Epigenetic Regulation by Z-DNA Silencer Function Controls Cancer-Associated ADAM-12 Expression in Breast Cancer: Cross-talk between MeCP2 and NF1 Transcription Factor Family

Metastatic spread of cancer is regarded as the greatest hurdle to cancer cure. In the metastatic cascade, multiple interrelated pathways are activated, which include proteolytic breakdown of the tumor membrane and spreading of cancer cells into the surrounding tissues, migration, and successful attachment of the escaped cancer cells at new sites and colonization and proliferation of cancer cells at secondary locations. Advances in cancer research indicate that genetic mutations along with epigenetic alterations also contribute to metastasis-related gene expression. In many human cancers, markedly high-level expression of a multifunctional protein, ADAM-12, is detected (1–7). In patients with breast and bladder cancer, increase of ADAM-12 is shown to correlate with disease progression and tumor stage (2, 5, 8) and in animal models, ADAM-12 is found to be required for aggressive tumor progression (3, 9).

ADAM-12 is capable of supporting several steps of the cancer-metastasis cascade. It proteolytically degrades several components of extracellular matrix (2), facilitates cell–cell and cell–extracellular matrix (ECM) attachments (10), and promotes cell proliferation by increasing bioavailability of growth factors (2, 5). Incidentally, 3 somatic mutations in ADAM-12 gene are frequently seen in breast cancers (11). These mutations cause mutant ADAM-12 proteins to be retained in the endoplasmic reticulum (ER) rather than at the cell surface (12). It is speculated that increased accumulation of ADAM-12 in the ER may be linked to tumor growth, which is yet to be experimentally determined. Normal cellular expression of ADAM-12 usually is very low and highly regulated. Previously, we reported a Z-DNA–forming negative regulatory element (NRE) that acts as transcriptional silencer of ADAM-12 expression (13). Here, we provide evidence that marked increase of ADAM- 12 level in breast cancer cells is, at least in part, due to loss of NRE-silencer function. The results reveal a novel mode of epigenetic regulation, which involves cross-talk between MeCP2, a prominent epigenetic regulator, and 2 members of the NF1 family of transcription factors and interaction of MeCP2:NF1 complex with Z-DNA–forming dinucleotide repeat sequences in regulating ADAM-12 expression.

MCF-10A, MDA-MB-231, MDA-MB-468, MCF-7, DU4475, and Hs578T cells were obtained from American Type Culture Collection (ATCC) in 2010, cultured, and stored following ATCC protocol. The cells have been authenticated by short tandem repeat DNA profiling method by using Cell ID system (Promega). PCR products of genomic DNA from each cell line were detected on a capillary electrophoresis equipment. The results were analyzed by using GeneMapper 4.0 software. Cells were tested at 4- to 5-month intervals and last tested in June 2012. Normal and cancer human breast tissue lysates were obtained from IMGENEX. Canine breast cancer tissues were obtained from the University of Missouri Veterinary Medical Teaching Hospital (Columbia, MO). Normal canine mammary tissues were obtained from cadavers. Histologic analysis confirms adenocarcinoma in canine breast cancer tissues (Supplementary Fig. S1). All procedures were approved by the Animal Care and Use Committee.

Chloramphenicol acetyltransferase (CAT) assay was conducted following transfection of cells with reporter plasmid and pSVβ-gal (Promega) DNA (for normalization of transfection efficiency), as described (13).

A human breast cDNA expression library in λgt11 (Clontech) was screened by ligand interaction method, using a ³²P-labeled concatenated ADAM-12 NRE (+100/+190) DNA, as described earlier (14). Positive clones were analyzed by DNA sequencing.

Total RNA isolated by guanidinium thiocyanate method was fractionated in a 1% agarose gel and after transfer onto a nitrocellulose membrane, hybridized to ³²P-labeled ADAM-12 and β-actin cDNA probes as described (13).

Two ADAM-12–CAT reporters, wt ADAM-12 containing ADAM-12 sequences from −1600 to +190 and ΔNRE ADAM-12 with deletion of sequences from +100 to +190, were described earlier (13). Two additional ADAM-12 promoter constructs were generated by PCR amplification. The sequence of non Z-DNA element that replaced Z-DNA at +124 to +159 was: 5′- GCATGCATTCAGGAACCATCGAACTTAGTCAATCGG-3′. The sequence of mutant NF1 oligonucleotide that replaced NF-1 binding region at +101 to +122 was: 5′-GTCAAGCGGGGCTCGTCCAGAA-3′. Underlined letters represent altered nucleotides.

