Genomic aberrations are common in cancers and the long arm of chromosome 1 is known for its frequent amplifications in breast cancer. However, the key candidate genes of 1q, and their contribution in breast cancer pathogenesis remain unexplored. We have analyzed the gene expression profiles of 1635 breast tumor samples using meta-analysis based approach and identified clinically significant candidates from chromosome 1q. Seven candidate genes including exonuclease 1 (EXO1) are consistently over expressed in breast tumors, specifically in high grade and aggressive breast tumors with poor clinical outcome. We derived a EXO1 co-expression module from the mRNA profiles of breast tumors which comprises 1q candidate genes and their co-expressed genes. By integrative functional genomics investigation, we identified the involvement of EGFR, RAS, PI3K / AKT, MYC, E2F signaling in the regulation of these selected 1q genes in breast tumors and breast cancer cell lines. Expression of EXO1 module was found as indicative of elevated cell proliferation, genomic instability, activated RAS/AKT/MYC/E2F1 signaling pathways and loss of p53 activity in breast tumors. mRNA–drug connectivity analysis indicates inhibition of RAS/PI3K as a possible targeted therapeutic approach for the patients with activated EXO1 module in breast tumors. Thus, we identified seven 1q candidate genes strongly associated with the poor survival of breast cancer patients and identified the possibility of targeting them with EGFR/RAS/PI3K inhibitors.
Citation: Muthuswami M, Ramesh V, Banerjee S, Viveka Thangaraj S, Periasamy J, Bhaskar Rao D, et al. (2013) Breast Tumors with Elevated Expression of 1q Candidate Genes Confer Poor Clinical Outcome and Sensitivity to Ras/PI3K Inhibition. PLoS ONE 8(10): e77553. https://doi.org/10.1371/journal.pone.0077553
Editor: Paul J. Galardy, Mayo Clinic, United States of America
Received: March 15, 2013; Accepted: September 3, 2013; Published: October 17, 2013
Copyright: © 2013 Muthuswami et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This project was funded by Council of Scientific and Industrial Research (CSIR), Govt. of India through the research grant 27(0223)/10/EMR-II to Kumaresan Ganesan, Madurai Kamaraj University. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Breast cancer is one of the most common malignancies in women worldwide. It is also one of the well explored human cancers with genome-wide technologies. In the past two decades, a number of breast cancer genomics investigations contributed to the understanding of the molecular portfolio of breast cancers [1,2]. Several cancer genes and gene signatures indicative of breast cancer sub-type, progression, prognosis, and disease aggressiveness have been derived from mRNA profiles of breast tumors [3,4]. Accumulating genome-wide profiles of various tumors in microarray repositories have revolutionized the field of cancer biology owing to their continuous contribution in addressing various questions in basic and translational research through meta-analysis based genomics approaches [5,6]. This possibility of dissecting and integrating cancer genomics and transcriptomics data in several possible contexts paved ways for identification of novel cancer biomarkers and to uncover various mechanisms involved in the process of carcinogenesis.
Genomic aberrations are the hallmarks of cancer genomes and breast cancer genomes have been characterized for copy number variations and associated biological and pathological features [7,8]. Prevalence of several genomic amplifications (1q, 8q, 17q, 20q) and deletions (5q, 16q, 8p) in breast cancers reflect the definite involvement of specific molecular factors of those loci and associated processes that contribute in cancer development . Aberrations in chromosome 1 are more frequent in various cancers . The long arm of chromosome 1 (1q) is known for its frequent copy number gains whereas 1p region often shows copy number loss . The most interesting aspect of 1q gain in breast cancer is its prevalence in almost all types of breast cancer like Estrogen Receptor (ER) positive, ER negative , Luminal A , Ductal carcinoma in situ (DCIS) and Invasive ductal carcinoma (IDC) . Recurrent 1q gain in breast cancers [11,15], and combined investigations of chromosome 1q gain with other amplifications have been reported [16,17]. Since 1q gain comprises several hundreds of genes, the functional consequences of this recurrent gain remains hard to establish . The potential 1q candidate genes and their specific contribution in breast cancer development remain un-identified. Therefore, in this study, we systematically examined the clinical significance of the expression of all 1q genes in breast tumors by meta analysis based integrative genomics approach and identified 7 potential candidate target genes. Motivated by the occurrence of underexplored candidacy of EXO1 from 1q, we investigated the upstream regulatory pathways and expression pattern across breast cancer sub-types. Further, consensus EXO1 co-expressing gene set was derived and is predicative of biological, clinical and pathological features of breast tumors. We also identified a possible therapeutic targeting approach for breast tumors with elevated EXO1 modular expression.
Materials and Methods
Datasets used in the study were collected from original references or microarray repositories Gene Expression Omnibus (GEO), ArrayExpress, etc. The expression profiles taken for the study were normalized while necessary and the expression values were log2 transformed in the case of single channel data and log ratio data from dual channel data was used as such. The probes were mapped to unique gene symbols with appropriate Affymetrix or Agilent annotation files. The expression values of genes with multiple probes were averaged and used for downstream analysis.
We considered relapse free survival and overall survival information of the breast cancer patient cohorts for predicting the clinical outcome. Hazard Ratios (HR) and significant p-values were calculated independently for each dataset using coxph function of Survival R package [19,20]. Genes with p-value < 0.005 from Wald statistic were considered significant. Combined HRs for each gene was estimated using inverse variance-weighted method with random effects model. Univariate and Multivariate Cox proportional hazards model for/with other clinical covariates was performed using Rcmdr package . Based on median value of EXO1 gene expression, the samples were stratified into two EXO1+ve and EXO1-ve groups, and are then used as one of the variables for univariate and multivariate analysis.
Kaplan Meier estimate was used for plotting survival curves and p-values were calculated using log-rank test. In case for the EXO1 module, average gene expression values were used for computing survival curve.
EXO1 gene expression values were extracted from normalized log2 transformed breast tumor profiles. The significant difference in gene expression between any two groups of breast tumor samples were calculated using Student’s t-test (two tailed) and while calculating for more than 2 groups (i.e. for grade), ANOVA was performed.
For defining EXO1 module, the Pearson correlation measure was calculated for each gene – EXO1 pair independently for all the datasets. With an assumption that effect sizes derived from correlation coefficients vary from dataset to dataset, we used random effects model for deriving the weighted average from correlation coefficients of individual datasets. A stringent cut-off of 0.6 and above with p-value<0.001 was fixed in defining the EXO1 module genes. Ontological terms for module genes were given based on DAVID function annotation tool and Cytoscape was used for network visualization .
Principal component analysis (PCA) was applied using Rcmdr package from CRAN. Transcription factor binding site analysis for single gene was performed using MAPPER database and for geneset DIRE tool was used. Significant over representation of EXO1 module genes to breast cancer signatures was estimated using hypergeometric distribution function.
Pathway activation analysis
Gene signatures representative of particular phenotype/condition were collected from MsigDB (http://www.broadinstitute.org/gsea/msigdb/genesets.jsp?collection=CGP) or from the original references. Detailed descriptions of the signatures and their sources were provided with Table S3. Each signature represented by corresponding up and down tags were scanned against the gene expression profiles of breast cancer profiles as mentioned earlier .
