Association between Acquired Uniparental Disomy and Homozygous Mutations and HER2/ER/PR Status in Breast Cancer

Background Genetic alterations in cellular signaling networks are a hallmark of cancer, however, effective methods to discover them are lacking. A novel form of abnormality called acquired uniparental disomy (aUPD) was recently found to pinpoint the region of mutated genes in various cancers, thereby identifying the region for next-generation sequencing. Methods/Principal Findings We retrieved large genomic data sets from the Gene Expression Omnibus database to perform genome-wide analysis of aUPD in breast tumor samples and cell lines using approaches that can reliably detect aUPD. aUPD was identified in 52.29% of the tumor samples. The most frequent aUPD regions were located at chromosomes 2q, 3p, 5q, 9p, 9q, 10q, 11q, 13q, 14q and 17q. We evaluated the data for any correlation between the most frequent aUPD regions and HER2/neu, ER, and PR status, and found a statistically significant correlation between the recurrent regions of aUPD and triple negative (TN) breast cancers. aUPD at chromosome 17q (VEZF1, WNT3), 3p (SUMF1, GRM7), 9p (MTAP, NFIB) and 11q (CASP1, CASP4, CASP5) are predictors for TN. The frequency of aUPD was found to be significantly higher in TN breast cancer cases compared to HER2/neu-positive and/or ER or PR-positive cases. Furthermore, using previously published mutation data, we found TP53 homozygously mutated in cell lines having aUPD in that locus. Conclusions/Significance We conclude that aUPD is a common and non-random molecular feature of breast cancer that is most prominent in triple negative cases. As aUPD regions are different among the main pathological subtypes, specific aUPD regions may aid the sub-classification of breast cancer. In addition, we provide statistical support using TP53 as an example that identifying aUPD regions can be an effective approach in finding aberrant genes. We thus conclude that a genome-wide scale analysis of aUPD regions for homozygous sequence alterations can provide valuable insights into breast tumorigenesis.

Understanding the molecular pathogenesis of cancer requires detailed cataloguing of all genetic and epigenetic lesions-not just identification of CN changes, but also detection of aUPD, DNA sequence, and methylation changes-in cancer cells. Because of the previous lack of high-throughput technology and analytical tools, to date very few reports have been published in breast cancer about either genome-wide aUPD analysis [5] or aUPD for specific genes, such as RB1 and TP53 [1]. We now know that aUPD can occur either on the entire chromosome or segmentally: a loss of one chromosome followed by duplication of the remaining chromosome leads to aUPD on the entire chromosome, whereas somatic recombination leads to segmental UPD [13][14][15]. Both mechanisms lead to the transmission to the daughter cell of a homozygous mutation from a heterozygous parental cell. The regions having aUPD are evident as large CNN stretches of somatically acquired homozygosity without any change in DNA content.
aUPD may result in two copies of an abnormal allele, which may give a growth advantage to the cell. Some of these abnormalities or mutations may affect mRNA-and proteinexpression levels. Homozygously mutated genes in aUPD regions that function in the initiation and progression of cancer may be associated with tumor type or subtype [8,27], risk of disease transformation [9], patient's survival time [9,28]. Inactivation of genes through different mechanisms may lead to or occur in different subtypes of disease. For example, in uveal melanomas, monosomy at chromosome 3 results in pigmented tumors, whereas aUPD at chromosome 3 results in unpigmented tumors [8]. Thus the dysfunction of cellular processes caused by deletion of a gene may affect a different cellular pathway than that affected by aUPD in the same gene.
As a result of all these findings, we hypothesized that aUPD is also a common feature found in breast cancer. Since genome-wide aUPD analysis by using high-resolution SNP arrays can pinpoint regions that carry homozygously mutated genes for nextgeneration gene sequencing, we hypothesized that identifying UPD regions can identify known and possibly novel mutated genes in breast cancer.
Breast cancers are routinely assessed for the expression of ER, PR and overexpression or amplification of the HER2/neu. Patients with HER2/neu-positive tumors (30%) respond to treatment with the anti-HER2 monoclonal antibody transtuzumab [29]. Patients with ER-or PR-positive tumors are candidates for hormonal therapy, including selective ER modulators such as tamoxifen for premenopausal women or aromatase inhibitors for postmenopausal women [30,31]. Patients with triple negative cancers (those negative for ER, PR and HER2) currently have no available targeted therapy and have relatively poor prognosis [32,33].
As the biology resulting in breast cancer pathological subtypes is different, we further hypothesized that specific aUPD regions might correlate with estrogen receptor (ER), progesterone receptor (PR), and/or HER2/neu status.
Our purpose in conducting this analysis was to identify aUPD regions in breast cancer samples. Such regions might be candidate regions for second-generation sequencing to identify novel mutated genes in breast cancer. This study is the first, to our knowledge, to describe high-resolution genome-wide UPD analysis of a large dataset and its integration with sequence alterations of TP53 in breast tumor samples.
The findings presented here provide strong evidence that mitotic recombination is a common molecular mechanism that results in an aUPD feature that occurs non-randomly in specific chromosomal locations, and that correlates with ER, PR and HER2/neu status of breast cancer and with homozygous mutation of specific genes.
We used Fisher's exact test calculated by STATAv10 (Stata-Corp, CollegeStation, TX, USA) and Spearman correlation analyses by SAS v9.2 (SAS, NC, USA) to evaluate correlations between aUPD regions and ER, PR and HER2/neu status, grade, lobular or ductal, invasive or infiltrating pathology. The Wilcoxon Mann-Whitney test was used for testing the association of aUPDscores with ER, PR and HER2/neu status. Stepwise logistic regression analyses was performed using SAS v9.2 (SAS, NC, USA) for prediction of pathology outcomes such as TN, ER, PR, and HER2/neu status, and Spearman correlation analyses were used for correlation between aUPD at chromosome 17p and TP53 mutation status.

