Array CGH Analysis of Paired Blood and Tumor Samples from Patients with Sporadic Wilms Tumor

Wilms tumor (WT), the most common cancer of the kidney in infants and children, has a complex etiology that is still poorly understood. Identification of genomic copy number variants (CNV) in tumor genomes provides a better understanding of cancer development which may be useful for diagnosis and therapeutic targets. In paired blood and tumor DNA samples from 14 patients with sporadic WT, analyzed by aCGH, 22% of chromosome abnormalities were novel. All constitutional alterations identified in blood were segmental (in 28.6% of patients) and were also present in the paired tumor samples. Two segmental gains (2p21 and 20q13.3) and one loss (19q13.31) present in blood had not been previously described in WT. We also describe, for the first time, a small, constitutive partial gain of 3p22.1 comprising 2 exons of CTNNB1, a gene associated to WT. Among somatic alterations, novel structural chromosomal abnormalities were found, like gain of 19p13.3 and 20p12.3, and losses of 2p16.1-p15, 4q32.5-q35.1, 4q35.2-q28.1 and 19p13.3. Candidate genes included in these regions might be constitutively (SIX3, SALL4) or somatically (NEK1, PIAS4, BMP2) operational in the development and progression of WT. To our knowledge this is the first report of CNV in paired blood and tumor samples in sporadic WT.


Introduction
Wilms tumor (WT) is the most common cancer of the kidney in infants and children, accounting for 6 to 7% of all children neoplasms, with a prevalence of 1:10.000 affected individuals [1,2]. This tumor originates from nephrogenic rests of pluripotent, potentially oncogenic, kidney precursors comprising non-differentiated blastemal cells, primitive epithelium structures and stromal components which present classical triphasic histology [3]. The average age of presentation is 42-47 months for children with unilateral WT and 30-33 months for bilateral WT [2]. Treatment, though differing between countries, is determined by clinical-pathologic stage and histological classification [4]. WT is frequently sporadic, occurring in only one member of a family or rarely affecting more than one (in familial WT).
WT has a complex etiology with genetic, constitutional and somatic factors playing dissimilar roles in its genesis and progression. This results in extensive genomic alterations including copy number variants (CNV) that may include oncogenes or tumor suppressor genes leading to dose-dependent, functional disruptions of gene expression. Identification of CNV regions may thus provide a better understanding of cancer development as well as insights for clinical management, diagnosis and therapeutic targets [5]. Although several tumors apparently show normal karyotypes regardless of stage and histopathology [6], the frequent loss of 11p13 in WAGR syndrome (Wilms tumor, aniridia, genital anomalies, mental retardation) led to the discovery of WT1, the first gene found in association with WT, with a relevant involvement in the development of kidney progenitor cells [7]. Germline missense mutations in WT1 exon 8 or 9 are associated to Denys-Drash syndrome (pseudo-or true hermaphroditism, nephropathy, and genital anomalies with an estimated risk for WT higher than 90%) [8] while germline, point mutations in the WT1 intron 9 donor splice site account for several cases of WT associated to Frasier syndrome (pseudohermaphroditism and progressive glomerulopathy) [9,10]. Moreover, WT has been found to be associated to 5% of cases of Beckwith-Wiedemann syndrome (macroglossia, overgrowth, omphalocele, neonatal hypoglycemia, ear creases/pits, adrenocortical cytomegaly and renal abnormalities) but this condition, however, is unrelated to germline WT1 mutations, but to 11p15.5 epigenetic alterations [4,11]. In contrast, in sporadic, non-syndromic WT, tumor predisposition is usually not the result of WT1 constitutive mutations [12].
Microarray Comparative Genomic Hybridization (aCGH) allows the identification, with high resolution, of CNV in tumor genomes which may identify biomarkers and genes involved in tumor progression [13]. aCGH studies in WT have identified whole, arm-and segmentalchromosome gains or losses and carried out comparative analyses of primary and relapsed tumors throughout their molecular evolution. They have also contributed to the understanding of genomic alterations of different histological profiles, their influence on prognosis, and identification of potential therapeutic gene targets [6,[14][15][16][17].
