Interaction Between P53 Mutation and a Somatic HDMX Biomarker Better Defines Metastatic Potential in Breast Cancer

Published OnlineFirst February 3, 2015; DOI: 10.1158/0008-5472.CAN-14-2637

Cancer Molecular and Cellular Pathobiology Research

Interaction between p53 Mutation and a Somatic HDMX Biomarker Better Deﬁnes Metastatic Potential in Breast Cancer Anna M. Grawenda1, Elen K. Møller2,3, Suzanne Lam4, Emmanouela Repapi1, Amina F.A.S. Teunisse4, Grethe I.G. Alnæs2, Anne-Lise Børresen-Dale2,3, Vessela N. Kristensen2,3,5, Colin R. Goding1, Aart G. Jochemsen4, Hege Edvardsen2, and Gareth L. Bond1

Abstract

TP53 gene mutation is associated with poor prognosis in utility that was comparable with a microarray-based prognos- breast cancer, but additional biomarkers that can further refine tic assay. Unexpectedly, the utility tracked independently of the impact of the p53 pathway are needed to achieve clinical DNA-damaging treatments and instead with different tumor utility. In this study, we evaluated a role for the HDMX-S/FL metastasis potential. Finally, we obtained evidence that this ratio as one such biomarker, based on its association with biomarker pair might identify patients who could benefitfrom other suppressor mutations that confer worse prognosis in anti-HDM2 strategies to impede metastatic progression. Taken sarcomas, another type of cancer that is surveilled by p53. We together, our work offers a p53 pathway marker, which both found that HDMX-S/FL ratio interacted with p53 mutational refines our understanding of the impact of p53 activity on status to significantly improve prognostic capability in patients prognosis and harbors potential utility as a clinical tool. Cancer with breast cancer. This biomarker pair offered prognostic Res; 75(4); 698–708. 2015 AACR.

Introduction steps of metastatic progression (4, 5). Given p530s roles in controlling the key cellular processes associated with the preven- The tumor suppressor p53, a central node of the cellular stress tion of malignant transformation and tumor progression, it is not response pathway, regulates transcriptional programs important 0 surprising that the loss of p53 s function is common in many in suppressing tumor formation and progression, and the cellular cancers. In approximately 50% of all tumors, wt p53 activity is lost response to certain therapies. The role of p53 in activating apo- by mutation of the TP53 gene (6, 7). The majority of these ptosis, cell-cycle arrest, and the DNA damage response is well mutations are point mutations leading to single amino acid established, associating not only with p530s tumor-suppressing substitutions and resulting in the expression of a mutant p53 properties, but also the cellular response to DNA damage-induc- protein. Mutant p53 can have dominant-negative effects over wt ing cancer therapies (1–3). In recent years, however, the less well p53, but can also acquire oncogenic gain of functions (5, 8). understood role of p53 in controlling cell migration and invasion Many attempts have been made to translate our vast knowledge has emerged, whereby both wild-type (wt) and mutant p53 are of the p53 tumor suppressor into personalization strategies and involved in processes of cell migration, cellular adhesion, cyto- targeted therapies to improve patient survival. One such approach skeletal organization, and angiogenesis, which are important 0 is to use p53 s high rate of mutation as a prognostic biomarker for overall survival, and a predictive marker for response to DNA damaging therapies, such as radiotherapy and many standard 1 fi Ludwig Institute for Cancer Research, University of Oxford, Nuf eld – Department of Clinical Medicine, Oxford, United Kingdom. 2Depart- chemotherapeutics (8 10). Indeed, multiple studies have dem- ment of Genetics, Institute for Cancer Research, Oslo University Hos- onstrated that patients with mutant p53 in their cancers do have 3 pital Radiumhospitalet, Oslo, Norway. KG Jebsen Center for Breast poorer outcomes (11). To date, breast cancer is one of the cancers Cancer Research, Institute for Clinical Medicine, Faculty of Medicine, TP53 fi University of Oslo, Oslo, Norway. 4Department of Molecular Cell Biol- wherein mutational status demonstrates the most signi - ogy, Leiden University Medical Center, Leiden, the Netherlands. cant prognostic impact (12). For example, in a study of 1,794 5Department of Clinical Molecular Biology (EpiGen), Medical Division, European patients with breast cancer, it was noted that the Akershus University Hospital, Lørenskog, Norway. presence of p53 mutations in tumors conferred an increased Note: Supplementary data for this article are available at Cancer Research relative risk (RR) of tumor-related death of 2.27 (13). The prog- Online (http://cancerres.aacrjournals.org/). nostic value of p53 mutations was shown to be independent of Corresponding Author: Gareth L. Bond, University of Oxford, The Ludwig other known prognostic factors, such as tumor size, node status, Institute for Cancer Research, Old Road Campus Research Building, Roosevelt and hormone-receptor status. Drive, Oxford OX3 7DQ, United Kingdom. Phone: 44-1865-617497; Fax: 44-1865- Although very significant, the prognostic value of TP53 muta- 617515; E-mail: [email protected] tional status in isolation is too small to dramatically affect clinical doi: 10.1158/0008-5472.CAN-14-2637 decisions for breast cancer or other cancers (14). One key reason is 2015 American Association for Cancer Research. clearly the fact that there are many ways in which a cancer cell can

