Identication of Molecular Biomarkers in Taiwanese Patients with Wilms Tumor Using a Methylation-Specic Multiplex Ligation-Dependent Probe Amplication (MS-MLPA)-Based Approach

Background: Wilms tumor is a solid tumor that frequently occurs in children. Genetic or epigenetic aberrations in WT1 and WT2 loci are implicated in its etiology. Moreover, tumor suppressor genes are frequently silenced by methylation in this tumor. Methods: In the present study, we analyzed the methylation statuses of promoter regions of 24 different tumor suppressor genes using a methylation-specic multiplex ligation-dependent probe amplication (MS-MLPA)-based approach in six Wilms tumors. Results: All six Wilms tumors showed methylation of RASSF1 specic to tumors, not in normal tissues. Moreover, methylated HIC1 was identied in stromal type Wilms tumors and methylated BRCA1 was identied in epithelial type Wilms tumors. Unmethylated CASP8, RARB, MLH1_167, APC, and CDKN2A were found only in blastemal predominant type Wilms tumors. Conclusions: Our results indicated that methylation of RASSF1 is the essential event in the tumorigenesis of Wilms tumor, which may inform its clinical and therapeutic management. In addition, mixed type Wilms tumors may be classied according to epithelial, stromal, and blastemal components via MS-MLPA-based approach.

Hypermethylation of CpG islands upstream of tumor suppressor genes has been reported for which methylation status is altered in one or more types of tumors [8,9]. It is interesting to note that Wilms tumor develops mainly through alterations in epigenetic regulation triggered by dedifferentiation [10]. However, the risk of Wilms tumor conferred by epigenetic changes associated with tumor suppressor genes is poorly characterized. In the present study, we investigated the methylation statuses of promoter regions of 24 different tumor suppressor genes in six Wilms tumors using a methylationspeci c multiplex ligation-dependent probe ampli cation (MS-MLPA)-based approach.

Study subjects
The six subjects have been described elsewhere [11][12][13]. Their para n-embedded Wilms tumor tissue samples were provided by the Department of Pediatrics of National Taiwan University Hospital. These patients had no history of Denys-Drash syndrome, Frasier syndrome, or Beckwith-Wiedemann syndrome. The study procedures were approved by the Institutional Review Board of Chung Shan Medical University Hospital (reference number CS2-16003). All procedures that involved human participants were conducted in accordance with the ethical standards of the institutional and/or national research committee and the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Histological examination
The tumor tissue samples (W7 to W12) were embedded in para n wax and cut into 5 um-thick slices. Then, sections were stained with hematoxylin and eosin and reviewed by two of the authors (T-C Hou and C-Y Kuo) under a microscope to con rm their diagnostic classi cations. Typing of Wilms tumor has been described in detail elsewhere [11,13]. W7 is of blastemal type and W8 is of epithelial type. In addition, W10 and W11 are of stromal type. W9 is a triphasic Wilms tumor comprised of three components: blastema, stroma, and epithelium, while W12 is of mixed blastemal and epithelial type.

DNA extraction
Genomic DNA was puri ed from para n-embedded tissues with the DNA FFPE Tissue Kit (Qiagen), according to the manufacturer's instructions and dissolved in 100 µl of TE buffer (10 mM Tris-HCl, pH 8.0, and 1 mM EDTA) as previously described [11,13]. UV-Vis measurements of DNA concentration of each sample were obtained using NanoDrop UV-VIS Spectrophotometer (Thermo Scienti c Nano-Drop 2000c).

Methylation analysis
MS-MLPA analysis was performed using Salsa MS-MLPA kit ME001-C2 Tumor suppressor-1 (MRC-Holland) according to the manufacturer's instructions. Samples were then subjected to capillary electrophoresis on an ABI PRISM 3130XL (Applied Biosystems). Twenty-six MS-MLPA probes were used to detect the methylation statuses of promoter regions of 24 different tumor suppressor genes by HhaI digestion (Table 1). MLPA results were analyzed using GeneMarker version 3.2.1 (SoftGenetics, LLC) to determine copy numbers and methylation statuses of the HhaI sites. The internal methylation ratio was obtained by comparison of the HhaI digested aliquot ( Fig. 1B) with the paired undigested aliquot (Fig. 1A) from each sample with intra-sample data normalization according to the manufacturer's instructions [14]. Methylation was assessed by comparing the probe methylation percentages of the test sample with the percentages of the 5 normal reference samples. Copy number ratio of 1.0 and methylation ratio of 0 were expected in most genes on normal reference, meaning that the methylation compared to normal reference was unlimited (∞). If methylation ratios of test sample and normal reference samples were appropriate, methylation compared to normal reference was around 1.0.