MCF7, T47-D, ZR75, MDA-MB-231, MDA-MB-453, MDA-MB-468, HBL-100 and AGS cells were obtained from National Centre for Cell Sciences (NCCS), Pune, India. Hs578T, Kato III and SKBR-3 were from American Type Culture Collection (ATCC), Manassas, USA. YCC16 cell line was from Yonsei Cancer Centre, Korea . The cells were cultured in the specified media (HiMedia); MCF-7 and YCC16: Minimal Essential Media( MEM), T47-D, ZR 75 & Kato III: RPMI 1640, Hs 578T: DMEM, SK-BR-3 & HBL-100: McCoy’s 5a medium, MDA-MB-231, MDA-MB-453 and MDA-MB-468: Leibovitz, AGS: DMEM-F12, with the supplements L-glutamine (2 mM), sodium pyruvate (1 mM), sodium bicarbonate (1.5 g/L), non-essential amino acids (0.1 mM), penicillin (100 μg/ml), streptomycin (100 μg/ml) (HiMedia) and 10% foetal bovine serum (20 % for Kato III) (Sigma).
Drug Treatment, RT-PCR & Western Blotting
For drug exposure experiments 5x105 cells/well were seeded in a 6-well cell culture dishes. All drug treatments were done upon cells reaching 80 % confluency. For silymarin treatment, the cells were grown in serum-free media for 24 hours prior to drug treatment and treated with 50 μM, 100 μM and 200 μM concentrations. Salirasib was treated at three different concentrations 25 μM, 75 μM and 150 μM. Alkylating agents carboplatin (5, 10 and 15 μg/ml), Cyclophosphamide (2.5, 5 and 10 μM) and Ifosfamide (0.01, 0.05 and 0.1 μg/ml) were used for the treatment in MCF-7 cell lines. All drug treatments were performed for 24 hours and the total RNA was isolated by Trizol (Invitrogen) as per the manufacturer’s protocol. 2 μg of RNA was used to synthesize cDNA using reverse transcriptase (Invitrogen) and the cDNA was used for semi-quantitative PCR analysis. Comparative analysis of the relative expression of EXO1 protein across the panel of breast cancer cell lysates was performed by standard Western blotting using Anti-EXO1 Antibody (LSBIO, LS-B3818 and Sigma, WH0009156M1). 30 μg whole cell lysate were resolved in 12 % gel and 1:100 dilution (LSBIO, LS-B3818) or 1:500 dilution (Sigma, WH0009156M1) of EXO1 antibody was used. Vinculin was probed as loading control for the blotting.
Luciferase Reporter Assays
For luciferase reporter experiments, 75,000 cells/well were seeded in 24-well cell culture dishes. After 24 hours, E2F, MYC, FOXO3 firefly reporter plasmids along with renilla reporter plasmid (SA Biosciences) as internal control in a ratio of 40:1 were transfected using Fugene transfection reagent (Promega). The transfection was performed as per the manufacturer’s instructions. In case of promoter activity assay, E2F or Myc ORF plasmid (Addgene) and EXO1 promoter reporter in the ratio of 1:1 along with renilla reporter plasmid (40:1) was used for transfection. After 48 hours of transfection, the cells were harvested and the luciferase activity was measured using dual luciferase assay protocol  in SpectraMax L (Molecular Devices). The normalization was performed by dividing the firefly reporter value by renilla reporter value to obtain relative luciferase activity. Fold change was obtained by dividing the normalized values of the respective pathway reporters with the negative control reporter. Student’s t-test (two tailed) was used for the analysis of significance.
Identification of 1q genes predictive of poor survival in breast tumors
Searching for prevalent genomic amplifications in breast cancers with the genome-wide copy number profiles in progenetix database has shown gain in chromosome 1q as prevalent and occurs in 40 - 50 % of breast cancer patients in multiple cohorts (Figure 1A). Recurrent copy number gain regions in cancer genomes often harbour genes that facilitate tumor development and progression. Despite being the most frequent chromosomal amplification in breast cancer, the candidate cancer genes of 1q amplicon have not been systematically analyzed. By a systematic integrative genomics analysis work-flow (Figure 1B), based on the gene’s association with the survival of breast cancer patients, we short-listed the clinically significant 1q candidate genes. Totally, mRNA expression profiles of 1635 breast tumor samples from 6 independent studies were investigated (Table S1). Cox regression analysis of 498 genes from 1q locus primarily filtered 10 genes that are consistently associated with survival of the patients in at least 3 of the 6 cohorts with the p-value <0.005. The subsequent filtering with the combined p-value <0.005 across 6 datasets, yielded 7 candidate genes ASPM, KIF14, NEK2, DTL, CENPF, CKS1B and EXO1 (Figure 1C) that are significantly associated with poor clinical outcome of the breast cancer patients (Table S2). Since, the physical neighbour genes of a chromosomal locus would sometime have coordinated pattern of expression , we mapped the candidate genes to their corresponding genomic location in chromosome 1q and the mapping identified that these genes are not tightly clustered at a specific locus in 1q (Figure 1D). Investigation of the expression pattern of these candidate genes reveals their elevated expression in breast tumor samples when compared to normal breast tissues (Figure 1E). Thus, the frequent 1q amplification, consistent elevated expression of seven 1q genes and their association with poor survival suggests their involvement in breast cancers.
(A) High frequency amplification of 1q region in multiple cohorts of breast tumors. Graphs I – VI represent inferred copy-number of probes at chromosome 1 from the aCGH studies [61-66]. These graphical representations were the outcome of data visualized through Progenetix [67,68]. Copy number gain and loss are represented with green and red colors respectively. In all these five cohorts, 1q is amplified in 40 - 60 % of tumors. (B) Meta analysis – workflow for the identification 1q genes associated with patient survival. (C) Cox regression coefficient, hazard ratio and p-value of the shortlisted 1q genes, (D) Chromosomal map showing the locations of clinically significant 1q genes. (E) The identified 1q candidate genes showing an elevated expression in breast tumors when compared to the non-cancerous breast tissues in 3 different cohorts.
Elevated EXO1 expression is associated with poor clinical outcome in breast cancer
While the breast cancer candidacy of CENPF, KIF14, NEK2, DTL, CKS1B, ASPM and EXO1 genes have been identified earlier [27-33], there is only one prominent report indicating the candidacy of EXO1 in breast cancers  and remains to be investigated. Apart from the reported polymorphisms [34-36] and elevated expression of EXO1 in Ductal carcinoma in situ , also there are contradicting reports relating EXO1 loss of function to increased susceptibility to lymphomas , urging the need for further investigation. Therefore, the current identification of the consistently elevated expression of EXO1 gene and its association with poor survival in breast cancers in multiple co-horts is a significant observation. Apart from the identified association between EXO1 expression and poor survival, we further confirmed its clear association with poor clinical outcome using overall survival as end point. Hazard ratio of EXO1 gene expression across 6 independent breast tumor profiles with relapse free survival as endpoint (Figure 2A) and Kaplan Meier survival curves for EXO1 expression with overall survival information as endpoint (Figure 2B-2C) clearly implies the poor clinical outcome associated with higher level EXO1 expression in breast cancer patients. Further, univariate and multivariate Cox proportional hazards model was performed in 3 breast tumor profiles for which maximal clinical variables were available. Univariate analysis revealed the significant association of EXO1 expression with relapse of breast tumors comparable to other individual clinical variables such as age, tumor size, grade, ER status, and lymph node status. In multivariate analysis also EXO1 retained the statistical significance with p-value <0.05 in all 3 analyzed profiles (Table S3). This indicates EXO1 to be an independent predictor of survival, and needs to be investigated in larger cohorts. Since EXO1 expression is associated with the poor prognosis, there arises a question on EXO1 gene expression in the context of specific subtype(s)/group(s) of breast cancers. We addressed this by investigating the expression of EXO1 in 18 breast cancer transcriptome profiles that are available from microarray repositories GEO and Array Express (Table S1). Systematic analysis of EXO1 expression across these breast cancer profiles revealed consistent elevated expression of EXO1 in i) breast tumors when compared to normal breast tissues (Figure S1A), ii) higher grade breast tumors (Figure 2D), iii) ER negative while compared to ER positive tumors (Figure 2E), iv) PR negative subtype of breast tumors (Figure S1B), and v) basal subtype breast tumors when compared to luminal subtypes (Figure S2). We also found subtle elevation of EXO1 expression in invasive ductal breast carcinoma and metastatic breast tumors with modest statistical significance (Figure S1C-S1D). Thus, across multiple cohorts of breast tumors from various populations that were profiled across various microarray platforms, we observed highly consistent and elevated expression of EXO1 in high grade, basal, ER negative and PR negative subtypes. All these show a strong association between EXO1 expression and poor clinical outcome in breast cancer patients.