698 Cancer Res; 75(4) February 15, 2015

Biomarkers in the p53 Pathway and Metastatic Breast Cancer

inhibit p530s activity and still retain a wt gene. These include Tumor material was obtained before adjuvant radio- and/or common mutations of crucial upstream or downstream pathway chemotherapeutics were administered. The ULL cohort consists genes that also result in attenuation of p53-mediated tumor of 78 women with primary breast cancer recruited at the Ulleval suppression (8, 15). Thus, additional biomarkers that can identify University Hospital between 1990 and 1994, and was first cancers with wt TP53, but attenuated p53 signaling, will be described by Bukholm and colleagues (25). All patients were required to increase the prognostic value and maximize clinical treated according to Norwegian national guidelines at the time of utility. diagnosis. Patients receiving adjuvant systemic therapy were given A crucial upstream pathway gene is HDMX (MDM4). The nine courses of CMF (cyclophosphamide, methotrexate, 5-fluo- HDMX protein can bind to the N-terminal transactivation rouracil) and/or tamoxifen for 2 years. Dosage of radiation given domain of p53 and thereby suppresses its transactivating func- as adjuvant treatment was dependent on indication; after tion (16). In addition, it has been shown that HDMX can breast conserving therapy, the mammary gland was given 50 Gy modulate the translocation of p53 by HDM2 from the nucleus (2 Gy 25). The MicMa breast cancer cohort consists of 112 to the cytoplasm (16) and can stimulate HDM2-mediated women with early-stage breast cancer from the Oslo MicroMe- ubiquitination and degradation of p53 (17). The crucial role tastasis Project, and was first described by Wiedswang and col- of HDMX in the regulation of p53 is further highlighted by the leagues (26). Routine selection of patients to adjuvant treatment fact that germline inactivation of mdm4 in mice leads to was based upon prevailing National Guidelines, where postmen- embryonic lethality through an increased activity of p53 during opausal hormone receptor (HR)–positive patients received the early stages of development (18–21). Importantly, this tamoxifen only, postmenopausal HR-negative patients received phenotype is completely rescued by concomitant inactivation CMF and premenopausal patients, if HR positive, received CMF of the p53 gene (18–21). Recently, it was demonstrated that a followed by tamoxifen. Five patients received high-dose chemo- biomarker for HDMX expression (HDMX-S alternatively spliced therapy and another five, preoperative chemotherapy due to large transcript levels compared with HDMX full-length transcript tumor size. After completing primary therapy, the patients were levels, HDMX-S/FL ratio) associates with multiple common followed at 6 to 12-month intervals. somatic genetic lesions connected with p53 inhibition in cell line panels and sarcomas. The somatic lesions included TP53 Gene expression mutation and HDM2 overexpression, a key negative regulator The RNA from patients with breast cancer was extracted from of p53 (22). Specifically, cancer cell lines and sarcomas with a primary tumors using the TRIzol reagent (Invitrogen) from high HDMX-S/FL ratio associated with both lower levels of fresh frozen tumor material. The RNA from breast cancer cell HDMX protein and an enrichment of cell lines or tumors with lines was isolated using the SV Total RNA Isolation System an attenuated p53 pathway, either by direct TP53 gene muta- (Promega) according to the manufacturer's protocol. The RNA tion or overexpression of HDM2, a key inhibitor of p53. A extraction was followed by cDNA synthesis following stan- model was proposed that higher HDMX-S/FL ratios, and there- dard protocols. The qRT-PCR amplification was performed fore lower HDMX protein levels, can arise in cancer cells that in triplicate duplex reactions according to the manufacturer's already have inhibited p53 signaling through alterations of recommendations using FAM-labeled TaqMan probes, MDM4- other key p53 pathway genes (22). Consistent with this model, FL (Hs00967241_m1; exon 5–6), and custom designed probes patients, whose sarcomas contained higher HDMX-S/FL ratios for MDM4-S, HDM2-P1, and HDM2-P2: metastasized faster and had poorer survival rates. Families who inherit an attenuated p53 stress response by ways MDM4-S_F: 50-GCCCTCTCTATGATATGCTAAGAAAGAATC-30; of a mutant TP53 in their germlines, develop tumors at an MDM4-S_R: 50-TTCTGTAGTTCTTTTTCTGGAAGTGGAA-30; alarmingly high rate (23, 24). The two most frequent tumor types MDM4-S_M FAM: 50-CTGCACTTTGCTGTAGTAGC-30), developed in these families are sarcomas and breast cancer. Thus, HDM2-P1_F: 50-GACTCCAAGCGCGAAAACC-30; a biomarker such as the HDMX-S/FL ratio, which further defines HDM2-P1_R: 50-CACCATCAGTAGGTACAGACATGTT-30; p53 pathway attenuation and prognoses in sarcoma, is a good HDM2-P1_M FAM: 50-CACATTTGCCTGCTCCTC-30, candidate prognostic marker for breast cancer. In this study, we HDM2-P2_F: 50-GGACGCACGCCACTTTT-30; provide supportive evidence for this hypothesis, and demonstrate HDM2-P2_R: 50-CACCATCAGTAGGTACAGACATGTT-30; that these two p53 pathway biomarkers could offer similar HDM2-P2_M FAM: 50-CTGATCCAGGCAAATGT-30). prognostic utility for breast cancer survival as microarray-based molecular subtyping. Unexpectedly, we demonstrate that the Relative gene expression was normalized according to expres- prognostic utility is independent of DNA-damaging treatments, sion of GAPDH measured in the same reaction with VIC-labeled but due to differential metastatic potentials of patients, thereby TaqMan probe (4326317E; Applied Biosystems). The qRT-PCR offering further support for an important role of the p53 pathway with breast cancer cell lines cDNA was performed with the Applied in metastasis. Finally, we provide evidence that the p53 pathway Biosystems 7500 detector. The qRT-PCR with breast cancer biomarkers identify patients for whom anti-MDM2 agents could patients' cDNA was performed with the Applied Biosystems serve as metastatic preventative therapies. 7900HT detector where a Human Breast Total RNA (Ambion) was used as a reference to generate a standard curve. The Sequence Detection Systems Software v2.3 was used to calculate the Materials and Methods amount of transcript expressed for each sample from the standard Patient material curve. Clinicopathologic data of 190 patients with breast cancer from Expression profiling of 50 classifier genes and 5 control genes Oslo MicroMetastasis Project (MicMa) and Ulleval University (PAM50; ref. 27) defined five molecular subtypes in both the Hospital (ULL) cohorts are shown in Supplementary Table S5. MicMa and ULL cohort. Gene expression was measured, as

