Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Aberrant methylation of the Wnt antagonist SFRP1 in breast cancer is associated with unfavourable prognosis

Abstract

The canonical Wnt signalling pathway plays a key role during embryogenesis and defects in this pathway have been implicated in the pathogenesis of various types of tumours, including breast cancer. The gene for secreted frizzled-related protein 1 (SFRP1) encodes a soluble Wnt antagonist and is located in a chromosomal region (8p22–p12) that is often deleted in breast cancer. In colon, lung, bladder and ovarian cancer SFRP1 expression is frequently inactivated by promoter methylation. We have previously shown that loss of SFRP1 protein expression is a common event in breast tumours that is associated with poor overall survival in patients with early breast cancer. To investigate the cause of SFRP1 loss in breast cancer, we performed mutation, methylation and expression analysis in human primary breast tumours and breast cell lines. No SFRP1 gene mutations were detected. However, promoter methylation of SFRP1 was frequently observed in both primary breast cancer (61%, n=130) and cell lines analysed by methylation-specific polymerase chain reaction (MSP). We found a tight correlation (P<0.001) between methylation and loss of SFRP1 expression in primary breast cancer tissue. SFRP1 expression was restored after treatment of tumour cell lines with the demethylating agent 5-aza-2′-deoxycytidine. Most interestingly, SFRP1 promoter methylation was an independent factor for adverse patient survival in Kaplan–Meier analysis. Our results indicate that promoter hypermethylation is the predominant mechanism of SFRP1 gene silencing in human breast cancer and that SFRP1 gene inactivation in breast cancer is associated with unfavourable prognosis.

Introduction

The Wnt pathway plays a key role in embryonic development, cell differentiation and cell proliferation (reviewed in Cadigan and Nusse, 1997). Recently, it has been shown that this critical developmental pathway is deregulated in several human tumour entities, including breast cancer (Polakis, 2000). Wnt proteins are secreted molecules that interact with transmembrane receptors of the frizzled family. Binding of the Wnt ligand to its receptor activates a complex signalling cascade which leads to stabilisation of cytosolic β-catenin. After translocation into the nucleus β-catenin interacts with LEF/TCF transcription factors leading to downstream target gene activation (Behrens et al., 1996). Among the members of the Wnt responsive genes are several oncogenes like c-myc, cyclin D1 and c-jun (He et al., 1998; Mann et al., 1999; Tetsu and McCormick, 1999). Secreted frizzled-related proteins (SFRPs) are a family of Wnt antagonists that harbours a domain homologous to the frizzled receptors, the so-called cysteine-rich domain (Uren et al., 2000). SFRPs remain associated with the cell membrane and are able to bind Wnt proteins in the extracellular compartment, thereby inhibiting ligand–receptor interaction and signal transduction (Bafico et al., 1999).

SFRP1 is a putative inhibitor of Wnt signalling (Finch et al., 1997) that is abundantly expressed in normal breast tissue. Its expression is very frequently (>70%) lost or strongly reduced in breast cancer. This has been shown on the RNA level by in situ hybridisation (Ugolini et al., 1999) and on the protein level by immunohistochemistry (Klopocki et al., 2004). Loss of SFRP1 protein expression is associated with poor overall survival (OS) in patients with early breast cancer (pT1 tumours) indicating a putative tumour suppressor gene function of SFRP1 (Klopocki et al., 2004). The SFRP1 gene is located in a chromosomal region (8p22–p12) that is frequently deleted in breast cancer (Seitz et al., 1997) and is thought to harbour a tumour suppressor gene (Lai et al., 2003).

The cause of SFRP1 loss in breast cancer has not been analysed so far. Recently, SFRP1 gene silencing mediated by promoter hypermethylation has been described in colon cancer (Caldwell et al., 2004; Suzuki et al., 2004), ovarian cancer (Takada et al., 2004), bladder cancer (Stoehr et al., 2004; Marsit et al., 2005), human mesothelioma (Lee et al., 2004), prostate cancer (Lodygin et al., 2005) and lung cancer (Fukui et al., 2005). Suzuki et al. (2004) demonstrated a direct link between the epigenetic inactivation of the SFRP1 gene and constitutive Wnt signalling in colorectal cancer. Caldwell et al. (2004) described a low frequency of protein-truncating mutations in the SFRP1 gene in colon cancer.

In this study, we have addressed the question how SFRP1 expression is silenced in breast cancer. To this end we have analysed the human SFRP1 gene in primary breast cancer by mutation and methylation analysis. Furthermore, we performed a parallel expression and promoter methylation analysis in normal and malignant breast cell lines and primary breast cancer. Using methylation-specific polymerase chain reaction (MSP) and reverse transcription PCR (RT–PCR) we found a clear correlation between methylation and loss of SFRP1 expression in both breast cancer and breast cell lines. Clinico-pathologic patient characteristics were statistically correlated with expression and methylation data and revealed an unfavourable prognosis in case of aberrant methylation. Our results indicate that promoter hypermethylation is the predominant mechanism of SFRP1 gene silencing in human breast cancer.