Correlation between the regions of aUPD and ER, PR, HER2/neu and Status and pathological features of Breast Tumors
We next examined the data to identify any correlation between the recurrent aUPD regions of diverse clinical parameters (including ER, PR and HER2/neu status, grade, lobular or ductal subtype and invasive or infiltrating cancer), and characterized by a distinct aUPD profile and whether specific aUPD regions in each group of tumors harbor narrow regions of aUPD that indicate regions harboring homozygous mutation that may be used as a marker or a therapeutic target in each group of breast cancer. We found correlation with ER, PR and Her2/neu status ( Table 1), grade and invasive type of cancer, but we could not find any correlation between aUPD regions and lobular or ductal type and infiltrating breast cancer (Table S2).
Data on ER, PR, and HER2/neu status were available for 468 cases. We tested those data for any correlation with the presence of the most recurrent aUPD regions at chromosome 2q, 3p, 5q, 9p, 9q, 10q, 11q, 13q, 14q, 17q and total aUPD scores. We also assessed whether there was any correlation between the presence of recurrent aUPD regions and triple-negative tumors (n = 111) compared to tumors expressing at least one of the three receptors (ER, PR or HER2/neu) (n = 356), aUPD at chromosome 17q and 13q revealed highly statistically significant association with ER-negative, PR-negative, HER2/ neu-negative and TN cases (P,0.001) ( Table 1), while all recurrent aUPD regions and total aUPD-score were highly statistically significant correlation with TN cases (P,0.001) ( Table 1). Other recurrent aUPD regions had less significant association with ER-negative, PR-negative, and HER2/neunegative cases (Table 1, Figure 1-3).
In addition, at a significance level of p,0.001 we observed differences in aUPD scores between ER-negative, PR-negative and HER2/neu-negative samples compared to their respective positive counterparts ( Table 1, Table S3). Higher aUPD-scores were also found in TN-negatives compared to receptor positive counterparts (p,0.001) ( Table 1). Similar results to those found in the clinical breast cancer specimens were seen in the breast cancer cell lines ( Figure S2, Figure S3).