www.aacrjournals.org Cancer Res; 75(4) February 15, 2015 699

Grawenda et al.

previously reported, using the Stanford 42K cDNA microarray and tumors were found to have a TP53 mutation. Specifically, 80% of Agilent catalog design whole human genome 4 44K one color all TP53 mutations in both cohorts were missense mutations oligo array for the ULL and MicMa cohorts, respectively (28, 29). located in exon 4, 10% frame-shift deletions, 5% nonsense mutations, 3% splice-site mutations, and 2% were in-frame Cell culture and lentiviral transductions insertions. Breast cancer cell lines were a gift from Mieke Schutte/John We first determined the prognostic value of TP53 muta- Martens (Erasmus MC, Rotterdam, the Netherlands). The origin tional status in both the ULL and MicMa cohorts, whereby we of all breast cancer cell lines and verification of their individual compared the breast cancer survival of patients whose tumors identity has been described previously, including a full descrip- had wt TP53 with those whose tumors had mutant TP53.The tion of the TP53 gene sequencing (30, 31). MCF-10A1, MCF- breast cancer survival of the 19 patients from the ULL cohort 10AT, and MCF-10CA1a were obtained from Dr. F. Miller whose tumors had mutant TP53 was significantly shorter than (Karmanos Cancer Center, Detroit, MI). These cell lines were those 59 patients whose tumors had wt TP53 (P ¼ 0.005, log- authenticated by STR typing within the last 6 months; STR profile ranktest,Fig.1A).Indeed,mutantTP53 tumors associated with was found to match the profiles as shown on the ATCC website. All a 2.8-fold higher RR of tumor-related death, as derived from a cell lines were maintained in DMEM/F-12 medium supplemented Cox multivariate regression analysis adjusted for known breast with 5% horse serum (HS) and antibiotics (all purchased from cancer prognostic factors: pathologic node status and adjuvant Invitrogen), 20 ng/mL epidermal growth factor (Upstate Biotech- systemic therapy (P ¼ 0.015, Table 1). Similarly, the breast nology), 100 mg/mL cholera enterotoxin, 5 mg/mL hydrocorti- cancer survival of the 42 patients from the MicMa cohort with sone, 10 mg/mL insulin (all purchased from Sigma-Aldrich). Cells mutant TP53 was shorter than those 70 patients whose tumors were treated with 10 mmol/L Nutlin-3 (Cayman Chemical) where had mutant TP53, although this failed to reach statistical indicated. The lentiviral shRNA expression vector targeting p53 significance (P ¼ 0.111, log-rank test, Fig. 1A, Table 1). was described previously (32). For lentiviral transductions, cells To begin to explore a potential interaction with TP53 were seeded 24 hours before infection and a multiplicity of mutational status and the HDMX-S/FL ratio, we determined infection of 2.0 was used. Cells were transduced overnight in the HDMX-FL and HDMX-S expression levels by performing qRT- presence of 8 mg/mL polybrene (Sigma-Aldrich). The next day, PCR using TaqMan gene expression assays designed to detect medium was replaced with medium containing 0.5 mg/mL puro- full-length (FL) and splice isoform S transcripts of HDMX.The mycin. Transduced cells were seeded for experimental purposes 3 HDMX-FL probe detects the exon 5–6 boundary of the HDMX to 4 days after transduction. gene, which is present in HDMX-FL transcript and only one out of six alternative transcripts, namely HDMX-A. The HDMX-S – Spheroid invasion assay assay was