To date, aCGH studies in WT have only been performed in tumor tissue samples. In this work, we report molecular studies in paired, blood and tumor samples from 14 patients with sporadic WT, with the aim of identifying germline and somatic anomalies associated with this tumor. We herein describe a novel set of constitutional abnormalities and point to candidate genes that might be associated with WT development.
features, Patient 6, showed 5th finger clinodactyly, 2-3 syndactyly and alopecia areata, while Patient 13 showed a single cafe au lait spot, a frequent clinical traits not indicative of a syndromic condition. Patients 4 and 11 showed clearly dysmorphic features unrelated to any specific syndromic condition (Table 1).
All tumor samples were collected from primary tumors from patients treated with the same neoadjuvant chemotherapy following the International Society of Paediatric Oncology (SIOP) WT 2001 trial protocol [18]. Age at diagnosis ranged from 7 to 136 months, with an average age at diagnosis of 53 months. Bilateral tumors were present in 3/14 ( This study was approved by the Instituto Nacional de Câncer (Brazil) Ethics Committee and the Clinical Research Ethics Committee of Hospital La Paz (Madrid, Spain), and was performed in accordance to the ethical standards prescribed in the 1964 Helsinki Declaration. All parents or guardians on behalf of the children enrolled in this study were given the option to accept or reject participation. Following acceptance, a signed, informed consent was provided. Informed consents are kept in patients' files with copies provided to parents/guardians. Details that might disclose the identity of the subjects under study have been omitted in the text of this article.
Genomic DNA isolation DNA was isolated from peripheral blood and tumor samples following standard saline [19] and phenol/chloroform [20] procedures.

Mutational screening
All patients were previously screened for mutation in WT1, WTX and exon 3 of CTNNB1 [21].

CNV screening by WT-Customized aCGH
We selected 60-mer oligonucleotide features from Agilent's eArray probe library (Agilent, https:// earray.chem.agilent.com/earray) in a customized, high resolution format of 4x180K. This array comprised specific probes covering WT1, WTX, CTNNB1, 16 regions involved in WT and the most relevant microdeletion and duplication syndromes as well as probes for the library backbone. Average probe spacing between regions of interest was 5 kb. Regions reported as non-pathogenic in the Genomic Variant Database (http://projects.tcag.ca/variation/) were excluded from the design to exclude CNV lacking apparent clinical relevance.
Array experiments were performed as recommended by the manufacturer (Agilent Technologies, Santa Clara, CA, USA). Briefly, DNA (500 ng) from the study specimen and a samesex reference (Yoruba Male NA18507 and Yoruba Female 18517, Coriell Institute, USA) were double-digested with RsaI and AluI for 2h at 37°C. Following heat inactivation of enzymes at 65°C for 20 min, each digested sample was labeled by random priming (Genomic DNA Enzymatic Labeling Kit Agilent) for 2 h with Cy5-dUTP for patient DNA and Cy3-dUTP for reference DNA. Labeled products were purified in columns (Microcon Ym-30 filters, Millipore Corporation). Following probe denaturation and pre-annealing with Cot-1 DNA, hybridization was performed at 65°C under rotation for 24 h. After two washing steps, microarrays were A detailed description of the statistical algorithms is available in the manual provided by Agilent Technologies. The Aberration Detection Method 2 (ADM-2) quality weighted interval score algorithm identified aberrant intervals in samples showing consistently high or low log ratios based on their statistical score. The score represented the deviation of the weighted average of the normalized log ratios from its expected value of zero calculated with Derivative Log2 Ratio Standard Deviation algorithm. A Fuzzy Zero algorithm was applied to incorporate quality information on each probe measurement. Our threshold settings for the CGH analytics software to make a positive call were 6.0 for sensitivity, 0.35 for minimum absolute average log ratio per region, and 5 consecutive probes with the same polarity were required for the minimum number of probes per region.

