Hypomethylation of the Synuclein γ Gene CpG Island Promotes Its Aberrant Expression in Breast Carcinoma and Ovarian Carcinoma¹

Recently, methylation of DNA at CpG dinucleotides has been recognized as an important mechanism for regulation of gene expression in mammalian cells. Methylation of cytosines in the CpG sequence located in the promoter region or exon 1 is thought to ensure the silencing of certain tissue-specific genes in nonexpressing cells. Aberrant methylation is now considered an important epigenetic alteration occurring in human cancer. Hypermethylation of normally unmethylated tumor suppressor genes correlates with a loss of expression in cancer cell lines and primary tumors (1, 2, 3, 4). On the other hand, failure to repress genes appropriately by abnormal demethylation of tissue-restricted genes or by hypomethylation of proto-oncogenes could result in the loss of tissue specificity and could promote cancer formation (5, 6). Numerous investigations suggest that hypermethylation of promoter CpG islands correlates with transcriptional inhibition in neoplasmas. Most of the hypermethylated CpG islands are located in the promoter region of tumor suppressor genes or DNA repair genes, and DNA methylation is associated with the loss of gene expression in cancer cell lines and primary tumors. However, compared with an extensive list of tumor suppressor genes or cell cycle-regulated genes that are silenced in cancer cells to date, only a few genes have been shown transcriptionally reactivated by DNA demethylation in cancer (7). Although the global hypomethylation in cancers has been observed for several years, it has not received much attention until recently. Initially, hypomethylation of human growth hormone, α-globin, and γ-globin in cancers was observed by Feinberg and Vogelstein (8). These genes are methylated in normal tissues and become hypomethylated in cancers. Later on, the study was extended to various grades of tumors like benign and malignant colon neoplasmas, and hypomethylation was observed in both types of tumors (9). The hypomethylation and overexpression of proto-oncogenes c-Myc and c-Jun has been detected in chemically induced tumors in mouse liver (10). Rosty et al. (11) has recently reported hypomethylation of S100A4 gene in pancreatic cell carcinomas. A correlation of hypomethylation and expression of MN/CA9 (a tumor-associated antigen) was reported by Cho et al. (12) in renal cell carcinomas. Additionally, DNA hypomethylation and overexpression has been shown for MDR in myeloid leukemias (13), BCL-2 in chronic lymphocytic leukemias (14), MAGE-1 in melanomas (15), and c-Ha-RAS in gastric carcinomas (10). Recently, Strichman-Almashanu et al. (16) have identified unique CpG islands that are methylated specifically in normal tissues and not in cancers.

SNCG,³ also referred to as the BCSG1 (breast cancer specific gene 1), is a member of a neuronal protein family synuclein and its expression is highly tissue specific (17, 18, 19, 20, 21). The SNCG protein is abundantly expressed in the peripheral nervous system such as primary sensory neurons, sympathetic neurons, and motor neurons. However, this tissue specificity was apparently lost during the disease progression of breast cancer and ovarian cancer, because this gene, normally silent in breast tissue and ovary, became abundantly expressed in the vast majority of the advanced staged breast carcinoma and ovarian carcinoma.

The involvement of SNCG in human neoplastic diseases first came to light 5 years ago when SNCG was isolated from a human breast tumor cDNA library and was shown to be overly expressed in infiltrating ductal carcinomas (22). By using in situ hybridization, Jia et al. demonstrated a stage-specific expression pattern of SNCG mRNA, varying from virtually no detectable expression in normal or benign breast tissues to low level and partial expression in low-grade ductal carcinoma in situ (DCIS) to high expression in advanced infiltrating carcinomas. Immunohistochemical studies to examine SNCG protein expression showed a similar pattern in that it was not detected in normal breast tissues but was detected in a high percentage of stage III/IV breast ductal carcinomas (23). SNCG expression in advanced breast carcinomas is not merely adventitious, but plays a positive role in the process of invasion and metastasis. Our groups and others have demonstrated that exogenous expressions of SNCG induced a more aggressive and invasive phenotype in the breast cancer cell line MDA-MB435 (24), and the inhibition of SNCG with SNCG antisense mRNA reversed the malignant phenotypes of T47D cells (25).