(A) Hazard ratio for EXO1 gene expression with relapse free survival as end point in 6 different breast tumor datasets shown as forest plot. B & C) Kaplan Meier curve based on EXO1 gene expression with overall survival as end point in breast tumor profiles, Vijver and GSE7390 respectively. The log-rank test was used for computing significance level for the survival curves. (D) and (E) Comparative analysis of EXO1 gene expression across different grades and subgroups of breast tumors reveal elevated EXO1 expression in high grade (D) and ER negative subgroups of breast tumors (E). The total number of samples in each group is denoted in parenthesis and the GEO accessions of the profiles are shown on the top of each boxplot.
Identification of the possible upstream regulators of EXO1 in Breast tumors
Elevated expression of EXO1 in breast cancer patients with poor clinical outcome indicates the need for investigating the pathways and factors regulating EXO1 expression. In order to identify the possible upstream regulators of EXO1, gene signature based pathway activation pattern was investigated in the mRNA expression profiles of breast cancer samples. 26 signatures or gene sets representing various pathways and molecular cellular processes were collected from MsigDB and original literatures and used for this analysis (Table S4). Two breast cancer profiles representative of breast tumors [GSE7930], and breast cancer cell lines [E-TABM-157] were analyzed to score the activation of pathway signatures using in-silico gene-set based pathway activation prediction approach described earlier [23,38]. Hierarchical clustering (Figure 3A and Figure S3A), regression analysis (Table S5-S6) and principal component analysis (Figure 3B and Figure S3B) of signature based activation pattern of pathways revealed a significant positive association of EXO1 gene expression with the activation of MYC, RAS, EGFR, Genomic instability, and E2F pathways in breast tumors and cell lines. On the other hand, Estrogen Receptor (ESR1), p53, and BRCA pathways showed negative association with EXO1 expression (Figure 3A - 3B, Figure S3, Table S5-S6). This shows that EGFR, RAS, MYC and E2F1 could be the possible upstream regulators of EXO1.
(A) The activation status of oncogenic signaling pathways and other cancer signatures in 198 breast tumor samples shown as heatmap. Different pathway or cellular process specific gene-sets were analyzed and the comprehensive activation (red) and inactivation (green) pattern across samples is shown. (B) Principal component analysis of the pathway activation pattern in breast tumors shown in Figure A. (C) Occurrence of E2F, Myc and p53 transcription factor binding sites in EXO1 promoter. (D) Western blot showing the expression of EXO1 in a panel of breast cancer cell lines. (E) RT-PCR showing the expression of EXO1 in a panel of breast cancer cell lines (top panel). In vitro pathway specific reporter assays showing relative activation of Myc and E2F transcription factors in three of the higher level EXO1 expressing cell lines (ER –ve) while compared to two lower level EXO1 expressing cell lines (ER +ve) (Lower panel). These assays have been performed thrice with triplicates. (F) Luciferase reporter assay showing the activation of EXO1 promoter upon the transfection of E2F1 or c-Myc cDNA in MDA-MB231 cells (p-values were 0.013 and 0.009 respectively). pGL3-EXO1 is EXO1 promoter construct in pGL3 enhancer reporter vector. The experiment has been performed thrice with triplicates.
EXO1 expression is regulated by Myc & E2F transcription factors
In order to identify suitable cell lines for the in vitro experiments, expression of EXO1 across a panel of breast cancer cell lines was probed by western blotting and RT-PCR. Overall, there is a concordance in EXO1 expression pattern identified by Western blotting, RT-PCR and available microarray data (Figure 3D, 3E and Figure S4). Multiple associations indicate the possible regulation of EXO1 by EGFR, RAS, MYC, and E2F1. First, the analysis of transcription factor binding sites in EXO1 promoter region revealed the presence of MYC and E2F1 binding sites (Figure 3C). Second, we investigated the association between EXO1 expression and inherent signalling / transcription factor activity of MYC and E2F transcription factors in a panel of breast cancer cell lines comprising few higher level EXO1 expressing cells (HBL100, SKBR3 and MDA-MB-231) and a couple of lower level EXO1 expressing cells (MCF7 and T47D) by in vitro reporter assay (Figure 3E). In five different breast cancer cell lines, the MYC and E2F reporter plasmids were transfected and assayed for the inherent un-induced transcriptional activity. The normalized luciferase reporter assay result shows a positive association between MYC and E2F reporter activity and EXO1 expression in these cell lines with 5-20 folds higher activation of MYC and E2F in majority of EXO1 expressing cell lines (Figure 3E).
To confirm the MYC and E2F mediated transcriptional regulation of EXO1, -2 kb promoter region of EXO1 was cloned in pGL3-Enhancer reporter vector (designated as pGL3-EXO1). In MDA-MB-231 cells, pGL3-EXO1 was transfected along with MYC and E2F cDNA (ORF cloned in mammalian expression vector) encoding plasmids and assayed for luciferase activity. The results implied a positive regulation of EXO1 promoter by the transcription factors MYC and E2F1 (Figure 3F). Further, chemical inhibition of E2F activity by E2F inhibitor (Silymarin) reduced the expression of EXO1 in MDA-MB-231 cell lines (Figure 4A). All these results demonstrate that MYC and E2F regulates the expression of EXO1 in breast cancer cells.
(A - D) Reduced expression of EXO1 in breast cancer cells upon chemical inhibition of E2F, Ras, EGFR, and PI3K/AKT pathways using the specific inhibitors Silymarin, (A) Salirasib (B), Geftinib (C), and LY-294002 (D) respectively. While E2F & Ras inhibition are from MB-231 cells (A-B), PI3K/AKT & EGFR inhibition experiments (C-D) are from MB-468 cells. (E) RT-PCR showing that EXO1 expression is relatively high in gastric cancer cell lines with KRAS mutation, while compared to the cell line without KRAS mutation. (F) FOXO3 reporter assay (inversely proportional to AKT pathway activity) showed relatively higher level activity of AKT in cell lines with higher level EXO1 expression (YCC16 and AGS) compared to the cell line with relatively less folds of AKT activity (Kato III). (G - I) Elevated expression of EXO1 in MCF7 cells upon exposure to DNA damaging agents carboplatin (G), Ifosfamide (H), and Cyclophosphamide (I). The luciferase assays have been performed thrice with triplicates. All semi-quantitative RT-PCRs were performed thrice.
