BRCA1-mutated breast cancer displayed a hypermethylated E2F1 motif and promoter region

To investigate DNMT1 transcriptional regulation through epigenetic mechanisms, we assessed the DNA methylation status of the DNMT1 core promoter region (from -99 to +521, with +1 denoting the transcription initiation site) in BRCA1-mutated and non-mutated breast cancer, and their adjacent normal breast tissues. As shown in Figure1Aiii and1B, BRCA1-mutated breast cancer exhibited global promoter hypermethylation (P = 0.044; Figure1D), especially around the E2F1 motif (P = 0.017; Figure1C). No similar changes were observed in non-BRCA1-mutated breast cancer (Figure1Aii).

Low DNMT1 transcript levels showed a significant inverse correlation with hypermethylated E2F1 motif in BRCA1-mutated breast cancer

Real-time PCR and immunohistochemical analysis showed that the levels of DNMT1 mRNA and protein were decreased in BRCA1-mutated breast cancer, compared to adjacent normal breast tissues (P < 0.05; Figure2C and D), DNMT1 protein levels was further confirmed by western blotting [see Additional file1]. Although there was no significant difference in mRNA levels between non-BRCA1-mutated breast cancer and adjacent normal breast tissues (Figure2A, P = 0.513), the protein expression of DNMT1 was upregulated (P < 0.05; Figure2B). In addition, we analyzed the correlation between DNMT1-positive cells or mRNA levels and the number of methylated sites within the E2F1 motif (+182) or DNMT1 core promoter region (-99 to +521) in BRCA1-mutated breast cancer and adjacent normal breast tissue (Figure2Ei-Eiv). It is, however, interesting to note that only a significant inverse correlation was observed between DNMT1 mRNA levels and the methylated E2F1 motif (R = 0.755, P < 0.001; Figure2Eiv).

Hypermethylated E2F1 motif is a key regulatory mechanism for DNMT1 transcription in BRCA1-mutated breast cancer

Based on the results above, we next explored the relationship between E2F1 motif methylation and DNMT1 expression in a large number of BRCA1-mutated breast cancer specimens. Similar results were observed using methylation-specific PCR (Figure3Ai), and BRCA1-mutated cancer specimens were classified into unmethylated and methylated groups for comparison of the protein expression of DNMT1. The methylated group showed significantly lower expression of DNMT1 in comparison to the unmethylated group (P = 0.0167; Figure3Aii). To further confirm the role of cytosine located in the E2F1 motif, a point mutation of cytosine to thymine was inserted in the E2F1 element (Figure3Bi). Then, the consensus E2F1 (con.E2F1) and mutated E2F1 (mut.E2F1) were transiently transfected into 293 T cells, and primary non-mutated and BRCA1-mutated breast cancer and their corresponding normal breast cells. Notably, only in BRCA1-mutated breast cancer cells was the E2F1 motif found to be the critical element for DNMT1 transcription (Figure3Bii).

Recently, a substantial body of evidence has been reported which suggests that most of the known genes contain specific motifs, such as G-quadruplex in their promoter regions, which can modulate gene transcription by affecting the binding of histones[14]. Therefore, the CD spectra were used to gain information on whether the methylated E2F1 motif can influence the structure of the DNMT1 promoter. However, our results showed that the methylation of cytosine located in a CpG within the E2F1 motif may not affect the structure of DNMT1 (Figure3C).

Loss of H3K9ac and E2F1 enrichment around the hypermethylated E2F1 motif in BRCA1-mutated breast cancer

To obtain further understanding of the regulatory potential of the crosstalk between DNA methylation and histone modification in controlling DNMT1 transcription, we examined the active histone markers H3K9ac, H3K18ac, H3K27ac, H3K4me1, H3K4me2, H3K4me3, H3K36me3 and H3K79me, and the repressive histone markers H3K9me, H3K9me2, H3K9me3, H3K27me, H3K27me2 and H3K27me3 in the core promoter of DNMT1, especially around the E2F1 motif; we also focused on the enrichment of the transcription factor E2F1, due to the fact that the hypermethylated E2F1 motif was observed in BRCA1-mutated breast cancer (Figure1Aiii). Chromatin immunoprecipitation analysis indicated that the levels of H3K9ac and E2F1 around the E2F1 motif were only significantly decreased in BRCA1-mutated breast cancer (Figure4A and B). These results, together with the methylation data in Figures1,2 and3, suggest that DNMT1 transcription is associated with changes in epigenetic features, including decreased H3K9ac and E2F1 enrichment, and the hypermethylated E2F1 motif. However, no differences exist in histone markers around the E2F1 motif between normal and cancer tissues in BRCA1 wild type cases [see Additional file2].