The first piece of evidence suggesting a possible role of SNCG in the development of ovarian carcinoma came from an analysis of the expressed sequence tag database. Lavedan et al. (20) noticed that 37% of the human SNCG sequences were originated from an ovarian tumor library, and the rest of the sequences were from brain and breast tumor libraries, thereby predicting that SNCG may also be overexpressed in ovarian tumor. This speculation was confirmed later by Bruening et al. (23), who conducted a study to examine SNCG expression in normal ovarian tissue samples and tissues from ovarian carcinoma. This study found that ovarian epithelial cells or ovarian stromal cells from normal ovaries were not stained with anti-SNCG antibody. In contrast, Bruening et al. showed that 33 (73%) of 45 ovarian carcinomas strongly reacted with anti-SNCG antibody, and the immunoreactivity was exclusively in malignant ovarian epithelial cells.

To elucidate the mechanisms that underlie the abnormal expression of SNCG in breast cancer cells, a 2.2 kb-fragment of human SNCG gene including 1 kb of the 5′-flanking region (−1260 to −170), exon 1 (−169 to +121), and intron 1 (+122 to +935), was isolated in the VA Palo Alto laboratory (26). Our previous studies of SNCG promoter activity in two SNCG-positive and one SNCG-negative breast cancer cell lines suggested that SNCG transcription is primarily controlled by regulatory sequences located in intron 1 and exon 1 but not in the 5′ flanking region.

The intron 1 contains two closely located AP1 recognition sequences. Deletion of these motifs greatly diminished the SNCG promoter activity, suggesting that AP1 is an important transactivator for SNCG transcription in breast cancer cells. Sequence analysis identified a CpG island in exon 1 that contains 15 CpG sites, covering the region −169 to +81, relative to the translation start codon. By using the sodium bisulfite DNA sequencing technique that examined the in vivo methylation pattern of the exon 1 region, we found that the CpG sites within the CpG island and its vicinity were partially and heterogeneously methylated in SNCG-negative MCF-7 cells but unmethylated in SNCG-positive SKBR-3 and T47D cells (26).

These initial observations suggest that demethylation of SNCG exon 1 could play a causative role for the expression of SNCG in cell culture. To determine whether demethylation of the exon 1 is a common molecular determinant responsible for the abnormal expression of SNCG in breast carcinoma and ovarian carcinoma, in the present study, we extensively examined the methylation status of exon 1 and SNCG expression in a panel of breast cancer and ovarian cancer cell lines, and in primary breast tumor and normal breast tissues.

The breast cancer cell lines, AU565, MCF-7, MDA-MB435, MDA-MB231, MDA-MB 468, T47D, SKBR-3, were cultured in RPMI 1640 with 10% FBS. H3922 was grown in Iscove’s modified Dulbecco’s medium in the presence of 10% FBS and BT-20 was grown in DME H-16 50% and F-12 50% media supplemented with 10% FBS. The normal mammary epithelial-derived cell line MCF10A was obtained from American Type Culture Collection (ATCC) and cultured according to the instruction provided by ATCC. Two normal human primary mammary epithelial cell lines 184 and 048R with finite life span were cultured in 1:1 mixture of DME/F-12 supplemented with 0.5% FBS, 10 μg/ml insulin, 5 ng/ml epidermal growth factor, 0.1 μg/ml hydrocortisone, and 1 ng/ml cholera toxin. The third finite life span primary cell line derived from the organoid 240L was cultured in complete MEGM medium (Clonetics) supplemented with 10⁻⁵ m isoproterenol. All HMECs and organoids, and the organoid-derived cell line (240L) were kindly provided by Dr. Martha R. Stampfer at the Lawrence Berkeley National Laboratory, in Berkeley, CA.

The ovarian carcinoma-derived cell lines (A2780 and OVCAR-3, -4, -5, -8) were maintained in DMEM supplemented with 10% FBS, glutamine, and insulin (0.2 IU/ml pork insulin; Novagen; Ref. 27). Human ovarian surface epithelial cell lines were derived as described previously (28). Cells were maintained in a 1:1 mixture of Media 199 and MCDB-105 media, supplemented with 4% FBS and 0.2 IU/ml insulin (29). The life spans of the human ovarian surface epithelial cells (HIO-103, -105, -107, -135) have been extended by ectopically expressing SV40 large T-antigen. These cell lines are nontumorigenic with the exception of HIO-118, which has been shown to form tumors in mice.⁴