designed to recognize the exon 5 7 bound- fi The spheroid invasion assay was performed as described pre- ary created by the deletion of exon 6, which is speci ctothe HDMX-S viously (33). Briefly, spheroids were formed from 1,000 cells transcript. Next, we divided patients from both HDMX-S/FL and embedded in a 1:1 mixture of Type I collagen (PureCol; cohorts into two groups of high and low ratios. HDMX-S/FL fi HDMX- Advanced BioMatrix) and methylcellulose (Sigma-Aldrich) onto a High ratio was de ned as tumors with S/FL HDMX-S/FL collagen-coated 96-well flat-bottom plate. Invasion was moni- ratios above the mean levels of found in tored for 2 days and images were quantified by measuring the area all 78 and 112 tumors from ULL and MicMa cohorts, respec- occupied by cells using the ImageJ software. tively. Twenty-two patients from the ULL cohort and 41 patients from the MicMa cohort had tumors with high HDMX-S/FL ratios, whereas 56 patients from the ULL cohort and 71 Statistical analysis patients from MicMa cohort had low HDMX-S/FL ratios. Survival analyses were performed using the Kaplan–Meier Interestingly, and in both cohorts, the TP53 mutational status analysis with log-rank test and the Cox multivariate propor- associated with differential breast cancer survival only in patients tional hazards regression model with the SPSS 21.0 software with low HDMX-S/FL ratios. Specifically, in ULL patients with low (SPSS Inc.; IBM). The Shapiro–Wilk test was used to assess a HDMX-S/FL ratios, patients with wt p53 tumors associated with normal distribution of the sample. The ANOVA test with a an almost 15-fold better survival rate than patients with p53- Bonferroni correction for multiple comparisons, Kruskal–Wallis mutant tumors (P ¼ 0.003 log-rank test, Fig. 1B; RR, 14.8; P ¼ 3.6 test, Mann–Whitney test, and unpaired t test were applied to 10 4, Cox analysis, Table 1). In the MicMa cohort, patients determine the statistical significance in the differences between with wt TP53 tumors associated with better survival rate than the means. The Fisher exact test was used to compare the patients with mutant TP53 tumors (P ¼ 0.027 log-rank test, Fig. differences in frequency distributions. Statistical significance 1B; Table 1). However, in both cohorts, there were no significant was regarded as P < 0.05. differences in survival rates between patients with either wt and mutant TP53 tumors when the tumors also had high HDMX-S/FL Results ratios (Fig. 1C, Table 1). The HDMX-S/FL ratio interacts with TP53 mutational status Another biomarker, estrogen receptor (ER) status, and a to further define breast cancer survival clinical parameter, adjuvant systemic treatment, have been We began to explore whether the HDMX-S/FL ratio can interact suggested to interact with TP53 mutational status to increase with TP53 mutational status to improve the prognostic value in its prognostic value (35). Thus, we next compared the effects of two different breast cancer cohorts, namely 78 patients from these factors on breast cancer survival in our patients and Ulleval University Hospital (28) and 112 patients from the Oslo explored potential interactions of them with the HDMX-S/FL MicroMetastasis Project (26, 34). TP53 mutational status was ratio. To do this, we first combined all patients from both determined in all tumors: 19 (24%) ULL and 42 (37%) MicMa cohorts and defined the high and low HDMX-S/FL ratios, as

