Main

Cystatin M (CST6) is a member of a family of proteins that function as physiological inhibitors of lysosomal cysteine proteases, and control target proteases by forming high-affinity reversible complexes.1, 2 Lysosomal cysteine proteases are involved in the degradation of components of connective tissues and basement membranes in vitro, and aberrant expression and activity of these proteases accompany cancer invasion and metastasis in vivo.1, 3 Thus, loss of CST6 expression might contribute to increased proteolysis of tissue architecture, facilitating the spread of cancer cells.1, 4 The majority of human breast cancer cell lines derived from metastatic breast tumors lack CST6 expression, whereas normal and premalignant cells express abundant levels of CST6.2, 5 Consistent with a role in suppression of metastasis, CST6 has been suggested to function as a breast tumor suppressor gene.2 Exogenous expression of CST6 in MDA-MB-435S breast cancer cells results in the suppression of cell proliferation, migration, matrix invasion, and tumor-endothelial cell adhesion in vitro.1 No deletions or structural rearrangements of CST6 have been characterized, suggesting that loss of gene expression my be the result of transcriptional silencing in breast cancer cells.4, 6 CST6 contains a large CpG island that spans the proximal promoter and exon 1, encompassing the start site for transcription. CpG islands sited in gene promoters represent a major target for DNA hypermethylation, which impairs transcription related to regional or specific methylation events.7, 8, 9 Promoter CpG island hypermethylation contributes to gene silencing by inhibiting the binding of certain transcription factors to their recognition sequence,8, 9 attracting methylated DNA-binding proteins,10, 11, 12, 13 and/or through chromatin remodeling.14 In a previous microarray-based analysis of differential gene expression in MCF-7 breast cancer cells, we identified CST6 as a methylation-sensitive gene and demonstrated an inverse relationship between CST6 mRNA expression and methylation of its promoter CpG island.15 Consistent with this observation, recent studies have shown CST6 to be silenced by methylation in select breast cancer cell lines and primary breast tumors.16 Furthermore, CST6 was identified as a methylation-sensitive gene in glioma cell lines and primary brain tumors.17 These findings combine to suggest that loss of CST6 expression in multiple tumor systems may be a direct consequence of methylation-dependent gene silencing.

The aim of the present study was to investigate the direct association between CST6 CpG island methylation and expression status in multiple breast cancer cell lines in order to definitively determine if aberrant DNA methylation represents an important mechanism for the loss of CST6 during breast tumorigenesis and/or progression. Gene expression analysis revealed diminished expression of CST6 in the majority of breast cancer cell lines and treatment with the demethylating agent 5-aza-2′-deoxycytidine (5-aza) produced a significant increase in CST6 expression, suggesting that CST6 is subject to methylation-dependent silencing. Consistent with this suggestion, bisulfite sequencing demonstrated extensive CpG island methylation in CST6-negative cell lines, while CST6-positive cell lines lack CpG island methylation. The extent of methylation within the CST6 CpG island was inversely correlated with gene expression. In particular, hypermethylation of the proximal promoter was significantly associated with CST6 silencing, and methylation of several individual CpGs was found to be significantly associated with diminished or absent expression of CST6. These results establish a strong link between CST6 CpG island hypermethylation and transcriptional silencing in breast cancer cell lines, and suggest that methylation-dependent gene repression represents an important mechanism leading to diminished CST6 expression during breast carcinogenesis and/or cancer progression.

