Expression Analysis of DNA Methyltransferases 1, 3A, and 3B in Sporadic Breast Carcinomas

DNA methylation is an epigenetic modification with an important role in the control of gene expression in mammalian cells. This regulation results from covalent addition of a methyl group to the 5-position of cytosine, predominantly within CpG dinucleotides of gene promoter regions (1). Methylation of CpG-rich DNA regions known as CpG islands is thought to silence gene transcription by recruiting protein complexes involved in nucleosome remodeling (2). Gene up-regulation associated with promoter hypomethylation has also been suggested, but this aspect of transcriptional regulation is far less well documented than the silencing mechanism (1, 3).

The transfer of a methyl group is catalyzed by three functional DNMTs² with distinctive roles. DNMT1 is mainly involved in maintaining the preexisting methylation pattern during replication, whereas DNMT3A and 3B are mainly involved in de novo methylation (1).

Abnormal DNA methylation is thought to be a major early event in the development of tumors characterized by widespread genome hypomethylation leading to chromosome instability and localized DNA hypermethylation; the latter may be important in tumorigenesis by silencing tumor suppressor genes (4, 5).

Recently, Esteller et al. (6) established the promoter hypermethylation profile in various tumor types. They showed that promoter hypermethylation affected all immortalization and transformation molecular pathways, some changes in promoter gene being shared by all tumor types (CDKN2A/P16), and others being specific for a given tumor type (BRCA1 in breast and ovarian carcinomas; TP73 and CDKN2B/P15 in hematological malignancies). Several studies have shown DNMT overexpression (mainly DNMT1 and DNMT3B) in a variety of cancers (7, 8, 9, 10, 11, 12, 13, 14). Induced DNMT1 overexpression in cultured cell lines gradually induces CpG hypermethylation and cell transformation (15). However, no clear relationship between DNMT overexpression and target gene underexpression has been demonstrated in vivo (7, 9, 13, 14, 16).

DNMT expression in breast cancer has only been studied in cell lines (17). To gain more insight into DNMT expression in breast tumors, we quantified the mRNA expression of the three known functional DNMT genes, namely DNMT1, 3A, and 3B, in a series of 130 patients with unilateral invasive primary breast tumors, by means of real-time quantitative RT-PCR assays (18). We then sought relationships between the expression of DNMT3B, the only markedly dysregulated DNMT gene in our tumor series, and classical clinical, pathological, and biological parameters, including patient outcome. Clinical outcome was also examined according to DNMT3B expression status in an additional well-characterized series of 98 postmenopausal patients exclusively treated with tamoxifen after surgery.

Finally, we sought relationships between the expression of DNMTs and (a) well-known genes subject to promoter hypermethylation, namely ERα, MLH1, RB1, E-cadherin, CDKN2B/P15, CDKN2A/P16, ARF/P14, BRCA1, and MASPIN (19, 20, 21); (b) eight genes which expression has been shown to be dysregulated in an in vitro model of loss of DNMT expression (22); and (c) genes involved in the DNMT downstream pathway, namely MECP2 (23), HDAC1 (24), and MBD2 (25).

We analyzed samples from two series of women with primary unilateral nonmetastatic breast carcinoma.

The first series consisted of 130 women whose breast tumors were excised at Centre René Huguenin from 1977 to 1989. The patients (mean age 58.5 years, range 34–91) had ERα-positive or -negative breast tumors (100 and 30, respectively) and were both pre and postmenopausal (45 and 85 patients, respectively). Seventy-five patients received adjuvant therapy, consisting of chemotherapy alone in 22 cases, hormone therapy alone in 18 cases, and both treatments in 35 cases. The standard prognostic factors have been reported elsewhere (26). The median follow-up was 8.1 years (range 1–15.9 years). Forty-seven patients relapsed. The first relapse events consisted of local and/or regional recurrences in 13 patients, metastases in 30 patients, and both events in four patients.

