Senescence sensitivity of breast cancer cells is defined by positive feedback loop between CIP2A and E2F1

Abstract

Senescence induction contributes to cancer therapy responses and is crucial for p53-mediated tumor suppression. However, whether p53 inactivation actively suppresses senescence induction has been unclear. Here, we show that E2F1 overexpression, due to p53 or p21 inactivation, promotes expression of human oncoprotein CIP2A, which in turn, by inhibiting PP2A activity, increases stabilizing serine 364 phosphorylation of E2F1. Several lines of evidence show that increased activity of E2F1-CIP2A feedback renders breast cancer cells resistant to senescence induction. Importantly, mammary tumorigenesis is impaired in a CIP2A-deficient mouse model, and CIP2A-deficient tumors display markers of senescence induction. Moreover, high CIP2A expression predicts for poor prognosis in a subgroup of patients with breast cancer treated with senescence-inducing chemotherapy. Together, these results implicate the E2F1-CIP2A feedback loop as a key determinant of breast cancer cell sensitivity to senescence induction. This feedback loop also constitutes a promising prosenescence target for therapy of cancers with an inactivated p53-p21 pathway.

Significance: It has been recently realized that most currently used chemotherapies exert their therapeutic effect at least partly by induction of terminal cell arrest, senescence. However, the mechanisms by which cell-intrinsic senescence sensitivity is determined are poorly understood. Results of this study identify the E2F1-CIP2A positive feedback loop as a key determinant of breast cancer cell sensitivity to senescence and growth arrest induction. Our data also indicate that this newly characterized interplay between 2 frequently overexpressed oncoproteins constitutes a promising prosenescence target for therapy of cancers with inactivated p53 and p21. Finally, these results may also facilitate novel stratification strategies for selection of patients to receive senescence-inducing cancer therapies.

Figure 1. Wild type p53 negatively regulates CIP2A expression

A) Representative immunohistochemical stainings of CIP2A and p53 expression in human breast cancer tumors. B) CIP2A expression positively correlates with p53 expression and with proliferation marker Ki67 in human breast tumors (n=1228). P value calculated by chi-squared test. C) Western blot analysis of CIP2A expression in mouse embryonal fibroblasts (MEFs) 48h after transfection with scrambled (SCR) or two different p53 siRNAs (p53.1 and p53.2). D) Western blot analysis of CIP2A, p53 and p21 expression in MCF-7 cells treated with 2, 5 or 10μM of Nutlin-3 for 36h. E) p21 and CIP2A mRNA expression in MCF-7 cells treated with 2, 3, and 5μM of Nutlin-3 for 24h. Shown is mean +SEM of two independent experiments. F) CIP2A protein expression in MDA-MB-231 human breast cancer cells harboring DNA-binding deficient p53 treated with 5 μM of Nutlin-3 for 24 hours. G) Western blot analysis of CIP2A expression in MDA-MB-231 cells 48h after transduction either with control (CTL) or wild-type p53 expressing (p53) adeno virus using different MOIs. H) p21 and CIP2A mRNA expression from wild-type (WT) and p53-/- HCT-116 cells treated with 0,2μg/μl of doxorubicin for 0 (ctrl), 24 or 48h. Shown is mean +SD of two experiments analyzed by qBasePLUS 1.0 analysis software. I) Representative western blot analysis of CIP2A expression from tamoxifen-inducible Eu-Myc:p53ER lymphomas treated systemically either with vehicle (−) or with tamoxifen (+). J) Quantitation of CIP2A protein levels from figure (I). CIP2A protein expression normalized to β-actin. Shown is mean +SD of three vehicle and four tamoxifen treated lymphoma lysates. K) Ingenuity transcription factor analysis of CIP2A regulated gene expression changes in HeLa cells. C,D,G) Shown is a representative result of two independent experiments with similar results.