Results

Loss of SFRP1 expression in breast cancer

Loss of SFRP1 expression in breast cancer has previously been demonstrated by RNA in situ hybridisation (Ugolini et al., 2001) and by immunohistochemistry (Klopocki et al., 2004). In both studies the amount of tumours exhibiting significant or complete loss of SFRP1 was approximately 75%. We have further analysed SFRP1 expression by dot blot analysis using a CPA containing 103 cDNAs from 50 breast cancer patients, that is, from 50 primary breast cancers, 50 matching normal breast tissues, and three breast cancer lymph node metastasis specimens (Figure 1). Downregulation was defined as the percentage of normal/tumour tissue pairs exhibiting an at least twofold reduction of SFRP1 expression in the tumour. According to this criterion, the CPA showed downregulation of SFRP1 mRNA in 46 out of 50 primary breast tumours (92%), as well as in all three metastatic lymph nodes, as compared to matched normal breast tissue. SFRP1 was upregulated in only one out of 50 primary breast tumours (2%), as compared to corresponding normal tissue, while in three cases (6%) there was no difference in SFRP1 expression between the tumour and the matched normal breast tissue sample.

Figure 1
figure1

Downregulation of SFRP1 in breast cancer. Expression profiles were determined using the Clontech cancer profiling array containing cDNA pairs derived from 50 primary breast cancer, 50 normal breast tissue, and three breast tumour lymph node metastasis specimens. Rows T1 (A to Z and aa to ff) and T2 (A to U) represent breast cancer, and rows N1 (A to Z and aa to ff) and N2 (A to G, I, K and M to U) represent normal breast tissue. The outlined groups represent primary tumour (T2-G, T2-I and T2-K) metastatic lymph nodes (T2-H, T2-J and T2-L) and normal breast tissue (N2-G, N2-I and N2-K) from the same patient (T=tumour, N=normal).

LOH and mutation analysis of the SFRP1 gene

Allelic loss of the SFRP1 gene locus was analysed in 43 matched normal/tumour DNA samples using the intragenic microsatellite marker D8S532 and the D8S268 marker in the close vicinity of the SFRP1 gene. Allelic loss was shown in 33.0% (D8S532) and 43.8% (D8S268) of informative cases, in total 11 tumours were identified displaying LOH at least with one of the two markers on 8p11.21 (representative examples are shown in Figure 2). These 43 tumours were further analysed for SFRP1 mutations by genomic sequencing of SFRP1 exons and splice sites. No pathogenic mutations in the SFRP1 gene could be detected in any of the 43 breast tumour samples.

Figure 2
figure2

Loss of heterozygosity analysis at the SFRP1 gene locus with microsatellite markers D8S268 and D8S532 (intragenic). Printouts from the automated DNA sequencer (a+b: marker D8S268) (c+d: marker D8S532) displaying PCR products of normal DNA (N: upper lane) and tumour DNA (T: lower lane), respectively. a+c: Both alleles are retained in the tumour DNA (no LOH) whereas allelic loss of the longer allele, respectively, is demonstrated in b and d.

Hypermethylation and expression of SFRP1 in breast cell lines

Since mutations in the SFRP1 gene could not account for its abundant loss in breast cancer we started analysis of epigenetic alterations in the SFRP1 promoter. We used the MSP technique and the highly specific MSP primers described by Suzuki et al. (2002). Two non-malignant cell lines (MCF10A, MCF12A) and eight malignant cell lines (BT20, BT474, MDA-MB-231, MDA-MB-453, MCF7, SKBR3, T47-D, ZR75-1) were assayed (Figure 3a). While the non-malignant cell lines exhibited a complete lack of SFRP1 methylation, the malignant cell lines showed hypermethylation of the SFRP1 promoter region. Next we asked whether SFRP1 methylation of tumour cell lines is associated with a loss of SFRP1 mRNA expression in these cells. The RT–PCR experiment demonstrated that SFRP1 mRNA is strongly expressed in MCF10A and MCF12A cells but is not detectable in any of the malignant cell lines harbouring SFRP1 methylation (Figure 3b).

Figure 3
figure3

SFRP1 methylation, expression and re-expression analysis of mammary cell lines. (a) MSP was performed with bisulphite-treated DNA of either benign (MCF10A, MCF12A) or malignant breast cell lines (BT20, BT474, MDA-MB-231, MDA-MB-453, MCF7, SKBR3, T47-D, ZR75-1). DNA bands in lanes labelled with U indicate PCR products amplified with primers recognising unmethylated promoter sequences. DNA bands in lanes labelled with M represent amplified products with methylation-specific primers; water served as ‘no template control’ (NTC). Only unmethylated DNA molecules were detected in the two benign cell lines MCF10A and MCF12A with signals in the U-reaction, but not for the M-reaction. In contrast, all cancerous cell lines showed aberrant hypermethylation of the SFRP1 promoter indicated by a clear signal in the M reaction. (b) RT–PCR expression analysis of SFRP1 mRNA in mammary cell lines. SFRP1 mRNA was only detectable in the non-malignant cell lines MCF10A and MCF12A indicating a perfect association of promoter methylation and loss of SFRP1 expression in breast cancer cell lines. The parallel analysis of GAPDH expression by RT–PCR served as a control for equal starting amounts of cDNA. (c) Re-expression of SFRP1 after treatment of breast cancer cell lines with 5-aza-2′-deoxycytidine (DAC). Breast cancer cell lines BT20, MCF7, SKBR3 and T47-D were treated with 1 μ M of DAC for 3 days. RT–PCR data for cell lines treated (+) and untreated controls (−) are shown. All cell lines restored expression of SFRP1 expression compared to the control cells that were not treated with the demethylating agent.