The genomic DNA was isolated from various cell lines by using Promega’s wizard DNA isolation kit according to the manufacturer’s instructions. Primary breast tumor tissues and normal breast tissues were obtained after surgical resection and stored frozen at −80° C. The tissues were incubated at 55°C in homogenization buffer containing 50 mm Tris (pH 8.0), 1 mm EDTA, 0.5% Tween 20, and 5 mg/ml proteinase K for 3 h, and then genomic DNA was isolated using Promega’s DNA isolation kit. Donors of tissue specimens agreed to allow their specimens to be used for research purposes. Breast tumors were staged following standard American Joint Committee on Carcinoma/International Union Against Carcinoma tumor-node-metastasis (TNM) methodology.

Two μg of genomic DNA from each sample was modified by sodium bisulfite as described previously (2). The modified DNA was amplified with primer SNCG-S2F and SNCG-S2R covering the region −275 to +140. PCR reactions were performed in a volume of 50 μl containing 1× PCR buffer, 1.5 mm MgCL₂, 0.2 mm dNTP, 25 pm of each primer, and 2.5 units of platinum Taq polymerase (Life Technology Inc.). PCR reaction was carried out at 94°C for 1 min, and 35 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and finally 72°C for 5 min. The 415-bp PCR product was gel purified and ligated into PCR2.1 Topo cloning vector (Invitrogen, Carlsbad, CA). After transformation, individual colonies were picked, and the insert was PCR amplified as described above and sequenced using SNCG-S2R as the primer. Table 1 provide the sequences of the oligonucleotide primers used in this study.

For cell lines, Ultraspec RNA reagent (Biotecxs Laboratory, Houston, TX) was directly added to the monolayer cell culture grown in culture dishes. For isolation of RNA from tumor tissues, 50–200 mg tissues were homogenized in 2–3 ml of Ultraspec RNA reagent on ice using polytron homogenizer at the setting of 3 with three 10-s bursts. RNA was then isolated from the lysate according to the vendor’s protocol. The RT was conducted with random primers (Promega) using Superscript II (Invitrogen). The PCR reaction was carried out at 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s, with initial activation of the enzyme at 94°C for 1 min. Thirty-eight cycles were performed for SNCG and 26 cycles for GAPDH. The PCR was performed using primers SNCG-RT 5′ and SNCG-RT 3′for SNCG and primers GAPDH-RT 5′ and GAPDH-RT 3′ for GAPDH.

Total cell lysates were isolated from cells as described previously (30). Fifty μg of protein from total cell lysate per sample was separated on 15% SDS PAGE, transferred to nitrocellulose membranes, blotted with goat anti-SNCG polyclonal antibody (E-20, sc-10698; Santa Cruz Biotechnology) at 1:200 dilution using an enhanced chemiluminescence (ECL) detection system (Amersham). Membranes were stripped and reblotted with anti-β-actin monoclonal antibiotic (Sigma) to normalize the amount of protein loaded on gels.

Two μg of genomic DNA isolated from ovarian cell lines was treated with sodium bisulfite and purified using Promega’s DNA clean-up kit. The treated DNA was dissolved in 20 μl of water, and 2 μl was used for MSP. The primers specific for unmethylated DNA were SNCG-U1CF and SNCG-U2R; the primers for methylated DNA were SNCG-M1F and SNCG-M2R. These two sets of primers were designed to amplify the same region of exon 1 from −139 to −37, covering the CpG sites 2–11 and yielding a product of 102 bp. The PCR conditions for both sets of primers were as follows: first cycle at 94°C for 1 min to activate the hot start enzyme, then 30 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 30 s, and a final elongation at 72°C for 5 min.

The SNCG-negative breast cancer cell lines (MCF-7 and MDA-MB435) and ovarian cell lines (HIO-135, OVCAR4, and OVCAR8) were cultured in medium containing 0, 1, 5, or 10 μm of 5-Aza-C for 6 days. The medium and drug were replaced every 24 h.

MCF-7 cells were transfected with pCIneo vector alone (mock transfected) or with pCIneo-SNCG, and the pooled transfectants were selected by adding 300 μg/ml of G418. The expression of SNCG was confirmed by Western blotting and the cells were further used for proliferation assays.