EGFR and RAS are the upstream regulators of EXO1
Apart from E2F and Myc, EGFR and RAS also showed positive association with the expression of EXO1 in pathway pattern correlation analysis (Figure 3A - 3B). Therefore, we hypothesized RAS being the key downstream regulator of EGFR which in turn might govern the expression of EXO1. In order to investigate this association, we took a KRAS mutant breast cancer cell line, MDA-MB-231 and treated with RAS inhibitor, salirasib. This resulted in the inhibition of EXO1 expression (Figure 4B). Subsequently to address the association between EGFR and EXO1 gene expression, we treated EGFR positive breast cancer cell line, MDA-MB-468, with EGFR inhibitor, gefitinib. As expected, EGFR inhibition resulted in reduced expression of EXO1 in MDA-MB-468, thereby illustrating the involvement of EGFR/RAS cascade in the regulation of EXO1 gene expression in breast cancer cells (Figure 4C). The involvement of RAS/PI3K signalling cascade was confirmed owing to the reduction in EXO1 expression upon treatment with PI3K/AKT inhibitor LY-294002 in the breast cancer cell line MDA-MB-468 (Figure 4D). We also found reduction in the expression of EXO1 upon the same EGFR and PI3K/AKT inhibitor treatment in MDA-MB-231 and ZR-75 cells (data not shown). These show that EXO1 is regulated through RAS/PI3K/AKT signalling in breast cancer cells. Interestingly, even in a panel comprising KRAS mutant and wild type gastric cancer cell lines, EXO1 showed higher expression in KRAS mutant cell lines (YCC16 and AGS) when compared to wild type cells (KATOIII) (Figure 4E). Further, we analyzed the FOXO3 firefly luciferase reporter activity (a negative indicator of RAS/PI3K/AKT pathway activation) in the same gastric cancer cell lines and identified FOXO3 reporter activity to be higher in wild type cells (Kato III) while compared to KRAS mutant cell lines (AGS and YCC16) (Figure 4F). All these indicate that EGFR, Ras, PI3K, Myc, and E2F are involved in the regulation of EXO1. However, among these the direct and indirect regulations remain to be determined.
We also addressed the involvement of EXO1 in DNA repair pathway in cancerous conditions since EXO1 is known to play a vital role in DNA repair process by mismatch mediated repair mechanism . This was addressed by treating MCF7 cell line with different alkylating agents carboplatin, cyclophosphamide and ifosfamide as they are well known to induce DNA repair [40-42]. The experiments showed EXO1’s increased expression with increasing concentration of these alkylating agents in a dose dependent manner thus supporting its role in DNA repair processes (Figure 4G-4I). Apart from DNA repair, another prime factor associated with EXO1 expression is genomic instability (Figure 3A-3B). Since these alkylating agents also would elevate the genomic instability in cancer cells upon exposure , these results also indicate that EXO1 expression is indicative of elevated genomic instability in cancer cells.
EXO1 co-expressing genes are predictive of poor clinical outcome in breast cancer patients
Since we identified EXO1 expression as indicative of poor prognosis and genomic instability with activated RAS/PI3K/AKT/MYC/E2F cascade, this observation needs to be explored for possible diagnostic and therapeutic implications. However, rather than single gene, a cluster of genes would be better predictors of a phenotype . Therefore, we derived a set of genes which have the expression pattern very similar to EXO1. Since coexpressing genes would have the same pattern of gene regulation, correlation coefficients of genes coexpressing with EXO1 were computed (Table S7) in 2479 samples from 9 different breast tumor transcriptome profiles (Table S1) and defined a module of genes that are tightly coexpressed with EXO1 in breast tumors (Figure 5A). Further analysis on the features of EXO1 modular genes mirrored EXO1 in i) occurrence of enriched E2F binding sites in the promoter of EXO1 modular genes (Figure S5) and ii) showing significant association with poor survival of the breast cancer patients (Figure 5B-5C). Another striking aspect of EXO1 module is that it includes all seven 1q candidate genes that were originally identified as genes associated with poor survival in breast cancer patients. Analysis of the expression of EXO1 modular genes in a comprehensive panel of 51 breast cancer cell lines showed higher expression in basal and invasive breast cancer cell lines (Figure 5D). EXO1 modular genes showed higher expression in ER negative and higher grade breast tumors (Figure S6). However, a sub-set of ER positive tumors also showed elevated expression of EXO1 modular genes. This shows that EXO1 modular genes are capable of predicting aggressive breast tumors irrespective of ER status.
: (A) Identification of EXO1 modular genes using meta-analysis of correlation coefficients from multiple breast tumor datasets. Schematic EXO1 centred network shows that EXO1 co-expressed genes are involved in cell cycle and DNA repair. (B) and (C) Elevated expression of EXO1 modular genes was associated with poor clinical outcome in breast tumor cohorts. (D) Heatmap showing the expression pattern of EXO1 modular genes in 51 breast cancer cell lines. EXO1 modular genes are highly expressed in aggressive and basal cell lines. I-Invasive, N-Non-invasive, B-Basal, L-Luminal and ER status of the cell lines are shown on the top.
PI3K/AKT inhibition is the potential therapeutic strategy for EXO1 module activated breast tumors
With the identified clinically significant molecular pathological and disease stratification features of EXO1 expression in breast cancer, the identification of suitable therapeutic strategy for targeting the breast cancer cells with elevated EXO1 expression would be useful for the development of novel breast cancer therapeutic options. In breast tumor mRNA profiles (GSE7390), the genes differentially expressed between elevated EXO1 expressing breast tumors (>60%) and reduced EXO1 expressing breast tumors (<10%) were short-listed. The derived genes were analyzed in connectivity map  and obtained a list of drugs that could inhibit the expression of genes which are expressed in breast cancer patients with elevated EXO1 expression (Figure 6A). This analysis revealed many PI3K inhibitors like sirolimus, LY-294002, and wortmanin to be potent inhibitors for the potential reversal of the expression of EXO1 associated gene-set (Figure 6B). This is also in agreement with the delineated regulatory mode of EXO1 which involves Ras/PI3K/AKT. In order to further assess this possibility, we analysed the expression of EXO1 modular genes in the gene expression profiles of PI3K / RAS inhibitor treated cells which were readily available in microarray repository. This analysis in a mammary epithelial cell (MCF10A) and few non-breast originated cells (A549 and SHEP), upon treatment with the PI3K inhibitor (LY294002) and Ras inhibitor (FTS), showed tremendous inhibition of EXO1 modular genes (Figure 6C). Similarly, inhibition of Ras by Salirasib also resulted in reduced expression of EXO1 modular and 1q candidate genes NEK2, CKS1B, DTL and KIF14 in MDA-MB-231 cells (Figure 6D). On the other hand, inhibition of PI3K activity in MDA-MB-468 cell lines with LY-294002 also showed decreasing EXO1 expression in a dose dependent manner. All these support Ras/PI3K inhibition as a possible therapeutic strategy for EXO1 modular activated breast tumors.
(A) Workflow employed for the identification of suitable chemo therapeutic agent for EXO1 module expressing breast tumors. (B) Results of Connectivity map analysis showing that while patients expressing EXO1 co-regulated genes, Pi3K inhibition could be the possible targeted therapeutic approach. (C) Heatmap showing the inhibition of EXO1 modular genes in GSE33403 downloaded from GEO that corresponds to MCF10A cells (PIK3CA E545K mutation) treated with LY294002 for 4 hours and 24 hours. Similarly, the profiles of A549 and SHEP cells treated with the Ras inhibitor FTS were obtained from Blum et al., 2007. (D) RT-PCR showed reduced expression of 5 out of 7, 1q candidate genes NEK2, CKS1B, DTL, KIF14 and EXO1 in MB231 cells upon chemical inhibition of RAS pathway with salirasib.
