Background
Epithelial ovarian cancer is the most lethal gynecological cancer among women worldwide. Metastasis occurs in most patients before diagnosis, and the majority of patients have high subsequent recurrent rate after completing surgery and chemotherapy [
1]. Survival is at about 3 years for patients in advanced stages [
2]. The poor prognosis of ovarian cancer is due to the difficulty in early diagnosis and the detrimental processes of invasion and metastasis. A better understanding of the molecular mechanisms of cancer development and progression will help to improve the diagnosis and treatment of the disease.
Insulin-like growth factor binding proteins (IGFBPs) are circulating transport proteins for IGF, with IGFBP-3 being the predominant IGFBP in circulation [
3]. IGFBP-3 can regulate cell growth and death, either dependent or independent of its interaction with IGF [
3,
4]. Recently, we have further identified
IGFBP-3 as an invasion suppressor gene using an established ovarian cancer cell line OVTW59-P0 and its sublines P1 to P4, which were obtained from a gradual invasion model with sequential increase in invasiveness [
5]. Many tumor suppressor genes involved in cancer formation and progression are frequently epigenetically silenced through aberrant hypermethylation at their promoter regions [
6]. In ovarian cancer, genes with promoter hypermethylation are frequently found to be related to cancer progression [
7,
8]. Aberrant promoter hypermethylation of
IGFBP-3 and gene silencing are observed in many cancers, such as lung, hepatocellular, gastric, colorectal, breast, and ovarian cancers [
9‐
13]. However, the association of
IGFBP-3 promoter hypermethylation with poor clinical outcome was identified only at early stages in lung and ovarian cancers [
9,
10].
p53 is a known transcription factor for IGFBP-3 expression [
9]. Induction of IGFBP-3 by p53 has been shown to cause cell apoptosis in an IGF-independent manner [
14]. Eleven p53 binding sites have been identified within IGFBP-3 gene based on the homology to the p53 binding consensus sequence, and confirmation by electrophoretic mobility shift analyses [
15]. Promoter hypermethylation at these p53 binding sites caused gene silencing and resistance to p53 [
16]. Therefore, it has been suggested that p53 could mediate cross-talk to the IGF axis through IGFBP-3 regulation [
14].
Pathologically, epithelial ovarian cancer is classified into four major histological subtypes: serous, mucinous, endometrioid and clear cell carcinoma. Each subtype is associated with distinct molecular alterations [
17]. In our previous study, we found 51.4% of the ovarian endometrioid carcinoma (OEC) subtype showing lower IGFBP-3 expression, which is associated significantly with poor patient outcome [
5]. From literature review, p53 overexpression is more frequently reported in the serous subtype of ovarian cancer [
18‐
20]. The association of altered p53 expression in tumor tissue with patient survival in ovarian cancer is still under debate [
19‐
21], but a significant correlation is often reported in the serous subtype [
18,
19]. Despite the close molecular regulation between p53 and IGFBP-3 being known for more than a decade [
9], little information concerning p53 regulation of IGFBP-3 or the effect of epigenetic inactivation of IGFBP-3 exist with defined clinical endpoints. In the current study, we analyzed the association between p53 and IGFBP-3 expression in ovarian cancer progression. We explored the clinical significance of aberrant promoter hypermethylation at the p53 binding sites of
IGFBP-3 in OEC. Functional analysis of p53 regulation on IGFBP-3 expression was further assessed using the OVTW59-P0 and P4 cell lines. Our study indicates that methylation at the p53 transcription factor binding sites in the
IGFBP-3 promoter can silence IGFBP-3 expression and hence lead to cancer progression in OEC.
Discussion
We demonstrated previously that IGFBP-3 is an invasion suppressor in OEC [
5]. In this study, we evaluated the molecular mechanism of IGFBP-3 regulation in cancer cells and found that methylation at the CpG sites of the
IGFBP-3 promoter was significantly correlated with IGFBP-3 silencing in OEC tumors. Clinically, patients with lower IGFBP-3 expression and higher
IGFBP-3 promoter methylation have significantly lower PFS and OS. This association was strongest in cases with normal p53 expression. In an
in vitro study using OEC cell line OVTW59, we identified a quantitative correlation between IGFBP-3 suppression and methylation in the first 200 base pairs surrounding the transcription start site of
IGFBP-3. Our data suggest that IGFBP-3 silencing through aberrant hypermethylation at its promoter is a major regulatory pathway for OEC tumorigenesis.