Materials and methods

Breast Cancer Cell Line Culture

Human breast cancer cell lines were obtained from the Tissue Culture Core Facility of the UNC Lineberger Comprehensive Cancer Center at the University of North Carolina at Chapel Hill: BT-20 (ATCC#HTB-19), BT549 (HTB-122), Hs578T (HTB-126), MCF-7 (HTB-22), MDA-MB-231 (HTB-26), MDA-MB-415 (HTB-128), MDA-MB-435S (HTB-129), MDA-MB-436 (HTB-130), MDA-MB-453 (HTB-131), MDA-MB-468 (HTB-132), SK-BR-3 (HTB-30), and ZR-75-1 (CRL-1500). Normal breast epithelial cell lines, MCF12A (CRL-10782) and MCF10-2A (CRL-10781), were obtained from the ATCC (Manassas, VA, USA). All cells were propagated according to recommendations from the ATCC. Cell lines that lack expression of CST6 (including Hs578T, MDA-MB-435S, MDA-MB-436, MDA-MB-453, MCF-7, and ZR-75-1) were treated with the demethylating agent 5-aza (Sigma Chemical Company, St Louis, MO, USA), as previously described.15 Briefly, two treatment groups were established from a single founding cell population: (i) control medium and (ii) medium containing 250 nM 5-aza. Cells in the treatment group were exposed to 5-aza for 3 weeks, with weekly subcultivation, followed by a 5-week recovery period in control growth medium with weekly subcultivations during the last 3 weeks. Control cells were subcultivated once per week during the 8-week cell culture period. Cell cultures were fed fresh growth medium three times weekly. The concentration of 5-aza utilized in this study was 4–6-fold lower than traditional methods,18, 19, 20, 21 eliminating the typically encountered cytotoxic effects22, 23 and allowing prolonged exposure of breast cancer cells to the demethylating drug.

Semiquantitative RT-PCR

Total RNA was isolated from cultured cells using a modification of the method of Chomczynski and Sacchi24 utilizing TRIzol Reagent (Invitrogen Life Technologies, Carlsbad, CA, USA), according to the manufacturer's protocol. RNA (2 μg) from control cells or cells treated with 5-aza was reverse-transcribed into cDNA using Superscript II Reverse Transcriptase (Invitrogen Life Technologies) and oligo(dT) as the primer for 60 min at 42°C, according to standard methodology. CST6 oligodeoxynucleotide primers were synthesized by the UNC Oligodeoxynucleotide Synthesis Core Facility (Chapel Hill, NC, USA) based on known cDNA sequence (Genbank, www.ncbi.nih.gov): 5′-AAGACCAGGGTCACTGGAGA-3′ and 5′-CGGGGACTTATCACATCTGC-3′ (163 bp). Verification of equal template concentration was accomplished using primers that amplify a portion of β-actin mRNA (5′-AGAGATGGCCACGGCTGCTT-3′ and 5′-ATTTGCGGTGGACGATGGAG-3′). PCR reactions were performed in a 50 μl total volume of buffer containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 0.001% gelatin, 200 μM of each dNTP (EasyStart Micro 50 PCR-mix-in-a-tube, Molecular BioProducts, San Diego, CA, USA), 0.4 μM of each primer, and 2.5U AmpliTaq enzyme (Perkin Elmer/Cetus, Foster City, CA, USA). Amplifications were carried out in a Perkin–Elmer 9700 Thermocycler using a step-cycle program consisting of 25–30 cycles of 94°C for denaturing (1 min), 58°C for annealing (1 min), and 72°C for extension (2 min).

Quantitative Real-Time PCR

Total RNA samples (20 μg) from control and treated cells were DNAase treated (Promega, Madison, WI, USA), purified using the Qiagen Rneasy mini-kit (Qiagen, Valencia, CA, USA), and reversed transcribed using the High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol. Real-time primers and probes for CST6 and β-actin were purchased from Applied Biosystems. Reactions were carried out using TaqMan Universal PCR Master Mix (Applied Biosystems) and the following PCR conditions: 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min. Gene expression levels were normalized using β-actin for each cell line and differences in CST6 gene expression were determined using the comparative Ct method described in the ABI Prism 7700 User Bulletin #2 (Applied Biosystems).