DNA-binding assays were conducted as described earlier (13) using 10 μg of nuclear extracts (NE) prepared from MCF10A, MDA-MB-231, canine cancer, and normal breast tissues. ³²P-labeled ADAM-12 NRE DNA was used as probe. DNA probe was methylated by CpG methyltransferase (New England BioLab) following manufacturer's protocol. In some binding assays, 100-fold molar excess of unlabeled nonspecific (a 30-mer ds-DNA containing random sequences) or homologous-specific (ds-DNA containing same sequence as the probe or smaller fragment thereof as indicated in figure legends) competitor DNA or antibodies against NF1-C (N. Tanese, New York University, Langone Medical Center, New York, NY), NF1-X (U.S. Singh, Uppsala University, Uppsala, Sweden) MeCP2 (P. L. Jones, Boston Biomedical Research Institute, Watertown, MA), ADAR1, DAI/DLM1/ZBP1, and normal IgG were added. Western blots were conducted using 1:3,000 dilution of NF1-C, NF1-X, MeCP2, p-STAT3, and β-actin (Santa Cruz Biotechnology) antibodies. For dephosphorylation assay, protein extracts were treated with alkaline phosphatase (Fermentas).

For chromatin immunoprecipitation (ChIP) assays, lysed solutions were incubated with NF1/C, NF1-X, MeCP2 antibody or control IgG. For PCR, specific primers as described earlier (13) that amplify the nucleotide position from +68 to +193 and as a negative control, sequences from −619 to −328, were used. Re-ChIP assays were conducted by following a method described earlier (15).

Both mRNA and protein analyses revealed that ADAM-12 expression is markedly (5–6 fold) increased in metastatic breast cancer cells as compared with normal mammary epithelial cells (Fig. 1A–C). To evaluate possible role of regulatory region of the gene in its expression, we transfected these cells with plasmids carrying a reporter gene, chloramphenicol acetyl transferase (CAT), whose expression was driven by either wild-type ADAM-12 promoter (−1600/+190) or a truncated promoter (−1600/+100) lacking the Z-DNA–forming NRE sequences (Fig. 1D). Results showed that NRE caused about 5-fold suppression of ADAM-12 expression in MCF-10A cells (Fig. 1E). However, the same NRE caused significantly less transcriptional suppression, 1.5- and 1.8-fold, respectively, in MDA-MB-231 and MDA-MB-468 cells. This finding suggested that the cancer cells probably lack some of the regulatory factors that normally bind to the NRE to suppress transcription of ADAM-12 in normal mammary epithelial cells or binding of some additional factors present in breast cancer cells that may cause reduction of suppression. To test these possibilities, we conducted an NRE-DNA–binding assay.

DNA-binding assay revealed a lower interaction of protein(s) in MDA-MB-231 cells as compared with MCF-10A cells (Fig. 2A, compare lanes 2 and 3). Similarly, proteins in canine breast cancer tissue, which expresses high level of ADAM-12 (Supplementary Fig. 1B), exhibited lower DNA-binding activity (Fig. 2A, compare lanes 7 and 8). These findings suggest that interaction of specific proteins to NRE may be necessary for suppression of ADAM-12 expression, and breast cancer cells may have lost this regulatory process. Identification of NRE-binding proteins may, therefore, provide a valuable clue regarding this regulatory mechanism.

As NRE of ADAM-12 forms Z-DNA structure (11), antibodies against the 2 known mammalian Z-DNA–binding proteins, ADAR1 (16) and DAI/DLM-1/ZBP1 (17) were used in a DNA-binding assay to test whether these proteins are involved in binding to ADAM-12 NRE. No effect of either of these 2 antibodies on the DNA–protein complex (Fig. 2B) suggested that the ADAM-12 NRE-specific complex does not involve these proteins.

To search for NRE-binding proteins, we screened a human breast cDNA expression library with a ³²P-labeled concatenated ADAM-12 NRE DNA and isolated several clones. DNA sequencing revealed MeCP2, NF1-C, and NF1-X as NRE-DNA binding factors.

MeCP2 is a methyl-CpG–binding protein (18–20) and NF1 represents a family of transcription factors that function either as a transcriptional repressor or a transcriptional inducer (20). Interestingly, a conserved NF1-binding element TGGCTTGTGCCA was located within nucleotide positions +100 to +122 of ADAM-12 (Fig. 3A). It is located adjacent to the Z-DNA–forming dinucleotide repeat element that is present within sequences +123 to +160 (Fig. 3A). Given the long-known implication of MeCP2 as repressor of gene expression via recruitment of mSin3A, HDAC1 and histone methyltransferase (18–20), and NF1 in chromatin-mediated transcriptional control (21, 22), we elected to address possible involvement of MeCP2 and NF1 proteins in binding to ADAM-12 NRE.

