Constitutional and somatic methylation status of DMRH19 and KvDMR in Wilms tumor patients

The most frequent epigenetic alterations in Wilms tumor (WT) occur at WT2, assigned to 11p15. WT2 consists of two domains: telomeric domain 1 (DMRH19) that contains the IGF2 gene and an imprinted maternally expressed transcript (H19) and centromeric domain 2 (KvDMR) that contains the genes KCNQ1, KCNQ1OT1 and CDKN1C. In this work, we used pyrosequencing and MS-MLPA to compare the methylation patterns of DMRH19/KvDMR in blood and tumor samples from 40 WT patients. Normal constitutional KvDMR methylation indicated that most of the epigenetic alterations in WT occur at DMRH19. Constitutional DMRH19 hypermethylation (HM DMRH19) was observed in two patients with Beckwith-Wiedemann syndrome. Pyrosequencing and MS-MLPA showed HM DMRH19 in 28/34 tumor samples: 16/34 with isolated HM DMRH19 and 12/34 with concomitant HM DMRH19 and KvDMR hypomethylation, indicating paternal uniparental disomy. With the exception of one blood sample, the MS-MLPA and pyrosequencing findings were concordant. Diffuse or focal anaplasia was present in five tumor samples and was associated with isolated somatic HM DMRH19 in four of them. Constitutional 11p15 methylation abnormalities were present in 5% of the samples and somatic abnormalities in the majority of tumors. Combined analysis of DMRH19/KvDMR by pyrosequencing and MS-MLPA is beneficial for characterizing epigenetic anomalies in WT, and MS-MLPA is useful and reliable for estimation of DNA methylation in a clinical setting.


Introduction
Wilms tumor (WT) is the most frequent extra-cranial solid cancer in childhood, accounting for 6-7% of all childhood neoplasias (Miller et al., 1995). In 80% of cases, WT is diagnosed in children less than five years old. In Brazil, the age-adjusted incidence for WT in children up to 14 years old varies from five to 18 cases per million (De Camargo et al., 2011). Bilateral presentation is observed in center mutations that result in isolated epigenetic defects or paternal uniparental disomies (UPD) (Scott et al., 2008b). WT2 consists of two domains that contain several genes: the telomeric domain 1 contains the insulin-like growth factor 2 (IGF2) gene and an imprinted maternally expressed transcript (H19), while the centromeric domain 2 contains the potassium voltage-gated channel gene (KCNQ1), the KCNQ1 opposite strand/antisense transcript 1 gene (KCNQ1OT1) and the dependent kinase inhibitor 1C gene (CDKN1C) (Smith et al., 2007;Scott et al., 2008a).
In domain 1, IGF2 is normally expressed only by the paternal allele, in contrast to H19 that transcribes an untranslated messenger RNA that is only maternally expressed. The expression of these genes is regulated by the differentially methylated region H19 (DMRH19) located upstream of the H19 promoter; this region is normally methylated only in the paternal allele. Abnormal methylation of DMRH19 leads to loss of IGF2 imprinting control and abnormal IGF2 expression by the maternal allele; expression of the latter gene is normally inhibited by ligation to the zinc-finger CCCTC-binding factor known as CTCF. The centromeric (domain 2) differentially methylated region (KvDMR) is located in exon 10 of KCNQ1OT1, along with the promotor region of this gene that is expressed in a tissue-specific manner normally only from the paternal allele. The KCNQ1OT1 transcript apparently cis-regulates the expression of other genes in domain 2. Loss of methylation (LOM) in KCNQ10T1 results in abnormal expression of the maternal allele, observed in 50-60% of BWS cases (Smith et al., 2007;Nativio et al., 2011;Romanelli et al., 2011).
In this work, we used a novel, rapid method of pyrosequencing, a technique that enables high resolution quantitative analysis of DNA methylation, and methylation sensitive-multiplex ligation-dependent probe amplification (MS-MLPA), a technique that allows the identification of changes in methylation patterns, to compare the constitu-tional and somatic methylation patterns of DMRH19 and KvDMR in 40 WT patients.