Sodium Bisulfite Modification of Genomic DNA

Genomic DNA from 2 × 106 cultured cells was isolated using the Puregene DNA Purification Kit (Gentra Systems, Minneapolis, PA, USA). Sodium bisulfite modification of genomic DNA was performed by a procedure adapted from Grunau et al25 as previously described.15 Sodium bisulfite-converted DNA was amplified using primers that targeted two regions of the proximal promoter and exon 1 of CST6: the first region encompasses nucleotides +242 to −228, which contains 55 CpG dinucleotides (5′-GGTTGGAATGTTGTAGTGGT-3′ and 5′-CTACCCATATTATAACTAACC-3′), and the second region encompasses nucleotides +118 to −228, which contains 46 CpG dinucleotides (5′-GGTTTTTTGGGTTTTTTGAATTT-3′ and 5′-CTACCCATATTATAACTAACC-3′). PCR amplification was accomplished using a step-cycle program consisting of 40 cycles of 94°C for denaturing (1 min), 52°C for annealing (1.5 min), and 72°C for extension (2 min). PCR products were fractionated on 2% agarose gels containing 40 mM Tris-acetate/1.0 mM EDTA (pH 8.0) and visualized by ethidium bromide staining. A portion of each PCR product (1 to 5 μl) was cloned into pGEM-T Easy Vector (Promega, Madison, WI, USA). Three to five colonies were selected per gene segment and expanded in liquid culture. Plasmid DNA was purified using the Wizard Plus Miniprep DNA purification kit (Promega), before digestion with NcoI and NdeI (New England Biolabs, Beverly, MA, USA) to confirm the presence and size of the cloned insert. Validated clones were sequenced using the universal M13R3 primer with an Applied Biosystems automated sequencer at the UNC Genome Analysis Facility (Chapel Hill, NC, USA). The bisulfite conversion efficiency was calculated for each sequenced clone based on the ratio of converted Cs (non-CpG) to total number of Cs (non-CpG) in a given gene segment. Only clones determined to have a conversion efficiency of >95% were included in the present study. The results of methylation analyses were expressed as total methylation index (TMI). This measure of methylation can be applied to single CpG dinucleotides, select groups of CpG dinucleotides, or to continuous groups of CpG dinucleotides in a given gene segment. TMI was calculated for each cell line and clone by dividing the number of methylated CpGs observed by the total CpGs analyzed and expressed as percent methylation. For instance, in an analysis of a gene segment containing 55 CpG dinucleotides and three clones sequenced, TMI would be calculated based upon 165 possible CpG methylation events (3 × 55).

Statistical Analysis

Values included in the text, table, and Figures represent averages±s.e.m. that were calculated using the statistical function of KaleidaGraph Version 3.5 (Synergy Software, Essex Junction, VT, USA). An unpaired t-test was performed to determine statistical significance of 5-aza treatment-related differences in gene expression in CST6-negative cell lines. Likewise, an unpaired t-test was used to examine the association between CST6 methylation status and gene expression levels among subsets of breast cancer cell lines.

Results

CST6 is Differentially Expressed among Breast Cancer Cell Lines

CST6 gene expression was examined by semiquantitative RT-PCR in MDA-MB-231, MDA-MB-415, MDA-MB-435S, MDA-MB-436, MDA-MB-453, MDA-MB-468, BT-549, BT-20, Hs578T, MCF-7, SK-BR-3, and ZR-75-1 breast cancer cell lines and two normal mammary epithelial cell lines, MCF10-2A and MCF12A. Five of the 12 breast cancer cell lines (42%) express detectable levels of CST6 mRNA: MDA-MB-468 and SK-BR-3 cells express low levels of CST6, MDA-MB-231 and MDA-MB-415 cells express moderate levels of CST6, and BT-20 cells strongly express CST6. However, no CST6 mRNA was found in the remaining seven cell lines (58%) (Figure 1a). Both MCF10-2A and MCF12A cells transcribe CST6 at low levels, and β-actin mRNA was expressed evenly across all cell lines examined (Figure 1a). These results are consistent with the recently published studies on MCF-7, SK-BR-3, and MDA-MB-231 cells.16 This analysis identified subsets of breast cancer cell lines that differentially express CST6 mRNA, providing the cellular reagents for examination of methylation-dependent epigenetic regulation of CST6 in breast cancer cells.

Figure 1
figure 1

CST6 expression in human breast cancer cell lines and normal mammary epithelial cells. Representative agarose gels of RT-PCR products are shown. (a) Differential expression of CST6 among breast cancer cell lines and two normal mammary epithelial cell lines (MCF10-2A and MCF12A). The CST6 RT-PCR product is 163 bp in size. β-Actin was utilized as a sample control. (b) Lane 1 corresponds to cDNA from the indicated untreated breast cancer cell lines; lanes 2 and 3 correspond to cells after 5-aza treatment or treatment withdrawal, respectively. β-Actin RNA was evenly expressed in all samples (data not shown).