700 Cancer Res; 75(4) February 15, 2015 Cancer Research

Biomarkers in the p53 Pathway and Metastatic Breast Cancer

Figure 1. The HDMX-S/FL ratio interacts with TP53 mutational status to further deﬁne breast cancer survival in two patient cohorts. A, Kaplan–Meier plots depicting the breast cancer survival of 78 patients from the ULL cohort (left) and 112 patients from the MicMa cohort (right), each separated in groups based on the tumor TP53 mutational status. B, Kaplan–Meier plots depicting the breast cancer survival of 56 patients from the ULL cohort (left) and 71 patients from the MicMa cohort (right) whose tumors had low HDMX- S/FL ratios. C, Kaplan–Meier plots depicting the breast cancer survival of 22 ULL cohort patients (left) and 41 MicMa patients (right) whose tumors had high HDMX-S/FL ratios. Also noted for each plot are the P values that were derived from a log-rank test.

described above, whereby 62 tumors were noted to contain high The HDMX-S/FL ratio and TP53 mutational status define breast HDMX-S/FL ratios and 128 tumors low. As expected, the TP53 cancer survival in a similar manner to microarray-based mutational status associated with differential breast cancer molecular subtypes survival rates in all 190 patients (P ¼ 0.005, log-rank test, When the 190 patients are stratified into four groups based Supplementary Fig. S1, Supplementary Table S1) and this asso- on the two p53 pathway biomarkers (Fig. 2A), it becomes clear ciation was greater in those 128 patients whose cancer had low that these four groups fall into three different categories of HDMX-S/FL ratios (P ¼ 1.44 10 4, log-rank test, Supplemen- prognoses with the wtTP53-lowHDMX-S/FL having good prog- tary Fig. S1, Supplementary Table S1) and absent in those 62 noses, wtTP53-highHDMX-S/FL and mutTP53-highHDMX-S/FL patients with cancers with high HDMX-S/FL ratios (P ¼ 0.996, having intermediate prognoses, and mutTP53-lowHDMX-S/FL Supplementary Fig. S1 and Supplementary Table S1). Interest- having poor prognoses (P ¼ 0.003, log-rank test, Fig. 2B). As ingly, the HDMX-S/FL ratio was the only factor that significantly expected, those 102 patients without either biomarker for p53 interacted with TP53 mutational status to affect breast cancer pathway attenuation (good prognosis, wtTP53-lowHDMX-S/ survival as measured by the Cox proportional hazards model FL) had the longest survival times compared with the 62 (P ¼ 0.031, Table 2) and no additional interactions with the patients from the intermediate group (wtTP53-highHDMX-S/FL HDMX-S/FL ratio and these factors were noted. and mutTP53-highHDMX-S/FL; P ¼ 0.039, log-rank test,

www.aacrjournals.org Cancer Res; 75(4) February 15, 2015 701

Grawenda et al.