Subjects
Forty patients (24 males, 16 females) with documented WT, treated at the Instituto Nacional de Câncer (Rio de Janeiro, Brazil), were included in this study. Peripheral blood DNA samples from 31/40 patients and 34 fresh tumor DNA samples (from 29/40 patients) were used. All tumor samples were derived from primary tumors. For five patients more than one fresh tumor sample per patient (obtained at different times) was included: in two cases the first tumor sample was obtained from a biopsy while in three patients the samples were obtained from partial nephrectomies. All of the patients received with the same neoadjuvant chemotherapy treatment based on the International Society of Paediatric Oncology (SIOP) WT 2001 trial protocol (Bhatnagar, 2009). The patients were examined by a medical geneticist trained in dysmorphology. This study was approved by the Instituto Nacional de Câncer Ethics Committee (protocol number 87/08). All of the patients provided informed consent that was signed by parents or tutors.

Methods
DNA was extracted from peripheral blood and fresh tumor samples, essentially as described by Miller et al. (1988) and Sambrook et al. (1989). Pyrosequencing reactions were done with 2 mg of DNA previously modified with bisulfite (EZ DNA Methylation kit, Zymo Research, CA, USA) according to the manufacturer's instructions. Primers complementary to the bisulfite-modified CpG islands in DMRH19 and KvDMR domains were designed with PSQ Assay Design Software v. 1.0.6 (Pyrosequencing AB, Uppsala, Sweden). Fragments spanning both domains were amplified with the following primers: DMRH19-F: GGGGTTATTTGGGAATAGGATAT; DMRH19-R: TAACTTAAATCCCAAACCATAACA; KvDMR-F: TGTTTAGGTTAGGTTGTATTGTTG; KvDMR-R: CCCATCTCTCTAAAAAAATTT. Reverse primers were biotinylated. The PCR mixtures contained 30 ng of modified DNA, 7.5 pmol of forward and reverse primers (Roche Applied Science ® , Spain), 7.5 mM dNTPs, 3 mL of PCR buffer (Roche Applied Science ® ) and 0.5 U of Taq DNA polymerase (Roche Applied Science ® ) in a final volume of 30 mL. The PCR conditions for the DMRH19 assay consisted of 95 ºC for 5 min, 45 cycles at 95 ºC for 30 s, 60 ºC for 30 s, 72 ºC for 30 s and 72 ºC for 10 min. The corresponding conditions for the KvDMR assay were 95 ºC for 5 min, 45 cycles at 95 ºC for 30 s, 54 ºC for 30 s, 72 ºC for 30 s and 72 ºC for 10 min. Single-stranded DNA templates and pyrosequencing were prepared and run according to the Biotage AB ® protocol (Uppsala, Sweden) in a PSQHS96A system (Pyrosequencing AB ® , Uppsala, Sweden). Pyrosequencing was done with sequencing primers for each PCR product: GAATAGGATATTTATGGGAG for DMRH19-SF and GGGTATATAGTTTATTTTAGTA for KvDMR-SF (Roche Applied Science ® ). The CpG peaks were analyzed and the final methylation index (MI) was calculated from the average of the percentage for each CpG peak in the pyrogram. The data were analyzed with Pyro Q-CpG software v. 1.0.9 (Pyrosequencing AB). Standard deviations (SD) were calculated based on the average MI and the distribution of MI values for the patients and healthy controls.
The MS-MLPA 11p15 region kit SALSA ME030-B2 BWS/SRS (MRC-Holland, Amsterdam, Holland) contains 45 probes, 27 of which are specific for the BWS/SRS (Silver Russel Syndrome) region in 11p15. This kit was used with 300 ng of genomic DNA and the methylationsensitive restriction enzyme HhaI (Promega Corporation, Spain) according to the manufacturer's protocol. The MS-MLPA PCR products were analyzed in an ABI 3130 automated platform (Applied Biosystems ® , Foster City, CA, USA). The raw data were analyzed with Excel-based in-house software (Meth-HULP v. 1.1) developed by INGEMM (Instituto de Genética Médica y Molecular, Hospital Universitario La Paz, Madrid, Spain) (Romanelli et al., 2011). Normalization of the peak areas of undigested and digested samples from patients and controls (blood and kidney DNA samples) allowed visualization of the MI for each locus and sample. Methylation analysis was done by comparing restriction digested aliquots with paired undigested aliquots. Standard deviations were calculated based on the average MI and the distribution of MI values for the patients and healthy controls for each region (DMRH19 and KvDMR) analyzed.