Full size image

5-Aza Treatment Induces CST6 Expression in Breast Cancer Cell Lines

CST6-negative cell lines Hs578T, MCF-7, MDA-MB-435S, MDA-MB-436, MDA-MB-453, and ZR-75-1 were treated with 5-aza as described previously,15 to determine if gene silencing was the likely result of DNA methylation. Four of six (67%) of these cell lines expressed significantly increased levels of CST6 mRNA in response to 5-aza treatment (Figure 1b). The significant induction of CST6 mRNA in response to 5-aza treatment in MCF-7 cells observed in this study is consistent with similar published studies.15, 16 In contrast, 5-aza exposure of MDA-MB-453 and MDA-MB-436 cells resulted in a modest, but detectable increase in CST6 mRNA (Figure 1b). The 5-aza-induced increase in CST6 expression was completely reversible. Following a period of recovery after treatment withdrawal, CST6 mRNA diminished to control levels (corresponding to untreated cells) in all cell lines examined (Figure 1b). The 5-aza treatment-related increases in CST6 expression, combined with the loss of CST6 expression following treatment withdrawal, suggests that CST6 may be subject to methylation-dependent silencing in these breast cancer cell lines.

Expression of CST6 was analyzed using real-time PCR in CST6-negative cell lines Hs578T, MCF-7, MDA-MB-436, MDA-MB-453, and ZR-75-1 to quantitate the 5-aza treatment-related increases in gene expression. Consistent with the RT-PCR results, untreated cell lines express extremely low levels of CST6 mRNA, and exposure to 5-aza resulted in significant increases in CST6 mRNA (Figure 2). Exposure of Hs578T cells to 5-aza led to a 49-fold increase in CST6 expression (P=0.0036 compared to control) and withdrawal of 5-aza resulted in a significant reduction (P=0.0021 compared to 5-aza treated) of gene expression to a level that approaches that of control (untreated) cells (Figure 2). Likewise, 5-aza treatment of MCF-7 and MDA-MB-436 cells produced 80-fold (P=0.0058) and 90-fold (P=0.0042) increases in CST6 expression, respectively, and withdrawal of the 5-aza treatment lead to significant reductions in CST6 expression in both cell lines (P=0.0062 and P=0.0043, respectively) (Figure 2). Of note, the induction of CST6 mRNA in MDA-MB-436 cells in response to 5-aza treatment was consistently demonstrable with all methods, but the magnitude of increased gene expression detected by real-time RT-PCR was greater than that detected using RT-PCR, possibly due to the increased sensitivity of the real-time method. Exposure of MDA-MB-453 and ZR-75-1 cells to 5-aza produced modest increases in CST6 mRNA (three- and eight-fold, respectively) that were statistically significant (P=0.0017 and P=0.0479), and withdrawal of treatment resulted in significant decreases in CST6 expression to levels that were comparable to untreated cells (P=0.0070 and P=0.0393, respectively) (Figure 2). These data show that 5-aza treatment of CST6-negative cell lines results in statistically significant increases in CST6 mRNA, and suggests strongly that CST6 is subject to methylation-dependent silencing in a variety of breast cancer cell lines.

Figure 2
figure 2

Quantitive real-time PCR analysis of CST6 in breast cancer cell lines. β-Actin was used to normalize gene expression levels for each cell line and differences in CST6 expression were determined using the comparative Ct method. CST6 gene expression for cell lines treated with 5-aza (black bars) and after treatment withdrawal (gray bars) are expressed as relative fold-change compared to control values (set at 1.0). Error bars reflect s.e.m. Values for gene expression that do not show error bars reflect data where the s.e.m. could not be depicted graphically. *Denotes statistical significance at P<0.05 compared to control values. **Denotes statistical significance at P<0.04 compared to 5-aza treatment values.