The second series consisted of 98 postmenopausal women whose breast tumors were excised at Centre René Huguenin from 1980 to 1994. The patients (mean age 70.7 years, range 54–86) all had ERα-positive breast tumors and received postoperative adjuvant hormone therapy consisting of tamoxifen (20 mg daily for 3–5 years) and no other treatment. The standard prognostic factors are reported in Table 1. The median follow-up was 6 years (range 1.5–17.5 years). Thirty-three patients relapsed. The first relapse events consisted of local and/or regional recurrences in three patients, metastases in 26 patients, and both events in four patients.

Both series of tumor samples were placed in liquid nitrogen until total RNA extraction immediately after surgery. Complete clinical, histological, and biological information was available; no radiotherapy or chemotherapy was given before surgery, and full follow-up took place at Centre René Huguenin. The histological type of the tumor and the number of positive axillary nodes were established at the time of surgery. The malignancy of infiltrating carcinomas was scored according to SBR’s histoprognostic system (27). ERα status was determined at both the protein level by the dextran-coated charcoal method until 1988 and enzymatic immuno-assay thereafter and at the mRNA level by real-time quantitative RT-PCR assay (26, 28).

Specimens of normal breast tissue, adjacent to the malignant breast tumor from six patients, were used as sources of normal RNA. The normal breast tissue samples were checked histologically and selected on the basis of their cellular composition, i.e., a large proportion of epithelial cells. Samples with a majority of stroma or inflammatory cells were excluded from the study.

Correlations between DNMTs and target genes mRNA levels were investigated in a subgroup of 46 breast tumors from the first series of 130 patients with a similar range of DNMT3B expression levels.

Reactions are characterized by the point during cycling when amplification of the PCR product is first detected, rather than the amount of PCR product accumulated after a fixed number of cycles. The larger the starting quantity of the target molecule, the earlier a significant increase in fluorescence is observed. The parameter Ct is defined as the fractional cycle number at which the fluorescence generated by SYBR Green dye-amplicon complex formation passes a fixed threshold above baseline.

The precise amount of total RNA added to each reaction mix (based on absorbance) and its quality (lack of extensive degradation) are both difficult to assess. We therefore also quantified transcripts of the gene coding for the TBP (a component of the DNA-binding protein complex transcription factor IID) as the endogenous RNA control and normalized each sample on the basis of its TBP content (18).

The relative target gene expression level was also normalized to a calibrator, or 1 × sample, consisting of a pool of normal breast tissue specimens.

Final results, expressed as N-fold differences in target gene expression relative to the TPB gene and the calibrator, termed “N_target,” were determined as follows:

where ΔCt values of the calibrator and sample were determined by subtracting the Ct value of the target gene from the Ct value of the TBP gene.

The primers for the DNMT, TBP, and other target genes were chosen with the assistance of the Oligo 4.0 computer program (National Biosciences, Plymouth, MN). We conducted BLASTN searches against dbEST and nr (the nonredundant set of the GenBank sequence database) to confirm the total gene specificity of the chosen nucleotide sequences and the absence of DNA polymorphisms. The primer nucleotide sequences for the DNMT and TBP genes are shown in Table 2. The primers for the others 21 genes tested in this study are available on request. To avoid amplification of contaminating genomic DNA, one of the two primers was placed at the junction between two exons, or in a different exon, e.g., the upper primer of TBP was placed at the junction between exons 5 and 6, whereas the lower primer was placed in exon 6.

Total RNA was extracted from tissue specimens by using the acid-phenol guanidinium method (29). The quality of RNA samples was determined by electrophoresis through agarose gels and staining with ethidium bromide; the 18S and 28S RNA bands were visualized under UV light.

RNA was reverse transcribed in a final volume of 20 μl containing 1× RT-PCR buffer [0.5 mm each deoxynucleotide triphosphate, 3 mm MgCl₂, 75 mm KCl, and 50 mm Tris-HCl (pH 8.3)], 20 units of RNasin RNase inhibitor (Promega, Madison, WI), 10 mm DTT, 100 units of Superscript II RNase H⁻ reverse transcriptase (Invitrogen, Carlsbad, CA), 1.5 mm random hexamers (Pharmacia, Uppsala, Sweden), and 1 μg of total RNA (patient samples). Samples were incubated at 20°C for 10 min and 42°C for 30 min, and reverse transcriptase was inactivated by heating at 99°C for 5 min and cooling at 5°C for 5 min.