Re-expression of SFRP1 after treatment with demethylating agents

To further support the hypothesis that aberrant SFRP1 methylation is the cause of SFRP1 gene silencing in breast cancer cell lines we treated four representative breast cancer cell lines (BT20, MCF7, SKBR3, T47D) with the demethylating agent DAC. SFRP1 expression of these cells collected at 24 h after the final addition of DAC was analysed by RT–PCR (Figure 3c). We found that expression of SFRP1 was restored in all four cell lines indicating that promoter demethylation can abrogate the apparent block of SFRP1 transcription in breast cancer cell lines.

Methylation of the SFRP1 promoter in primary breast cancers

Owing to SFRP1 promoter methylation found in breast cancer cell lines, we analysed whether promoter methylation may contribute to SFRP1 gene silencing in human primary breast cancer as well. Since it has been reported that normal human tissue may exhibit some degree of age-related gene methylation (Waki et al., 2003), we analysed, whenever possible, matching samples of tumour and normal breast tissue by MSP. 130 mammary tumour samples, of which 26 had matching normal breast epithelium, were analysed. Representative results are shown in Figure 4. Altogether, in 107/130 (82%) breast tumour samples (e.g. patients #14, #18, #19, #21, #27 in Figure 4) a PCR product was amplified with methylation-specific primers indicating a methylated SFRP1 promoter. Other tumours (e.g. #9, #20 in Figure 4) showed no evidence of aberrant SFRP1 promoter methylation since no methylation-specific product was detectable. Some of the methylated breast tumours showed only a very weak but reproducible methylation signal (see e.g. lane M of tumour #19 in Figure 4). Although MSP is usually not considered as a quantitative technique, these very weak methylation signals were scored as an independent third group beside unmethylated and (clearly) methylated tumours. To distinguish very weak versus clear methylation we defined a threshold signal as described in Materials and methods. Overall, clear SFRP1 promoter methylation was found in 79 of 130 (61%) breast tumours, very weak methylation was detectable in 28/130 (22%) of breast tumours and no methylation was detectable in 18% of tumours (23/130). In the normal breast epithelium, we detetected a very weak methylation signal in 2/26 (8%) of analysed normal breast tissues (data not shown).

Figure 4
figure4

SFRP1 methylation analyses of primary breast cancer specimens. Methylation-specific PCR (MSP) was performed on bisulphite-treated DNA from breast cancer (T) and matching normal primary breast tissue (N). MSP results from seven representative patients are shown. DNA bands in lanes labelled with U indicate PCR products amplified with primers recognising the unmethylated promoter sequence. DNA bands in lanes labelled with M represent amplified products with methylation-specific primers. Human placenta (PL) and in vitro methylated DNA (IVD) served as positive controls for the U and M reaction, respectively. Water was used as template in the negative control (NTC). Note that tumour tissue usually displayed a PCR product in the U-reaction as well, due to contaminating normal tissue (stromal cells, endothelial cells) present in the tumour specimens as has also been described by Suzuki et al. (2004).

Correlation analysis of SFRP1 expression and methylation in primary breast cancers

Next we examined SFRP1 mRNA expression by real-time PCR in the same cohort of tumours (n=130) used for methylation analysis. cDNA was available for 86 of those 130 tumour samples. SFRP1 mRNA was downregulated in 75 of these 86 tumours (87%) compared to normal breast tissue (by a fold change of >2). We found a highly significant correlation (P<0.001; two-tailed Mann–Whitney U-test) between methylation of the SFRP1 promoter and downregulation of the SFRP1 transcript (Figure 5). Grouping of the tumour samples according to their methylation status (no methylation, very weak methylation, clear methylation) revealed that very weakly methylated tumours exhibit a significant degree of SFRP1 mRNA downregulation (P=0.01), compared to the non-methylated tumours, however, the clearly methylated tumours exhibit a much stronger, highly significant SFRP1 mRNA downregulation (P<0.001) compared to the weakly methylated tumours. Compared to the non-methylated breast tumours (set equal to 1) the very weakly methylated tumours exhibited on average a 2.1-fold downregulation, however, the strongly methylated breast tumours showed on average a 10.1 fold downregulation compared to the non-methylated tumours (Figure 5).

