Cytogenomic Changes in Sporadic Colorectal Cancer and Surrounding Nonneoplastic Tissues: The Relevance of Subtelomeric Copy Number Variations

The purpose of this study was to investigate the relevance of subtelomeric cytogenomic changes in patients with sporadic colorectal cancer (CRC) using multiplex ligation-dependent probe amplication (MLPA) and single nucleotide polymorphism arrays. The results revealed pathogenic genomic alterations in the TNFRS18 (1p), CHL1 (3p), TRIML2 (4q), FBXO25 (8p), NKX3-1 (8p), RECQL4 (8q), DOCK8 (9p), ZMYND11 (10p), KDM5A (12p), PSPC1 (13q), ADPRTL2 (14q), MTA1 (14q), DECR2 (16p), GAS8 (16q), THOC1 (18p), CTDP1 (18q), SOX12 (20p), ADRM1 (20q), UCKL1 (20q), OPRL1 (20q), IL17RA (22q), and SYBL1 (Xq) genes. We detected copy number variations (CNVs) with frequencies greater than 40% in the probes located in 20q, which contains very important genes in the study of tumors. These ndings showed instability in the tumor genome and altered regions associated with cell migration, transcription activation, apoptosis, and immune system deregulation. Unexpectedly, we detected concomitant pathogenic CNVs in tumors and surrounding tissues. Our data suggest that characterizing the genomic CRC prole is an important contribution to better understanding instability as a mechanism of carcinogenesis in CRC patients.


Introduction
Investigations of cytogenomic changes are focused on genomic variations, such as copy number variations (CNVs) and architecture at the microscopic and submicroscopic levels. These encompass several pathogenic conditions, including many of the most common cancers [1,2].
Genomic arrays are a useful tool for studying slight variations between whole genomes and play a role in research into both constitutional disorders and cancer [1,3]. Different cytogenomic techniques have been used to detect these changes, including multiplex ligation-dependent probe ampli cation (MLPA) and single nucleotide polymorphism (SNP) arrays [3,4].
MLPA is a useful method based on polymerase chain reaction (PCR) principles for the detection of different genomic abnormalities (aneuploidies, gene deletions or duplications, subtelomeric rearrangements and methylation status), allowing the detection of speci c deletions, and duplications in regions of interest, whereas arrays can assess the CNVs and SNPs present in the whole genome of a patient in a single reaction [3,5]. MLPA probes are evenly distributed over the investigated region, and the speci c or overlapping genomic locations depended on the design of each set assay for a MLPA probe [1,2,4].
The study of genomic variations related to hereditary colorectal cancer (CRC) can provide strong clues to individual's sporadic CRC development predisposition [5,6]. The presence of pathogenic CNVs may also disturb gene dosage and be a relevant source of cytogenomic instability in human cancers [1].
Despite extensive knowledge of chromosomal instability (CIN), microsatellite instability (MSI) and CpG island methylator phenotype (CIMP) pathways in CRC, less is known about the role of genomic changes [2].
In sporadic cancer, CIN is characterized by gains and losses of small genomic segments or entire chromosomal arms, which are mainly caused by chromosome breaks [5,6]. CIN can result from defects in chromosomal segregation, telomere instability, or the response to DNA damage [2]. Approximately 85% of CRC cases have accumulated chromosomal imbalances and runs of homozygosity (ROH), which lead to altered expression of tumor suppressor genes and oncogenes [2][3][4][5][6][7]. MSI is a hypermutable event caused by the loss of DNA mismatch repair activity, whereas a CIMP is produced by hypermethylation of promoter CpG island sites [7,8]. Both mechanisms result in the inactivation of several tumor-related genes in CRC tissues [8][9][10].
The cytogenomic instability of cancer cells is also characterized by a high rate of telomere loss and double-strand breaks in subtelomeric regions; therefore, cytogenomic instability could contribute to chromosomal instability and tumor cell progression in CRC tissues [11]. Authors agree that ROH are an important genetic clue for understanding tumorigenesis and that the SNP array is currently the best resolution method for analyzing these changes [12].
The aim of this study was to investigate the role of subtelomeric genomic changes in neoplastic and nonneoplastic surrounding colorectal tissues in patients with sporadic CRC.

