Loss of Heterozygosity at the Calcium Regulation Gene Locus on Chromosome 10q in Human Pancreatic Cancer

Loss of heterozygosity (LOH) is a common genetic alteration in cancer genomes, which derives from heterozygous deletion of one of the two alleles, or duplication of a maternal or paternal chromosome or chromosomal region and concurrent loss of the other allele (Frampton and King, 2013). LOH on chromosomal regions containing key tumor suppressor genes has been considered as a significant contributor to drive cancer progression by inactivating the suppressor genes of tumor (Baker et al., 2009), including pancreatic cancer. For example, p53 tumor suppressor gene is identified to experience LOH accompanied by mutation in pancreatic cancer cell lines (Butz et al., 2003). Cooperating with mutation and/or loss of p53, the LOH of BRCA1 and BRCA2 may contribute to the progression of pancreatic cancer (Lucas et al., 2013). Besides, LOH in THBS2 (6q), p16 (9p) and APC (5q) are also demonstrated to carry the worst prognosis for resected pancreatic ductal and ampullary adenocarcinomas (Franko et al., 2008). By using genome-wide LOH maps in pancreatic cancer, LOH is found with high frequency at various chromosomes


Introduction
Loss of heterozygosity (LOH) is a common genetic alteration in cancer genomes, which derives from heterozygous deletion of one of the two alleles, or duplication of a maternal or paternal chromosome or chromosomal region and concurrent loss of the other allele (Frampton and King, 2013). LOH on chromosomal regions containing key tumor suppressor genes has been considered as a significant contributor to drive cancer progression by inactivating the suppressor genes of tumor (Baker et al., 2009), including pancreatic cancer. For example, p53 tumor suppressor gene is identified to experience LOH accompanied by mutation in pancreatic cancer cell lines (Butz et al., 2003). Cooperating with mutation and/or loss of p53, the LOH of BRCA1 and BRCA2 may contribute to the progression of pancreatic cancer (Lucas et al., 2013). Besides, LOH in THBS2 (6q), p16 (9p) and APC (5q) are also demonstrated to carry the worst prognosis for resected pancreatic ductal and ampullary adenocarcinomas (Franko et al., 2008). By using genome-wide LOH maps in pancreatic cancer, LOH is found with high frequency at various chromosomes
Hybridization to single-nucleotide polymorphism (SNP) arrays is an efficient method to detect genome-wide cancer LOH by identifying the absence of heterozygous loci (Beroukhim et al., 2006;Staaf et al., 2008). Thus, the goal of this study was to further identify candidate genes with LOH in pancreatic cancer by performing a genomewide analysis of LOH using the SNP arrays deposited in public database by Donahue et al. (2012), followed by annotation analysis and transcription factors screening in an attempt to explain the involvement of these genes in pancreatic cancer.

SNP expression profiling data
The SNP expression profiling data GSE32682 (Donahue et al., 2012) of human pancreatic samples including 25 human pancreatic cancer and 7 nonmalignant pancreas samples snap-frozen during surgery were downloaded from the GEO (Gene Expression Omnibus) database that is developed as a repository of microarrays, chips, hybridization arrays and high throughput gene expression data. The gene expressions of these samples were investigated by Affymetrix Genome-Wide Human SNP 6.0 Array (GPL6801).

Data processing and screening of candidate genes with LOH
Affymetrix CEL files were analyzed using Genotype Console software (version 4.0; Affymetrix) for initial intensity quality control (QC) with the criterion of recommended contrast QC>0.4, followed by generating SNP genotype calls using the Affymetrix Birdseed algorithm and filtrating SNPs loci with the excluded thresholds of no-call rate ≥ 10%, minor allele frequency <0.05 and Hardy Weinberg Equilibrium (HWE) P-value 0.001. Then, copy-neutral LOH (CN-LOH) analysis was performed with the threshold of MAPD<0.04 to identify the SNP loci of LOH that existed in over 50% of the pancreatic cancer samples while not in control samples. The candidate genes with LOH were screened based on the genotype calls, SNP loci of LOH and dbSNP database (Day, 2010).

Gene annotation of Candidate genes
The candidate genes with LOH were annotated by using Gene database in NCBI (the National Center for Biotechnology Information) and the unannotated genes were inputted into DAVID (the Database for Annotation, Visualization and Integration Discovery) for Gene Ontology (GO), INTERPRO, PFAM and SMART annotation (Sherman et al., 2007).

Transcription factors screening of unannotated genes
UCSC Genome Browser track that displays all analyzed transcription factor binding sites (TFBS) was performed to identify the transcription factors of the unannotated genes using DAVID database (Fujita et al., 2010), which was then visualized by constructing regulatory network using Cytoscape software (Kohl et al., 2011).

Intensity QC filtration
The intensity QC of each sample was analyzed using Genotype Console software ( Figure 1). The samples with contrast QC<0.4 (GSM811149_17T.CEL and GSM811154_14T.CEL) were excluded in the following research.