Figure 5
figure5

Box plot analysis illustrating loss of SFRP1 mRNA expression in relation to SFRP1 promoter methylation. Factor of SFRP1 mRNA downregulation relative to the normal breast standard (a normal breast tissue sample containing approximately 40% of epithelial cells) is shown on the Y axis (fold change N/T). Unmethylated breast tumours exhibit a very moderate SFRP1 downregulation, very weakly methylated tumours exhibit a somewhat stronger downregulation, however, only clearly methylated breast tumours exhibit strong SFRP1 downregulation. This correlation between SFRP1 methylation and SFRP1 mRNA downregulation is highly significant (P-values of the Mann–Whitney U-test are given for the two comparisons). Horizontal lines: group medians; boxes: 25–75% quartiles, range, peak and minimum.

Statistical analysis of clinipathologic patient data

For descriptive data analysis, clinico-pathologic characteristics were associated with the SFRP1 methylation status. Since the weakly methylated tumours exhibited an SFRP1 expression value more similar to the unmethylated tumours, we categorised into two groups: weak or no methylation and clear methylation (Table 3). A hypermethylated gene promoter was significantly associated with tumour stage (P=0.032), and loss of SFRP1 RNA transcription (P<0.001), but not with age at diagnosis, lymph node status, grading, oestrogen/progesteron receptor status and histologic type of invasive breast cancer. OS was compared between clearly methylated versus very weakly and unmethylated SFRP1 alleles by univariate log-rank statistics (Table 4). Clear hypermethylation was significantly associated with shorter OS (P=0.045) as shown by Kaplan–Meier analysis (Figure 6). Patients with very weak or no SFRP1 methylation exhibited comparable survival fractions (data not shown). Cox regression analysis was used for the assessment of OS. Only SFRP1 methylation, age at diagnosis, tumour stage, grade and lymph node status were included in the analysis. Limit for stepwise reverse selection procedures has been set to P=0.2. After reverse selection, only SFRP1 and tumour grade remained in the model of which SFRP1 methylation was significant (P=0.047). SFRP1 methylation was an independent risk factor for OS (hazard ratio 1.565; 95% confidence interval 0.992–3.657; P=0.047).

Table 3 Clinico-pathologic and immunohistochemical parameters in relation to SFRP1 promoter methylation
Table 4 Univariate analysis of factors regarding overall survival (OS)
Figure 6
figure6

Kaplan–Meier analysis of patients' overall survival (OS) with SFRP1 methylation. Cumulative survival is presented on the Y axis. The variables consisted of either unmethylated/weak methylated samples (upper graph) or strongly methylated samples (lower graph).

Discussion

It has previously been demonstrated by RNA in situ hybridisation and immmunohistochemistry that SFRP1 expression is lost in about 70% of breast cancer specimens (Ugolini et al., 1999; Klopocki et al., 2004). Since these two techniques are sometimes difficult to quantify, we have determined SFRP1 expression in breast cancer by quantitative cDNA dot blot analysis of 50 matching normal/tumour tissue pairs. According to the well established ‘fold change two approach’ (FC2) SFRP1 downregulation in breast cancer was determined to be 92%. Furthermore, applying real-time PCR techniques to an additional set of primary cancers (n=86) we detected loss of SFRP1 mRNA in 87% of tumours by the FC2 approach. These results clearly indicate that the reported frequencies of SFRP1 loss in breast cancer could be well confirmed by two additional methods in our analysis.

Chromosome 8p is a site of frequent genetic alterations in several tumour entities including breast cancer (Seitz et al., 1997). Particularly the region 8p22–p12 is thought to harbour a tumour suppressor gene that may be involved in the progression of breast cancer (Yokota et al., 1999). Since the SFRP1 gene is located in this chromosomal region and acts as a Wnt antagonist, it may exert a tumour suppressor gene function. Therefore we analysed 43 primary breast cancers for mutations in the SFRP1 coding sequence and the SFRP1 gene splice acceptor and donor sites. Although LOH with the intragenic microsatellite marker D8S532 showed a considerable frequency of loss (44%) we could not detect any pathogenic mutations within the three SFRP1 exons or splice sites. This indicates that genomic mutations are not a major cause of SFRP1 gene inactivation in breast cancer.

Next we analysed whether epigenetic inactivation mediated by promoter hypermethylation could be involved in SFRP1 gene silencing. Hypermethylation of CpG islands near gene promoter regions has been shown to be associated with transcriptional silencing and represents, in addition to genetic aberrations, an important mechanism of gene inactivation in tumorigenesis. A variety of genes involving fundamental cellular pathways have been demonstrated to undergo aberrant DNA methylation in human cancer (Esteller et al., 2002; Herman and Baylin, 2003; Egger et al., 2004). We detected complete SFRP1 promoter methylation in eight breast cancerous cell lines, whereas no methylation was detectable in the non-malignant cell lines. SFRP1 expression was found only in MCF10A and MCF12A cells without SFRP1 promoter methylation. The breast cancer cell lines with methylated SFRP1 promoter sequences did not transcribe detectable amounts of SFRP1 mRNA indicating an epigenetic inactivation of the putative tumour suppressor gene SFRP1. After treatment with the DNA methyltransferase inhibtor DAC, SFRP1 expression was restored in all four cell lines, supporting the hypothesis of methylation-mediated SFRP1 gene silencing in breast cancer.