All PCR reactions were performed using an ABI Prism 7700 Sequence Detection System (Perkin-Elmer Applied Biosystems, Foster City, CA).

For each PCR run, a master mix was prepared on ice with 1× SYBR Green buffer; 5 mm MgCl₂; 0.2 mm dATP, dCTP, and dGTP; 0.4 mm dUTP; 0.3 μm each primer; and 1.25 units of AmpliTaq Gold DNA polymerase (Perkin-Elmer Applied Biosystems). Five microliters of each diluted reverse transcriptase sample were added to 20 μl of the PCR master mix. The thermal cycling conditions comprised an initial denaturation step at 95°C for 10 min and 50 cycles at 95°C for 15 s and 65°C for 1 min.

RFS was determined as the interval between diagnosis and detection of the first relapse (local and/or regional recurrence and/or metastasis). Clinical, histological, and biological parameters were compared using the χ² test. Spearman’s rank correlation test was used to study relationships between continuous variables. Survival distributions were estimated by using the Kaplan-Meier method (30), and the significance of differences between survival rates was determined using the Log-rank test. Multivariate analysis using Cox’s proportional hazards model was used to assess the independent contribution of each variable to RFS (31). Differences between groups were judged significant at confidence levels >95% (P < 0.05).

To determine the cutoff point for altered DNMT expression in breast cancer tissue, the N_DNMT value for each DNMT was determined in six normal breast RNA samples. N_DNMT values fell between 0.65 and 1.61 (1.15 ± 0.34) for DNMT1, between 0.63 and 1.53 (1.09 ± 0.31) for DNMT3A, and between 0.94 and 1.37 (1.20 ± 0.22) for DNMT3B. As N_DNMT values for overexpression cutoffs (determined as the mean ± 5 SD) were 2.9, 2.6, and 2.3 for DNMT1, DNMT3A, and DNMT3B, respectively, a common N_DNMT value of 3 was chosen to define overexpression of all of the DNMTs in breast tumor RNA samples.

Transcripts of the three DNMTs were detected and quantifiable in all of the breast tumors. Means, medians, and ranges of mRNA values are shown in Table 3. The highest expression range was observed with DNMT3B (81.8-fold between the tumor with the lowest and the tumor with the highest DNMT3B mRNA levels); DNMT1 and DNMT3A had similar, more moderate expression ranges (16.6- and 14-fold, respectively).

We sought links between continuous mRNA values of each DNMT by using the Spearman rank correlation test. mRNA levels of the three DNMTs correlated strongly with each other, as follows: DNMT1 with DNMT3A (r = +0.421, P = 2.3 × 10⁻⁶), DNMT1 with DNMT3B (r = +0.363, P = 4.3 × 10⁻⁵), and DNMT3A with DNMT3B (r = +0.521, P < 10⁻⁷).

Using the overexpression cutoff of N_DNMT > 3, DNMT1 and DNMT3A overexpression was only observed in 7 (5.4%) and 4 (3.1%) tumors, respectively, whereas DNMT3B overexpression was observed in 39 tumors (30%; Table 3). No DNMT mRNA underexpression (N_DNMT < 2 SD below the mean for normal breast tissue) was observed.

As few tumors overexpressed DNMT1 or DNMT3A, we focused on the DNMT3B gene and sought links between DNMT3B mRNA expression status and clinical, pathological, and biological parameters (Table 4). DNMT3B overexpression was significantly associated with SBR histological grade III (P = 0.002), ERα negativity (P = 0.0015), and strong expression of the proliferation marker MKI67 (P = 3 × 10⁻⁶). We found no link between DNMT3B expression status and age, menopausal status, lymph node involvement, macroscopic tumor size, or ERβ status.

In the series of 130 patients, the 39 patients whose tumors overexpressed DNMT3B tended to have shorter RFS (P = 0.12, data not shown).

