifications are first established, and over time, an increased density of DNA methylation creeps into the promoter region of target genes to produce a permanent mode of epigenetic silencing. Epigenetic changes evolve over time and the compilation of methylation profiles from tumors captured at different stages of tumor development can recreate this slow cumulative process. Therefore, we used the MSO microarray to map the methylation status of a 4-kb region surrounding the RASSF1A promoter/first exon core. The composite profiles suggest that progressive accumulation of DNA methylation occurs in the core region during breast cancer development. This reasoning is based on the narrowing of the unmethylated zone in the core region across tumors with increasing methylation level in this gene. In addition, we observe that the extent of methylation in the RASSF1A promoter core region of normal adjacent tissues is usually higher than those of unrelated controls but lower than their corresponding tumors (see Fig. 2C). We speculate that there is a localized “field” of RASSF1A hypermethylation in these tumors and their associated normal tissues. This phenomenon is consistent with the so-called methylation spreading theory in which the flanking methylation barriers of a CpG island promoter are broken down in neoplastic cells, resulting in gradual invasion of methylation to the promoter core (11, 34, 39, 40).

In addition to the spreading of DNA methylation within a promoter CpG island, we are also interested in another “geographic” event, the “field defect” of carcinogenesis. It is known that breast cancer can recur/self-metastasize locally as well as to distant sites. We are interested, as are other researchers (41, 42), in discovering a biomarker of field defect, as it will be useful in risk assessment and in early disease detection. Epigenetic modifications are believed to be early events in cancer development. This coupling with the heritable nature of DNA methylation marks renders them more traceable and thus good candidates to study the field defect of cancerization. Shen et al. (41) proposed that such a defect, which is not observed in normal specimens, should be frequently encountered in cancer samples and observed in subjects that are at risk of development cancer. Once established in premalignant tissues, the extent of such defect will accumulate as the disease progresses. It is not certain how field defect arises. Norton (43) proposed the self-metastasis model whereby a small tumor becomes a large tumor by becoming a conglomerate mass of tiny tumors, each may be started by a seed/stem, progenitor cell. This seeding concept may also explain the field defect around the primary tumor. Maekita et al. (42) also discussed the possibility of stem cell methylation profile (due to H. pylori infection) influencing the methylation in gastric mucosa long after the infection is discontinued. Here we observed a phenomenon in a subset of the studied tumor/adjacent normals that suggests the primary tumor may serve as an epicenter from which methylation density progressively diffuses outwards to surrounding tissues (similar to the seeding process previously described). Sample sets 4810 and 4817 exhibited a clear methylation gradient that was not evident in the remaining four tumor sets (4974, 4768, 4879, and 5037). It is possible that samples showing an epicentric effect are caught early in this seeding process whereas the methylation patterns not showing such an effect may have a wider sphere of methylation gradient (>4 cm away from the epicenter). However, more samples need to be analyzed to provide credence to such an observation.

Survey of normal tissues from the contralateral breast of two of these patients (4974 and 4768) also showed the presence of higher RASSF1A methylation relative to the control samples. This finding corroborates the previous notion that promoter hypermethylation of critical tumor suppressor genes may arise early during neoplastic transformation (44, 45). In this regard, methylation spreading of RASSF1A is already evident in precancerous cells with hypermethylation progressively infiltrating the core region during the development of neoplastic lesions. This results in gradual attenuation of gene transcription. However, tumorigenesis requires multiple genetic and epigenetic abnormalities in the affected area to produce invasive carcinomas. We therefore used differential methylation hybridization to screen additional promoter loci, the hypermethylation of which may occur early during breast cancer development. Using this methylation microarray method, we have uncovered a group of loci frequently hypermethylated in normal tissues adjacent to tumors. Among these, hypermethylation of three promoters (CYP26A1, KCNAB1, and SNCA) was commonly detected in adjacent tissues (∼70%) of the tumor sets analyzed by quantitative methylation-specific PCR. This high degree of concordant promoter methylation in primary tumors and adjacent normal tissues suggests that the prevalence of promoter hypermethylation in normal tissue adjacent to tumors is more extensive than previously noted (12, 41).