For HMEC-184, 1.2 × 10⁴ cells were seeded in 24-well plate and treated with 50 ng/ml of OM for 1 and 3 days. [³H]thymidine incorporation was done as described previously (31). For mock- and SNCG-transfected MCF-7 cells, 1 × 10³ cells were seeded in black 96-well plates in RPMI medium containing 10% FBS and were harvested at indicated intervals of time. The total DNA content was estimated using cyquant cell proliferation assay kit (Molecular Probes) as per the manufacturer’s instructions.

MCF7-SNCG or MCF7-neo cells were seeded at a density of 2 × 10⁵ in 100-mm dishes in RPMI containing 10% serum. After 24 h, the medium was replaced with RPMI having 0.5% serum. The cells were harvested after 48 h, and DNA content was analyzed as described earlier (32).

Comparisons of experimental data were analyzed by a two-tailed Student’s t test. A P < 0.05 was considered to indicate a statistically significant difference.

Total RNA was isolated from 10 breast cancer cell lines, and RT-PCR was conducted using specific primers for SNCG and a housekeeping gene, GAPDH. SNCG mRNA was detected in 6 of 10 cell lines including AU565, BT-20, SKBR-3, T47D, H3922, and MDA-MB231 (Fig. 1,A, and data not shown). The remaining four cell lines either showed no expression of SNCG mRNA such as MCF-7, H3396, and MDA-MB435 or expressed this gene at a very low level (Fig. 1,A, MDA-MB468). In contrast, RT-PCR detected equal expression of GAPDH mRNA in all of the cell lines. Western blot analysis (Fig. 1 B), using specific anti-SNCG antibody, demonstrated SNCG protein expression in the same six cell lines that express the mRNA but not in the four cell lines that express no mRNA or a trace amount, thereby providing a solid validation for the results of the RT-PCR.

We further examined SNCG mRNA and protein expression in one normal mammary epithelium-derived cell line, MCF10A, and 2 primary normal HMEC lines, HMEC-184 and HMEC-048R. Unlike MCF10A cells that can grow in culture indefinitely, the HMEC lines have limited life span. Fig. 1 C shows that MCF10A expressed a trace amount of SNCG mRNA, as detected by RT-PCR, but SNCG protein could not be detected by immunoblotting. Unexpectedly, both of the HMEC cells showed reasonable levels of SNCG mRNA and protein, which contradicted previous studies by in situ hybridization and Western blot that neither detected the mRNA or the protein of SNCG using normal human breast tissues, suggesting that culturing in vitro may induce SNCG gene expression

To identify the correlation between DNA methylation of exon 1 and expression of SNCG, we examined the in vivo methylation status of all of the cell lines by genomic bisulfite sequencing. Genomic DNAs were treated with sodium bisulfite, and the modified DNAs were amplified with the primer SNCG-S2F and SNCG-S2R. This primer set specifically amplified the modified sense strand of SNCG gene from −275 to +140, covering the entire exon 1 and its vicinity. The results obtained from 12 cell lines are summarized and are shown schematically in Fig. 2. The data demonstrated a specific and consistent pattern of methylation in four of five SNCG-negative cell lines. In MCF10A cells, the CpG sites, 2, 5, 7, and 10–15 CpG, were consistently methylated, a pattern identical to the other three cancer cell lines (MDA-MB435, MDA-MB468, and MCF-7). In contrast to SNCG-negative cells, all SNCG-positive cell lines contain unmethylated exon 1. H3396 is the only cell line that has unmethylated CpG island but did not express the SNCG gene. In this cell line, repression of transcription by other factors such as lack of transcriptional activator or expression of repressor, may contribute to the loss of SNCG expression despite promoter demethylation. Nevertheless, the results presented in Figs. 1 and 2 clearly demonstrate that the silencing of SNCG expression in mammary epithelial cell lines is correlated predominantly with methylation of exon 1 at specific CpG sites regardless of whether they were originated from breast tumor or from normal breast tissue.