Figure 2. E2F1 upregulates CIP2A expression downstream of inactivated p53

A) MCF-7 cells transfected either with CIP2A promoter (-1802CIP2Aluc) or epidermal growth factor receptor promoter (EGFRluc) luciferase reporter plasmid were treated with Nutlin-3 (2μM) for 24h and luciferase activity was measured. Shown is mean +SD of two independent experiments. B) Putative p53 responsive elements in CIP2A −1802 promoter according to Genomatix and ConTra softwares (tss= transcription start site). C) ChIP was performed with p53 antibody from HCT-116 cells treated with 0,2 μg/ml of doxorubicin for 0 (ctrl), 6, 12 or 24 hours. ChIP DNA was analyzed by real-time PCR with two different set of primers against putative p53 binding sites in CIP2A promoter and as a positive control against p53 binding site in p21 and Mdm2 promoters. Results were analyzed by qBasePLUS 1.0 analysis software and shown is mean +SD from a representative of two independent experiments. D) Western blot detecting p53, p21, phosphorylated (Ser807/Ser811) Rb (ppRb), E2F1 and CIP2A expression from MCF-7 cells treated with 3μM of Nutlin-3 for 8 hours. Irrelevant data has been removed from the original graph. E) p21 and CIP2A mRNA expression analyzed by RT PCR from isogenic wild-type (WT) and p21-/- HCT-116 cells treated with 0,2μg/ml of doxorubicin for 48 h. Shown is mean +SD of two experiments analyzed by qBasePLUS 1.0 analysis software. F) Western blot analysis of CIP2A, p21, E2F1 and β-actin expression from MDA-MB-231 cells transduced either with control (CTL) or p21 expressing adeno vectors (p21) with MOI 80 for 1, 2 or 3 days. Irrelevant data has been removed from the original graph. G) Western blot analysis of CIP2A and E2F1 expression in MCF-7 cells 24 hours after transfection either with scrambled (SCR) or E2F1 siRNA. Irrelevant data has been removed from the original graph. H) CIP2A mRNA expression in doxicycline-inducible wild-type E2F1 expressing Saos-2 cells treated for 24 hours with doxicycline (DOX). I) E2F1 ChIP was performed in Saos-2 cells transfected either with empty CMV vector or with CMV vector expressing E2F1 (E2F1 CMV). Shown is mean +SD of replicates from a representative of two experiments with similar results. J) Schematic model of CIP2A regulation by p53 activity. Inactive molecules and functions are shown in grey. D,F,G) Shown is a representative result of two independent experiments with similar results.

Figure 3. Inhibition of CIP2A expression is a prerequisite for p53-mediated senescence induction

A) SA-beta-gal staining of MCF-7 cells 5 days after transfection either with scrambled (siSCR) or CIP2A siRNA (siCIP2A). B) Western blot analysis of senescence marker DcR2 expression in MCF-7 cells 5 days after transfection either with scrambled (SCR) or with CIP2A (CIP2A) siRNA. C) SA-beta-gal staining of MDA-MB-231 cells 5 days after transfection either with scrambled (siSCR) or CIP2A siRNA (siCIP2A). D) Western blot analysis of CIP2A and p21 expression in either control (AdCTL) or CIP2A (AdCIP2A) transduced (MOI=40) MCF-7 cells treated with Nutlin-3 (3μM) for 3 days. Irrelevant data has been removed from the original graph. E) SA-beta-gal staining of AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with Nutlin-3 (3μM) for 3 days. F) Percentage of SA-β-gal positive and morphologically flattened cells of AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with Nutlin-3 (3μM) for 3 days. Shown is mean +SD from two experiment. ** p=0,0022 by Student’s t-test. G) RT-PCR analysis of cellular senescence associated p53-regulated genes IFNG, IRF5, IGFBP5, RBL2 and SOD2 from AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with Nutlin-3 for 3 days. Shown is log2 fold change +SEM of two replicates from a representative experiment of (E). H) Western blot analysis of CIP2A, p21 and E2F1 expression in MCF-7 cells treated with Doxorubicin (Doxo) with indicated concentrations. I) SA-beta-gal staining of AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with doxorubicin (2uM) for 3 days. J) Percentage of SA-β-gal positive and morphologically flattened cells of AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with doxorubicin (2μM) for 3 days. Shown is mean +SD of three replicates from a representative experiment in (I) . ** p=0,0082 by Student’s t-test. A,B,C,D,E,H,I) Shown is a representative result of at least two independent experiments with similar results.