The present study has practical implication for predicting the future risk of local recurrence in breast cancer patients undergoing lumpectomy. As part of the breast conservation strategy, lumpectomy is a surgical procedure preferred by some patients with small, early tumors whereby only the tumor-containing breast tissue is removed while sparing the residual breast tissue (1). Because most recurrences (86%) occur in tissues near or in the same quadrant as the lumpectomy site, an issue of critical importance is to ensure that the surgical margins are cancer-free (2). Currently, this evaluation relies on pathologic examination of margins free of gross tumor tissues. The use of epigenetic markers can be an important adjunct for predicting local recurrence in histologically normal margins. In this regard, simultaneous screening of multiple methylation markers can be done by the aforementioned differential methylation hybridization technique. Because amplicons are prepared from small amounts of DNA, sufficient DNA samples can be obtained from cytology touch imprints (46) made by gently pressing the corresponding margin to the glass slide. Methylation profiling will be used to determine the prevalence of promoter hypermethylation in the margin. This type of analysis will provide oncologists useful information in determining whether patients should undergo radiological therapy or other treatments to prevent future recurrence.

In summary, the present study provides, for the first time, a comprehensive analysis of promoter hypermethylation in normal tissues adjacent to breast cancers. The results indicate that premalignant epigenetic changes radiating from the tumor epicenter could be widespread. Future studies will be oriented to the analysis of whether increased density of methylation within a promoter as well as increased number of hypermethylated loci in these adjacent normal tissues is a factor of local recurrence. As such, hypermethylation of multiple gene promoters can be used as a surrogate biomarker to define the molecular margin of lumpectomy in the future.

Grant support: National Cancer Institute grants U54 CA11300 (T.H-M. Huang, P.S. Yan, and C. Eng), R01 CA069065 and R33 CA 094441 (T.H-M. Huang and P.S. Yan), and P30 CA16058 and R01 CA098522-03 (T.J. Yeatman), and the Doris Duke Distinguished Clinical Scientist Award (C. Eng).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

P.S. Yan and C. Venkataramu are co-first authors. T.J. Yeatman and T.H-M. Huang contributed equally to this work.