In the DNA-binding assay, the NRE-specific DNA–protein complex was efficiently supershifted by MeCP2, NF1-C, and NF1-X antibodies (Fig. 3B). Specificity of DNA–protein interaction was verified by competitor oligonucleotide and specific antibodies (Fig. 3B). Normal mammary tissue extracts from canine showed similar pattern indicating similarities between the cultured cells and mammary tissues with regard to MeCP2, NF1-C, and NF1-X proteins (Fig. 3B, lanes 13–18). Super-shift of the same DNA–protein complex by each of these antibodies suggested that MeCP2, NF1-C, and NF1-X proteins cooperatively interact with the ADAM-12 NRE DNA. This finding was further substantiated by the combined antibodies that resulted in further or higher super-shift of the NRE-specific DNA–protein complex (Fig. 3C, compare between lanes 1–7).

As DNA methylation is prevalent in cancer, we tested whether methylation of CpG element of ADAM-12 NRE would enhance the binding of cancer cell–derived MeCP2. Methylated DNA did not increase protein binding to ADAM-12 NRE in either normal or cancer cells (Fig. 3D). Lack of increased binding suggested that function of MeCP2 for suppression of ADAM-12 is not altered by DNA methylation in breast cancer cells, which may explain why ADAM-12 expression remains high in cancer cells.

As MeCP2 and NF-1–binding sites are present side-by-side in the NRE of ADAM-12, to determine their relative contribution, we divided NRE in 2 parts, +100/+122 and +123/+190, containing the NF1- and MeCP2-binding sites, respectively. Surprisingly, there was very little binding of both proteins even in MCF-10A cells that show more avid binding when both sites are present (compare Figs. 4A and B with C). But, as competitors, these small DNA units inhibited DNA–protein complex formation by the full-length probe (Fig. 4C, lanes 4 and 5). Also, specific mutations of NF-1 (+100/+122) and MeCP2 (+123/+190) binding sites markedly affected NRE-mediated transcriptional suppression activity (Fig. 4D). Together, these results indicated that loss of either NF1 or Z-DNA (MeCP2-binding site) sequences severely compromises interaction of both NF1 and MeCP2 proteins. These data also suggest that MeCP2 and NF1, after occupying their respective DNA-binding sites, most likely cooperatively stabilize NRE-specific DNA–protein complex. Consistent with these findings, ChIP assay revealed that both MeCP2 and NF1 proteins interact with the ADAM-12 NRE in vivo (Fig. 5A). Simultaneous presence of MeCP2 and NF1 proteins in the ADAM-12 promoter was examined by re-ChIP assays, which showed the presence of both proteins at the ADAM-12 NRE in MCF-10A cells (Fig. 5B).

Reduced DNA–protein complex with NRE in breast cancer cells (Figs. 2–4), raised the possibility that the levels of MeCP2 and NF1 are altered in breast cancer cells. Western blot analysis revealed 2 closely migrating MeCP2 bands in normal MCF-10A cells but only one in MDA-MB-231 cancer cells (Fig. 6A). Further analysis showed only the upper MeCP2 band in several breast cancer cells (Fig. 6B). In human breast cancer tissue, MeCP2 level is also very low compared with the adjacent normal breast tissue, but unlike cell lines, only a single protein band was detected (Fig. 6B, lanes 6 and 7). Quantitative evaluation of MeCP2 levels revealed a significant reduction of the protein in cancer cells (Fig. 6C, MeCP2-b) and tissues (Fig. 6C, columns 6 and 7). Detection of different MeCP2 proteins in normal and breast cancer cells, to the best of our knowledge, has not been reported earlier. We investigated whether phosphorylation, which often generates multiple bands for many proteins in a PAGE, accounts for the difference in the protein pattern seen in Fig. 6A and B. Dephosphorylation reaction did not change MeCP2 migration (Fig. 6D) suggesting that a mechanism, other than phosphorylation, might be involved for the difference of MeCP2 in MCF-10A and the breast cancer cells and addressed in Discussion.

A 50 kDa NF1-X protein was detected only in MCF-10A cells and human normal breast tissue (Fig. 6E), whereas an approximately 74–75 kDa NF1-C protein was seen only in cancer cells and human breast cancer tissue (Fig. 6F). Incidentally, reduced NF1-X level has been linked to increased ADAM-12 expression during heat-induced stress in U-251 MG glioblastoma cells (23) and a 74 kDa NF1-C protein has been detected during early mammary gland involution but not in lactating mammary gland (24).