Full size image

Methylation-Dependent Silencing of CST6 in Select Breast Cancer Cell Lines

To facilitate a correlative analysis of CST6 gene expression and CST6 CpG island methylation status, we analyzed 55 CpG dinucleotides from a segment of the proximal promoter region and exon 1 (+242 to −228) in normal breast epithelial cells (MCF12A), two breast cancer cell lines (BT-20 and SK-BR-3) that express CST6, and five breast cancer cell lines (Hs578T, MCF-7, MDA-MB-435S, MDA-MB-436, and MDA-MB-453) that lack expression of CST6. Multiple clones (n=3–5) corresponding to the CST6 promoter and exon 1 from each cell line were analyzed by sodium bisulfite sequencing and individual CpGs were scored for methylation status. In MCF12A cells, 35/55 (64%) CpGs were not methylated, 18/55 (33%) CpGs were methylated at an intermediate level, and 2/55 (3%) CpGs were 100% methylated, producing a TMI for the promoter/exon 1 of 16% (Figure 3). BT-20 cells exhibit sparse methylation of CST6: 53/55 (96%) CpGs were unmethylated in all clones analyzed, resulting in a TMI of 1%. The CST6 gene in SK-BR-3 cells was significantly more methylated than MCF12A or BT-20 cells (especially within exon 1): 39/55 (71%) CpGs show some level of methylation, producing a TMI of 46% (Figure 3). Overall, the CST6-positive cell lines (MCF12A, BT-20, and SK-BR-3) exhibit low levels of methylation within the proximal promoter/exon 1 of CST6 (mean TMI=21±13%) (Table 1). In contrast, breast cancer cell lines that do not express CST6 exhibit hypermethylation of the CST6 promoter/exon 1 region, with TMI values ranging from 72 to 98% (Figure 3). MDA-MB-435S and MDA-MB-453 cells were 100% methylated at 52/55 (95%) and 51/55 (93%) CpGs, respectively (Figure 3). In MCF-7 cells, 42/55 (76%) CpGs were 100% methylated, 12/55 (22%) CpGs were methylated at an intermediate level, and 1/55 (2%) CpGs was unmethylated (Figure 3). Hs578T and MDA-MB-436 cells were 100% methylated at 19/55 (35%) and 13/55 (24%) CpGs, respectively, with the remaining CpGs methylated at an intermediate level. The average TMI for the CST6-negative cell lines was 87±5% (n=5), reflecting CST6 promoter/exon 1 hypermethylation among these cells (Table 1). When the methylation status of CST6 promoter/exon 1 was compared between groups of cell lines that differentially express CST6, a significant association (P=0.0227) between CST6 promoter/exon 1 region methylation (CpGs 1–55) and CST6 gene expression was found (Table 1). However, hypermethylation of the proximal promoter of CST6 (CpGs 1–23 and 1–31, P<0.001) was more strongly associated with loss of CST6 expression status than methylation involving exon 1 (CpGs 24–55, NS) (Table 1). These observations suggest that hypermethylation of the CST6 CpG island contributes to the silencing of CST6 expression in breast cancer cell lines, and that hypermethylation of the proximal promoter is most important for downregulation of CST6 gene expression.

Figure 3
figure 3

Methylation analysis of the CST6 proximal promoter and exon 1 in breast cancer cell lines and normal mammary epithelial cells that differentially express CST6. A summary of the methylation analysis of the CST6 promoter (23 CpGs) and exon 1 (32 CpGs) is shown. The black arrow indicates the start of transcription between CpGs 23 and 24. Black circles correspond to fully (100%) methylated CpGs, gray circles correspond to CpGs with intermediate methylation, and open circles correspond to unmethylated CpGs. TMI values for the entire promoter/exon 1 region (55 CpGs) are given for each breast cancer cell line. MCF12A, SK-BR-3, and BT-20 cells express CST6, while the remaining cell lines lack CST6 expression.

Full size image
Table 1 Segmental CpG methylation analysis of the CST6 promoter and proximal exon 1 among CST6-postive and CST6-negative breast cancer cell lines
Full size table