Because the above data confirmed a role of demethylation in SNCG gene expression in breast cancer cells in culture, we wanted to know whether the demethylation-dependent gene expression occurs under in vivo conditions as well. First, using RT-PCR assays, we examined SNCG mRNA expression in 10 breast tumors, most of which were diagnosed as invasive ductal carcinoma with histological grade of II and above. The pathological characteristics of the tumor samples are listed in Table 2. The expression level of SNCG mRNA in the tumor-adjacent normal tissue from each patient was also determined. In this experiment, T47D cells were used as a positive control. Fig. 3. top panel, show the results of RT-PCR and the next panel down presents the relative SNCG mRNA levels after normalization with the signal of GAPDH. These results demonstrate that of 10 patient samples, 9 patients displayed SNCG expression in tumor samples. SNCG mRNA was not detected in five tumor-adjacent normal tissues (N1, N5, N7, N8, and N10), whereas low levels were detected in other four normal samples (N2, N3, N6, and N9), and a high level was found in N4. The detection of SNCG expression in the five tissue samples, supposed to be normal, prompted us to reexamine the original tissue slices; we found that all of the adjacent “normal” tissues contained regions of infiltrating tumor cells, which were the likely source of SNCG mRNA detected in these tissue samples.

Second, to determine whether the expression of SNCG in breast tumors is associated with hypomethylation, we selected five patient samples of both tumor and normal tissues for genomic bisulfite sequencing. The sequencing results are summarized in Table 3. The N/T pair number 7, which were SNCG-negative for both the normal tissue and the tumor, showed methylation in five of five clones for normal tissue and four of six clones for tumor. Importantly, the methylation occurred at the CpG sites 2, 5, 7, and 10–15, exactly matching the methylation pattern seen in cell lines. The N/T pair number 8, in which the normal tissue did not show any expression and the tumor showed low level of SNCG mRNA, the exon 1 was methylated in all of the clones from the normal tissue and was partially demethylated in the tumor tissue. In N/T pair number 6, the exon 1 was unmethylated for both normal and tumor tissues, which correlated well with the SNCG expression. In N/T pairs numbers 1 and 5, the exon 1 was unmethylated in tumors as well as in normal tissues, albeit with a lack of SNCG expression in normal tissues, suggesting that demethylation could occur before the gene expression. Taken together, the results from breast tumor tissues agreed to a large extent with the results from breast cancer cell lines and provided critical in vivo evidence to support the role of DNA methylation in the control of SNCG expression.

It is possible that SNCG mRNA detected in the tumor-adjacent normal tissues was produced by infiltrated tumor cells instead of normal cells. Alternatively, the genetic background of noncancerous tissue adjacent to the tumor cells has been changed, which could activate SNCG expression. We evaluated these possibilities by an examination of SNCG expression in 10 invasive tumors and in 6 normal breast tissues obtained from reduction mammoplasty of healthy individuals. RT-PCR analysis showed that of 10 tumor tissues, 8 displayed high expression of SNCG (Fig. 4,A, left panel). In contrast, under the same conditions, RT-PCR did not detect SNCG expression in normal samples (Fig. 4,A, right panel). The bisulfite genomic sequencing was done for five tumors and the results showed that all of the tumor samples contained mostly unmethylated SNCG gene, whereas the normal breast tissues contained methylated as well as unmethylated SNCG gene. The percentage methylation for normal breast and tumor samples is presented in Fig. 4 B, and the difference in methylation was found to be statistically highly significant. (P ≤ 0.003). Because the majority of genetic materials from normal breast tissue were derived from fat cells and fibroblasts, and only a small portion was actually from the epithelium, it is possible that the unmethylated allele belonged to other cell types.

We further extended our study to organoids. They were epithelial clumps obtained by digestion of normal breast tissue from one donor with collagenase and hyaluronidase at 37°C for 24–72 h. Fig. 5 showed the results of RT-PCR from the organoids before and after extensive cell culture to become finite HMECs. Whereas a very low level of SNCG mRNA was detected in the organoids in which epithelial cells were incubated in culture medium containing growth factors for only 24–72 h, a high level of SNCG mRNA was shown in the matching HMECs. Sequencing results showed that the organoids contained mostly methylated clones (75%) but the finite life span HMECs were completely unmethylated. These data combined with Fig. 1 C suggest that SNCG is not expressed in vivo in normal mammary epithelial cells, however, during the initial selection and culturing in vitro, the SNCG gene expression is induced through demethylation by growth factors present in the culture medium. SNCG expression may stimulate these primary cultures of cells to proliferate.