We report a novel finding that delineates mechanism of ADAM-12 overexpression in breast cancer cells. Our investigation has revealed a unique epigenetic regulatory process in which Z-DNA element plays an essential role. The loss of Z-DNA–mediated silencer function in breast cancer cells is a major cause for marked increase of the expression of ADAM-12 leading to cell proliferation and metastasis. Furthermore, the results have revealed that MeCP2, a prominent epigenetic regulator, in association with NF1 family of transcription factors, interacts with the Z-DNA element possibly in a methylation-independent manner.

Z-DNA is an unusual left-handed conformation formed in the DNA by a stretch of alternating purine-pyrimidine dinucleotide repeats, such as GC, TG, or TA repetitive sequences (n ≥ 12 units; ref. 25). Z-DNA elements are predominantly concentrated near the transcription start sites (26) and have been implicated in gene regulation (27–30), chromatin remodeling (31), recombination (32, 33) and large-scale deletions (34). It is interesting to note that many cancer-associated genes, including EGFR, ER-α, Cyr61, ACCA, prolactin, MMP-9, heme oxygenase 1, and HMGA2, contain dinucleotide repeat elements that have been identified as regulatory elements (Supplementary Table S1); but how these sequences regulate gene expression remained poorly understood. The findings, reported here, provide a regulatory mechanism that may explain how these cancer-associated genes could be regulated in malignant cells.

MeCP2 generally acts as a transcriptional repressor. Two global mechanisms of gene regulation, DNA methylation, and histone deacetylation can be linked by MeCP2 (18, 19). It also links histone methyltransferase and the DNA methyltransferase DNMT1 (20, 35) and thus acts as a mechanistic bridge between DNA methylation, histone deacetylation, and histone methylation. It also associates with the BAF/SWI/SNF chromatin remodeling complex to repress gene expression (36). The NF1 family of transcription factors has been shown to interact with hormone receptor (22), histones (37, 38), epigenetic modifiers, such as histone deacetylase (HDAC; refs. 28, 39) and BAF/SWI/SNF (23) during regulation of gene expression. We show that binding of MeCP2 at Z-DNA site and recruitment of NF1 to the adjacent site (Fig. 3) are 2 events that lead to the suppression of ADAM-12 expression.

Our finding of 2 MeCP2 bands in normal MCF-10A breast epithelial cells, but only the slower migrating MeCP2 band in several breast cancer cells, is intriguing. Phosphorylation of MeCP2 does not seem to play any role in the appearance of these alternate forms (Fig. 6C). Two MeCP2 bands could be the translation product of MeCP2_e1 and MeCP2_e2 splice variants, which differ by 12 amino acids (40). Although the longer MeCP2_e1 isoform is more abundant in brain (40), we find the shorter isoform seems to be predominant in normal breast epithelial cells (Fig. 6A). Similarly, in normal breast tissue lysate, only one MeCP2 band that comigrates with the shorter isoform (MeCP2-b) was detected. This isoform is markedly reduced in cancer cells. Its low level in breast cancer cells offers an interesting possibility that this MeCP2 isoform might be involved in suppressing ADAM-12 expression in normal breast cells.

In summary, we have uncovered a novel interplay between Z-DNA, epigenetic regulator MeCP2, and NF1 family of transcription factors in regulating gene expression. A close association between MeCP2 and NF1 proteins at Z-DNA element of ADAM-12 is found to be necessary for the suppression of ADAM-12 expression. MeCP2 deficiency in breast cancer cells results in the loss of this crucial suppression mechanism leading to overexpression of ADAM-12. Such a phenomenon has not been reported earlier. Although Z-DNA–forming sequences have been identified in many prominent cancer-related genes (Supplementary Table S1), how these sequences regulate gene expression remained unknown. Our results may provide a molecular basis for interpretation of Z-DNA–forming dinucleotide repeat length polymorphism-associated cancer susceptibility and unravel the ultimate relevance of this epigenetic mechanism to cancer.

No potential conflicts of interest were disclosed.

Conception and design: A. Ray, B.K. Ray

Development of methodology: A. Ray, S. Dhar, B.K. Ray, A. Rich

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.J. Henry, A. Ray

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Ray, B.K. Ray, A. Rich

Writing, review, and/or revision of the manuscript: C.J. Henry, A. Ray, B.K. Ray

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.J. Henry, S. Dhar, A. Ray, B.K. Ray

Study supervision: A. Ray, B.K. Ray

The authors thank N. Tanese, P.L. Jones and U.S. Singh for generous gift of NF1/C, MeCP2, and NF1-X antibodies, respectively.

This study was supported partly by grants from U.S. Army Medical Research and Materiel Command, University of Missouri Research Board, and University of Missouri, College of Veterinary Medicine.

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.