In order to identify CpG dinucleotides that are critical in CST6 silencing, 5 CST6-negative cell lines (Hs578T, MCF-7, MDA-MB-435S, MDA-MB-436, and MDA-MB-453) were treated with 5-aza, and 46 CpG dinucleotides from the proximal promoter region and exon 1 (+118 to −228, CpGs 10–55) were evaluated by sodium bisulfite sequencing (Figure 4). The CST6 promoter/exon 1 became significantly demethylated in response to 5-aza treatment in each cell line examined (P<0.0001), resulting in lower values for TMI (33±10%, range 1–64%) (Figure 4a). Following withdrawal of 5-aza treatment, remethylation of the CST6 promoter/exon 1 occurred in MDA-MB-436, Hs578T, and MCF-7 cells (Figure 4b). The extent of methylation after 5-aza withdrawal was indistinguishable from that of controls for MDA-MB-436 and Hs578T cells, reflecting a complete remethylation of the promoter/exon 1 (P<0.0001) (Figures 3 and 4b). Likewise, significant remethylation of CST6 occurred in MCF-7 cells after treatment withdrawal, with methylation levels approaching that of control (TMI of 64 vs 89%) (Figures 3 and 4b). In contrast, there was no apparent change in CST6 methylation in MDA-MB-435S cells after treatment withdrawal based upon calculated TMIs (42 vs 40%), but there was a significant qualitative change in the methylation pattern (Figure 4b). MDA-MB-435S cells exhibit a loss of CST6 expression when the demethylating treatment was withdrawn, suggesting that some or all of the 12 CpGs that were differentially remethylated (CpGs 13, 15, 19, 27, 31, 33, 39, 42, 46, 47, 50, and 54, Figure 4) may be critical for CST6 silencing. CST6 expression after 5-aza withdrawal was diminished in MDA-MB-453 cells in the absence of a dramatic increase in CST6 methylation, although loss of expression was accompanied by a redistribution of methylation across the CST6 promoter and exon 1 (Figure 4). There is a significant association between CST6 methylation status and expression among CST6-negative cell lines treated with 5-aza for both the promoter/exon 1 region (CpGs 10–55, control vs 5-aza, P=0.0033), as well as exon 1 alone (CpGs 24–55, control vs 5-aza, P=0.0127) (Table 1).

Figure 4
figure 4

Methylation analysis of the CST6 proximal promoter and exon 1 in CST6-negative breast cancer cell lines that have been exposed to demethylating treatment. A summary of the methylation analysis of the CST6 promoter (14 CpGs) and exon 1 (32 CpGs) is shown. Black circles correspond to fully (100%) methylated CpGs, gray circles correspond to CpGs with intermediate methylation, and open circles correspond to unmethylated CpGs. TMI values for the promoter/exon 1 region (46 CpGs) are given for each breast cancer cell line. (a) CpG methylation analysis of the CST6 promoter/exon 1 region after treatment with 5-aza. (b) CpG methylation analysis of the CST6 promoter/exon 1 region after withdrawal of 5-aza treatment.

Full size image

A comparative analysis of CpG dinucleotides in the promoter and exon 1 regions of CST6-positive and -negative cell lines was performed to identify methylation events involving individual CpGs or regions of CpGs that are important for the silencing of CST6. Average TMI values for individual CpG dinucleotides were calculated for CST6-positive (BT-20, MCF12A, and SK-BR-3) and CST6-negative cell lines (Hs578T, MCF-7, MDA-MB-435S, MDA-MB-436, and MDA-MB-453) (Figure 5). The CST6-positive cell lines exhibit a low level of methylation (TMI ≤33%) for the 15 CpG dinucleotides contained in segment 1 (corresponding to −50 to −200). Of note, six CpGs (CpGs 2, 4, 5, 7, 12, and 13) were not methylated in these cell lines, suggesting a possible role for these CpGs and/or the sequences containing these CpGs in the positive regulation of CST6 expression (Figure 5b). In contrast, a high level of methylation for CpGs in segment 1 (TMI ≥73%) was found in CST6-negative cell lines, including CpG 14, which was fully (100%) methylated in all cell lines (Figure 5b). The extent of methylation of individual CpG dinucleotides correlated with CST6 gene expression for 14/15 CpGs in segment 1 (P=0.0197 to P<0.0001). Segment 2 spans the transcriptional start site of CST6 (+50 to −50), and consists of CpG dinucleotides 16–31 (Figure 5c). CST6-positive cell lines have a relatively low level of methylation at individual CpG dinucleotides (TMI range: 0–44%), whereas CST6-negative cell lines contain high levels of methylation in this region of the CST6 promoter/exon 1 region (TMI range: 66–93%) (Figure 5c). The extent of methylation of individual CpG dinucleotides was significantly associated with CST6 gene expression for all CpGs in segment 2 (P=0.0343 to P=0.0046). Segments 3 and 4 encompass CpG dinucleotides 32–55 of CST6 exon 1 (+242 to +50). CpG dinucleotides in this region were moderately methylated in CST6-positive cell lines with TMI values ranging from 11 to 67% (Figure 5d). Among CST6-negative cell lines, the calculated TMI values ranged from 80 to 100%, reflecting exon 1 hypermethylation in both segments. With few exceptions, there was no significant correlation between CST6 gene expression and methylation status of individual CpG dinucleotides in segments 3 and 4.