1
Kaplan J, Giron G, Tartter PI, Bleiweiss IJ, Estabrook A, Smith SR. Breast conservation in patients with multiple ipsilateral synchronous cancers.
J Am Coll Surg
2003
;
197
:
726
–9.
2
Fredriksson I, Liljegren G, Palm-Sjovall M, et al. Risk factors for local recurrence after breast-conserving surgery.
Br J Surg
2003
;
90
:
1093
–102.
3
Deng G, Lu Y, Zlotnikov G, Thor AD, Smith HS. Loss of heterozygosity in normal tissue adjacent to breast carcinomas.
Science
1996
;
274
:
2057
–9.
4
Ellsworth DL, Ellsworth RE, Love B, et al. Outer breast quadrants demonstrate increased levels of genomic instability.
Ann Surg Oncol
2004
;
11
:
861
–8.
5
Li Z, Moore DH, Meng ZH, Ljung BM, Gray JW, Dairkee SH. Increased risk of local recurrence is associated with allelic loss in normal lobules of breast cancer patients.
Cancer Res
2002
;
62
:
1000
–3.
6
Ellsworth RE, Ellsworth DL, Neatrour DM, et al. Allelic imbalance in primary breast carcinomas and metastatic tumors of the axillary lymph nodes.
Mol Cancer Res
2005
;
3
:
71
–7.
7
Larson PS, de las Morenas A, Bennett SR, Cupples LA, Rosenberg CL. Loss of heterozygosity or allele imbalance in histologically normal breast epithelium is distinct from loss of heterozygosity or allele imbalance in co-existing carcinomas.
Am J Pathol
2002
;
161
:
283
–90.
8
Larson PS, de las Morenas A, Cupples LA, Huang K, Rosenberg CL. Genetically abnormal clones in histologically normal breast tissue.
Am J Pathol
1998
;
152
:
1591
–8.
9
Lynch ED, Ostermeyer EA, Lee MK, et al. Inherited mutations in PTEN that are associated with breast cancer, Cowden disease, and juvenile polyposis.
Am J Hum Genet
1997
;
61
:
1254
–60.
10
Widschwendter M, Berger J, Daxenbichler G, et al. Loss of retinoic acid receptor β expression in breast cancer and morphologically normal adjacent tissue but not in the normal breast tissue distant from the cancer.
Cancer Res
1997
;
57
:
4158
–61.
11
Yan PS, Shi H, Rahmatpanah F, et al. Differential distribution of DNA methylation within the RASSF1A CpG island in breast cancer.
Cancer Res
2003
;
63
:
6178
–86.
12
Lewis CM, Cler LR, Bu DW, et al. Promoter hypermethylation in benign breast epithelium in relation to predicted breast cancer risk.
Clin Cancer Res
2005
;
11
:
166
–72.
13
Holst CR, Nuovo GJ, Esteller M, et al. Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia.
Cancer Res
2003
;
63
:
1596
–601.
14
Bird A. DNA methylation de novo.
Science
1999
;
286
:
2287
–8.
15
Nephew KP, Huang TH. Epigenetic gene silencing in cancer initiation and progression.
Cancer Lett
2003
;
190
:
125
–33.
16
Yan PS, Chen CM, Shi H, et al. Dissecting complex epigenetic alterations in breast cancer using CpG island microarrays.
Cancer Res
2001
;
61
:
8375
–80.
17
Lehmann U, Berg-Ribbe I, Wingen LU, et al. Distinct methylation patterns of benign and malignant liver tumors revealed by quantitative methylation profiling.
Clin Cancer Res
2005
;
11
:
3654
–60.
18
Umbricht CB, Evron E, Gabrielson E, Ferguson A, Marks J, Sukumar S. Hypermethylation of 14-3-3 sigma (stratifin) is an early event in breast cancer.
Oncogene
2001
;
20
:
3348
–53.
19
Belinsky SA, Nikula KJ, Palmisano WA, et al. Aberrant methylation of p16(INK4a) is an early event in lung cancer and a potential biomarker for early diagnosis.
Proc Natl Acad Sci U S A
1998
;
95
:
11891
–6.
20
Yan PS, Wei SH, Huang TH. Methylation-specific oligonucleotide microarray.
Methods Mol Biol
2004
;
287
:
251
–60.
21
Gitan RS, Shi H, Chen CM, Yan PS, Huang TH. Methylation-specific oligonucleotide microarray: a new potential for high-throughput methylation analysis.
Genome Res
2002
;
12
:
158
–64.
22
Yan PS, Wei SH, Huang TH. Differential methylation hybridization using CpG island arrays.
Methods Mol Biol
2002
;
200
:
87
–100.
23
Huang TH, Perry MR, Laux DE. Methylation profiling of CpG islands in human breast cancer cells.
Hum Mol Genet
1999
;
8
:
459
–70.
24
Aldred MA, Ginn-Pease ME, Morrison CD, et al. Caveolin-1 and caveolin-2, together with three bone morphogenetic protein-related genes, may encode novel tumor suppressors down-regulated in sporadic follicular thyroid carcinogenesis.
Cancer Res
2003
;
63
:
2864
–71.
25
Boehm D, Herold S, Kuechler A, Liehr T, Laccone F. Rapid detection of subtelomeric deletion/duplication by novel real-time quantitative PCR using SYBR-green dye.
Hum Mutat
2004
;
23
:
368
–78.
26
Marsh DJ, Dahia PL, Coulon V, et al. Allelic imbalance, including deletion of PTEN/MMACI, at the Cowden disease locus on 10q22-23, in hamartomas from patients with Cowden syndrome and germline PTEN mutation.
Genes Chromosomes Cancer
1998
;
21
:
61
–9.
27
Kurose K, Hoshaw-Woodard S, Adeyinka A, Lemeshow S, Watson PH, Eng C. Genetic model of multi-step breast carcinogenesis involving the epithelium and stroma: clues to tumour-microenvironment interactions.
Hum Mol Genet
2001
;
10
:
1907
–13.
28
Heisler LE, Torti D, Boutros PC, et al. CpG Island microarray probe sequences derived from a physical library are representative of CpG Islands annotated on the human genome.
Nucleic Acids Res
2005
;
33
:
2952
–61.
29
Nouzova M, Holtan N, Oshiro MM, et al. Epigenomic changes during leukemia cell differentiation: analysis of histone acetylation and cytosine methylation using CpG island microarrays.
J Pharmacol Exp Ther
2004
;
311
:
968
–81.
30
Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression.
Proc Natl Acad Sci U S A
2002
;
99
:
6567
–72.
31
Ellsworth DL, Ellsworth RE, Liebman MN, Hooke JA, Shriver CD. Genomic instability in histologically normal breast tissues: implications for carcinogenesis.
Lancet Oncol
2004
;
5
:
753
–8.
32
Ellsworth DL, Ellsworth RE, Love B, et al. Genomic patterns of allelic imbalance in disease free tissue adjacent to primary breast carcinomas.
Breast Cancer Res Treat
2004
;
88
:
131
–9.
33
Pieper U, Groll DH, Wunsch S, et al. The GTP-dependent restriction enzyme McrBC from Escherichia coli forms high-molecular mass complexes with DNA and produces a cleavage pattern with a characteristic 10-base pair repeat.
Biochemistry
2002
;
41
:
5245
–54.
34
Strunnikova M, Schagdarsurengin U, Kehlen A, Garbe JC, Stampfer MR, Dammann R. Chromatin inactivation precedes de novo DNA methylation during the progressive epigenetic silencing of the RASSF1A promoter.
Mol Cell Biol
2005
;
25
:
3923
–33.
35
Ballestar E, Esteller M. The epigenetic breakdown of cancer cells: from DNA methylation to histone modifications.
Prog Mol Subcell Biol
2005
;
38
:
169
–81.
36
Santoro R, Grummt I. Epigenetic mechanism of rRNA gene silencing: temporal order of NoRC-mediated histone modification, chromatin remodeling, and DNA methylation.
Mol Cell Biol
2005
;
25
:
2539
–46.
37
Wade PA. SWItching off methylated DNA.
Nat Genet
2005
;
37
:
212
–3.
38
Leu YW, Yan PS, Fan M, et al. Loss of estrogen receptor signaling triggers epigenetic silencing of downstream targets in breast cancer.
Cancer Res
2004
;
64
:
8184
–92.
39
Thakur N, Kanduri M, Holmgren C, Mukhopadhyay R, Kanduri C. Bidirectional silencing and DNA methylation-sensitive methylation-spreading properties of the Kcnq1 imprinting control region map to the same regions.
J Biol Chem
2003
;
278
:
9514
–9.
40
Turker MS. Gene silencing in mammalian cells and the spread of DNA methylation.
Oncogene
2002
;
21
:
5388
–93.
41
Shen L, Kondo Y, Rosner GL, et al. MGMT promoter methylation and field defect in sporadic colorectal cancer.
J Natl Cancer Inst
2005
;
97
:
1330
–8.
42
Maekita T, Nakazawa K, Mihara M, et al. High levels of aberrant DNA methylation in Helicobacter pylori-infected gastric mucosae and its possible association with gastric cancer risk.
Clin Cancer Res
2006
;
12
:
989
–95.
43
Norton L. Conceptual and practical implications of breast tissue geometry: toward a more effective, less toxic therapy.
Oncologist
2005
;
10
:
370
–81.
44
Esteller M. Dormant hypermethylated tumour suppressor genes: questions and answers.
J Pathol
2005
;
205
:
172
–80.
45
Belinsky SA. Silencing of genes by promoter hypermethylation: key event in rodent and human lung cancer.
Carcinogenesis
2005
;
26
:
1481
–7.
46
Gaston SM, Soares MA, Siddiqui MM, et al. Tissue-print and print-phoresis as platform technologies for the molecular analysis of human surgical specimens: mapping tumor invasion of the prostate capsule.
Nat Med
2005
;
11
:
95
–101.