Results
In this study, we analyzed CNV with a WT-Customized aCGH platform, using paired, blood/ tumor DNA samples from 14 patients with sporadic WT. Our findings, together with clinical and histological data, are shown in Table 1. Raw data were deposited in the GEO repository with the reference GSE69971. Gains or losses involving whole chromosomes or chromosome arms were not observed in blood DNA but segmental alterations were observed in 4/14 patients. Gains in 4/14 (28.6%) patients included the following regions within which some genes were identified: 2p21 with SIX3 (1/14), 3p22.1 with CTNNB1 (1/14), 4p16.3 with FGFR3 (1/14), and 20q13.3 with SALL4 (1/14). A single segmental loss was identified involving 19q13.31 with ZNF227, ZNF233 and ZNF235 in 1/14 patient. All these alterations were also found in matched tumor samples (Fig 1).
Among the three patients with bilateral WT, a 1q gain in tumor samples was found in Patients 8 and 10. Patients 4 and 8 also presented PLNR, and both of them showed gains of chromosomes 6 and 12. Patient 4 presented several chromosome gains but no losses, while tumor sample from Patient 8 presented one arm loss and fewer gains. Two focal, anaplastic WT (in Patients 4 and 5) showed different extensive gains or losses but both patients showed, in their paired samples, alterations of well-known WT-related genes: CTNNB1 within a 3p22.1 gain in Patient 4, and a p.R560W hemizygous WTX mutation in Patient 5. Chromosome-, arm-or segmental alterations were unrelated to histological profiles and mutations affecting WT1, WTX and CTNNB1.

Discussion
In this study, blood and tumor DNA samples from 14 patients with sporadic WT were analyzed for CNV, with the aim of identifying germline and somatic anomalies associated to this tumor. To our knowledge, this is the first aCGH analysis of paired samples of this tumor. CNV were apparently unrelated with the heterogeneous histological profiles of the samples herein studied. Among chromosome abnormalities, 9 of 41 (22%) were novel (Table 1) pointing to a likely role in WT pathogenesis.
All alterations detected in blood were segmental and were also present in their corresponding tumor samples, comprising 4 gains (3p22.1, 4p16.3, 2p21 and 20q13.3) and one loss (19q13.31) in 4/14 (28.6%) patients (Fig 1). Two of these gains, 3p22.1 and 4p16.3, have been previously described as part of larger segmental gains, and only in tumor samples, respectively [6,22]. Constitutive alterations coexisted with atypical phenotypes unrelated to defined clinical syndromes suggesting their possible association to congenital abnormalities and predisposition to cancer.
Gain of 3p22.1 in Patient 4 included CTNNB1, a gene that encodes β-catenin and an essential participant of the Wnt signaling pathway whose deregulation is critical for tumor development. Somatic mutations affecting β-catenin are present in >10% of WT and are frequently associated to the presence of WT1 mutations [6,23,24]. This patient showed dysmorphic features, with triangular face and syndactyly of fingers 2 and 3. Large constitutive gains and losses containing this chromosome region are registered in DECIPHER database v8.6 (Database of Chromosomal Imbalance and Phenotype in Humans using Ensembl Resources) [25] and Database of Genomic Variants (DGV) (http://dgv.tcag.ca/), albeit unrelated to WT. In DECIPHER, complete gains and losses of CTNNB1 were registered while, in DGV, a duplication including CTNNB1 was part of a complex rearrangement. These alterations, however, differed from the partial duplication of CTNNB1 exon 8 and 9 found by us, and thus may have different functional effects. The 3p22.1 gain thus represents a novel alteration associated to WT.
Gain of 4p16.3 in Patient 13 included FGFR3 exon 11 and part of exon 12. This gene is a member of a receptor tyrosine kinase family regulating cell proliferation, differentiation and migration, and also involved in selective apoptosis during embryogenesis. FGFR3 mutations had been associated with congenital skeletal disorders and several cancers but not with WT [26]. Large CNVs of this region were registered in ClinVar (http://www.ncbi.nlm.nih.gov/ clinvar/)-a database that aggregates information about genomic variation and its relationship to human health-and DECIPHER, which included the complete FGFR3 gene while, in DGV, deletions accounted only for internal alterations of this gene. The 4p16.3 gain including FGFR3 exon 11 and part of exon 12 is thus a novel alteration associated to WT.