Genomic abnormalities remain a characteristic feature of cancer cells. Despite being the frequent chromosomal aberration in breast cancer genomes, the candidate target genes of 1q remain unexplored. Understanding of the pathological roles of target genes may further lead to the development of novel breast cancer stratification and targeted therapeutics and would have implications in the effective management of breast cancer patients with 1q amplification. Through a meta-analysis work-flow, by analysing the clinical outcome of breast cancer patients in association with the expression of the genes individually, we scored the candidacy of all genes from 1q. Genome-wide mRNA profiles and survival data available for 6 different breast tumor cohorts were used for this analysis and we identified 7 candidate genes from chromosome 1q region. Earlier, these genes were independently noticed for their significant involvement in breast cancers. For instance, CENPF (Centrosome protein F) and NEK2 (NIMA (never in mitosis gene a) - related kinase 2) were reported for their association with poor prognosis and chromosomal instability in breast cancer [27,29]. In our current filtering of 1q genes related to survival, CENPF (HR = 1.616; p-value = 4.90E-05) ranked first followed by NEK2 (HR = 1.641; p-value = 0.000163) which demonstrates the reliability of the strategy and the results obtained in the study. Commonality with most of the short-listed genes (CENPF, KIF14, NEK2, CKS1B and ASPM) is their positive association with proliferation marker Ki-67 [26,27,30,45,46], thereby, indicating their possible role in cell cycle related dysregulations and the resultant proliferation of breast cancer cells. However, it has been reported that simultaneous 1q gain/16q loss was related to low Ki-67 level (low proliferation) and high p27 expression of breast cancer cells . While simultaneously considering 1q gain and 16q loss, low proliferation rate which otherwise the consequence of 16q loss also might have masked the significance of 1q gain. Our results show clear association between 1q candidate gene expression and proliferation.
The candidate gene which is relatively underexplored in breast cancers is Exonuclease I (EXO1), a Rad2 family member possessing 5’-3’ exonuclease activity and well established for its role in mismatch repair and DNA recombination. Being a DNA mismatch repair gene which is known to play a role in maintaining genomic integrity, its strong candidature with poor clinical outcome among breast cancer patients need further investigation. EXO1 mutant mice were reported to have reduced survival and increased susceptibility to lymphoma development . Studies evaluating single nucleotide polymorphisms in DNA repair related genes have emphasized the role of EXO1-K589E allele as a biomarker potentially linked with carcinogenesis [34-36]. A recent study showed the elevated expression of EXO1 in ductal carcinoma in situ samples and is the first hint wherein the role of EXO1 in breast cancer was revealed . In the current integrative genomic investigation, analysis of EXO1 gene expression across various groups and subtypes of breast cancer reveal EXO1’s higher expression in higher grade, basal and ER negative subtypes. In the same manner, a pattern of elevated expression in ER negative breast cancer was reported for 2 of the 1q candidate genes CENPF  and KIF14 . NEK2 was reported for marked expression with both ER positive and negative subtypes . Our analysis across breast cancer and non-cancerous breast mRNA profiles showed all 7 candidate genes to have similar expression pattern, apart from their association with poor clinical outcome. Since this pattern identification is from 4 different cohorts of breast tumors from 1371 breast cancer tissues, this would be more reliable than the previously mentioned single cohort based studies.
Due to 1q amplification and very similar expression pattern, we hypothesized all 7 genes to have a similar regulatory pattern and investigated EXO1 as a representative candidate from 1q region. Unravelling the factors regulating EXO1 by signaling pathway focused gene set activation pattern prediction revealed closer association of activated RAS, EGFR, MYC and E2F pathways with EXO1’s elevated expression. In addition, genomic instability and telomerase activations showed positive association with EXO1 gene expression. This association is prominent in breast tumors and breast cancer cells. RAS oncogenes (KRAS, HRAS and NRAS) harbour activation mutation in about 20% of human tumors and confers uncontrolled cell proliferation advantage . In breast cancers, RAS is often activated by HER2 [ErbB2/epidermal growth factor receptor (EGFR) 2/Neu] receptor tyrosine kinase and is over expressed and persistently activated in approximately 25 % of cancers . We observed a striking concordance between the expression pattern of EXO1 (Figure 3D) and RAS activation pattern in breast cancer cell lines reported earlier . This strongly suggests the intimate positive association between EXO1 expression and RAS activation. Apart from RAS being a possible regulator of EXO1, the striking observation is, ‘the expression of EXO1 is indicative of RAS activation’.The effector signals downstream of RAS could be ERK/MAPK signalling or PI3K/AKT signalling which are involved in cell survival and proliferation . PI3K/AKT signalling is also reported to stabilize c-MYC expression in GSK3B dependent manner and the transcription of E2F1 gene thereby leading to S-phase progression in cell cycle . Ras also regulates the expression of E2F . First, in the light of these literature evidences, second, with closer association of activated RAS, EGFR, MYC and E2F signaling with EXO1 expression, and third with the series of in vitro experiments (Figure 3 & 4), for the first time we report the regulation of EXO1 in breast cancer cells through EGFR/RAS/PI3K/AKT/MYC/E2F signaling cascades (Figure 7). However, the direct upstream regulator of EXO1 needs to be determined. Since EXO1 is identified from analysing a frequently aberrated genomic region, its association with genomic instability and loss of p53 activity is not surprising and indeed supports the delineated regulatory modes of EXO1. Earlier report shows that frequent loss of PTEN could contribute to genomic instability in triple negative breast tumors . This also strengthens the involvement of RAS/PI3K/MYC/E2F pathways in conferring genomic instability, and EXO1 modular expression is indicative of that.
EGFR, RAS, PI3K / AKT, MYC, and E2F were identified as the upstream regulators of these genes. The sequential arrangement of pathways is from the well established literature [49-51]. Apart from the pathways, genomic instability, telomerase activation and loss of p53 are also associated with the expression of 1q candidate genes.
We also assessed the involvement of EXO1 in DNA repair in breast cancer cells. Exposure of breast cancer cells to alkylating and DNA damaging agents induces the expression of EXO1. In view of the observed correlation between elevated EXO1 expression and activated genomic instability related gene-set in breast tumors (Figure 3A), this observation suggests the possibility that prolonged EXO1 expression in breast cancer is indicative of un-rectified DNA repairs. This shows that apart from being a poor survival indicator, EXO1 expression in breast cancers is also indicative of cancers with elevated genomic instability. However, the functional role of EXO1 in DNA repair needs to be investigated further. Our pathway focused analysis also supports the observation of overall elevated genomic instability in ER negative, basal and aggressive breast tumors [52,53]. Further, activated RAS/PI3K/MYC/E2F signaling is also the feature of these ER negative, basal and aggressive breast tumors. Importantly, EXO1 expression is indicative for all these features in breast cancers.
Using the concept of co-expression, we identified the common thread connecting EXO1 gene with other genes in breast cancer (EXO1 module) to be cell cycle progression and proliferation (Figure 5A). Cellular proliferation stands the traditional marker for prognosis that has different predictive values in ER positive and ER negative cancers , and works far beyond ER status . Strikingly, in the derived EXO1 modular gene-set, all seven 1q candidate genes were observed. This also implies that the delineated regulation of EXO1 is applicable to all EXO1 modular genes and in particular to all the seven 1q genes identified in this investigation. Supporting this notion, 44 out of 63 EXO1 module genes were earlier identified to be regulated by RAS . This also illustrates the possible 1q amplification independent regulation of EXO1 module in breast cancers. Since only 7 out of the 498 genes located in 1q amplicon are selectively expressed in breast cancers, 1q amplification independent and specifically regulated EXO1 modular expression is quite possible in breast cancers and needs to be investigated.