Table 1. Cox proportional regression analysis to predict the risk of tumor- fully determine the molecular subtype for the tumors from related death for patients from individual MicMa and ULL cohorts RNA derived from the tumors of 176 patients using the PAM50 95% CI assay (27). As expected, the five different breast cancer molec- df P RR Lower Upper ular subtypes fell into three different categories of prognoses ULL P ¼ TP53 1 0.015 2.8 1.226 6.618 ( 0.016,log-ranktest,Fig.2E).Interestingly,thedifferences pN 1 0.82 1.1 0.411 3.075 in survival times between the different categories were remark- Adjuvant systemic therapy 1 0.749 0.8 0.301 2.37 ably similar to differences found in the groupings based on ULL; low HDMX-S/FL ratio the two p53 pathway biomarkers. For example, the good 4 TP53 1 3.6 10 14.8 3.372 65.355 prognosis group (Luminal A subtype) had a 3-fold lower RR pN 1 0.108 3.403 0.763 15.174 of tumor-related death compared with the poor groups (Lumi- Adjuvant systemic therapy 1 0.367 0.563 0.162 1.962 þ P ¼ ULL; high HDMX-S/FL ratio nal B, Basal, and ERBB2 ; 0.001, Cox analysis, Fig. 2F). TP53 1 0.851 0.9 0.224 3.438 This association is very similar compared with the 4-fold lower pN 1 0.886 1.1 0.195 6.637 RR noted between the good and poor prognoses groups as Adjuvant systemic therapy 1 0.774 0.8 0.118 4.899 defined by TP53 mutational status and the HDMX-S/FL ratio. MicMa Together, these results suggest that these two p53 pathway TP53 1 0.066 1.8 0.962 3.441 biomarkers could offer a similar prognostic utility for breast pN 1 0.022 2.8 1.156 6.649 Adjuvant systemic therapy 1 0.586 0.3 0.315 1.922 cancer survival as the more complex microarray-based molec- MicMa; low HDMX-S/FL ratio ular subtyping. TP53 1 0.07 2.2 0.939 5.104 pN 1 0.313 1.7 0.598 4.985 The HDMX-S/FL ratio interacts with TP53 mutational status Adjuvant systemic therapy 1 0.484 1.5 0.457 5.216 to further define metastasis-free survival HDMX-S/FL MicMa; high ratio Mutant p53 is known to increase a cancer's ability to both TP53 1 0.796 0.8 0.258 2.829 pN 1 0.023 5.2 1.251 22.016 metastasize and become resistant to DNA-damaging therapies Adjuvant systemic therapy 1 0.055 0.2 0.057 1.029 (38). To begin to assess whether the associations of the p53 Abbreviations: CI, confidence interval; df, degrees of freedom; pN, pathologic pathway biomarkers with differential breast cancer survival are node status. a consequence of differential metastatic progression or differential responses to the chemo- and radiotherapy, we performed the breast cancer survival analyses on patients who did not Fig. 2B) and compared with the 26 patients from poor prog- receive any type of adjuvant treatment or radiotherapy. Of the nosis group (mutTP53-lowHDMX-S/FL; P ¼ 1.44 10 4 log- 181 patients from the ULL and MicMa cohorts for whom the rank test, Fig. 2B). For example, after 5 years, 13% of good treatment status was known, 110 patients had been treated with prognosis (wtTP53-lowHDMX-S/FL) patients had died of breast adjuvant- and/or radiotherapies (Fig. 3A). The remaining 71 cancer, compared with 51% of poor prognosis (mutTP53- patients, who did not receive any form of adjuvant- or radio- lowHDMX-S/FL) patients, and 30% of the intermediate therapies, were, as expected, enriched for node-negative and fi group (wtTP53-highHDMX-S/FL and mutTP53-highHDMX- early-stage cases, but did not signi cantly differ in the p53 S/FL; Fig. 2B). Indeed, the good prognosis group had a 1.9- biomarker grouping (Fig. 3B). Interestingly, when we separated fold lower RR of tumor-related death compared with the the 71 patients into the four groups based on both p53 pathway intermediate group (P ¼ 0.033, Cox analysis, Fig. 2C) and a biomarkers, they still fell into the same three different catego- 4.1-fold lower risk relative to the poor group (P ¼ 4.4 x10 5, ries of prognoses as seen in the breast cancer survival analysis of fi TP53 HDMX-S/FL Cox analysis, Fig. 2C). all patients. Speci cally, the wt -low group had TP53 HDMX-S/FL TP53 To begin to assess the potential prognostic strength of these the best prognoses, the wt -high and mut - HDMX-S/FL two p53 pathway biomarkers, we compared them to the well- high groups had intermediate prognoses, and the TP53 HDMX-S/FL P ¼ utilized breast cancer molecular subtyping. Microarray-based mut -low group had the poorest prognoses ( gene expression profiling has been very successful in develop- 0.017, log-rank test, Fig.3C, Supplementary Table S2). These results ingamolecularclassification system for breast cancer and suggest that the association of these p53 biomarkers with differ- prognostic multigene classifiers (36, 37). Briefly, expression ential times of survival after breast cancer diagnosis is not due to profiling of 50 classifier genes and 5 control genes (PAM50) differential responses to DNA-damaging therapies. is able to define minimally five different subtypes of breast We next explored potential associations of these p53 path- cancer, which are known to have either good (Luminal A), way biomarkers with differential times to metastasis after intermediate (normal-like), or poor prognoses (Luminal B, diagnosis. For the 190 patients from the ULL and MicMa Basal, and ERBB2þ; Fig. 2D; ref. 27). We were able to success- cohorts, we had information on the occurrence of distal metas- tases after diagnosis for 185 patients. When we separated the 185 patients into the four groups based on both p53 pathway Table 2. The interactions of TP53 and the HDMX-S/FL ratio with other known biomarkers, and compared their metastasis-free survival (MFS), prognostic factors are known to alter breast cancer survival they still fell into the same three different categories of prog- P nosis as seen in the breast cancer survival analyses. Specifically, TP53 HDMX-S/FL ratio the wtTP53-lowHDMX-S/FL group had the best prognoses, the TP53 NA 0.031 wtTP53-highHDMX-S/FL and mutTP53-highHDMX-S/FL groups HDMX-S/FL ratio 0.031 NA had intermediate prognoses, and the mutTP53-lowHDMX-S/FL ER status 0.870 0.494 group had the poorest prognoses (P ¼ 0.028, log-rank test, Fig. Adjuvant systemic therapy 0.057 0.066 3D, Supplementary Table S1). These same trends in MFS were