Figure 5
figure 5

Methylation analysis for individual CpG dinucleotides in CST6-positive and CST6-negative breast cancer cell lines. (a) Distribution of CpG dinucleotides proximal to the transcriptional start site in the promoter (0 to −1400 nucleotides) and exon 1 (0 to +294 nucleotides) of CST6 are depicted schematically. Vertical lines represent the relative position of individual CpG dinucleotides and the segmented horizontal lines (designated 1–4) indicate the location of individual CpG dinucleotides depicted in (b) (segment 1), (c) (segment 2), and (d) (segment 4). The results for CpGs in segment 3 are not shown. Representative CpG dinucleotides are indicated by lollipops corresponding to various segments of the promoter or exon 1 (+242 to −228). TMI values represent averages of the three CST6-positive and five CST6-negative breast cell lines: black circles correspond to 100% methylated CpGs, partially-filled circles correspond to >40% methylated CpGs, gray circles correspond to <40% methylated CpGs, and white circles correspond to unmethylated CpGs. †Denotes statistical significance at P<0.001; **Denotes statistical significance at P<0.01; and *Denotes statistical significance at P<0.03, when comparing individual CpG methylation to CST6 gene expression status. (b) CpGs 1–15 from segment 1 (−50 to −200) within the promoter region. (c) CpGs 16–31 from segment 2 (+50 to −50) spanning the transcriptional start site. (d) CpGs 43–55 from segment 4 (+242 to +138) within exon 1.

Full size image

Discussion

CST6 (which encodes cystatin M) was originally identified as a gene whose expression is lost in metastatic breast cancer, suggesting a possible role for this gene in suppression of the invasive/metastatic phenotype.5 Consistent with this suggestion, normal human breast epithelial cells express high levels of cystatin M, while invasive ductal carcinoma cells do not express or express very low levels of this protein.2 When exogenously expressed in human MDA-MB-435S breast cancer cells, CST6 significantly alters the neoplastic phenotype in vitro, resulting in diminished cell proliferation, loss of cell migration, inhibition of Matrigel invasion, and reduced endothelial cell adhesion.1 Furthermore, expression of CST6 in MDA-MB-435S cells delays tumorigenesis by transplanted cells and suppresses spontaneous formation of liver and lung metastases.2 Cystatin M is a member of a family of proteins that function as inhibitors of lysosomal cysteine proteases, which include the cathepsin proteases.4, 26 Increased levels of cathepsin protease activity have been noted in breast cancer,27 and predict poor prognosis among these patients.28 Thus, breast cancer invasiveness may be a direct consequence of inappropriate cathepsin protease activity in the absence of their inhibitory molecules, which include cystatin M.29 While CST6 expression is known to be lost in many primary breast tumors and cancer cell lines, the mechanism that accounts for loss of CST6 expression has only recently been investigated. Given a role for CST6 in tumor suppression and/or metastasis suppression, possible mechanisms for loss of expression include genetic alterations (mutation or deletion) and epigenetic silencing. Southern blotting in breast cancer cell lines failed to identify gross structural rearrangements of the CST6 gene or deletion of the gene locus,5 leading to the more recent suggestion that CST6 expression may be lost due to gene silencing, either in response to a transcriptional repressor protein or as a consequence of promoter hypermethylation.4

The CST6 gene contains a large CpG island that encompasses the proximal promoter and exon 1, consisting of 55 CpG dinucleotides distributed over a 470 bp region (CpGPlot, http://www.ebi.ac.uk/emboss/cpgplot/). In a previous microarray-based gene expression study, we identified CST6 as a target for methylation-dependent gene silencing in MCF-7 breast cancer cells, and showed that loss of CST6 expression in these cells is related to hypermethylation of its CpG island.15 More recently, Ai et al16 has shown aberrant methylation of CST6 in select breast cancer cell lines and primary breast tumors. Additionally, Kim et al17 reported CST6 to be frequently methylated in glioma cell lines and primary brain tumors. In the current study, we established a direct association between CST6 promoter hypermethylation and gene silencing in a panel of human breast cancer cell lines that differentially express CST6. The majority of CST6-negative cell lines examined were originally isolated from invasive/metastatic breast neoplasms,30, 31, 32 whereas the CST6-positive BT-20 cells (which express very high levels of CST6) were derived from a primary breast carcinoma.33 These results suggest strongly that (i) CpG island hypermethylation contributes to CST6 silencing in breast cancer cell lines and (ii) the loss of CST6 expression is associated with the invasive/metastatic phenotype of the breast cancer cell line.