We next analysed whether SFRP1 promoter methylation also occurs in human primary breast cancers. SFRP1 promoter methylation has been detected in a variety of solid human tumours, including colorectal cancer (Suzuki et al., 2004), ovarian cancer (Takada et al., 2004), mesotheliomas (Lee et al., 2004), and most recently in bladder cancer (Marsit et al., 2005), lung cancer (Fukui et al., 2005) and prostate cancer (Lodygin et al., 2005), indicating that loss of SFRP1 expression might be a common mechanism to aberrantly activate Wnt signalling in solid tumours. In the breast tumour samples we detected a clear SFRP1 methylation in 61% (79 out of 130) of cases. An additional 22% (28 out of 130) of tumours exhibited only a very weak methylation signal that we did not incorporate in either of the two other groups (no methylation, clear methylation). Since the U signal from cells without SFRP1 methylation was always strong in these weakly methylated tumours we can rule out technical problems or a limited DNA quality. Therefore, we interpret these specimens as breast tumours having only a small fraction of SFRP1 methylated tumour cells. The intermediate downregulation of the SFRP1 mRNA in these tumours is in accordance with this interpretation.

Our study shows for the first time a striking correlation (P<0.001) between SFRP1 promoter hypermethylation and SFRP1 mRNA downregulation. Furthermore, multivariate Cox regression analysis showed that SFRP1 promoter hypermethylation may represent an independent adverse prognostic factor for OS in breast cancer. Although we have to address this issue in a larger prospective study, the presented Kaplan–Meier analysis demonstrates that clear SFRP1 promoter methylation is associated with unfavourable prognosis. Interestingly, a similar relationship between loss of SFRP1 promoter methylation and unfavourable disease prognosis has recently been demonstrated for bladder cancer as well (Marsit et al., 2005). Our data further imply that the correlation between SFRP1 methylation and OS in breast cancer is dependent on a gene dose effect. OS may only be affected if sufficient tumour cells have lost SFRP1 expression due to promoter methylation. Future projects including cloning and sequencing of bisulphite-treated DNA from methylated breast tumours should address this interesting question, potentially leading to the definition of a cutoff level that could be used for breast cancer prognosis.

In breast cancer, several putative tumour suppressor genes have recently been described as being downregulated due to CpG methylation, for example, ADAM32 (Costa et al., 2004), DSC3 (Oshiro et al. 2005) or TPM1 (Varga et al., 2005). The loss of their corresponding proteins may lead to major aberrations in important cellular networks, thus promoting tumorigenesis, for example, by affecting cell adhesion (ADAM23 and DCS3) or cell migration (TPM1). Loss of SFRP1 may affect cell proliferation via activation of the Wnt pathway, thereby potentially enhancing tumour growth and promoting malignant transformation.

Since the initial identification of Wnt1 as an effective oncogene in mouse mammary tumour virus-infected mice (Tsukamoto et al., 1988), several lines of evidence have suggested that activation of the canonical Wnt signalling pathway promotes tumorigenesis in human mammary tissues: β-catenin is actively stabilised in over 50% of breast cancers and its nuclear localisation correlates with poor patient prognosis (Lin et al., 2000). Several target genes of the Wnt signalling pathway like cyclin D1 are activated in a significant proportion of breast tumours (reviewed in Brown, 2001). However, mutational inactivation of Wnt pathway components such as APC and Axin is a rare event in breast cancer compared to other tumour entities like colon or hepatocellular carcinomas. In this study, we found that hypermethylation of the SFRP1 promoter is the predominant cause of SFRP1 gene silencing in breast cancer. Our data raise the hypothesis that methylation of the SFRP1 gene might have comparable effects as mutational inactivation of other negative Wnt regulators (e.g. APC, Axin) resulting in constitutive activation of this signalling pathway in breast cancer. Additional functional studies on the role of the SFRP1 protein and its interactions with other regulatory proteins of the Wnt pathway are needed. These experiments should clarify and extend the importance of the Wnt signalling pathway for the pathogenesis of human breast cancer.

Materials and methods

Clinical material

Matched tumour/normal samples of breast cancer specimens (n=130) analysed in this study were obtained from patients treated by primary surgery for breast cancer at the Departments of Gynecology at the University Hospitals of Aachen, Jena, Düsseldorf and Charité, Berlin, Germany. Tumour material used for expression analysis was snap frozen in liquid nitrogen immediately after surgery. Haematoxylin- and eosin-stained sections were prepared for assessment of the percentage of tumour cells, only samples with >70% tumour cells were selected for analysis. Frozen tissue samples were homogenised in liquid nitrogen and dissolved in lysis buffer followed by DNA isolation using the blood and cell culture DNA kit (Qiagen, Hilden, Germany) or by RNA isolation using TRIZOL (Gibco-BRL, Glasgow, UK) according to the protocols supplied by the manufacturers. For patients' characteristics see Table 1.