702 Cancer Res; 75(4) February 15, 2015 Cancer Research

Biomarkers in the p53 Pathway and Metastatic Breast Cancer

found for all four groups in those patients that did not receive molecules to inhibit the HDM2-p53 interaction and reactive adjuvant treatments or radiotherapy (P ¼ 0.048, log-rank the pathway to promote tumor clearance (10, 39, 40). Our data test, Fig. 3E, Supplementary Table S2). Together, these observa- would suggest that, together with the TP53 mutational status, tions suggest that the associations of the p53 pathway biomar- the HDMX-S/FL ratio could help identify those patients for kers with differential survival are a consequence of differential whom anti-HDM2 agents could serve as metastatic preventative metastatic progression, and not differential responses to the therapies. As a test of this hypothesis, we studied the effects of chemo- and radiotherapy. HDM2 inhibition by the small molecule Nutlin-3, and its effectsonaninvasivephenotypeofabreastcancercellline High HDMX-S/FL ratios associate with p53 inhibition in breast with wt TP53 (MCF-10A). The invasive phenotype was mea- cancer sured using a spheroid invasion assay (33). Importantly, reduc- As mentioned above, it was previously demonstrated that the ing p53 levels using short hairpin RNA promoted invasion of HDMX-S/FL ratio associates with multiple common somatic normal (MCF-10A1), premalignant (MCF-10AT), and metastat- genetic lesions connected with p53 inhibition in cell line panels ic (MCF-10CA1a) breast epithelial cells (Fig. 4A and B), thereby and sarcomas (22). The somatic lesions included TP53 muta- underlining the importance of wt p53 in inhibiting metastatic tion and HDM2 overexpression, a key negative regulator of p53. progression of breast cancer. In contrast, the stabilization and Specifically, cancer cell lines and sarcomas with a high HDMX- activation of p53 by addition of Nutlin-3 inhibited invasion S/FL ratio associated with both lower levels of HDMX protein (Fig. 4C and D). Together, these data lend support to the and an enrichment of cell lines or tumors with an attenuated hypothesis that anti-HDM2 agents could serve as metastatic p53 pathway, either by direct TP53 gene mutation or over- preventative therapies in those patients with breast cancer with expression of HDM2. To gain a better understanding of the shorter MFS times, such as those defined by a wt TP53 gene and molecular underpinnings of these highly significant asso- high HDMX-S/FL ratios. ciations of the HDMX-S/FL ratio and TP53 mutational status in the metastatic progression of breast cancer, we explored if similar associations could be noted in the breast tumors of Discussion the ULL and MicMa cohorts, as well as in a breast cancer cell There is great heterogeneity between individuals in their panel consisting of 36 well-characterized cell lines, wherein it cancer risk, progression, and responses to therapy. This hetero- had previously been shown that the HDMX-S/FL ratio was a geneity is a major obstacle in designing uniformly effective biomarker for overall HDMX protein levels (22). prevention, screening and treatment strategies, and motivates HDM2 levels in tumors were determined through qRT-PCR the large effort to personalize them. The utilization of multiple measurements of mRNA transcripts from both HDM2 promoters, pathologic and molecular biomarkers, particularly HR status, namely the P1 promoter and the p53-responsive P2 promoter. has improved breast cancer treatment management and sur- The average levels of HDM2-P1 transcript were 0.062 and ranged vival. However, the vast majority of patients have yet to benefit from 0.06 to 1.8, whereby 113 tumors had HDM2-P1 