Epigenetic silencing of gene expression is a consequence of DNA hypermethylation and/or chromatin remodeling related to direct interference with the binding of transcription factors to their recognition sequences,8, 9 indirect mechanisms associated with recruitment and binding of methylated DNA-binding proteins,10, 11, 12, 13 and/or modification of histone proteins and alteration of chromatin structure.34 In the current study, we generated evidence for CpG island hypermethylation in the epigenetic silencing of CST6 in human breast cancer cell lines. However, the precise mechanism and the temporal order of events related to CST6 gene silencing have not been determined. Nonetheless, we were able to gain significant insight into the process through comparison of the natural methylation pattern of CST6-positive and -negative cell lines. CST6-negative cell lines are characterized by extensive CpG island methylation, suggesting the possibility that regional methylation across the CST6 proximal promoter and exon 1 may be required for gene silencing. However, a subset of CST6-positive cell lines contain a significant level of methylation in exon 1, indicating that methylation in this portion of the CpG island does not negatively impact on CST6 transcription. In contrast, methylation of CpG dinucleotides in the proximal promoter is strongly associated with loss of CST6 expression, suggesting that regional methylation or specific methylation events affecting this portion of the promoter contribute to gene silencing. There is evidence that discrete methylation events within a larger methylation target (CpG island) can negatively affect gene expression. The AP-2α tumor suppressor gene is an example of a gene that is silenced in response to CpG methylation of a discrete region that is contained within a larger CpG island.7 Discrete methylation events may negatively impact on promoter function by direct interference with transcription factor binding or through indirect interference related to binding by methylated DNA-binding proteins. Among the 55 CpGs that comprise the CST6 CpG island, seven CpGs in the proximal promoter were found to be unmethylated in all cell lines that express the gene, and these CpGs were frequently methylated in CST6-negative cell lines. Specific methylation events affecting these CpG dinucleotides may be required for CST6 gene silencing, but it is not clear if these specific methylation events occur in isolation or if they always take place in conjunction with more extensive regional methylation. Methylation of these CpGs may directly impact on the binding of specific transcription factors to their recognition sequence. Analysis of the CST6 promoter using ProSpector (http://prospector.nci.nih.gov) identified 16 CpGs within the proximal promoter that directly impinge upon transcription factor binding sites and are associated with loss of CST6 expression when methylated, including sequences for AP-2, AP-4, Egr-1, MEIS1, NF-κB, Sp1, and YY1. Methylated CpGs in the binding sites for AP-2 and Sp1 transcription factors have been shown to directly downregulate gene expression.35, 36 Of note, CpG dinucleotide 13, which impinges on three transcription factor binding sites, is never methylated in CST6-positive cell lines. The transcription factor requirements for expression of the CST6 promoter have not been determined. Nonetheless, CpG methylation of the proximal promoter of CST6 is likely to inhibit or impair gene transcription by either direct or indirect interference with the transcription machinery.

The differential CpG island methylation profile among CST6-postive and -negative breast cancer cell lines indicates that certain methylation events and/or specific promoter regions are strongly associated with gene silencing. Nevertheless, CST6-negative cell lines tend to be heavily methylated across the entire promoter/exon 1 CpG island. We posit that there is a succession of methylation events that progressively leads to CST6 gene silencing in metastatic breast cancer cell lines: (i) individual CpG dinucleotides within the promoter region are preferentially methylated resulting in transient silencing of CST6, (ii) methylation spreads throughout the promoter/exon 1 CpG island, (iii) chromatin remodeling occurs resulting in stable silencing of CST6. Alternatively, chromatin remodeling might occur before the completion of regional methylation affecting the entire CpG island. Additional studies will be required to establish the temporal order of events and the nature of chromatin alterations that accompany CST6 silencing in breast cancer cell lines. Likewise, additional studies will be required to determine if methylation-dependent gene silencing accounts for loss of CST6 expression in primary breast tumors and their metastatic lesions.