A constitutive 2p21 gain was also found in Patient 13. Constitutional gains or losses of 2p21 were described in association with a wide range of dysmorphic manifestations, skeletal abnormalities, holoprosencephaly and intellectual disability [27,28]. One patient, with a larger genomic gain (27.21 Mb) of this region, abnormality of the kidney, intellectual disability and prenatal short stature was reported, albeit without WT [25]. The 2p21 gain reported herein included the terminal region of SIX3 exon 2. This gene is a regulator of the Wnt signaling pathway [29] whose alterations are associated with holoprosencephaly [30]. SIX3, together with SIX1 and SIX2, are members of the transcription factor family SIX, with important roles in regulation of cell proliferation, survival, migration and invasion [31]. SIX1 promotes mesenchymal-to-epithelial transition in different tumor types [32]. On the other hand, embryonic, renal mesenchyme contains pluripotent progenitor cells characterized by expression of SIX2 that suppresses cellular differentiation [33]. Recently, somatic mutations have been described in a hotspot of the SIX1 and SIX2 homeo-domain responsible for DNA binding and protein interaction in WT. Both mutant and wild type SIX1/2 alleles have been found to be highly expressed, a reason why these mutations were considered to be activating, resulting in failure of mesenchymal-to-epithelial transition and continued proliferation of the metanephric mesenchyme [34,35]. DECIPHER registered manifold 2p21 CNVs including several genes, among which the complete SIX3, unrelated to WT. In ClinVar, two gains of the terminal region of SIX3 exon 2 were associated to clinically benign conditions. The 2p21 gain reported herein is thus a novel alteration associated to WT. A 20q13.3 gain, found in a patient with a high number of dysmorphic features (Patient 11), included part of the zinc finger transcription factor SALL4, a gene involved in the maintenance and the pluripotency of embryonic stem cells. This gene has been found to be expressed in WT and developing fetal kidney [36] while SALL4 deletions are a common cause of Okihiro, acrorenal-ocular, and SALL4-related Holt-Oram syndromes [37], all of which associated with renal morphological malformations rather than WT. In DECIPHER and ClinVar, 20q13.3 CNVs were not associated to WT but rather to developmental delay and cardiac abnormalities while, in DGV, a complete SALL4 loss has been registered. The 20q13.3 gain herein reported is thus a novel anomaly associated to WT, suggesting that SALL4 might be relevant for the genesis of this tumor.
Loss of 19q13.31 in Patient 6 included three zinc finger protein genes (ZNF227, ZNF233 and ZNF235), which could not be associated with clinical manifestations. Although this loss has not been previously associated with WT, functional studies of these zinc finger genes in WT might provide future insights on their role in tumor development. In Clinvar, one 19q13.31 duplication associated to cardiac alterations has been reported while, in DECIPHER, deletions of this region have not been registered. In DGV, CNVs did not include the three genes involved in the 19q13.31 loss reported herein. This makes this loss a novel alteration associated to WT.
A strong association has been described between gain of 1q and loss of 1p (P<0.001), and between gain of 1q and loss of 16q (P<0.001), this latter reflecting the occurrence of a typical t (1;16) [14]. Somatic gain of 1q21-q25 was associated with poor prognosis in WT patients with favorable histology [17]. In this study, all segmental gains including 1q21-q25 occurred in tumor samples. Loss of 1p, one of the most frequent chromosome alterations in WT samples and crucial for tumor development [13,[40][41][42], was not found in our patients. However, a small segmental loss (1p22.2-p36.3) was observed in one case of stage V, bilateral, triphasic WT, suggesting that this loss might be associated with tumor progression rather than carcinogenesis.