EXO1 module represents a highly conserved set of interactions in breast cancer. Comparing EXO1 module with different sets of gene signatures revealed very less overlap with clinically used gene signatures like MammaPrint and Oncotype and other breast cancer derived prognostic gene expression signatures (Figure S7). The observed minimal overlaps among the other signatures were pointing the proliferation related genes as common denominator for several prognostic gene expression signatures analysed in this study. Wound response and CIN70 gene signatures, which were earlier demonstrated as powerful predictors of metastasis in multiple cancers [56,57] showed higher overlap with EXO1 module in terms of gene content (Table S8). Accordingly, EXO1 module genes’ expression is also observed to be higher in metastatic and basal breast tumors. EXO1 module genes are comprised of cellular proliferation related genes and thus represent i) cellular proliferation, ii) activated RAS/PI3K/MYC/E2F signaling, iii) elevated genomic instability and loss of p53 functionality, and iv) better predictor of metastasis and poor clinical outcome.
Elevated expression of EXO1 in ER negative and high grade breast tumors indicate the possible development of targeted therapeutics by targeting EXO1 module or its upstream regulators. Connecting the gene expression with possible chemical inhibition through the concept of connectivity map showed that RAS/PI3K inhibitors could inhibit the expression of EXO1 modular genes. With evidences from the connectivity map and in-vitro gene expression analysis upon treatment with PI3K or RAS inhibitor, it seems possible to target PI3K or RAS in order to inhibit EXO1 modular expression in breast cancer cells. While considering the identified upstream regulators of EXO1, targeting EGFR, RAS or PI3K is also possible. A number of clinical trials targeting EGFR, PI3K and RAS downstream signals were performed recently. Anti-EGFR therapies yielded Gefitinib and Erlotinib, anti-PI3K therapies yielded Rapamycin and Tipifarnib that inhibits farnesylation of RAS and other proteins involved in signal transduction pathways [47,58,59]. Despite being the activator of RAS signalling pathway, EGFR derives less attention in this aspect due to activating mutations in KRAS, which are significantly associated with lack of response or resistance with some of the EGFR inhibitors like cetuximab in colorectal and lung cancers [58,60]. Therefore, as per the predictions and in vitro validation, RAS, or PI3K inhibition could be considered for evaluating the potential inclusion in the regimen for EXO1 over expressing group of patients. However, in prior, the KRAS and PIK3CA mutations status also needs to be considered. It is worth mentioning that EXO1 over expressing group of patients also could be denoted as 1q candidate gene over expressing group of patients.
In this study, we scanned chromosome 1q genes for their significant association with survival of the patients and identified 7 potential candidate genes. These genes were found to consistently over express in high grade and aggressive breast tumors with poor clinical outcome. We delineated the upstream regulators of EXO1, an underexplored candidate gene in breast cancer. By integrative functional genomics and molecular cell biological approaches, we showed the involvement of EGFR, RAS, PI3K / AKT, MYC, E2F signaling in the regulation of these selected 1q genes in breast tumors. Expression of EXO1 module, the gene set derived from co-expressed genes of EXO1 gene, was found as indicative of elevated cell proliferation, genomic instability, activated RAS/AKT/MYC/E2F1 signaling pathways and loss of p53 activity in breast tumors. We also suggest the inhibition of RAS/PI3K as possible therapeutic option for the patients with elevated expression of the EXO1 module.
Investigation of EXO1 gene expression in different categories of breast tumors. Expression pattern of EXO1 in (A) breast tumors while compared to normal breast tissues. (B) PR positive and negative tumors, (C) metastatic tumors against non-metastatic groups, (D) Invasive Ductal Carcinoma (IDC), Invasive Lobular Carcinoma (ILC) and mixed sub-types.
EXO1 expression is elevated in basal type of breast tumors. Expression pattern of EXO1 across basal, luminal and HER2 subtypes of breast cancer in 3 different datasets (A-C).
Pathway activation pattern in breast cancer cell lines. (A) Heatmap showing pathway activation pattern in breast cancer cell lines (E-TABM-157). (B) Principal component analysis of pathway activation scores in breast cancer cell lines.
Gene expression pattern of EXO1 in a panel of breast cancer cell lines. EXO1 expression in breast cancer cell lines as extracted from E-TABM-157 profile was showed as bar chart.
NFY and E2F transcription factors are enriched in EXO1 module. Analysis on transcription factor binding sites enrichment in EXO1 module revealed the higher percentage of NFY and E2F transcription factors.
EXO1 module genes show higher expression in high grade, aggressive breast tumors. Expression pattern of EXO1 module in two different breast tumor profiles (A) GSE25066 and (B) GSE7390 is depicted as heatmap.
Comparison of the overlap between EXO1 modular genes and other prognostic gene sets of breast tumors.
List of datasets used in the study.
Results of Meta-analysis on breast cancer survival data for 1q genes.
Univariate and Multivariate analysis with EXO1 gene expression and other clinical covariates reveals the prognostic significance of EXO1 in breast cancer.
List of pathway signatures used for predicting upstream regulators of EXO1.
Regression analysis of EXO1 gene expression with pathway activation status in 198 breast tumor samples (GSE7390).
Regression analysis of EXO1 gene expression with pathway activation status in 51 breast cancer cell lines (E-TABM-157).
Results of Meta-analysis of correlation coefficients.
DST-INSPIRE fellowship support to Muthulakshmi Muthuswami and NET-CSIR fellowship support to Vignesh Ramesh are acknowledged. Instrumentation supports of UGC-CEGS, UGC-CAS, DBT-IPLS, DST-PURSE and UGC-NRCBS programme supported central facilities of SBS, MKU are acknowledged.
Conceived and designed the experiments: KG MM SB VR. Performed the experiments: VR SVT JP DBR GDB SR. Analyzed the data: MM VR SB KG. Wrote the manuscript: KG MM SB VR. Performed data mining and statistical analysis: MM VR SB KG Approved the final manuscript: KG MM VR SB SVT JP DBR GDB SR.
- 1. Perou CM, Sørlie T, Eisen MB, van de Rijn M, Jeffrey SS et al. (2000) Molecular portraits of human breast tumours. Nature 406: 747-752. doi:https://doi.org/10.1038/35021093. PubMed: 10963602.
- 2. Sørlie T, Perou CM, Tibshirani R, Aas T, Geisler S et al. (2001) Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc_Natl_Acad_Sci_U_S_A 98: 10869-10874. PubMed: 11553815.
- 3. van 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA et al. (2002) Gene expression profiling predicts clinical outcome of breast cancer. Nature 415: 530-536. doi:https://doi.org/10.1038/415530a. PubMed: 11823860.
- 4. Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP et al. (2005) Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet %19-25;365: 671-679.
- 5. Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R et al. (2004) Large-scale meta-analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression. Proc_Natl_Acad_Sci_U_S_A 101: 9309-9314. PubMed: 15184677.
- 6. Wirapati P, Sotiriou C, Kunkel S, Farmer P, Pradervand S et al. (2008) Meta-analysis of gene expression profiles in breast cancer: toward a unified understanding of breast cancer subtyping and prognosis signatures. Breast Cancer Res 10: R65. doi:https://doi.org/10.1186/bcr1949. PubMed: 18662380.
- 7. Kao J, Salari K, Bocanegra M, Choi YL, Girard L et al. (2009) Molecular profiling of breast cancer cell lines defines relevant tumor models and provides a resource for cancer gene discovery. PLOS_ONE 4: e6146. PubMed: 19582160.
- 8. Neve RM, Chin K, Fridlyand J, Yeh J, Baehner FL et al. (2006) A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell 10: 515-527. doi:https://doi.org/10.1016/j.ccr.2006.10.008. PubMed: 17157791.