Results
The methylation statuses of the HhaI sites in 24 different tumor suppressor genes were determined by 26 MS-MLPA probes on MS-MLPA analysis (Fig. 1). If the target site was unmethylated, the DNA-probe complex was digested to prevent exponential ampli cation. No signal was detected after fragment analysis (Fig. 1B).
Methylation of RASSF1 and CDKN2B was found in all six Wilms tumors (Fig. 1A, Table 2). Methylation was observed in both probes in RASSF1 gene. Internal methylation ratio was not 0 and methylation compared to normal reference was unlimited (∞). CDKN2B gene was methylated in normal reference. In comparison with normal reference, methylation was not unlimited (∞). W7 to W11 methylation was 3.72, 8.85, 2.58, 3.15, and 5.40-fold that of normal reference, respectively. However, W12 methylation was only 1.05-fold that of normal reference.    VHL, MLH1_463, FHIT1, CHFR, and BRCA2 were unmethylated in all six Wilms tumors (Fig. 1B). Moreover, there was methylation of HIC1 in the Wilms tumors of stromal type (W9, W10, and W11) and methylation of BRCA1 in the Wilms tumors of epithelial type (W8, W9, and W12). CASP8, RARB, MLH1_167, APC, and CDKN2A were unmethylated only in W7, which was of blastemal predominant type ( Table 2). In other words, there was methylation of CASP8, RARB, MLH1_167, APC, and CDKN2A in stromal and epithelial type Wilms tumors, but not in blastemal type Wilms tumors.
In addition to these 13 tumor suppressor genes (15 MS-MLPA probes), there were differential methylation patterns for the other 11 tumor suppressor genes among the six Wilms tumors. The six Wilms tumors are listed in order from minimum number to maximum number of these methylated genes: W7 and W11 (2), W12 (4), W9 and W10 (6), and W8 (10). Only two tumor suppressor genes, KLLN and CD44, were methylated in W7. Only one tumor suppressor gene, CD44, was unmethylated in W8.