H3K9ac and E2F1 present in the E2F1 motif are responsible for the transcriptional regulation of DNMT1 in BRCA1-mutated breast cancer

As described previously by Jin[15], we observed that only the combined knockdown of GCN5 and PCAF can specifically induce a decrease of H3K9ac around the E2F1 motif (Figure5Ci). Interestingly, knockdown of E2F1 showed a similar effect to knockdown of GCN5 and PCAF (Figure5Ci); therefore, an alternative possibility may be that E2F1 can recruit GCN5 and PCAF to the E2F1 motif, which was confirmed by co-immunoprecipitation analysis (Figure5D). As shown in Figure5Cii, knockdown of E2F1 was an effective way to reduce E2F1 enrichment, but did not change the levels of H3K9ac around the E2F1 motif. Meanwhile, we observed that knockdown of GCN5, PCAF and E2F1 had no detectable effect on the cell morphology and proliferation (Figure5A and B). Of particular note, as described in Figure4, the hypermethylated E2F1 motif of the DNMT1 promoter was accompanied by loss of H3K9ac and E2F1 enrichment in BRCA1-mutated breast cancer. After the deletion of H3K9ac and/or E2F1 enrichment in BRCA1-mutated breast cancer, the transcription of DNMT1 was significantly down-regulated (Figure5Ev).

Correlation of E2F1 motif methylation with clinicopathological characteristics in BRCA1-mutated breast cancer

Multivariate and univariate analysis of overall survival for patients with BRCA1-mutated breast cancer

We analyzed the overall survival to assess the prognostic significance. Multivariate Cox regression analysis indicated that lymph node metastasis was an independent prognostic factor for predicting the overall survival of BRCA1-mutated breast cancer patients (Table2, P = 0.038). We also performed Kaplan-Meier analysis and log-rank tests for overall survival to define prognostic subgroups. The results revealed that the significant prognostic factors were histological grade (Figure6D, P = 0.004), lymph node metastasis (Figure6E, P = 0.004), E-cadherin (Figure6K, P = 0.015) and DNMT1 methylation (Figure6L, P = 0.016). Moreover, patients with premenopausal (Figure6B, P = 0.155), estrogen receptor-negative (Figure6F, P = 0.170) and p53-negative (Figure6I, P = 0.089) breast cancer showed a trend for poor overall survival, although this was not statistically significant. No significant difference in overall survival was found among patients with different ages at diagnosis, tumor size, progesterone receptor status, c-erbB-2 status or Ki67 status (Figure6A, C, G, H and J).

Variable	RR	95%CI	P
Age at diagnosis (> 50 y vs ≤ 50 y)	1.059	0.982-1.143	0.137
Menstrual status (Post vs Pre)	0.258	0.043-1.542	0.138
Tumor size (> 5 cm vs ≤ 5 cm)	0.389	0.107-1.413	0.151
Histological grade (III vs I-II)	2.628	0.642-10.761	0.179
LN status (Pos vs Neg)	3.511	1.074-11.481	0.038
ER status (Pos vs Neg)	0.561	0.123-2.554	0.455
PR status (Pos vs Neg)	1.558	0.420-5.775	0.507
c-erbB-2 status (Pos vs Neg)	1.429	0.422-4.832	0.566
p53 status (Pos vs Neg)	0.437	0.112-1.703	0.233
Ki67 status (Pos vs Neg)	0.628	0.120-3.271	0.580
E-cadherin status (Pos vs Neg)	0.651	0.158-2.678	0.552
DNMT1 methylation (M vs UM)	1.386	0.337-5.699	0.651

Post, postmenopausal; Pre, premenopausal; LN, lymph node; ER, estrogen receptor; PR, progesterone receptor; Pos, positive; Neg, negative; M, methylated; UM, unmethylated; RR, relative risk; CI, confidence interval.

Patients and tissue collection

This study was approved by the Institutional Review Board at China Medical University. Eighty-five invasive ductal carcinomas from BRCA1 mutation carriers and fifteen invasive ductal carcinomas from non-BRCA1 mutation carriers were enrolled between 2007 and 2009; all patients gave informed consent. Fresh breast cancer and adjacent normal breast tissues were obtained at the time of primary surgery before any chemotherapy or radiotherapy had been administered. Hematoxylin and eosin staining of the samples for histopathological diagnosis and grading were performed by three staff pathologists using the Nottingham Combined Histologic Grade. The tumor stages were classified according to the National Comprehensive Cancer Network guidelines. All patients were screened for BRCA1 mutations by multiplex polymerase chain reaction with complete sequence analysis as previously reported[22–24]. Their characteristics are given in Table1.