We then investigated the prognostic value of DNMT3B expression status according to adjuvant treatment. In the subgroup of patients who received chemotherapy-based adjuvant treatment (57 patients, 35.1% of whose tumors overexpressed DNMT3B), patient outcome was not significantly associated with DNMT3B expression status (Log-rank test, P = 0.18; Fig. 1,A). In contrast, among the 53 women who received adjuvant hormone treatment, the 17 whose tumors overexpressed DNMT3B had significantly shorter RFS than the 36 patients with normal DNMT3B expression (Log-rank test, P = 0.0064; Fig. 1 B) [5-year RFS, 58.8% (46.9–70.7) versus 88.6% (83.2–94); 10-year RFS, 34.3% (21.4–47.3) versus 70.4% (61.9–78.9)].

To confirm the prognostic value of DNMT3B expression status in patients who receive adjuvant hormone therapy, we studied an independent well-characterized series of 98 postmenopausal women with ERα-positive breast tumors who exclusively received tamoxifen as hormone therapy.

DNMT3B overexpression was again associated with shorter RFS in this independent series (Log-rank test, P = 0.0036; Fig. 1 C) [5-year RFS, 55% (43.88–66.12) versus 78.7% (73.77–83.63); 10-year RFS, 38.5% (27.36–49.64) versus 71% (64.76–77.24)].

Using a Cox proportional hazards model, we assessed the prognostic significance of DNMT3B expression status in combination with the three other prognostic parameters that were significant in this series, namely histopathological grade, lymph node status, and macroscopic tumor size (Table 1). The prognostic significance of histopathological grade and lymph node status persisted, whereas that of macroscopic tumor size and DNMT3B status disappeared (data not shown).

We then sought relationships between mRNA levels of the three DNMTs and genes involved in the DNMT pathway in a subgroup of 46 breast tumors from the series of 130 patients (Table 5).

mRNA levels of the DNMTs and genes with CpG islands showed a single negative relationship, between DNMT3B and ERα. DNMT1 mRNA levels correlated positively with those of E-cadherin, whereas DNMT3A mRNA levels correlated positively with those of MASPIN and RB1.

Among the genes whose expression has been found to be dysregulated in an in vitro model of DNMT1 expression loss (22), DNMT3A mRNA levels correlated positively with those of JUNB and MYCN, whereas DNMT3B mRNA levels correlated positively with those of EREG.

Among the genes involved in the DNMT downstream pathway, MECP2 mRNA levels correlated positively with those of both DNMT1 and DNMT3A, whereas HDAC1 only correlated with DNMT3A.

Finally, MKI67 mRNA levels correlated positively with those of both DNMT3A and 3B but not with those of DNMT1. MKI67 mRNA levels also correlated positively with EREG mRNA levels but negatively with ERα.

Epigenetic events such as DNA methylation are thought to be a major early event in tumor development (1). During breast tumorigenesis, dysregulated expression of the three functional DNMTs, which catalyze cytosine methylation, may therefore be of importance in dysregulating gene expression and especially that of tumor suppressor genes.

Here, we used real-time quantitative RT-PCR assays to study the mRNA expression of the three functional DNMTs in a series of 130 primary unilateral nonmetastatic breast carcinomas. A major advantage of real-time RT-PCR in this setting is its wide linear dynamic range, which is suitable for analyzing genes that show a wide range of expression among tumors.

We found that DNMT1, 3A, and 3B mRNA levels correlated positively with each other, suggesting a common regulation pathway. mRNA levels of DNMT3B showed the highest expression range (81.8-fold), DNMT3B overexpression status being observed in 30% of the 130 patients; in contrast, only 5.4 and 3.1% of the patients’ tumors overexpressed DNMT1 and DNMT3A, respectively. Previous studies have shown that DNMT overexpression is common in different tumor types (7, 8, 9, 10, 11, 12, 13, 14). Our results, which are consistent with these previous reports, point to a predominant role of DNMT3B in breast tumorigenesis, relative to DNMT1 and DNMT3A. We also investigated the mRNA expression of the recently discovered DNMT3L gene, which may regulate the DNA methylation process (32). DNMT3L mRNA levels were very low, only detectable but not quantifiable by means of real-time RT-PCR in a subgroup of 46 breast tumors from this series (data not shown), arguing against a role of DNMT3L in breast tumorigenesis.