Results
In this study, we investigated the constitutional and somatic 11p15 methylation patterns in 24 males and 16 females with WT. There were four cases of bilateral tumor. The age at diagnosis ranged from five months to 137 months, with an average age at diagnosis of 43 months for all 40 patients and 44 months for patients with unilateral tumor. Diffuse or focal anaplasia was observed in five tumors ( Table 1) and four of them also presented DMRH19 hypermethylation (Table 2). Perilobar nephrogenic rests (PLNR) 716 Methylation of DMRH19/KvDMR in Wilms tumor   were observed in two patients and intralobar rests (ILNR) in one patient. One patient with PLNR (P38) had macrosomia since birth and mild indentations on the ear lobes, but no other findings compatible with the clinical features of BWS. Eight patients had phenotypic abnormalities: one had WAGR syndrome (P1), two had BWS (P2 and P35), two had macrosomia defined by stature and weight above the 97 th centile for age and sex (P22 and P38), two had hemihypertrophy (P7 and P26), one of whom (P26) also presented with Moebius sequence (bilateral VII cranial nerve paralysis), and the eighth patient (P6) had an undiagnosed dysmorphic syndrome (Table 1).
Pyrosequencing of DMRH19 in 31 blood DNA samples detected hypermethylation in the two BWS patients (Table 2), with MI of 71% and 57%, respectively; these values were higher than in healthy blood controls (MI = 45.4 ± 6.2%) (Figure 1). Pyrosequencing of KvDMR in 31 blood samples yielded normal MI estimates (Table 2, Figure 2), with the corresponding MI for healthy blood (control) being 38.9 ± 3.1%.
Pyrosequencing of DMRH19 in 34 fresh tumor samples detected hypermethylation in 28 samples (Table 2). In five patients with more than one tumor sample, analysis of the two samples obtained at different times yielded similar   Figure 3). The corresponding MI for healthy renal tissue (control) was 42.6 ± 15.3%. KvDMR pyrosequencing in 34 fresh tumor DNA samples detected LOM in 12 samples ( Table 2). The MI in healthy renal tissue (control) was 39.2 ± 10.8%. Pyrosequencing showed that these 12 tumor samples with KvDMR LOM also had DMRH19 hypermethylation that was characteristic of paternal uniparental disomy (UPD) ( Table 2).
MS-MLPA was done in 31 peripheral blood samples and in all 34 tumor samples (Table 2). DMRH19 hypermethylation was observed in only one peripheral blood sample (P2); the remaining blood samples were normal ( Figure 4A,B). MS-MLPA analysis of tumor samples showed isolated DMRH19 hypermethylation in 16/34 cases and in 12/34 samples a characteristic pattern of paternal UPD was observed (Table 2, Figure 4C). Isolated DMRH19 LOM was detected in one tumor sample (P15) while a normal methylation pattern was observed in the remaining 5/34 samples (Table 2).

Discussion
In this study, we used pyrosequencing and MS-MLPA to analyze the constitutional and somatic methylation patterns in the DMRH19 and KvDMR regions of 40 WT patients. Data from tumor histopathology and physical examination were compared with the methylation status in 11p15. The comparison of peripheral blood and fresh tumor samples confirmed that the epigenetic changes associated with WT were mostly somatic. Phenotypic abnormalities characteristic of a syndromic form of WT were observed in 20% of patients (8/40). The average age at diagnosis in our cohort was 43 months, which was similar to the 38 months reported by Scott et al. (2012). Cardoso et al. 719 for DMRH19 in patients P2 and P35 was 71% and 57%, respectively. Panel (C) shows DMRH19 methylation in blood from healthy controls (average MI = 41.7%). The average MI of all blood samples from healthy controls used in DMRH19 pyrosequencing was 45.4 ± 6.2%. The MI is indicated above peaks (gray columns) corresponding to the CpG islands in this region.
Four patients had bilateral WT and tumor samples were available from two of them (P17 and P27). Two tumor samples from each of these patients were studied, one from each kidney (left and right). Sample A from patient P17 had a predominantly epithelial histology while sample B, which was collected from a subsequent nephrectomy on the contralateral kidney, showed focal anaplasia that probably reflected evolution of the WT. Both samples had DMRH19 hypermethylation and the DNA methylation indices were similar in the two samples from each of these patients.
Constitutional DMRH19 hypermethylation was observed in only two patients (P2 and P35), both with the BWS phenotype. These patients fit into a small group (5-7%) of BWS patients with isolated DMRH19 hypermethylation reported by Bliek et al. (2001). Interestingly, these authors found that two of the four BWS patients who showed isolated DMRH19 hypermethylation also developed WT (Bliek et al., 2001). The normal constitutional methylation profiles shown by KvDMR pyrosequencing in the present study confirmed that most epigenetic alterations associated with WT involve DMRH19 but not KvDMR, as also reported by Priolo et al. (2008).