Table 1 Clinico-pathologic characteristics of primary carcinomas (n=130)

Cell lines

The human mammary epithelial cell lines MCF10A and MCF12A as well as the breast cancerous cell lines BT20, BT474, MDA-MB-231, MDA-MB-453, SKBR3, MCF7, T47-D and ZR75-1 were obtained from the ATCC (Rockville, MD, USA) and cultured under recommended conditions. MCF10A and MCF12A were additionally supplemented with 10 μg/ml insuline (Novo Nordisk, Bagsvaerd, Denmark), 500 ng/ml hydrocortisone and 20 ng/ml epidermal growth factor (Sigma-Aldrich, Deisenheim, Germany).

RT–PCR

Of the total RNA, 1 μg was reverse transcribed using the Reverse Transcription System (Promega, Madison, WI, USA). For PCR, 1 μl cDNA was amplified using SFRP1 and GAPDH primers given in Table 2. Cycle conditions applied for both genes were: 35 cycles of 95°C for 1 min, 60°C for 1 min, 72°C for 1 min and a final extension at 72°C for 10 min.

Table 2 Primer sequences and annealing temperatures used in this study

Expression analysis using the cancer profiling array (CPA)

The matched tumour/normal expression array used consists of 103 cDNAs, synthesised from human malignant and corresponding normal tissues, that is, 50 breast cancer, 50 normal breast tissue and three breast tumour lymph node metastasis specimens and has been described in Dahl et al. (2005). Hybridisation using 25 ng of a gene-specific 32P-labelled cDNA probe digested from Unigene cDNA clone (Accession Number W21306) was performed according to the manufacturer's recommendations. The tumour/normal intensity ratio was calculated using a STORM-860 phosphorimager (Molecular Dynamics, Sunnyvale, CA, USA) and normalised against the background.

Loss of heterozygosity (LOH) analysis

LOH analysis was carried out as described previously (Niederacher et al., 1997). Briefly, matched normal/tumour DNA samples were analysed for LOH using the two microsatellite markers D8S532 and D8S268. For primer sequences and annealing temperatures see Table 2. One primer of each primer pair was Cy5-labelled at the 5′ end. The target sequences were amplified by PCR in 50 μl of 1 × Taq polymerase reaction buffer containing 40 pmol of each primer, 1.5 mM MgCl2, 200 μ M each of dNTPs, 2.5 U Taq, and 20–50 ng of genomic DNA. Cycling conditions were as follows: 5 min denaturation at 95°C, 35 cycles of denaturation at 94°C for 1 min, annealing at 58°C for 1 min and extension at 72°C for 1 min, followed by a final extension for 8 min at 72°C. DNA amplification was performed in an Omnigene thermal cycler (Hybaid, Middlesex, UK).

PCR products were analysed on 6% polyacrylamide denaturing gels in 0.6 × TBE buffer in an automated laser-activated fluorescent DNA sequencer (ALFexpress, Pharmacia Biotech, Uppsala, Sweden). In all samples 5 μl of the diluted PCR reaction (1:20–1:80) were mixed with 5 μl of stop solution (90% formamide, 10 mM EDTA, 0.3% bromophenol blue). The mix was denatured at 95°C for 10 min, cooled on ice, loaded into each well of the preheated gel (40°C), and run for 3–4 h at 30 W. While the samples were undergoing electrophoresis, fluorescence was detected after laser activation. Fluorescent gel data were collected automatically during the electrophoresis and calculated using Fragment Manager (FM1.1) software (Pharmacia), which yields quantification of results in terms of peak size, height and area. The sizes of the two alleles were assigned to the peaks of greatest height; smaller peaks were interpreted as polymerase artefacts, so-called stutter bands. For assessment of allele loss the ratio of allele peak areas was calculated in matched normal and tumour DNA samples. The ratio obtained in tumour DNA divided by the allele peak ratio of paired normal DNA below 0.5, which means an allele signal reduction of 50%, was considered to be indicative of allele loss.

Mutation analysis of SFRP1

All three exons as well as splice sites of SFRP1 were examined for mutations by genomic sequencing analysis. Primer sequences and annealing temperatures are given in Table 2. The procedure has been described in Stoehr et al. (2004).

Semiquantitative real-time PCR

Semiquantitative PCR was performed using the LightCycler system together with the LightCycler DNA Master SYBR Green I Kit (Roche Diagnostics). Reaction volumes of 20 μl consisted of the following components: 25 mM MgCl2, 10 μ M forward primer, 10 μ M reverse primer, 2 μl LightCyler DNA Master SYBR Green I and 2 μl of cDNA as PCR template. Gene expression was quantified by the comparative CT method, normalising CT values to the housekeeping gene GAPDH and calculating relative expression values (Fink et al., 1998).