Methylation specific PCR (MSP) has been widely used to study promoter methylation in clinical specimens. By MSP, aberrant methylation at the
IGFBP-3 promoter was found in 44% of epithelial ovarian cancer and in 28.6% (12/42) of OEC subtype [
7,
12]. However, IGFBP-3 silencing has been reported unrelated to
IGFBP-3 promoter methylation [
7,
12]. A significant association between
IGFBP-3 promoter methylation and disease progression has been reported, but only in early-stage ovarian cancer [
12]. It was concluded that methylation at
IGFBP-3 promoter is a common tumorigenesis process in the early steps of ovarian cancer progression [
12]. By setting the calculated percentage of IGFBP-3 methylation at 3% as a cutoff point for qMSP analysis, we found methylation at the
IGFBP-3 promoter significantly correlated with IGFBP-3 silencing in OEC. Furthermore, we found patients with lower IGFBP-3 expression and higher
IGFBP-3 promoter methylation to be significantly associated with poor outcomes. A possible reason is that we have used a more sensitive method, qMSP, to study these epigenetic events. In addition, we have focused our study on the OEC subtype of ovarian cancer. It has been observed that promoter methylation of specific genes in cancer occur frequently in a tumor-type and cell-type specific manner [
23]. Our results demonstrate that the clinical significance of aberrant
IGFBP-3 promoter methylation is more commonly observed in the OEC subtype of ovarian cancer.
Ten cases in our study showed unparallel results between IGFBP-3 expression and
IGFBP-3 promoter methylation, i.e. either high IGFBP-3 and
IGFBP-3 promoter methylation > 3% or low IGFBP-3 and
IGFBP-3 promoter methylation ≤ 3%. Though threshold of > 3% as hypermethylation was selected based on ROC curve analysis that showed 0.86 sensitivity and 0.73 specificity of low IGFBP-3 expression, we found five (50%) of these unparallel cases were at the third quartile of methylation, i.e. high IGFBP-3 expression and promoter methylation levels between 3% and 6.36% (Table
2). This could be due to inadequate sensitivity and specificity of qMSP assay. Alternatively, the unparallel result could be explained by multiple mechanisms of IGFBP-3 regulation. IGFBP-3 could also interact with several other growth-inhibitory agents to mediate wide varieties of growth suppression signal in the absence of IGF [
24,
25]. In addition, other CpG methylation sites in the
IGFBP-3 promoter that were not studied might also contribute to IGFBP-3 silencing
We further explored the association between p53 and
IGFBP-3 promoter methylation and the mechanism of p53 regulation on IGFBP-3. Thirty-five percent of these cases showed p53 overexpression that was correlated with higher tumor grade at marginal significance. However, overexpression of p53 was not correlated with IGFBP-3 expression or
IGFBP-3 promoter methylation and it did not correlate with OEC patient survival. This suggested that p53 is not important in the progression of OEC, and that IGFBP-3 silencing through promoter methylation is a mechanism different from p53 overexpression. Since OEC is similar in histological pattern to the endometrioid (EC) subtype of endometrial cancer, we reviewed the molecular characteristics of different subtypes of endometrial cancer. In endometrial cancer, p53 alteration was specifically reported to be present in the non-endometrioid (non-EC) serous subtype, which has been classified as a type 2 endometrial cancer. The EC subtype of endometrial cancer is classified as type 1 endometrial cancer [
26]. Accordingly, Kurman
et al. proposed an epithelial ovarian cancer model composed of type 1 and type 2 tumors based on: the histological patterns, molecular features, and the process of tumor progression. Type 1 tumors are slow growing, genetically stable, and are characterized by mutations in KRAS, BRAF, PTEN, and beta-catenin. Type 2 tumors are highly aggressive, rapidly growing, genetically instable, and are characterized by mutation of TP53 [
20]. Based on this model, the OEC subtype is classified as a type 1 tumor and serous subtype, with characteristic TP53 mutations, is classified as a type 2 tumor [
20]. In our study, we confirmed that p53 alteration is not a characteristic of OEC. Instead, low IGFBP-3 and high
IGFBP-3 promoter methylation are major prognostic factors for OEC. By survival analysis, patient survival is lowest in cases with low IGFBP-3, high
IGFBP-3 promoter methylation, and normal p53. These suggest OEC is a distinct subtype of ovarian cancer that IGFBP-3 silencing through
IGFBP-3 promoter methylation could play an important role in cancer development and progression. We found p53 alteration is not important for the tumorigenesis of OEC, but IGFBP-3 silencing through
IGFBP-3 promoter methylation might subsequently interrupt the communication network between p53 and the IGF axis, and hence lead to cancer progression of OEC.