Study Participants
All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Research Ethics Committee of Faculdade de Medicina do ABC (808.569, approval date 26/09/2014).
We collected 33 CRC samples from patients with a mean age of 63 years old (49 to 84 years), including 16 males and 17 females, from the Division of Gastrointestinal Surgery of ABC Medical School (Santo André, São Paulo, Brazil). All patients underwent curative or palliative surgical resection of CRC by conventional or laparoscopic access. The information obtained from each patient included gender, age, family history of cancer, previous treatments with neoadjuvant therapy, primary tumor site, surgery performed, tumor histologic grade, metastasis (presence vs. absence), and staging [13].
Adult patients of both genders with the con rmed histological diagnosis of CRC con rmed by histopathological exam according WHO criteria [13,14], age greater than 49 years old, and high quality of extracted DNA were included regardless of ethnicity.
Patients who had undergone neoadjuvant therapy (chemotherapy or radiotherapy); patients with familial adenomatous polyposis, hereditary CRC syndromes, colorectal neoplasia other than carcinoma, or in ammatory bowel disease; patients with synchronous or metachronous tumors elsewhere; and those who had incomplete anatomopathological data were excluded. The criterion for exclusion of suspected family cancer for patients was the existence of at least 3 rst-degree family members and age of less than 50 years old with CRC or another related cancer, such as cancer of endometrium, stomach, kidney, ureter, biliary, or small intestine [14].
After applying the inclusion and exclusion criteria, we selected 10 patients (7 women and 3 men) and excluded 23. The mean age was 62.1 years (49 to 84 years). Primary tumors were located in the left colon in 5 cases, the right colon in 4, and the rectum in 1. The operations performed were rectosigmoidectomy in 4 cases, right colectomy in 4, and left colectomy in 2. Patients' characteristics are shown in Table 1.

Collection and Classi cation of Samples
To obtain patients' DNA, 5 ml of peripheral blood was collected immediately before the beginning of the surgical procedure, and tissue samples were collected at the time of surgery for tumor resection. Two tissue samples (approximately 1 cm 3 in size) were obtained: a sample of the macroscopically neoplastic region of the tumor, and another sample of the distant surrounding colorectal tissue located within 10 cm cranially of the tumor. Tissues were preserved in 400 µL of RNAlater® Stabilization Solution (Ambion® Life Technologies, Austin, Texas, Unites States).
Other samples of the areas considered to be tumoral and nontumoral were obtained, embedded in para n, stained with hematoxilin and eosin, and subjected to microscopic examination to con rm the presence or absence of neoplasia.

Cytogenomic Techniques
Following the manufacturer's instructions, we obtained genomic DNA from blood and tissues using the QIAamp® Midi DNA Blood and Tissue Kit (QIAGEN, Hilden, Mettmann, Germany) and stored it at -20 °C. DNA concentration and purity were evaluated using a spectrophotometer (NanoDrop 2000, Thermo Scienti c, Wilmington, Delaware, United States), Qubit 2.0 uorimeter (Invitrogen, Carlsbad, California, United States) and agarose gel electrophoresis (2%).
All DNA samples were analyzed using MLPA according to the manufacturer's recommendations (MRC-Holland, Amsterdam, Netherlands). Blood samples without cytogenomic changes were used to standardize the MLPA reaction and as a control in all assays with tumor and nonneoplastic surrounding tissue samples.
The P036 and P070 kits were used to investigate CNVs in subtelomeric regions. The P413-B1 (CCR-LOSS) and P146-B1 (CCR-GAIN) kits were used for the detection of speci c cytogenomic changes. Probes locations for each kit used in MLPA are shown in Supplemental Material 1. Details of the identi ed regions and probes are available at www.mlpa.com.
Ampli ed products were analyzed using the ABI 3500 Genetic Analyzer (Life Technologies) and GeneMarker® software (Softgenetics, LLC, State College, Pennsylvania).
The uttermost area of every one piece was related with controls samples. The results were considered altered when the peakto-height ratio was fewer than 0.75 (deletion) or superior than 1.25 (duplication).