Primer sequences for SFRP1 and GAPDH are listed in Table 2. The cycling conditions were set up to an initial denaturation at 95°C for 15 min, followed by 40 cycles with denaturation at 95°C for 20 s, annealing at 60°C for 20 s and elongation at 72°C for 30 s. To verify the specificity of the PCR products, melting curve analyses were performed. The relative SFRP1 expression levels were standardised to the expression level of a normal breast tissue sample that contained approximately 40% of epithelial cells (tumours generally contained >70% of tumour cells). To ensure experiment accuracy, all reactions were performed in triplicates.

Bisulphite-modification and MSP

Of the genomic DNA, 1 μg was bisulphite modified using the DNA modification Kit (Chemicon, Ternecula, CA, USA) according to the manufacturer's recommendations. The final precipitate was eluted in 30 μl of pre-warmed (55°C) TE-buffer. Methylation-specific PCR was performed according to Herman et al. (1996). In short, 1 μl of modified DNA was amplified using MSP primers (see Table 2) that specifically recognised either the unmethylated or methylated SFRP1 gene sequence after bisulphite conversion (Suzuki et al., 2004). Normal DNA from human placenta was bisulphite modified to serve as a control for the unmethylated promoter sequence. Normal human DNA was treated in vitro with SssI methyltransferase (New England Biolabs, Beverly, MA, USA) in order to generate a positive control for methylated alleles (Esteller et al., 1999). Amplification products were visualised on 3% low-range ultra agarose gel (Bio-Rad Laboratories, Hercules, CA, USA) containing ethidium bromide and illuminated under UV light. For the estimation of methylation intensity bisulphite modified in vitro methylated positive control DNA from placenta (IVD) was diluted 1:3 with bisulphite modified placenta DNA, thus representing a methylation status of approximately 30%. Tumour samples were scored as significantly methylated, when the amplification signal was at least as strong as the signal achieved by the diluted IVD probe in the same PCR reaction. Methylation signals were scored as weakly methylated when signal strength was below this cutoff of 30% methylation.

5-Aza-2′-deoxycytidine (DAC) treatment

Cells were seeded at a density of 3 × 104 cells/cm2 in a six-well plate on day 0. The demethylating agent DAC (Sigma-Aldrich, Steinheim, Germany) was added to a final concentration of 1 μ M in fresh medium on days 1, 2 and 3. Cells were harvested on day 4 for RNA extraction. Control cells were incubated without the addition of DAC and fresh medium was also supplied on days 1, 2 and 3.

Statistical analysis of clinico-pathologic patient data

Statistical analyses were completed using SPSS version 12.0. (SPSS, Chicago, IL, USA). Differences were considered statistically significant when P-values were <0.05. Contingency table analysis and two-sided Fisher's exact tests were used to study the statistical association between clinico-pathologic data and methylation status. OS curves comparing patients with or without any of the factors were calculated using the Kaplan–Meier method, with significance evaluated by two-sided log-rank statistics. OS was measured from time of surgery. Patients were censored at the time of their last tumour-free clinical follow-up appointment or at their date of death not related to the tumour. A multivariable Cox regression was employed, testing the independent prognostic relevance of SFRP1 methylation. The limit for reverse selection procedures was P=0.2. The proportionality assumption for all variables was assessed with log-negative-log survival distribution functions. Characteristics of all variables are summarised in Tables 3 and 4.

Accession codes

Accessions

GenBank/EMBL/DDBJ

References

  1. Bafico A, Gazit A, Pramila T, Finch PW, Yaniv A, Aaronson SA . (1999). J Biol Chem 274: 6180–6187.

  2. Behrens J, von Kries JP, Kuhl M, Bruhn L, Wedlich D, Grosschedl R et al. (1996). Nature 382: 638–642.

  3. Brown AM . (2001). Breast Cancer Res 3: 351–355.

  4. Cadigan KM, Nusse R . (1997). Genes Dev 11: 3286–3305.

  5. Caldwell GM, Jones C, Gensberg K, Jan S, Hardy RG, Byrd P et al. (2004). Cancer Res 64: 883–888.

  6. Costa FF, Verbisck NV, Salim AC, Ierardi DF, Pires LC, Sasahara RM et al. (2004). Oncogene 23: 1481–1488.

  7. Dahl E, Sadr-Nabavi A, Klopocki E, Betz B, Grube S, Kreutzfeld R et al. (2005). J Pathol 205: 21–28.

  8. Egger G, Liang G, Aparicio A, Jones PA . (2004). Nature 429: 457–463.

  9. Esteller M, Gaidano G, Goodman SN, Zagonel V, Capello D, Botto B et al. (2002). J Natl Cancer Inst 94: 26–32.

  10. Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG . (1999). Cancer Res 59: 793–797.