Of the eleven p53 binding sites in IGFBP-3 gene that has been identified [
22], Hanafusa
et al. observed that at least four sites between -210 to -150 are essential for p53 induced expression of IGFBP-3 in human hepatocarcinoma cell line HepG2. They also found that hypermethylation of these sequences could selectively suppress p53 induced IGFBP-3 expression [
16]. Using our OEC cell line, we identified four sites in the
IGFBP-3 promoter at the -210, -206, -183, and -179 loci, as methylation hot spots. Functional analyses have been performed to study the influence of methylation at these sites on the p53 regulation of IGFBP-3. Hanafusa
et al. showed diminished p53 binding and IGFBP-3 repression associated with hypermethylation of these sequences by luciferase assay after co-transfection of p53 and mutant IGFBP-3 promoter constructs, and by electrophoresis mobility shift assay [
16]. In our study, we used pifithrin-α to inhibit p53 and found a subsequent decrease in IGFBP-3 expression in OVTW59-P4 and 293T cell lines, both of which contain wt p53. These functional analyses support the observation that p53 can up-regulate IGFBP-3 expression [
27], and methylation at these sites suppress the p53 activation of IGFBP-3.
Furthermore, our results show a quantitative association between methylation at the IGFBP-3 promoter and IGFBP-3 expression. The amount of methylation at these hot spots were positively related to IGFBP-3 suppressions in the P0 and P4 sublines. Using site-directed mutagenesis, we demonstrated a linear correlation between IGFBP-3 suppression and the amount of methylation at these loci. Our in vitro data support our clinical qMSP finding that IGFBP-3 expression was correlated to IGFBP-3 promoter methylation in OEC tumors.
Our observation that normal p53 is a required factor along with lower IGFBP-3 and higher
IGFBP-3 promoter methylation for a significant survival result also suggests that wt p53 is an important molecular characteristic of OEC subtype. Recently, Kawasaki
et al. reported an inverse association between
IGFBP-3 promoter methylation and microsatellite instability in patients with methylator phenotype colorectal cancers, particularly under the condition of wt p53 [
28]. We postulate that the biological environment of
IGFBP-3 promoter methylation is strictly regulated, such that wt p53 and un-methylated
IGFBP-3 promoter region are necessary factors to maintain a homeostatic condition.
Progression of tumorigenesis involves multiple steps of genetic alteration. From OEC clinical specimens and cell lines, our data indicates that aberrant methylation at p53 consensus sequence binding region of IGFBP-3 promoter could contribute to low IGFBP-3 expression and subsequently to a poor patient survival outcome. Normal p53 plays an important role in this regulation. In conclusion, our studies show clinical evidence on methylation-dependent epigenetic silencing of IGFBP-3 expression regulated by p53. This regulatory pathway represents an important mechanism for loss of IGFBP-3 expression during OEC tumorigenesis and/or progression to metastasis.
Methods
Chemical reagents
General laboratory reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA) and cell culture reagents from Gibco/BRL (Grand Island, NY, USA), unless otherwise specified. 5-aza-dC was obtained from Sigma-Aldrich and pifithrin-α from Calbiochem (Gibbstown, NJ, USA).
Tumor specimen and cell lines
Sixty cases of OEC specimens were collected from the archive of Department of Pathology, National Taiwan University (NTU) Hospital from May, 1994 to September 2006. These patients received surgeries and chemotherapies according to our previous report [
29]. Informed consents were obtained for each patient before receiving operations. OEC cell lines OVTW59-P0 and P4 were established in our laboratory previously [
5]. Cell lines A549, H1299 and 293T were obtained from Taiwan National Health Research Institute cell bank. All cell lines were maintained in DMEM solution with 5% FCS.
Immunohistochemical staining
Staining was performed in 4 μm paraffin sections as described previously [
29]. In brief, sections were dewaxed, rehydrated, heated by microwave and then blocked with 1% H
2O
2 and normal horse serum. Antibodies against IGFBP-3 (Sigma-Aldrich) and p53 (clone DO-7; DAKO, Glostrup, Denmark), as the primary antibodies, were incubated, followed by biotinylated secondary antibodies, avidin-biotin-peroxidase complex and peroxidase substrate were then added for microscopic observation. Interpretations of immunostaining were performed independently by two authors (PLT and CTL). Immunostaining of IGFBP-3 was defined as low or high when less or more, respectively, than 25% of tumor cells exhibited strong staining of cytoplasm. Immunostaining of p53 was defined as normal or overexpressed when less or more, respectively, than 25% of tumor cells exhibited strong nuclear staining.