Association between aUPD Regions and Homozygously Mutated Genes in Breast Tumors
To date, mutations have been found in a number of genes in breast cancer. However, the most important problems interpreting mutations is the presence of numerous mutations that have no direct role in cancer; these may be called 'passenger mutations'. The other group of mutated genes, which affect protein function and involve tumor initiation and/or progression, may be called 'drivers.' Distinguishing the driver genes from the passengers is challenging, but the integration of aUPD analysis with mutation and functional data can overcome this problem.
If indeed aUDP pinpoints these aberrant genes, the latter findings with these integrated data indicate that more than one cell-signaling pathway is being interrupted in breast cancer.

Discussion
In this study, using a large representative cohort of patients with breast cancer (n = 656) and cell lines (n = 44), we have shown that aUPD is a common and non-random event in breast tumorigenesis. The frequency of aUPD is statistically significantly higher in TN breast cancer and in estrogen receptor negative than receptor positive tumors. We have characterized aUPD regions associated with the most reproducible breast cancer subtypes, defined by tumor ER, PR, and HER2/neu status. For clinical practice ER and PR status is generally established by immunohistochemistry (IHC), and HER2/neu status by IHC or fluorescence in situ hybridization. In addition, in the current study aUPD at 17q, 3p, 9p and 11q were found as predictors for TN cases, while aUPD at 17q, 13q and 3p were predictors of ER-negative disease. aUPD at 11q was predictive of PR-negative breast cancer, and aUPD at 17q and 13q marked HER2/neu-negative cases. Overall our findings indicate that each group has a different pattern and that specific aUPD regions clearly associated with ER, PR or HER2/neu status.
Until now sporadic breast tumors have shown mutations in different genes, with TP53 being the most frequently mutated (44% in tumor and 73-76% in cell lines) [39,55] particularly in BRCA1 and sporadic basal-like carcinoma [58,59]. It is also known that breast cancers in patients with BRCA1 germ-line mutations are more often triple negative than positive for HER2/neu, PR or ER [60], and the majority of basal-like carcinomas lack ER, PR, and HER2/neu expression. In concordance with this finding, our result ( Figure 2) provides strong evidence that in addition to mutation in TP53 and BRCA1, other genes in aUPD region at chromosome 17q, 3p, 9p and 11q may be mutated or otherwise suppressed in triple-negative tumors. Thus it is possible that other than TP53 and BRCA1 mutated genes in these regions contribute functionally to the development of triple-negative breast cancers; future studies, however, are needed to support this finding.
One of the candidate genes for mutation is VEZF1 at chromosome 17q, which is transcriptional regulatory zing finger protein 161. This gene regulates DNA methylation [61,62] and is involved in both normal and abnormal cellular proliferation and differentiation. WNT3 is in another aUPD region of chromosome 17q and is a member of WNT gene family. Gene expression studies suggest that this gene may play a key role in variety of human cancer including breast cancer through activation of the WNT-beta-catenin-TCF pathway, and the WNT pathway may be active in basal-like tumors relapsing to brain based on pathway analysis [63]. Another candidate gene is miR-31 which is affected by the focal homozygous deleted region at chromosome 9p. Overexpression of miR-31 inhibits breast cancer metastasis [64], suggesting that homozygous deletion of miR-31 may play role in metastasis of breast cancer. A final candidate for mutation is FGFR2 at chromosome 10q. FGFR2 is a member of the fibroblast growth factor receptor family, showed heterozygous mutation in FGFR2 in breast cancer [52], and recently showed that SNPs (rs2981582) in this gene associated with increased risk of breast cancer [65]. Allele-specific up-regulation of FGFR2 was associated with increasing susceptibility to breast cancer [66]. We found aUPD at FGFR2 region in chromosome 10q. Taken together, data indicates that FGFR2 may be a good candidate for homozygous mutation or imprinting. From all these data, we conclude that aUPD is a common and non-random molecular event in breast cancer. Identifying aUPD regions could be a very effective approach for discovering novel candidate genes for mutation screening. Our data also suggest that aUPD may be used for sub-classification of breast tumors. Finally, the integration of mutation data with aUPD data provides strong evidence that many more genes than previously thought to be aberrant in breast cancer and which await discovery and could include useful new therapeutic targets. aUPD may pinpoint regions with homozygous alterations and identifying those mutated genes will provide valuable insights.