To further study that SNCG expression, controlled by demethylation, contributes to the cell proliferation, a normal mammary epithelial cell line HMEC-184 was treated with cytokine OM, which has been previously shown to inhibit the proliferation of HMECs (31). Cells were treated with OM at a concentration of 50 ng/ml for the indicated intervals of time, and cell proliferation was determined by [³H]thymidine incorporation assay. Fig. 6,A shows that OM treatment led to 30% inhibition of cell growth after 1 day of treatment, whereas up to 60% of inhibition was observed after 3 days of treatment. RT-PCR analysis (Fig. 6 B) showed that levels of SNCG expression in OM-treated cells was concomitantly reduced by 70% as compared with control. With bisulfite genomic sequencing, four of six clones were found to be methylated in OM-treated cells, whereas all five of the clones sequenced from the untreated cells were unmethylated.

To directly examine the relationship between SNCG expression and cellular proliferation, SNCG was stably transfected into MCF-7 cells, and pool population was selected for both mock- and SNCG-transfected cells. Western blot detected an abundant amount of SNCG protein in MCF7-SNCG transfectants (Fig. 7,A). The proliferation rate of MCF7-SNCG was compared with mock-transfected MCF-7 cells by using Cyquant cell proliferation assay kit. Fig. 7,B shows that SNCG-transfected cells proliferate faster than mock-transfected cells, and a >2-fold increase in cell proliferation was observed after 5 days. Cell cycle analysis (Fig. 7 B, inset) showed that SNCG-transfected cells have more cells in S phase (33.11 + 3.2 in SNCG-transfected cells compared with 22.4 + 2.3 in mock-transfected cells) and less in G₂-M (15.06 + 2.8) compared with mock-transfected cells (22.22 + 1.5). Overall, these data suggest that SNCG is a stimulating factor for cell proliferation; the SNCG expression is correlated with demethylation of the exon 1 and that the inhibition of cell growth leads to an increased DNA methylation at the CpG island of the SNCG gene, resulting in reduced expression.

To determine whether demethylation of exon 1 is also a causal factor for SNCG expression in ovarian cancer cells, the experiments of RT-PCR and in vivo genomic sequencing were conducted in five ovarian tumor-derived cell lines and five normal ovarian epithelium-derived cell lines (HIOs). The results of RT-PCR are presented in Fig. 8, upper panel, and the summarized sequencing data are shown in Fig. 9. Among the five cancer cell lines, OVCAR3 and OVCAR5 expressed high levels of SNCG mRNA, and these cells contained fully demethylated exon 1, whereas A2780 expressed a moderate level of SNCG mRNA, and exon 1 in these cells was partially methylated. In contrast, OVCAR4 and OVCAR8 did not express SNCG mRNA at all, and all of the 15 CpG sites of the CpG island in these two cell lines were completely methylated. Thus, SNCG expression appears to correlate very well with the demethylation of SNCG gene in ovarian cancer cell lines. Interestingly, the methylation pattern in ovarian cancer cells is different from that in breast cancer cells. The specific methylation patterns of ovarian cancer cell lines and breast cancer cell lines are presented in Table 4. Among the five normal HIO cell lines, SNCG mRNA was detected only in HIO-107 cells. Within eight clones of HIO-107 that were sequenced, four clones were completely methylated at every CpG site of CpG island, whereas the four other clones were nearly unmethylated. Other HIO lines showed no expression of SNCG mRNA, although all four of the cell lines displayed demethylation of the CpG island to a different extent. Thus, these data suggest that in normal ovarian epithelium-derived cell lines, besides DNA methylation, there are other factors that repress SNCG expression. Because all of the CpG sites in the CpG island of SNCG in ovarian cancer cells were methylated in SNCG-negative cells and unmethylated in SNCG-positive cells, it is possible to determine the methylation status by using a MSP assay. To explore this possibility, the bisulfite-modified genomic DNAs were amplified with the methylation-specific primers SNCG-M1F or SNCG-M2R, or the primers corresponding to unmethylated sequence, SNCG-U1CF and SNCG-U2R (Table 1). The bottom panel of Fig. 8 shows that a strong band of 102 bp (SNCG-M), corresponding to methylated sequence of exon 1 (−139 to −37), was amplified with the methylated primers from SNCG-negative OVCAR4 and OVCAR8 cells, but this band was not amplified from SNCG-positive OVCAR3 and OVCAR5 cells. In contrast to the methylated primers, the same region was specifically amplified using unmethylated primers from OVCAR3 and OVCAR5 but was not amplified from OVCAR4 and OVCAR8. MSP detected both methylated and unmethylated alleles from A2780. Thus, the MSP produced similar results as the direct sequencing. Likewise, the results of MSP of five HIO cell lines agreed to a large degree with the sequencing data and showed that HIO-107 and HIO-108 contained unmethylated and methylated alleles and HIO-103 and -105 contained mostly unmethylated alleles, whereas HIO-135 contained mainly methylated SNCG gene. These data suggest that MSP can reliably distinguish the unmethylated SNCG from the methylated gene in ovarian cancer cells.