Bisulfite modification
Tumor parts from 10 μm paraffin sections were dissected under a stereomicroscope for DNA isolation. An estimation of greater than 80% purity of tumor cells was obtained in each case. Genomic DNA was extracted using QlAamp DNA kit (Qiagene, Valencia, CA, USA) and subjected to EpiTect Bisulfite kit (Qiagene) according to the manufacture's protocol. In brief, 2 μg of genomic DNA was subjected to a denature/incubation repeat cycle in sodium bisulfite solution.
Quantitative real time methylation specific PCR (qMSP)
Bisulphite converted DNA was subjected to real time qMSP using ABI StepOne real time PCR system (Applied Biosystems, Foster City, CA) as previously described [
30] with slight modification. In Brief, each reaction contained 12.5 μL of 2 × SYBR green PCR mix (Toyobo, Japan), 160 nM of each primer and 4 μl of bisulphite modified DNA in a total volume of 25 μl at 95°C for 10 min, 40 cycles of 95°C for 15 sec, 67°C for 30 sec, and 72°C for 30 sec. Primers targeting the
IGFBP3 promoter region were shown in Table
4.
β-actin (ACTB) was used to normalize for input DNA. A region of ACTB devoid of any CpG dinucleotide was amplified using the following primer sequences: forward, 5' TGGTGATGGAGGAGGTTTAGTAAGT and reverse, 5' AACCAATAAAACCTACTCCTCCCTTAA. The amount of methylated IGFBP3 and ACTB were determined by the threshold cycle number (Ct) for each sample against a standard curve generated by
Sss I-treated DNA (Millipore, Billerica, MA)-MSP cloned fragment. The sequence of the fragment was confirmed by sequencing reaction. The performance of the standard curve was shown in additional file
1, Figure S3. The percentage of IGFBP3 methylation was calculated as the IGFBP3:ACTB ratio of a sample divided by the same ratio of
SssI-treated sperm DNA (Millipore, Billerica, MA) and multiplying by 100
Table 4
Primer sequences for IGFBP-3 analysis.
Quantitative real time methylation-specific PCR
|
IGFBP3
| F: 5'- AGGTGATTCGGGTTTCGGGC -3' R: 5'- GACCCGAACGCGCCG -3' | 60 (40) | 223 |
Methylation-specific PCR
|
IGFBP3 _MSP | F: 5'-TCGGGTATATTTTGGTTTTTGTAG-3' R: 5'-AAACATATAAAATCCAAACAAAAA-3' | 55 (30) | 351 |
IGFBP3 _M (methythlated) | F: 5'CGAAGTACGGGTTTCGTAGTCG-3' R: 5'-CGAC CCGAACGCGCCGACC-3' | 66 (40) | 129 |
IGFBP3 _U (unmethylated) | F: 5'-TTGGTTGTTTAGGGTGAAGTATGGGT-3' R: 5'-CACCCAACCACAATACTCACATC-3' | 64 (40) | 158 |
Bisulphate-PCR
|
IGFBP3
| F: 5'-TTTGAGAGTGGAAGGGGTAAGGG-3' R: 5'-CCCACTACATAACACCTACAACC-3' | 53 (40) | 512 |
Mutagenesis
|
Wild type IGFBP3 | F: 5'-GGGCACACCTTGGTTCTTGTAG-3' R: 5'-TTCCTGCCTGGATTCCACAGCT-3' | 52 (40) | 316 |
Mutant (M12) IGFBP3 | F: 5'-ACAAGGTGACCCGGGCTCAGGGAGTGAGCACGAGGAGCAGGT-3' R: 5'-ACCTGCTCCTCGTGCTCACTCCCTGAGCCCGGGTCACCTTGT-3' | 55 (12) | |
Mutant (M34) IGFBP3 | F: 5'-GCACGAGGAGACGGTGCCAGGGAGAGTCTCAAGCTCCACGCC-3' R: 5'-GGCGTGCAGCTTGAGACTCTCCCTGGCACCTGCTCCTCGTGC-3' | 55 (12) | |
Methylation-specific PCR (MSP)
Primers for methylated and unmethylaled region of
IGFBP-3 promoter, as previously reported [
31], are listed in Table
3. The schematic diagram showing the locations of primers used are shown in Figure
1a. Twenty μL of reaction volume containing one-twentieth of the modified DNA, 4 deoxynucleoside triphosphates, PCR primers (-MSP) and HotStar TaqDNA MasterMix (Qiagen, Valencia, CA) were set for first PCR. The PCR product was then used as DNA template for a second methylated (-M) or unmethylated (-U) IGFBP-3 PCR reaction. Results were resolved by electrophoresis using 2% agarose gels. Cell line A549 treated with
Sss I was used as a positive methylated control.