Cell culture, lentiviral infection and cell proliferation assay

Detailed isolation and cultivation protocols were established as previously described[25]; breast cancer and normal cells were maintained in CnT-27 mammary epithelium medium (CELLnTEC, Bern, Switzerland). Human 293 T cells were maintained in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (Invitrogen, CA USA). Lentiviral vectors expressing short hairpin RNAs (shRNA) against BRCA1 (NM_007299) were obtained from Genechem Co., Ltd (Shanghai, China), and synthesized as follows: Forward: 5-CCGGAACCTGTCTCCACAAAGTGTGCTCGAGCACACTTTGTGGAGACAGGTTTTTTTG, and Reverse: 5-AATTCAAAAAAACCTGTCTCCACAAAGTGTGCTCGAGCACACTTTGTGGAGACAGGTT. The non-silencing siRNA sequence (TTCTCCGAACGTGTCACGT) was used as a negative control. For overexpression of BRCA1, the open reading frame of BRCA1 (NM_007299) was cloned into the lentiviral vector GV287 (Ubi-MCS-3FLAG-SV40-EGFP) (GeneChem). The shRNA lentiviral particles of GCN5 (sc-37946-V), PCAF (sc-36198-V) and E2F1 (sc-29297-V) were purchased from Santa Cruz Biotechnology (CA, USA). Transfections were performed using the polybrene and enhanced infection solution (Genechem) according to the manufacturer’s recommended protocol, the knowdown effiency for the double-knockdown and triple-knockdown were shown in Additional file6. After 48 hours of infection, cell proliferation was determined using the Cell-Light™ EdU Apollo®643 In Vitro Imaging Kit (Ribobio, Guangzhou, China) following the instructions provided by the manufacturer.

DNA methylation analysis

Genomic DNA from the breast cancer and adjacent normal breast tissues was extracted using a TIANamp Genomic DNA kit (Tiangen Biotech, Beijing, China). Sodium bisulfite conversion, polymerase chain reaction (PCR) amplification and general experimental procedure were performed as previously described[22]. The specific primer sequences for bisulfite sequencing and methylation-specific PCR are described in Additional file7. Global DNA methylation was measured as previously described[26].

Real-time PCR and immunohistochemistry analysis

Real-time PCR and immunohistochemistry were performed as previously described[22]. The specific primer sequences for real-time PCR are listed in Additional file7. The primary antibody for immunohistochemistry was rabbit anti-DNMT1 of human origin, and are listed in Additional file8 (1:200; Santa Cruz Biotechnology).

Chromatin immunoprecipitation (CHIP), site-directed mutagenesis, transfection and dual-luciferase reporter assay

CHIP, site-directed mutagenesis, transfection and dual-luciferase reporter assays were performed as previously described[27]. The specific primer sequences for site-directed mutagenesis and CHIP are provided in Additional file7. The specific antibodies for CHIP are provided in Additional file8.

Co-immunoprecipitation (CO-IP) and immunoblotting

CO-IP was performed using an immunoprecipitation kit (Invitrogen) according to the manufacturer’s recommended protocol. Then, the cell lysates and immunoprecipitates were analyzed by immunoblotting. The specific antibodies for CO-IP and immunoblotting are provided in Additional file8.

Circular dichroism (CD) spectra

The CD spectra were obtained on a Jasco J-810 spectropolarimeter at 25°C using a 0.1 cm path length cell; data were collected with a 2 nm slit width from 350 to 200 nm at 0.5 nm intervals and averaged over three scans. CD experiments were carried out on DNA samples (5 μM) using a buffer containing 0.2 M phosphate buffer (pH 7.0) in the presence of 100 mM Na⁺ or K⁺. The DNA samples were annealed by heating to 95°C for 5 min followed by cooling to room temperature over 10 h before analysis. The DNA sequence is as follows: 5-CCCCTCCCCATCGGTTTC“C (CH₃ or non-CH₃)” GCGCGAAAAGCCGGGGCGCC, and was synthesized by Sangon Biotech Ltd (Shanghai, China).

Statistical analysis

Regression analysis was used to examine the possible relationship between DNMT1 mRNA or protein levels, and the status of promoter methylation. The association between clinicopathological features and DNMT1 promoter methylation was determined using the Fisher’s exact test. Univariate analysis of survival was performed using the Kaplan-Meier method. Multivariate Cox regression analysis was performed to identify the independent prognostic factors for overall survival. The data are presented as means ± SD. Statistical differences in the data were evaluated by Student’s t test or one-way ANOVA as appropriate, and were considered significant at P < 0.05.