Candidate genes with LOH
The candidate genes with LOH identified in this study were MCU (mitochondrial Ca 2+ uniporter), MICU1 (mitochondrial calcium uniporter regulator 1) and OIT3 (oncoprotein induced transcript 3) on chromosome 10 (Partial results were shown in Table 1). The copy number (CN) in the samples were two or four in this study, probably suggesting LOH resulted from heterozygous deletion of one of the two alleles, or duplication of a maternal or paternal chromosome or chromosomal region and concurrent loss of the other allele.

Gene annotation
Gene annotation was performed to identify the functions of candidate genes with LOH using Gene database in NCBI. Accordingly, MCU (Gene ID: 90550) was found to encode a calcium transporter that localizes to the mitochondrial inner membrane and interacts with mitochondrial calcium uptake (Bick et al., 2012;Csordas et al., 2012;Raffaello et al., 2012;Curry et al., 2013;Patron et al., 2013). MICU1 (Gene ID: 10367) could encode an essential regulator of mitochondrial Ca 2+ uptake under basal conditions. The encoded protein interacts with the mitochondrial calcium uniporter and is essential in preventing mitochondrial Ca 2+ overload that could cause excessive production of reactive oxygen species and cell stress (Mallilankaraman et al., 2012;Arvizo et al., 2013;Chen et al., 2013;Hoffman et al., 2013;Logan et al., 2014). OIT3 didn't obtain any annotated results from Gene database, but get some other annotated messages from GO, INTERPRO, PFAM, SMART annotation analysis provided by DAVID (Table 2). OIT3, oncoprotein induced transcript 3, was correlated with calcium ion binding, ion binding and cation binding.

Transcription factors of OIT3
By performing UCSC TFBS analysis, this study identified a large number of transcription factors including STAT (signal transducer and activator of transcription), SOX9 (sex determining region Y-box 9), CREB (cAMP responsive element binding protein), NF-kB (nuclear factor kappa B), PPARG (peroxisome proliferatoractivated receptor gamma) and p53, which had regulatory effects on OIT3 (Figure 2).

Discussion
By mapping the SNPs showing LOH in the tumor versus matched normal samples, this study identified three candidate genes with LOH (MICU1, MCU and OIT3) on chromosome 10 in pancreatic cancer, implying an important role for these genes in pancreatic cancer.
MCU and MICU1 are two genes that have been demonstrated to play important roles in mitochondrial Ca 2+ uptake (Perocchi et al., 2010;Baughman et al., 2011;Csordas et al., 2013). The Ca 2+ handling by mitochondria is involved in cell life by triggering or preventing apoptosis probably functioning through the released pro-apoptotic factors such as Bax and Bak from the intermembrane space (Scorrano et al., 2003;Kroemer et al., 2007;Contreras et al., 2010). Thus, it could be speculated that MCU and MICU1 may be related to the cancer progression via affecting cellular apoptosis. As expected, previous study has reported the down-regulation of MCU targeted by cancer-related microRNA may increase cancer cell survival and contribute to tumorigenesis in various cancer cells (Marchi et al., 2013). Silencing MICU1 is also revealed to initiate the mitochondrial pathway for apoptosis by decreasing Bcl-2 expression together with increasing caspase-3 activity and cytosolic cytochrome c contents (Arvizo et al., 2013). Therefore, MCU and MICU1 may be oncogene or tumor suppressor gene and the LOH in them may lead to their lower expression, thus preventing or promoting pancreatic cancer cell apoptosis.
OIT3 located at 10q22.1 was another gene with LOH found in pancreatic cancer. This gene, also termed as liver-specific zona pellucida domain-containing protein (LZP), is related to hepatocellular function and could be used as a potential diagnostic biomarker for hepatocellular carcinoma (Xu et al., 2003). Also, OIT3 is reported to experience copy number losses and down-regulation in colorectal cancer (Yoshida et al., 2010). However, the details of the relationships between OIT3 and cancers are not clear. Herein, OIT3 was found to be associated with calcium ion binding, probably implying an involvement of this gene in cellular Ca 2+ homeostasis. Moreover, based on the regulatory network in this study, OIT3 was regulated by a large number of transcription factors including STAT, SOX9, CREB, NF-kB, PPARG and p53 which were previously reported to be associated with pancreatic cancer. STAT-3, PPARG and NF-kB signaling pathways are involved in apoptosis and cellular differentiation in pancreatic cancer cells (Elnemr et al., 2000;Sahu and Srivastava, 2009). CREB is related to activate STAT-3 and cyclin D1 expression in pancreatic cancer cells which lead to cell proliferation and tumor progression (Zhang et al., 2010). SOX9 could initiate and accelerate the formation of premalignant lesions of pancreatic cancer (Kopp et al., 2012). In addition, mutations in p53 tumor suppressor gene are considered to drive metastasis and contribute to the carcinogenesis of pancreatic cancer (Sato et al., 1996;Amaya et al., 2004;Morton et al., 2010). The LOH of OIT3 may suggest dysregulated functions of these transcription factors, which may contribute to the progression of pancreatic cancer.
In conclusion, our global genomic analysis of SNPs provides evidence of LOH in MICU1, MCU and OIT3 on chromosome 10 in pancreatic cancer. They may be involved in pancreatic cancer progression by regulating the calcium ion homogenizes and cell apoptosis. However, future mechanistic researches will be required to determine the molecular mechanisms.