We sought relationships between DNMT3B expression status and clinical, pathological, and biological parameters. We found significant relationships between DNMT3B overexpression and histological grade III (P = 0.002), ERα negativity (P = 0.0015), and strong MKI67 expression (P = 3 × 10⁻⁶), suggesting that DNMT3B may be a marker of both tumor aggressiveness and proliferation. However, as in vitro studies point to the involvement of DNMTs in proliferation and immortalization (33, 34), DNMT overexpression may be a cause rather than a consequence of tumor proliferation (7, 14, 35).

The in vitro study by Vertino et al. (15) of transfected cell lines showed a significant relationship between hypermethylated CpG islands and strong DNMT expression, but most in vivo studies investigating CpG island methylation status in various cancers have failed to confirm this relationship (7, 13, 14, 16). Here, we investigated the influence of DNMT mRNA expression status on target gene expression. We first sought relationships between each of the three DNMTs and well-known genes in which CpG islands are associated with the promoters. Few significant relationships were found. The expected relationship between DNMT overexpression and gene silencing, reflected by a negative correlation, was only observed between DNMT3B and ERα. However, this relationship may simply be attributable to: (a) the positive relationship between DNMT3B and MKI67; and (b) the negative relationship between MKI67 and ERα (Table 5). Three unexpected links were also observed, namely DNMT1 with E-cadherin and DNMT3A with MASPIN and RB1.

Recently, Jackson-Grusby et al. (22) showed that loss of DNMT1 expression was associated with up-regulation of E1A, EREG, GRO1, and JUNB but also with down-regulation of COX26, H19, IGF2, and MYCN. No such relationships were observed with DNMT1 in our breast tumor series, but DNMT3A and MYCN expression showed similar trends, in partial agreement with the in vitro model. We observed two positive correlations, namely DNMT3A with JUNB and DNMT3B with EREG. This latter relationship was probably caused by the positive correlation between EREG and MKI67. These results suggest that, although DNMTs may cause gene inactivation through CpG island hypermethylation, increased DNMT expression is not alone sufficient to repress target gene expression (in a dose-dependent manner) and that other factors are therefore involved in gene silencing.

Methyl CpG binding protein or histone deacetylase family genes, along with DNMTs, may also be of importance in tumorigenesis (36). Here, MECP2 expression was found to correlate positively with both DNMT1 and DNMT3A expression, whereas HDAC1 only correlated positively with DNMT3A. It is noteworthy that these positive links were independent of proliferation status, as MKI67 was not concomitantly up-regulated. These results suggest that DNMTs and genes involved in the regulation of nucleosome remodeling may cooperate in a specific manner.

The shorter RFS associated with DNMT3B overexpression was first observed in a subgroup of patients who had received at least one adjuvant hormone therapy (Log-rank test, P = 0.0064). Although the prognostic value of DNMT3B overexpression was confirmed in an independent well-characterized series of postmenopausal patients with ERα-positive breast tumors who exclusively received tamoxifen as adjuvant therapy (Log-rank test, P = 0.0036), it did not persist in multivariate analysis. This may be explained by the relationship between DNMT3B overexpression and histopathological tumor aggressiveness (SBR grade III; Table 1).

In conclusion, tumor DNMT expression status may have important therapeutic implications. Indeed, re-expression of promoter-methylated genes has been obtained with DNMT inhibitor treatment of cell lines in vitro, suggesting that DNMT pathway antagonists could be of therapeutic use (37). We found that DNMT3B was the most frequently overexpressed DNMT gene in primary breast tumors but failed to identify target genes with corresponding underexpression. Further study of expression pattern of other genes involved in the DNMTs pathway and the relationship with gene silencing during breast tumorigenesis is necessary.

We thank the Centre René Huguenin staff for assistance in patient care and specimen collection.