If the transcriptional silencing of SNCG can be attributed to the methylation of exon 1, then demethylating the gene with 5-Aza-C should lead to expression of SNCG in both breast cancer and ovarian cancer cells. To test this, we chose two SNCG-negative breast cancer cell lines (MDA-MB435 and MCF-7), two SNCG-negative ovarian cancer cell lines (OVCAR4 and OVCAR8), and one SNCG-negative HIO cell line (HIO-135). All of these cell lines contain methylated SNCG gene. Cells were exposed to different concentrations of 5-Aza-C for 4–6 days, and the medium and drug were replaced daily. Fig. 10,A shows that, in breast cancer cells, SNCG mRNA expression was induced by low doses of 5-Aza-C (0.6 μm for MCF-7 and 1 μm for MDA-MB435), and its level was increased by higher concentrations of 5-Aza-C in a dose-dependent manner. In the two ovarian tumor lines, higher doses (10 μm) of 5-Aza-C were required to induce SNCG expression as compared with 1 μm for the HIO-135 cells (Fig. 10 B). To confirm that reactivation of SNCG expression by 5-aza-C was the result of the demethylation of the exon 1, we isolated DNA from OVCAR4 and OVCAR8, which had been treated with a 10-μm concentration of 5-aza-C, and performed bisulfite sequencing. As expected, the CpG islands in these ovarian tumor cells became unmethylated (data not shown). In contrast to SNCG-negative cell lines, 5-aza-C treatment of OVCAR3 and OVCAR5 did not further increase the level of SNCG expression. Collectively, these data clearly demonstrate that reactivation of SNCG expression by 5-Aza-C was the direct effect of demethylation of exon 1 and was not a secondary effect caused by other factors the expression of which were changed by the treatment.

The onset of cancer is associated with the silencing of the tumor suppressor genes and activation of proto-oncogenes. Previous studies have suggested that SNCG could function as an oncogene in breast cancer cells. It has been demonstrated that exogenous expression of SNCG in breast cancer cells (MDA-MB435) led to a significant increase in cell motility and invasiveness in cell culture and to a profound augmentation of metastasis in nude mice (24). Furthermore, exogenous expression of SNCG in MCF-7 cells significantly stimulated the growth of MCF-7 cells under anchorage-dependent (Fig. 7) and anchorage-independent conditions (33), whereas blocking SNCG expression with SNCG antisense mRNA markedly reduced the colony formation of T47D cells (25). Although previous studies by our group and others have shown that SNCG was abnormally expressed in advanced-stage breast carcinoma and ovarian carcinoma and that it is likely to be important in the pathogenesis of these neoplastic diseases, the mechanism(s) for its aberrant expression was unclear. In this report, we extensively examined the relationship between demethylation or hypomethylation of SNCG CpG island, and the expression of this candidate oncogene in breast cancer and ovarian cancer under both in vitro and in vivo conditions. From our present study, four important findings emerged.

First, in breast cancer cell lines, SNCG expression correlates with complete demethylation of the exon 1 region. By contrast, three of four SNCG-negative cell lines showed specific methylation at the CpG sites 2, 5, 7, and 10–15, suggesting that methylation at these sites is sufficient to block the expression of this gene in cell culture. More importantly, these same CpG sites were found methylated in primary breast tumor tissues and in MCF-10A, a cell line that originated from normal mammary epithelium. Thus, these data argue that, in normal mammary epithelium, SNCG is methylated at the specific CpG sites, resulting in blocking of transcription.