- 9. Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM et al. (2012) The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486: 346-352. PubMed: 22522925.
- 10. Atkin NB (1986) Chromosome 1 aberrations in cancer. Cancer Genet_Cytogenet 21: 279-285. PubMed: 3513945.
- 11. Orsetti B, Nugoli M, Cervera N, Lasorsa L, Chuchana P et al. (2006) Genetic profiling of chromosome 1 in breast cancer: mapping of regions of gains and losses and identification of candidate genes on 1q. BrJCancer %20;95: 1439-1447.
- 12. Fang M, Toher J, Morgan M, Davison J, Tannenbaum S et al. (2011) Genomic differences between estrogen receptor (ER)-positive and ER-negative human breast carcinoma identified by single nucleotide polymorphism array comparative genome hybridization analysis. Cancer 117: 2024-2034. doi:https://doi.org/10.1002/cncr.25770. PubMed: 21523713.
- 13. Waddell N, Arnold J, Cocciardi S, da SL, Marsh A et al. (2010) Subtypes of familial breast tumours revealed by expression and copy number profiling. Breast Cancer Res_Treat 123: 661-677. doi:https://doi.org/10.1007/s10549-009-0653-1. PubMed: 19960244.
- 14. Gao Y, Niu Y, Wang X, Wei L, Lu S (2009) Genetic changes at specific stages of breast cancer progression detected by comparative genomic hybridization. JMolMed(Berl) 87: 145-152.
- 15. Bièche I, Champème MH, Lidereau R (1995) Loss and gain of distinct regions of chromosome 1q in primary breast cancer. Clin_Cancer Res 1: 123-127. PubMed: 9815894.
- 16. Stange DE, Radlwimmer B, Schubert F, Traub F, Pich A et al. (2006) High-resolution genomic profiling reveals association of chromosomal aberrations on 1q and 16p with histologic and genetic subgroups of invasive breast cancer. Clin_Cancer Res 12: 345-352. PubMed: 16428471.
- 17. Farabegoli F, Hermsen MA, Ceccarelli C, Santini D, Weiss MM et al. (2004) Simultaneous chromosome 1q gain and 16q loss is associated with steroid receptor presence and low proliferation in breast carcinoma. Mod_Pathol 17: 449-455. PubMed: 14976537.
- 18. Albertson DG, Collins C, McCormick F, Gray JW (2003) Chromosome aberrations in solid tumors. Nat_Genet 34: 369-376. PubMed: 12923544.
- 19. T T (2012). A Package Surviv Anal S.
- 20. Therneau M, [!(surname)!] , Grambsch M P (2000) Modeling Survival Data: Extending the Cox Model: Springer Verlag.
- 21. Fox J (2005) The R Commander: A Basic Statistics Graphical User Interface to R. J Stat Softw 14: 1-42.
- 22. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT et al. (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 13: 2498-2504. doi:https://doi.org/10.1101/gr.1239303. PubMed: 14597658.
- 23. Ooi CH, Ivanova T, Wu J, Lee M, Tan IB et al. (2009) Oncogenic pathway combinations predict clinical prognosis in gastric cancer. PLOS_Genet 5: e1000676. PubMed: 19798449.
- 24. Kim TM, Jeong HJ, Seo MY, Kim SC, Cho G et al. (2005) Determination of genes related to gastrointestinal tract origin cancer cells using a cDNA microarray. Clin Cancer Res 11: 79-86. PubMed: 15671531.
- 25. Dyer BW, Ferrer FA, Klinedinst DK, Rodriguez R (2000) A noncommercial dual luciferase enzyme assay system for reporter gene analysis. Anal_Biochem 282: 158-161. PubMed: 10860516.
- 26. Caron H, van SB, van der Mee M, Baas F, Riggins G et al. (2001) The human transcriptome map: clustering of highly expressed genes in chromosomal domains. Science 291: 1289-1292. doi:https://doi.org/10.1126/science.1056794. PubMed: 11181992.
- 27. O'Brien SL, Fagan A, Fox EJ, Millikan RC, Culhane AC et al. (2007) CENP-F expression is associated with poor prognosis and chromosomal instability in patients with primary breast cancer. Int_J_Cancer 120: 1434-1443. PubMed: 17205517.
- 28. Corson TW, Gallie BL (2006) KIF14 mRNA expression is a predictor of grade and outcome in breast cancer. Int_J_Cancer 119: 1088-1094. PubMed: 16570270.
- 29. Tsunoda N, Kokuryo T, Oda K, Senga T, Yokoyama Y et al. (2009) Nek2 as a novel molecular target for the treatment of breast carcinoma. Cancer Sci 100: 111-116. doi:https://doi.org/10.1111/j.1349-7006.2008.01007.x. PubMed: 19038001.
- 30. Ueki T, Nishidate T, Park JH, Lin ML, Shimo A et al. (2008) Involvement of elevated expression of multiple cell-cycle regulator, DTL/RAMP (denticleless/RA-regulated nuclear matrix associated protein), in the growth of breast cancer cells. Oncogene 27: 5672-5683. doi:https://doi.org/10.1038/onc.2008.186. PubMed: 18542055.
- 31. Wang XC, Tian J, Tian LL, Wu HL, Meng AM et al. (2009) Role of Cks1 amplification and overexpression in breast cancer. BiochemBiophysResCommun %20;379: 1107-1113.
- 32. Horvath S, Zhang B, Carlson M, Lu KV, Zhu S et al. (2006) Analysis of oncogenic signaling networks in glioblastoma identifies ASPM as a molecular target. Proc_Natl_Acad_Sci_U_S_A 103: 17402-17407. PubMed: 17090670.
- 33. Kretschmer C, Sterner-Kock A, Siedentopf F, Schoenegg W, Schlag PM et al. (2011) Identification of early molecular markers for breast cancer. Mol_Cancer 10: 15-10. PubMed: 21314937.
- 34. Wang HC, Chiu CF, Tsai RY, Kuo YS, Chen HS et al. (2009) Association of genetic polymorphisms of EXO1 gene with risk of breast cancer in Taiwan. Anticancer Res 29: 3897-3901. PubMed: 19846925.
- 35. Luo X, Hong XS, Xiong XD, Zeng LQ, Lim CE (2012) A single nucleotide polymorphism in EXO1 gene is associated with cervical cancer susceptibility in Chinese patients. Int_J_Gynecol_Cancer 22: 220-225. PubMed: 22146767.
- 36. Bau DT, Wang HC, Liu CS, Chang CL, Chiang SY et al. (2009) Single-nucleotide polymorphism of the Exo1 gene: association with gastric cancer susceptibility and interaction with smoking in Taiwan. Chin JPhysiol 52: 411-418. doi:https://doi.org/10.4077/CJP.2009.AMH076. PubMed: 20337148.
- 37. Wei K, Clark AB, Wong E, Kane MF, Mazur DJ et al. (2003) Inactivation of Exonuclease 1 in mice results in DNA mismatch repair defects, increased cancer susceptibility, and male and female sterility. Genes Dev 17: 603-614. doi:https://doi.org/10.1101/gad.1060603. PubMed: 12629043.
- 38. Lamb J, Crawford ED, Peck D, Modell JW, Blat IC et al. (2006) The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science 313: 1929-1935. doi:https://doi.org/10.1126/science.1132939. PubMed: 17008526.
- 39. Schmutte C, Marinescu RC, Sadoff MM, Guerrette S, Overhauser J et al. (1998) Human exonuclease I interacts with the mismatch repair protein hMSH2. Cancer Res 58: 4537-4542. PubMed: 9788596.