SNP Arrays
SNP arrays were used only in selected cases to con rm the MLPA results. We evaluated 8 samples of tumor tissue and surrounding nonneoplastic tissue from patients ID 13, ID 14, ID 23 and ID 26 (

MLPA results
The genomic imbalances found in samples of tumor tissue and tissue surrounding nonneoplastic tissue of 10 patients whose CRC was operated on using the MLPA technique as well as the imbalances correlations with the functions of genes, are shown in Table 2.
The speci c CRC kit results of the tumor samples showed genomic gains in the long arm of chromosome 20, including probes located in the ADRM1, OPRL1, MAPRE1, TPX2, and ZNF217 genes; genomic losses in the short arm of chromosome 8, including markers in the NKX3 and LPL genes; and alterations in the long arm of chromosome 13, including the RB1, DACH, PSPC, and ZMYM2 genes. The percentage graph, which shows the most relevant genomic changes found by the MLPA technique in all patients with CRC is shown in Fig. 1.
The protein networks associated with the relevant genes identi ed in our study by using the STRING database and the genetic ontology classi cation (con dence score > 0.9) is shown in Fig. 2.
The main cellular pathways were cellular apoptosis, natural killer (NK)-cell-mediated cytotoxicity pathway, cell cycle pathway, cell-adhesion-molecules, interleukin-4 and interleukin-13 signaling, regulation of TP53 activity signaling pathway by interleukins, immunoregulatory interactions between lymphoid cells and nonlymphoid cells, and cytokine signaling in immune system.

Snp Array Results
The SNP array results showed several different pathogenic CNVs, mainly associated with tumorigenesis and tumor necrosis factors, and corroborated the genomic instability in these tissues. The array results con rmed the MLPA ndings and unexpectedly revealed genomic alterations in the surrounding nonneoplastic tissue, which should be free from changes (Tables 3 and 4). SNP array results of ROH in surrounding tissue are shown in Supplemental Material 1.

Discussion
Our investigation revealed cytogenomic changes associated with both the pathogenesis and the progression of CRC. Additionally, it was possible to distinguish the complex and heterogeneous molecular characteristics of tumor tissues, clearly revealing the genomic instability pro le characteristic of these samples. We also revealed the presence of pathogenic CNVs in surrounding nonneoplastic tissues, which were not detectable in histopathological investigations. This may indicate an initial tumorigenesis process for these patients. Our results also showed concomitant genomic changes, such as THOC1 gene losses in 18p11.3 and UCKL1 gene duplications in 20q13.3 in both tumoral and nontumoral tissues. Silencing of the THOC1 gene inhibits the proliferation of some cell cancer strains, while the UCKL1 gene is related to maintaining a high cell proliferation rate in tumors [17,[19][20][21]. Furthermore, chromosome 20 is a genomic target of extreme importance in tumors [17]. We identi ed several genomic alterations associated with chromosome 20, including ROH regions and duplication and deletion of whole arms in one patient. Duplications in the 20q subtelomeric region are a marker of unfavorable prognosis in patients with CRC [22].
Lake et al. [23] studied the relationship of tumorigenesis in primary and metastatic conjunctival melanomas in their analysis of metastatic samples, nding that the gene OPRL1 (20q13.33) was frequently excluded. In our study, we recognized a genomic gain in the long arm of chromosome 20, including at a probe located in the OPRL1 by using MLPA. Additionally, they found by use of MLPA and SNP arrays that the ADRM1 gene (20q13.3), located on chromosome 20, was duplicated in colorectal tumor tissue. This gene encodes a plasma membrane protein that participates in cell adhesion and dysregulation of this protein has been implicated in carcinogenesis by inducing interferon gamma in some cancer cells [17,22]. Fejzo et al. [24] have shown that the ubiquitin proteasome ADRM1 receptor is ampli ed in cancers, including gastric, ovarian and colonic cancer. This ADRM1 gene directs the protein levels of speci c oncogenes, resulting in an increase in metastatic potential [24].
Some genomic changes not previously described in CRC were identi ed in our study, such as duplication of ADAP1 (7p22.3), DMRT1 (9p24.3), and DOCK8 (9p24.3), which may be important in the study of sporadic CRC pathophysiology, considering that all these genes are involved in cell cycle control. The ADAP1 gene codes a protein related to Arf6 signaling events and the B cell receptor signaling pathway. The DOCK8 gene has been identi ed as a putative gene associated with the progression of brain tumors, especially gliomas [9].
In addition to the CNVs detected in THOC1 and UCKL1, we found other genomic abnormalities concomitant in neoplastic and nonneoplastic surrounding tissues, including duplications in TNFRSF18 and deletions of the MTA1 and DECR2 genes [25]. The TNFRSF18 gene, a member of the tumor necrosis factor receptor superfamily, plays a key role in the self-regulation of cellular apoptosis and the immune system, coding and regulating T cells [25]. The MTA1 protein plays important roles in cell signaling processes, chromosomal remodeling and transcription that in turn participate in the progression, invasion and growth of metastatic cells [26].
The FBXO25 gene encodes a member of the F-box protein family, a subunit of an ubiquitin protein ligase complex hat functions in phosphorylation-dependent ubiquitination. A connection exists between CNVs and tumor suppressor genes such as FBXO25 with negative regulation of gene expression [27], which may help in further understanding the behavior of tumorigenesis and cancer progression. In the present investigation, we detected a deletion in the FBXO25 gene (8p23.3) in the tumor tissue using both techniques.
In this investigation, other genomic alterations detected in tumor samples included duplications in the PSPC1 gene (13q12.1), which encodes protein tyrosine kinase 6 and is involved in determining oncogenic subcellular translocations. Its positive regulation of PSPC1 is related to a prometastatic activator associated with a poor prognosis, while its negative regulation suppresses activated metastases and is a potential marker of improved cancer therapy outcomes [28].
In particular, duplication in 1p36 located in the TNFRSF18 gene in both tumoral and nontumoral tissues, was a relevant nding of our study. The tumor necrosis factor receptor-associated factor-6 protein (TRAF-6), which is encoded by one of the TNFRSF18 family genes, was abnormally expressed in positive CRC tissues and was closely linked to patient's prognosis [29][30][31]. Thus, TNFRSF18 ampli cation deserves special attention to clarify its clinical signi cance in this patient's pro le [29][30][31].
In the present investigation, SNP array results of the surrounding nonneoplastic tissue samples from patients with metastases revealed different structural variations at 1q42.3, comprising the LYST gene, which has been previously associated with cancer and autoimmune diseases. LYST has been associated with immunode ciency syndromes and with impaired cytotoxic lymphocytic function, especially among NK cells, which are very important in the defense against tumor growth [32].
When the amount of DNA damage is greater than the DNA repair capability, a checkpoint-signaling pathway is activated [29,[33][34][35][36][37]. One of the rst steps in the cellular response to DNA damage induced by exogenous agents is DNA repair protein activation [35][36][37]. The literature suggests that CNVs may be important genetic variants that explain tumor heterogeneity and genetic instability in CRC [38][39][40]. Therefore, we suggest that these CNVs are initialization markers of cellular abnormality.