Eight of our patients also had major phenotypic abnormalities, including two patients with BWS (P2 and P35), two patients with hemihypertrophy (P7 and P26), two with isolated macrosomia (P22 and P38), one patient with WAGR complex (P1) and one patient with an undiagnosed dysmorphic syndrome (P6). Except for Moebius syndrome, the occurrence of which in patient P26 may have been fortuitous, all of the other pathogenic conditions have been re-720 Methylation of DMRH19/KvDMR in Wilms tumor and (C) healthy control blood sample (average MI = 37.6%). The average MI of all blood samples from healthy controls used in KvDMR pyrosequencing was 38.9 ± 3.1%. The MI is indicated above peaks (gray columns) corresponding to the CpG islands in this region.
currently described in association with WT, indicating the heterogeneous etiology of this neoplasia (Dome and Huff, 2011). As already described in a proportion of these patients, constitutive DMRH19 hypermethylation was observed in both BWS patients and somatic UPD was present in patient P35. Perilobar nephrogenic rests were observed in one of the patients with macrosomia (P38), as described above, and her tumor sample showed DMRH19 hypermethylation. DMRH19 pyrosequencing showed isolated hypermethylation in 47% (16/34) of tumor samples, which was lower than the 71% previously reported with the COBRA assay in 28 WT samples (Brown et al., 2008). However, another study based on COBRA analysis  detected isolated somatic DMRH19 hypermethylation in 10/40 WT patients (25%). Contrary to our study, in which 12 patients showed somatic LOM of KvDMR, Brown et al. (2008) observed no differences in the extent of KvDMR methylation between normal and tumor samples. Scott et al. (2012) reported that the proportion of isolated DMRH19 hypermethylation observed with MS-MLPA was 31%, a value similar to our results.
One of our patients (P21) showed a difference in the extent of methylation between tumor samples A (79%) and B (42%). This discrepancy may reflect the fact that sample B was collected after successful chemotherapy with 90% tumor regression. DNA methylation levels are known to be influenced by various chemotherapeutic regimens. Among our patients, with exception of two biopsy samples, the remaining tumor samples were obtained from patients who had been submitted to the same chemotherapeutic SIOP protocol (Bhatnagar, 2009). Variations in the rate of tumor regression in response to chemotherapy may have an impact on DNA methylation levels, as observed in patient P21.
Isolated DMRH19 hypermethylation and paternal UPD result in IGF2 biallelic expression and, subsequently, tumor development (Beckwith, 1998   cating paternal UPD in 11 out of 29 patients (38%). This proportion was similar to a previous UPD estimate (45%) in 40 WT patients . We used MS-MLPA as a complementary procedure to analyze DMRH19 and KvDMR methylation. This assay yielded findings consistent with those obtained by pyrosequencing, except for one tumor sample (P15) that showed DMRH19 LOM, probably because of a technical artefact. In addition, MS-MLPA confirmed the somatic, paternal UPD in 12 tumor samples.
Few studies have compared constitutional and somatic 11p15 epigenetic alterations in WT patients. One of these used MS-MLPA to analyze blood DNA samples from 437 children with non-syndromic WT and identified constitutional defects in 13 patients (3%), including six patients with DMRH19 hypermethylation and six with paternal UPD (Scott et al., 2008b). Constitutional 11p15 defects were observed in 5% (2/40) of our patients. The concomitant finding of macrosomia and somatic isolated DMRH19 hypermethylation in two patients (P22 and P38; Tables 1 and 2) indicated that these patients may represent somatic mosaics for this epigenetic disturbance. Additionally, patient P38 also had PLNR, a condition previously reported to be associated with this phenotype (Dome and Huff, 2011).
To the best of our knowledge, this is the first study to compare constitutional and somatic DMRH19 and KvDMR epigenetic alterations in WT patients by pyrosequencing and MS-MLPA. Our findings highlight the benefits of the combined analysis of the DMRH19 and KvDMR regions. The MS-MLPA and pyrosequencing findings were highly consistent. MS-MLPA is a rapid, reliable technique that is less expensive than pyrosequencing and is adequate for accurate quantitative estimation of DNA methylation in a clinical setting.