The two deleted regions including 4q (4q32.5-q35.1 in tumor sample from Patient 14 and 4q35.2-q28.1 in tumor sample from Patient 8) contained several genes, including NEK1. This gene encodes a protein involved in cilium control and is essential for response to DNA damage, cell cycle progression, and correct transmission of chromosomes to daughter cells. NEK1 is developmentally regulated and strongly expressed in kidney, mainly in podocytes and tubular epithelial cells. NEK1 deficiency causes early abnormalities in kidney development, including excessive apoptosis and diminished proliferation, and has been associated with polycystic kidney disease [43]. On the other hand, increased NEK1 expression is associated to a decreased sensitivity to treatment of damaged DNA in renal cell carcinoma [44]. Patient 8 also carried a somatic CTNNB1 mutation, while Patient 14 showed a somatic duplication of chromosome 3 where this gene is located. Patient 4 carried a constitutive and a somatic 3p22.1 duplication, in agreement with previous studies showing gain-of-function, CTNNB1 mutations in WT [45]. NEK1 and CTNNB1, together with AXIN1, GSK3B and TSC2, participate in the Wnt signaling pathway in urological cancer cells and polycystic kidney disease (PKD) [46,47]. Whether loss of NEK1 might coexist with β-catenin stabilization, similarly to WT1 alterations, for the survival and proliferation of tumor cells [45] is an open question. This might be clarified by analyzing the expression profiles of CTNNB1-mutants versus CTNNB1 wild-type in WT with NEK1 alterations.
Chromosome region 11p13 was the first locus implicated in the development of WT whose alterations have been detected both by karyotyping and DNA arrays [48]. This region contains WT1, the first reported gene associated to the etiology of WT [49,50], which is affected by deletions and point mutations in 10-20% of WT [51,52]. WT1 encodes a transcription factor that is essential for survival and differentiation of renal cells and subsequent kidney development. Additionally, it has both tumor suppressor and oncogenic activities [53] while its epigenetic silencing can also contribute to sporadic WT [54]. In our study, one 11p13 gain and one loss were found in tumor samples from Patients 8 and 1, respectively.
Rearrangements encompassing 17p may result in loss of TP53 (in 17p13.1), resulting in alterations of cell cycle checkpoint controls, apoptosis, cell migration and invasion. Secondary to TP53 loss, structural rearrangements in 17q are strongly associated with an unfavorable anaplastic histology [39,40,42]. In our series, however, a WT with a favorable histology (in Patient 3) showed a novel segmental loss in 17p13.2-p11.2. Also, a novel structural somatic gain of 17q21.31-q25.3 was observed in Patient 8 presenting bilateral WT, PLNR, and histology with epithelium predominance. This histological profile was coincident with the previous finding that tumors originated from PLNR usually present a predominance of epithelium or blastemal components [55,56].
Several genes associated with carcinogenesis, but not with WT, are present in 2p16.1-p15 and 19p13.3. Patient 1 showed a somatic 19p13.3 segmental loss including PIAS4, a regulator of cellular senescence and apoptosis through activation of p53-mediated cell death, a pathway also positively regulated by WTX [57]. Patients 4 and 5 showed a 20p12.3 somatic duplication. This region contains BMP2, a gene that belongs to the transforming growth factor-beta (TGFB) superfamily. It is involved in mesenchymal progenitor cell chemotaxis through PI3K signaling induction [58] and has been identified as an epigenetically silenced growth inhibitor in rhabdomyosarcoma [59]. Furthermore, it also inhibits hepatocellular carcinoma growth and migration through down-regulation of the PI3K/AKT pathway [60]. As the RING domain of PIAS4 is also involved in the suppression of the BMP-signaling pathway [61], the BMP2 duplication carried out by Patients 4 and 5 might produce similar effects to the loss of PIAS4 in Patient 1.
To discard the new changes described in this study are small, previously undocumented CNVs that may also be present in the constitutional DNA of control individuals, in addition to public databases we have compared them with 3500 pseudo-controls studied in our Centre with a related customized aCGH [62] covering the regions included in 5 of the 9 new described CNVs. In none of the cases these alterations were detected. These suggests that these CNVs are associated with WT in our sample, and further studies should clarify whether they are recurrent and directly associated with carcinogenesis or if instead they are passenger alterations neutral to tumor cell selection.
To our knowledge, this is the first study using aCGH for detecting chromosomal alterations in paired blood and WT tumor samples. Although several of these alterations may not be recurrent but rather passenger alterations restricted to our sample, the novel chromosome abnormalities might be contributing to carcinogenesis, pointing to several candidate genes that might be constitutively (i.e. SIX3, SALL4) or somatically involved (i.e. NEK1, PIAS4, BMP2) in WT development and progression.