Bisulfite-PCR and sequencing (BSP)
The -372 ~+140 bp from the first exon of IGFBP-3 promoter region was amplified from one-twentieth of the modified DNA by PCR using HotStarTaq DNA Master Mix Kit (Qiagene, Valencia, CA). The primer sequences and PCR conditions are shown in Table
4. The pGEM-T Easy Vector (Promega Corp., Madison, WI, USA) was used for TA cloning and high-efficiency competent cells DH5α (Yeastern, Taipei, Taiwan) was used for transformation. Colonies were selected and rechecked by electrophoresis. The recovered plasmids were sequenced at the Sequencing Core Facility of the NTU College of Medicine. Cytosines in CpG dinucleotides that remained unconverted after bisulfite treatment were considered as methylated.
Western blotting
Twenty-fold concentrated conditioned media, collected from cells cultured in FCS-free DMEM media for 24 hours, were subjected to 8% SDS-PAGE (Millipore, Bedford, MA, USA) and then transferred to membranes (Schleicher & Schuell, Germany) using MilliBlot-SDS semi-dry electroblotting system (Millipore). After blocking, the membrane were probed with antibodies against IGFBP-3 (Research Reagents, Texas, USA). Reactions were amplified by biotinylated second antibody, followed by streptavidin-HRP. Results were exposed using ECL system.
cDNA preparation and quantitative real time reverse-transcriptase PCR (QRT-PCR)
Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and cDNA was prepared using SuperScript III First-strand Synthesis of the Oligo (dT) primer system (Invitrogen). The cDNA samples were subjected to QRT-PCR as described previously [
5] using 7700 Sequence Detector (Applied Biosystems, Foster City, CA, USA) and SYBR green Master Mix Kit (Alppied Biosystems). Primer sequences for IGFBP-3 were forward: 5' TGTGGCCATGACTGAGGAAA, reverse: 5' TGCCAGACCTTCTTGGGTTT; and GAPDH were forward: 5' TGGTATCGTGGAAGGACTCA, reverse: 5' AGTGGGTGTCGCTGTTGAAG. Each target gene was normalized to GAPDH mRNA expression for comparison.
Construction of wild and mutant expression vectors
The promoter region of IGFBP-3 (from -253 ~+61) and the site-directed mutagenesis constructs at 4 different mutant sites (C to A at -210, -206, -183 and -179) were generated by PCR amplification with primers and PCR conditions as shown in Table
4. Wild type pGL3-IGFBP-3 and the mutant constructs were generated by subcloning of the PCR fragments to pGL3-Basic plasmid (Promega Corp.). The mutant constructs were named as Mut A (mutation at -210 and -206), Mut B (mutation at -183 and -179) and Mut A+B (mutant at all of the four sites), respectively, as shown in Figure
3c. All constructs were confirmed by DNA sequencing.
Transient transfection and luciferase assays
Cells were transiently transfected using Arrest-In transfection reagents (Open Biosystems, Huntsville, USA). Cells at a density of 1 × 105/well were seeded in 6-well plate. Transfection was performed using 2 μg of plasmid DNA in serum free medium for 6 hours. Cells were then incubated in serum containing medium for another 48 hours. After washing twice with PBS, cells were collected for luciferase assays (Promega Corp.) using a dual luciferase reporter assay system according to the manufacture's instructions. The relative luciferase activity was normalized against renilla activity by co-transfection with 1 μg of pRL-TK (Promega Corp.).
Statistical analysis
The cutoff value of IGFBP-3 methylation as a threshold of hypermethylation was identified using ROC curve and AUC. The expressions of IGFBP-3 and p53, and IGFBP-3 methylation status were compared among patients with different clinical and pathologic features using t test and χ2 test. Associations among IGFBP-3 expression, p53 overexpression and IGFBP-3 promoter methylation with progression-free survival (PFS) and overall survival (OS) rates were assessed by Cox-regression analysis. Kaplan-Meier analysis with logrank test was used to estimate survival probabilities and to compare survival distributions categorized by IGFBP-3 expression, IGFBP-3 promoter methylation and p53 overexpression. Statistical analysis was carried out using Statistical Analysis System (SAS) version 8.0 (SAS Institute, Cary, NC). Probability values less than 0.05 were regarded as significant.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
PLT designed and directed the study, analyzed and interpreted all the data, and drafted the manuscript. PLT also performed IHC and the results were interpreted by PLT and CTL. CWL carried out the MSP, BSP, Western blot analysis, QRT-PCR and site directed mutagenesis assays. MWC and HWY carried out qMSP assays. CTL and SCH participated in the study coordination and helped draft the manuscript. All authors read and approved the manuscript.