  11. Finch PW, He X, Kelley MJ, Uren A, Schaudies RP, Popescu NC et al. (1997). Proc Natl Acad Sci USA 94: 6770–6775.

  12. Fink L, Seeger W, Ermert L, Hanze J, Stahl U, Grimminger F et al. (1998). Nat Med 4: 1329–1333.

  13. Fukui T, Kondo M, Ito G, Maeda O, Sato N, Yoshioka H et al. (2005). Oncogene 24: 6323–6327.

  14. He TC, Sparks AB, Rago C, Hermeking H, Zawel L, da Costa LT et al. (1998). Science 281: 1509–1512.

  15. Herman JG, Baylin SB . (2003). N Engl J Med 349: 2042–2054.

  16. Herman JG, Graff JR, Myohanen S, Nelkin BD, Baylin SB . (1996). Proc Natl Acad Sci USA 93: 9821–9826.

  17. Klopocki E, Kristiansen G, Wild PJ, Klaman I, Castanos-Velez E, Singer G et al. (2004). Int J Oncol 25: 641–649.

  18. Lai J, Flanagan J, Phillips WA, Chenevix-Trench G, Arnold J . (2003). Br J Cancer 88: 270–276.

  19. Lee AY, He B, You L, Dadfarmay S, Xu Z, Mazieres J et al. (2004). Oncogene 23: 6672–6676.

  20. Lin SY, Xia W, Wang JC, Kwong KY, Spohn B, Wen Y et al. (2000). Proc Natl Acad Sci USA 97: 4262–4266.

  21. Lodygin D, Epanchintsev A, Menssen A, Diebold J, Hermeking H . (2005). Cancer Res 65: 4218–4227.

  22. Mann B, Gelos M, Siedow A, Hanski ML, Gratchev A, Ilyas M et al. (1999). Proc Natl Acad Sci USA 96: 1603–1608.

  23. Marsit CJ, Karagas MR, Andrew A, Liu M, Danaee H, Schned AR et al. (2005). Cancer Res 65: 7081–7085.

  24. Niederacher D, Picard F, van Roeyen C, An HX, Bender HG, Beckmann MW . (1997). Genes Chromosomes Cancer 18: 181–192.

  25. Oshiro MM, Kim CJ, Wozniak RJ, Junk DJ, Munoz-Rodriguez JL, Burr JA et al. (2005). Breast Cancer Res 7: R669–80.

  26. Polakis P . (2000). Genes Dev 14: 1837–1851.

  27. Remmele W, Stegner HE . (1987). Der Pathol 8: 138–140.

  28. Seitz S, Rohde K, Bender E, Nothnagel A, Kolble K, Schlag PM et al. (1997). Oncogene 14: 741–743.

  29. Stoehr R, Wissmann C, Suzuki H, Knuechel R, Krieg RC, Klopocki E et al. (2004). Lab Invest 84: 465–478.

  30. Suzuki H, Gabrielson E, Chen W, Anbazhagan R, van Engeland M, Weijenberg MP et al. (2002). Nat Genet 31: 141–149.

  31. Suzuki H, Watkins DN, Jair KW, Schuebel KE, Markowitz SD, Chen WD et al. (2004). Nat Genet 36: 417–422.

  32. Takada T, Yagi Y, Maekita T, Imura M, Nakagawa S, Tsao SW et al. (2004). Cancer Sci 95: 741–744.

  33. Tetsu O, McCormick F . (1999). Nature 398: 422–426.

  34. Tsukamoto AS, Grosschedl R, Guzman RC, Parslow T, Varmus HE . (1988). Cell 55: 619–625.

  35. Ugolini F, Adelaide J, Charafe-Jauffret E, Nguyen C, Jacquemier J, Jordan B et al. (1999). Oncogene 18: 1903–1910.

  36. Ugolini F, Charafe-Jauffret E, Bardou VJ, Geneix J, Adelaide J, Labat-Moleur F et al. (2001). Oncogene 20: 5810–5817.

  37. Uren A, Reichsman F, Anest V, Taylor WG, Muraiso K, Bottaro DP et al. (2000). J Biol Chem 275: 4374–4382.

  38. Varga AE, Stourman NV, Zheng Q, Safina AF, Quan L, Li X et al. (2005). Oncogene 24: 5043–5052.

  39. Waki T, Tamura G, Sato M, Motoyama T . (2003). Oncogene 22: 4128–4133.

  40. Yokota T, Yoshimoto M, Akiyama F, Sakamoto G, Kasumi F, Nakamura Y et al. (1999). Cancer 85: 447–452.

Download references

Acknowledgements

The technical assistance of Sonja von Serenyi, Sevim Alkaya and Inge Losen is greatly appreciated. The study was supported by the German Ministry for Education and Research (BMBF Grant 01KW0404) as part of the German Human Genome Project (DHGP) and a grant from the RWTH Aachen (START program).

Author information

Affiliations

Authors

Corresponding author

Correspondence to E Dahl.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Veeck, J., Niederacher, D., An, H. et al. Aberrant methylation of the Wnt antagonist SFRP1 in breast cancer is associated with unfavourable prognosis. Oncogene 25, 3479–3488 (2006). https://doi.org/10.1038/sj.onc.1209386

Download citation

Keywords

  • breast cancer
  • tumour suppressor gene
  • Wnt pathway
  • SFRP1
  • methylation
  • epigenetics

Further reading

Search

Quick links