Semiquantitative RT-PCR, a highly sensitive detection method, failed to identify SNCG expression in six of six normal breast tissues from healthy women without cancer. This result provided new evidence to support original findings that SNCG expression was confined to malignant mammary epithelial cells. In vivo genomic sequencing, however, detected both methylated and unmethylated SNCG gene from the same normal breast tissue samples. We tentatively interpret this discrepancy thus: the unmethylated genes could be derived from other cell types such as fat cells, fibroblasts, or mononuclear leukocytes, and the lack of expression in these cells could be controlled by other mechanisms. However, our results cannot exclude the possibility that in normal mammary epithelium, SNCG is partially methylated and that partial methylation is sufficient to block its expression. Additional studies to examine SNCG expression and methylation in nonepithelial cell types of breast tissue will be needed to resolve this discrepancy. The unmethylated exon 1 in normal tissues adjacent to tumors suggests that demethylation may precede expression. A similar observation has been reported in benign and malignant colon neoplasms (9). Although the benign tumors do not express HGH, α and γ globin genes were hypomethylated in the promoter region of these genes just like malignant tissues, thereby suggesting that alteration in DNA methylation precedes malignancy.

The second important finding in this study is that DNA methylation also plays an important role in SNCG expression in ovarian cancer cells. Similar to breast cancer cells, an inverse relationship between exon 1 methylation and SNCG expression was found in various ovarian cancer and HIO cell lines. However, there are two characteristics that are unique and distinguish ovarian-derived epithelial cells from breast-derived epithelial cells. First, the 15 CpG sites in ovarian cancer cells were all methylated instead of being selectively methylated at the hot spots that were identified in breast cancer cells. Secondly, in ovarian cells, partial methylation permitted SNCG expression, albeit at a lower level, whereas SNCG expression could not be detected in breast cancer cells in which the exon 1 was partially and heterogeneously methylated. These differences suggest that, whereas partial methylation in the exon 1 is adequate to inhibit SNCG expression in breast cancer cells, complete methylation of the CpG island is required for silencing SNCG in ovarian cancer cells. The third finding of this study is that SNCG is expressed in three primary HMEC lines (184, 048R, and 240L) that have limited life span and that the expression correlates with hypomethylation of the exon 1. By contrast, SNCG is not expressed in normal mammary epithelial cells in vivo nor in the established cell line MCF-10A, and is detected at a very low level in organoids before extensive culturing. These observations suggest that SNCG expression and demethylation of the exon 1 are regulated possibly by growth factors that are present in culture medium during the establishment of HMEC cell lines. Consequently, SNCG gene product further stimulates cell proliferation of the primary mammary epithelial cells that normally have low growth potential. Although this is a hypothetical scenario, there is some evidence to support this hypothesis. Celis et al. (34) have reported that a group of genes that were not expressed in the original bladder transitional carcinomas became expressed when the tumor tissues were incubated in culture medium for a very short time (1–2 days). Synuclein was found within this group. Future studies to clearly define the function of this protein in neoplasm will provide insight to understand the molecular mechanisms that control the methylation status of SNCG gene.

The last important finding is that we provided direct evidence to demonstrate a stimulating role of SNCG in the growth of breast cancer cells. Transfection of SNCG into MCF-7 cells resulted in an increased proliferation rate of cells. Conversely, we also demonstrated a correlation between reduced growth rate and decreased SNCG expression. Importantly, we showed the coordinated changes in cell growth, SNCG mRNA level, and methylation status of the exon 1 of SNCG gene. When the HMECs, which express high levels of SNCG, are arrested by OM or by serum starvation (data not shown), the SNCG levels decrease, and the exon 1 becomes hypermethylated. At the present, the normal cellular functions of SNCG are largely unknown. Future studies to clearly define the function of this protein in neurons as well as in neoplasm will provide insight to understand the molecular mechanisms that control the methylation status of SNCG gene.

Currently, in the cancer research field, DNA hypermethylation has received considerable attention, and DNA hypomethylation is studied inadequately. In fact, the original observation of altered DNA methylation in cancer was hypomethylation (16). A recent study of genome-wide screening for normally methylated human CpG islands has found a considerable number of genes containing methylated CpG islands. We believe that our studies provide a clearly defined example supporting the hypothesis that abnormal hypomethylation contributes to cancer formation.

We thank Dr. Martha R. Stampfer for her valuable suggestions on HMEC culturing, Dr. Zhong-Zong Pan for his helpful input, and the Tissue Bank and Biosample Repository Core Facility’s (Fox Chase Cancer Center, Philadelphia, PA) staff members for providing tissue samples.