Figure 4. Positive feedback loop between CIP2A and E2F1 functions as a barrier for p21-mediated senescence induction

A) Western blot analysis of CIP2A, p53, p21, phosphorylated (Ser807/Ser811) Rb (ppRb) and E2F1 expression in either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=40) MCF-7 cells treated with Nutlin-3 (3μM) for 8 hours. Irrelevant data has been removed from the original graph. B) RT-PCR analysis of mRNA-expression of CIP2A and E2F1 from either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=40) MCF-7 cells (MOI=40) treated with Nutlin-3 (3μM) for 8 hours. Shown is mean +SEM of two independent experiments. C) Western blot analysis of pS364-E2F1, E2F1 and CIP2A expression in MCF-7 either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=40) MCF-7 cells. D) Western blot analysis of CIP2A, p21 and pS364-E2F1 expression in AdCTL or AdCIP2A transduced (MOI=40) MCF-7 cells treated with Nutlin-3 for 24 hours. E) Western blot analysis of p21, CIP2A and E2F1 expression in either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=40) MCF-7 cells treated with Nutlin-3 (3μM) for 24 hours. Irrelevant data has been removed from the original graph. F) Western blot analysis of B55α, pS364-E2F1, E2F1 and CIP2A expression in either scrambled (SCR), B55α or B56β siRNA transfected MCF-7 cells. G) Western blot analysis of p21, E2F1 and CIP2A expression in either scrambled (SCR), B55α or B56β siRNA transfected MCF-7 cells treated with Nutlin-3 (3μM) for 24 hours. H) SA-beta-gal staining of either non-targeting shRNA (shNTC) or E2F1shRNA (shE2F1) stable expressing MCF-7 cells treated with Nutlin-3 (2μM) for 3 days. I) Percentage of SA-β-gal stained and morphologically flattened cells in either shNTC or shE2F1 expressing MCF-7 cells treated with Nutlin-3 (2μM) for 3 days. Shown is mean of replicates +SEM from one representative experiment. ** (shNTC control vs. shNTC N3) p=0,0019, ** (shNTC control vs. shE2F1 control) p=0,0032 and n.s. p=0,1358 by Student’s t-test. J) SA-beta-gal staining of MDA-MB-231 cells 3 days after transduction with combination of indicated adeno viruses. AdCIP2A and AdCTL were transduced at MOI=80 and Adp21 and AdCTL at MOI=150. K) Percentage of SA-β-gal positive and morphologically flattened MDA-MB-231 cells 3 days after transduction with combination of indicated adenoviruses (I). Shown is mean +SEM from a representative experiment. ** p=0,0021 by Student’s t-test. A,C,D,E,F,G,H,J) Shown is a representative result of at least two independent experiments with similar results.

Figure 5. Inhibition of CIP2A inhibits growth and induces senescence in mouse embryonic fibroblasts

A) Western blot analysis of CIP2A expression in MEFs isolated from wild-type (WT) and CIP2A genetrap hypomorph (CIP2A^HOZ) mouse embryos. B) Growth curve presenting proliferation capacity of WT and CIP2A^HOZ MEFs. MEFs from 3 different WT or CIP2A^HOZ embryos were cultured for 46 days. Two CIP2A^HOZ MEF colonies ceased to proliferate after first passage and therefore their flat curves overlap in the graph. C) SA-beta-gal staining of wild-type (WT) and CIP2A^HOZ MEFs at passage 4. Shown is a representative of two independent experiments. D) Percentage of SA-β-gal stained WT and CIP2A^HOZ MEFs treated with Nutlin-3 (10μM) for 3 days. Shown is mean +SEM of two independent experiments. P values by Student’s t-test. E) Western blot analysis of CIP2A and E2F1 expression in either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=50) wild-type MEFs. Shown is a representative result of two independent experiments.

Figure 6. CIP2A inactivation induces senescence and growth arrest, and restricts tumorigenesis in a breast cancer mouse model