Accordingly, our results demonstrate tumor cells characterized by different CNVs in subtelomeric regions, clearly indicating
the presence of CIN con rmed by SNP test array results.
In sporadic tumors, CIN is characterized by gains and losses of small genomic segments or whole arms, that are mainly caused by chromosomal breaks [4,6,8,41]. CIN can result from defects in chromosomal segregation, instability or loss of telomeres, or errors during a response to DNA damage. Double-strand breaks are a common mechanism in tumor instability, and usually occur through nonhomologous end joining [5,7,8,38]. Furthermore, subtelomeric regions are more susceptible to rearrangement and have recently been implicated in a genomic rearrangement event known as chromothripsis, which has been reported in some types of cancer [41][42][43][44]. The aberrant chromosomal architecture -i.e., the variation of small insertions or deletions leading to major chromosomal changes such as premitotic defects, stress in replication and telomeric fusionshas an important role in CIN and is usually found in cancer genomes [6,41,[45][46][47][48]. Studies have reported aneuploidies of whole chromosomes in 70% of colorectal tumors [45,46]. Importantly, CIN may disturb the cell environment and immune signaling. The inhibition of the immune vigilance system has revealed altered expression of several genes involved in adaptive immunity and/or associated with cytotoxic cells and NK cells, suggesting a decrease in the level of immune cells and functioning as an immunosuppressant [45][46][47][48]  Approximately 85% of CRC cases demonstrated chromosomal and ROH imbalances, which led to changed expressions of tumors and oncogene suppressor genes [45][46][47][48][49][50]. We observed several insertions and deletions and ROH in the array results in surrounding nonneoplastic tissue. There results indicate there was an unstable microenvironment growing around the tumor.
Our ndings suggest the need for a more detailed molecular investigation of the altered genomic regions, considering an expanded study of the genomic pro le in a larger population, using this model to con rm the discovery of relevant prognostic markers for this disease. In