- 40. Fichtinger-Schepman AM, van Dijk-Knijnenburg HC, SD vdV-V, Berends F, Baan RA (1995) Cisplatin- and carboplatin-DNA adducts: is PT-AG the cytotoxic lesion? Carcinogenesis 16: 2447-2453.
- 41. Harrouk W, Codrington A, Vinson R, Robaire B, Hales BF (2000) Paternal exposure to cyclophosphamide induces DNA damage and alters the expression of DNA repair genes in the rat preimplantation embryo. Mutat_Res 461: 229-241. PubMed: 11056294.
- 42. Korkmaz A, Topal T, Oter S (2007) Pathophysiological aspects of cyclophosphamide and ifosfamide induced hemorrhagic cystitis; implication of reactive oxygen and nitrogen species as well as PARP activation. Cell Biol_Toxicol 23: 303-312. PubMed: 17225077.
- 43. Fonseca FL, Sant Ana AV, Bendit I, Arias V, Costa LJ et al. (2005) Systemic chemotherapy induces microsatellite instability in the peripheral blood mononuclear cells of breast cancer patients. Breast Cancer Res 7: R28-R32. doi:https://doi.org/10.1186/bcr1232. PubMed: 15642167.
- 44. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL et al. (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc_Natl_Acad_Sci_U_S_A 102: 15545-15550. PubMed: 16199517.
- 45. Wang S, Li W, Lv S, Wang Y, Liu Z et al. (2011) Abnormal expression of Nek2 and beta-catenin in breast carcinoma: clinicopathological correlations. Histopathology 59: 631-642. doi:https://doi.org/10.1111/j.1365-2559.2011.03941.x. PubMed: 22014044.
- 46. Hagemann C, Anacker J, Gerngras S, Kühnel S, Said HM et al. (2008) Expression analysis of the autosomal recessive primary microcephaly genes MCPH1 (microcephalin) and MCPH5 (ASPM, abnormal spindle-like, microcephaly associated) in human malignant gliomas. Oncol_Rep 20: 301-308. PubMed: 18636190.
- 47. Downward J (2003) Targeting RAS signalling pathways in cancer therapy. Nat_Rev_Cancer 3: 11-22. PubMed: 12509763.
- 48. Eckert LB, Repasky GA, Ulkü AS, McFall A, Zhou H et al. (2004) Involvement of Ras activation in human breast cancer cell signaling, invasion, and anoikis. Cancer Res 64: 4585-4592. doi:https://doi.org/10.1158/0008-5472.CAN-04-0396. PubMed: 15231670.
- 49. Castellano E, Downward J (2011) RAS Interaction with PI3K: More Than Just Another Effector Pathway. Genes Cancer 2: 261-274. doi:https://doi.org/10.1177/1947601911408079. PubMed: 21779497.
- 50. Schild C, Wirth M, Reichert M, Schmid RM, Saur D et al. (2009) PI3K signaling maintains c-myc expression to regulate transcription of E2F1 in pancreatic cancer cells. Mol_Carcinog 48: 1149-1158. PubMed: 19603422.
- 51. Berkovich E, Ginsberg D (2001) Ras induces elevation of E2F-1 mRNA levels. Biol_Chem 276: 42851-42856. PubMed: 11551910.
- 52. Hu X, Stern HM, Ge L, O'Brien C, Haydu L et al. (2009) Genetic alterations and oncogenic pathways associated with breast cancer subtypes. Mol_Cancer Res 7: 511-522. PubMed: 19372580.
- 53. Melchor L, Benítez J (2008) An integrative hypothesis about the origin and development of sporadic and familial breast cancer subtypes. Carcinogenesis 29: 1475-1482. doi:https://doi.org/10.1093/carcin/bgn157. PubMed: 18596026.
- 54. Desmedt C, Sotiriou C (2006) Proliferation: the most prominent predictor of clinical outcome in breast cancer. Cell Cycle 5: 2198-2202. doi:https://doi.org/10.4161/cc.5.19.3254. PubMed: 16969100.
- 55. Blum R, Elkon R, Yaari S, Zundelevich A, Jacob-Hirsch J et al. (2007) Gene expression signature of human cancer cell lines treated with the ras inhibitor salirasib (S-farnesylthiosalicylic acid). Cancer Res 67: 3320-3328. doi:https://doi.org/10.1158/0008-5472.CAN-06-4287. PubMed: 17409441.
- 56. Chang HY, Sneddon JB, Alizadeh AA, Sood R, West RB et al. (2004) Gene expression signature of fibroblast serum response predicts human cancer progression: similarities between tumors and wounds. PLOS_Biol 2: E7. PubMed: 14737219.
- 57. Carter SL, Eklund AC, Kohane IS, Harris LN, Szallasi Z (2006) A signature of chromosomal instability inferred from gene expression profiles predicts clinical outcome in multiple human cancers. Nat_Genet 38: 1043-1048. PubMed: 16921376.
- 58. Engelman JA (2009) Targeting PI3K signalling in cancer: opportunities, challenges and limitations. Nat_Rev_Cancer 9: 550-562. PubMed: 19629070.
- 59. Hobday TJ, Perez EA (2005) Molecularly targeted therapies for breast cancer. Cancer Contr 12: 73-81. PubMed: 15855890.
- 60. Lièvre A, Bachet JB, Le CD, Boige V, Landi B et al. (2006) KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Res 66: 3992-3995. doi:https://doi.org/10.1158/0008-5472.CAN-06-0191. PubMed: 16618717.
- 61. Habermann JK, Doering J, Hautaniemi S, Roblick UJ, Bündgen NK et al. (2009) The gene expression signature of genomic instability in breast cancer is an independent predictor of clinical outcome. Int_J_Cancer 124: 1552-1564. PubMed: 19101988.
- 62. Ghazani AA, Arneson N, Warren K, Pintilie M, Bayani J et al. (2007) Genomic alterations in sporadic synchronous primary breast cancer using array and metaphase comparative genomic hybridization. Neoplasia 9: 511-520. doi:https://doi.org/10.1593/neo.07301. PubMed: 17603634.
- 63. Park SC, Hwang UK, Ahn SH, Gong GY, Yoon HS (2007) Genetic changes in bilateral breast cancer by comparative genomic hybridisation. Clin_Exp_Med 7: 1-5. PubMed: 17380298.
- 64. Karlsson E, Danielsson A, Delle U, Olsson B, Karlsson P et al. (2007) Chromosomal changes associated with clinical outcome in lymph node-negative breast cancer. Cancer Genet_Cytogenet 172: 139-146. PubMed: 17213022.
- 65. Gronwald J, Jauch A, Cybulski C, Schoell B, Böhm-Steuer B et al. (2005) Comparison of genomic abnormalities between BRCAX and sporadic breast cancers studied by comparative genomic hybridization. Int_J_Cancer, 114: 230–6 %20;114: 230-236 PubMed: 15540206.
- 66. Weber-Mangal S, Sinn HP, Popp S, Klaes R, Emig R et al. (2003) Breast cancer in young women (< or = 35 years): Genomic aberrations detected by comparative genomic hybridization. Int_J_Cancer, 107: 583–92 %20;107: 583-592 PubMed: 14520696.
- 67. Baudis M, Cleary ML (2001) Progenetix. net: an online repository for molecular cytogenetic aberration data. Bioinformatics 17: 1228-1229. doi:https://doi.org/10.1093/bioinformatics/17.12.1228. PubMed: 11751233.
- 68. Baudis M (2012) Progenetix oncogenomic online resource. Retrieved onpublished at whilst December year 1111 from www.progenetix.net.