A) RT-PCR analysis of CIP2A mRNA expression from parental MMTVneu (neu/WT) and MMTVneu x CIP2A^HOZ (neu/HOZ) normal mammary glands and tumors. Shown is mean +SEM of mammary glands from 6 neu/WT and 8 neu/HOZ mice and 25 tumors from 9 neu/WT and 14 tumors from 10 neu/HOZ mice. P values by Mann-Whitney test. B) Representative Ki67 immunohistochemistry staining from 5 neu/WT and 4 neu/HOZ macroscopic tumor-free mouse mammary glands at the time of tumor appearance. C) Quantitation of Ki67 staining in (B). Shown is mean +SEM of Ki67 positive cells in a field at 20x magnification. p<0,0001 by Mann-Whitney test. D) Number of mammary gland tumors per mouse in neu/WT and neu/HOZ mice. Tumors were counted when mice were sacrificed due to 20 mm size of the largest tumor. Shown is mean +SEM in 9 neu/WT and 10 neu/HOZ mouse. p=0,0220 by Student’s t-test. E) Tumor growth was followed from the day of tumor appearance to the day when the mice had to be sacrificed due to 20 mm size of the largest tumor. Shown is tumor growth (days) of 6 neu/WT and 7 neu/HOZ mice. p=0,0030 by log-rank test. F) RT-PCR analysis of senescence markers from neu/WT and neu/HOZ mammary gland tumors at the time of tumor appearance. Shown is mean +SEM from two neu/WT and two neu/HOZ tumors.G) Representative DcR2 immunohistochemistry staining from 7 neu/WT and 3 neu/HOZ mammary gland tumors at the time of tumor appearance. H) Representative SA-beta-gal staining from isolated neu/WT and neu/HOZ mammary gland tumor cells after 3 days in culture. Experiment was performed twice with cells isolated from 2 different neu/WT and 2 different neu/HOZ mammary gland tumors with similar results. I) Representative western blot analysis of CIP2A and E2F1 expression in neu/WT and neu/HOZ mammary gland tumors isolated at the time of tumor appearance. J) Quantitation of E2F1 protein levels from figure (I). E2F1 protein expression normalized to β-actin. Shown is mean +SEM of 9 neu/WT and 6 neu/HOZ tumor lysates.

Figure 7. CIP2A confers resistance of human breast tumors to senescence-inducing chemotherapy

A) CIP2A expression in human breast cancer tumors in FinHer study. CIP2A is expressed in 79% of breast tumors and correlates with high p53 immunopositivity and with other poor prognostic factors. P values by chi-squared test, except for Ki-67 and tumor diameter Kruskal-Wallis test was used. B) CIP2A expression significantly correlates with survival of patients with HER2-negative tumors. CIP2A− = CIP2A negative tumor, CIP2A+ = moderately CIP2A positive tumor, CIP2A++ = high CIP2A expressing tumor. P=0,011 by log-rank test. C) Stratification scheme of patients with HER2-negative tumors to therapies including either vinorelbine followed by FEC (n=340) or docetaxel followed by FEC (n=343). D) CIP2A overexpression significantly associates with poor survival of vinorelbine + FEC treated HER2-negative patients. P=0,019 by log-rank test. E) SA-beta-gal staining of MCF-7 cells treated with vinorelbine (VRB, 30nM) for 5 days. F) Western blot analysis of E2F1 and CIP2A expression in MCF-7 cells treated with vinorelbine (VRB, 20nM and 30nM) for 24 hours. G) RT-PCR analysis of p53, p21, E2F1 and CIP2A mRNA expression in MCF-7 cells treated with vinorelbine (VRB, 20nM and 30nM). Shown is mean +SEM of two independent experiments. H) Western blot analysis of E2F1 and CIP2A expression in either scrambled (SCR) or CIP2A siRNA transfected MCF-7 cells treated with vinorelbine (VRB, 20nM and 30nM) for 12 hours. Quantitation of E2F1 expression normalized to β-actin expression is shown below the graph. I) Western blot analysis of E2F1 in either control (AdCTL) or CIP2A (AdCIP2A) adenovirus transduced (MOI=40) MCF-7 cells treated with vinorelbine (VRB, 10nM and 20nM) for 12 hours. Quantitation of E2F1 expression normalized to β-actin expression is shown below the graph. (E, F,H,I) Shown is representative result of at least two experiments with identical results. J) Schematic presentation of the positive feedback loop between E2F1 and CIP2A in regulation of cellular senescence sensitivity downstream of p53. Inactive molecules and functions are shown in grey. In non-transformed cells (left panel), either oncogene-or chemotherapy-induced p53 activity inhibits E2F1 expression resulting in subsequent inhibition of CIP2A expression. CIP2A inhibition further inhibits E2F1 protein expression by post-translational mechanism involving PP2A. Loss of E2F1-CIP2A positive feedback activity provokes cellular senescence and hence tumor suppression. In tumorigenic cells (right panel) in which p53 activity is inhibited either by mutations or by enhanced proteolytic degradation, E2F1-CIP2A positive feedback loop is active, resulting in inhibition of senescence induction and hence tumor progression. Importantly, in addition to p53 inactivation, activity of CIP2A-E2F1 feedback may be stimulated by ETS1 and MYC that enhance CIP2A expression.