Discussion
Traditional methods of methylation detection involve modifying DNA by methylation-speci c restriction enzymes or sodium sul te, combined with Sanger sequencing, PCR, or hybridization analysis. This process is cumbersome and di cult, with low reproducibility and insu cient sensitivity. MLPA is a PCR ampli cation reaction. ME001 can simultaneously detect changes in the degree of methylation of up to 24 different genes in a single reaction tube. Due to MLPA's easy operation, low cost, and wide range of applications, our team has published several studies based on this method [13,15,16]. However, MLPA analysis of tumor samples only provides information regarding the "average" situation in the cells from which the DNA samples were puri ed [14].
The results of this study regarding methylation of RASSF1 gene are consistent with the ndings of previous studies [17,18]. RASSF1 gene on 3p21.31 encodes a cytosolic RASSF1A protein similar to Ras effector proteins [19]. De novo methylation of the RASSF1 promoter is one of the most frequent epigenetic inactivation events in human cancer and leads to silencing of RASSF1 expression [20,21]. Association of RASSF1 promoter methylation with Wilms tumor has been reported [22]. The methylation status of RASSF1 might be a novel biomarker for predicting outcome of Wilms tumor patients [23]. It appears that methylation of RASSF1 is the essential event in the tumorigenesis of Wilms tumor, which may inform its diagnostic, clinical, and therapeutic management. CDKN2B gene hypermethylation was observed in W7 to W11, but not in W12. CDKN2B gene on 9p21.3 encodes the p15 INK4B protein that binds to CDK4 or CDK6 and inhibits its activation [24]. Hypermethylation of CDKN2B CpG islands occurs in the majority of leukemia patients [25].
Hypermethylation at CpG islands in the 5′ ends of tumor suppressor genes is controversial and di cult to interpret. HIC1 gene on 17p13.3 encodes a transcriptional repressor for p21 [26]. Hypermethylation of HIC1 gene is found in 3% of Wilms tumors [27]. CASP8 gene on 2q33.1 encodes Caspase-8 that is an apoptosis-related cysteine peptidase [28]. In 43% of Wilms tumors there is methylation at CASP8 [22]. MLH1 gene on 3p22.2 encodes proteins that detect and repair DNA mismatches [29]. A small proportion of Wilms tumors might be associated with the presence of microsatellite instability [30]. CDKN2A gene on 9p21.3 encodes two proteins that regulate two critical cell cycle regulatory pathways, the p53 pathway and the RB1 pathway [31]. Arcellana-Panlilio et al. demonstrated methylation of the CpG island in the 5′ region of CDKN2A (p16) in seven out of seven Wilms tumors exhibiting decreased CDKN2A expression [32]. This is inconsistent with the nding of negligible methylation of the 5′ CpG island of CDKN2A by Erlich et al. [21]. BRCA1 gene on 17q21.31 encodes a nuclear phosphoprotein that plays a role in maintaining genomic stability [33]. However, BRCA1, which demonstrates promoter hypomethylation, is over-expressed in Wilms tumor [34]. RARB2 gene on 3p24.2 encodes retinoic acid receptor beta that is a type of nuclear receptor activated by all-trans retinoic acid and 9-cis retinoic acid [35]. Our results were inconsistent with those of Morris et al. in which promoter methylation was absent at RARB2 [22]. APC gene on 5q22.2 encodes a 312-kDa protein that acts as an antagonist of the Wnt signaling pathway [36]. Activation of the Wnt/β-catenin pathway is common in Wilms tumor, but rarely through β-catenin mutation and APC promoter methylation [37]. An important nding of this study is the possibility to further classify mixed type Wilms tumors using genetic results of epithelial, stromal, and blastemal components based on MS-MLPA-based approach. In particular, the methylation statuses of these 9 genes make them candidate molecular markers of metastasis in Wilms tumors.
In Wilms tumor, the differentiation arrest of renal progenitor cells is not complete, allowing for maturing lineages of varying proportions [7]. The outcome for stromal and epithelial predominant Wilms tumors is generally excellent [38]. Histological classi cation of Wilms tumor is not always possible based on morphology alone [39]. From our results, methylation of HIC1 in stroma, BRCA1 in epithelium, and CASP8, RARB, MLH1_167, APC, and CDKN2A in either stroma or epithelium can be used to identify stromal predominant and epithelial predominant Wilms tumors which are associated with a good outcome [38]. In the situation that the subtype of Wilms tumor is di cult to diagnose, unmethylated CASP8, RARB, MLH1_167, APC, and CDKN2A are potential diagnostic markers to diagnose blastemal type Wilms tumors which are associated with a poor outcome [40]. Strati cation of Wilms tumor by epigenetic analysis of these genes highlights the bene ts of methylation status analysis of important tumor suppressor genes. Is it possible that epigenetic modi cations in Wilms tumor provide potential therapeutic options? To answer this question, additional studies based on a larger number of cases and tissue microdissection are necessary.

Conclusions
In summary, all six Wilms tumors showed methylation of RASSF1 speci c to tumors, not in normal tissues. Moreover, methylation of HIC1 was identi ed in the Wilms tumors of stromal type and methylation of BRCA1 was identi ed in the Wilms tumors of epithelial type. Unmethylated CASP8, RARB, MLH1_167, APC, and CDKN2A was found only in the Wilms tumor that was of blastemal predominant type. Our results indicated that methylation of RASSF1 is the essential event in the tumorigenesis of Wilms tumor, which may inform its clinical and therapeutic management. In addition, mixed type Wilms tumors may be classi ed using genetic results of epithelial, stromal, and blastemal components based on MS-MLPA-based approach. However, a larger number of cases are necessary to further re ne the molecular classi cation and pathogenesis of Wilms tumors.