below from the existing biomarkers and still either succumb to their average and 76 tumors had HDM2-P1 above. The average levels of disease or are overtreated, underlining the need for additional HDM2-P2 transcript in wt TP53 tumors were 1.08 and ranged prognostic and predictive biomarkers (36). A potential prog- from 0.1 to 15.5, whereby 98 tumors had HDM2-P1 below nostic and/or predictive breast cancer biomarker is certainly the average and 32 tumors had HDM2-P1 above average. The HDM2 mutational status of the TP53 tumor-suppressor gene. As men- levels of the cell line panel had been previously reported (32). tioned above, the prognostic value of TP53 mutational status With these data and together with the TP53 mutational status of alone is too small to dramatically affect clinical decisions for both tumors (Supplementary Table S3) and cell lines (Supple- breast cancer or other cancers (14). Here, we demonstrated in mentary Table S4), we were able to determine that like in sarco- two different breast cancer cohorts that an additional p53 mas, higher levels of HDMX splicing associate with biomarkers for pathway biomarker, the HDMX-S/FL ratio, can interact with p53 pathway attenuation in breast tumors and breast cancer– TP53 mutational status to further identify patients with signif- derived cell lines. Specifically, we observed that the 62 tumors icantly different prognoses. We provided evidence that the with high HDMX-S/FL ratios were significantly enriched for TP53 associations are independent of DNA-damaging treatments, mutation or high expression of HDM2 in wt TP53 tumors. but due to differential metastatic potentials of patients. Specifically, 81% of tumors with high HDMX-S/FL ratios had Indeed, there is a growing body of evidence that the p53 either mutant TP53 or above-average HDM2 transcripts levels, pathway can play key roles in suppressing metastatic progres- whereas only 66% of tumors with low HDMX-S/FL ratios had sion (41–43). For example, p53 can activate E-cadherin, the key mutant TP53 or above-average levels of either HDM2 tran- cell adhesion mediator and suppressor of EMT-dependent scripts (P ¼ 0.0359, Fisher exact test, Table 3). Interestingly, we cancer cell invasion, through the facilitation of HDM2-medi- observed the similar enrichment of p53 pathway attenuation ated ubiquitination and the subsequent degradation of proin- biomarkers in the 18 breast cancer cell lines with high HDMX- vasive Zinc-finger transcription factor SLUG, a transcriptional S/FL ratios. Specifically, 100% of the cell lines with high HDMX- suppressor of E-cadherin (41). S/FL ratios had either mutant TP53 or wt TP53 and the highest It was previously demonstrated that the HDMX-S/FL ratio levels of HDM2 (the top 4), in contrast with only 78% in those associates with multiple common somatic genetic lesions cell lines that had low HDMX-S/FL ratios (P ¼ 0.033, Fisher exact connected with p53 inhibition in cell line panels and sarco- test, Table 3). mas (22). The somatic lesions included TP53 mutation and Overexpression of HDM2 and subsequent attenuation of p53 HDM2 overexpression, a key negative regulator of p53. In this signaling in p53 wt cancers is frequently observed. These report, we offer evidence that like in sarcomas, higher HDMX- observations have motivated a large effort to develop small S/FL ratios associate with these biomarkers for p53 pathway

www.aacrjournals.org Cancer Res; 75(4) February 15, 2015 703

Grawenda et al.

704 Cancer Res; 75(4) February 15, 2015 Cancer Research

Biomarkers in the p53 Pathway and Metastatic Breast Cancer

Figure 3. The HDMX-S/FL ratio interacts with TP53 mutational status to further define MFS. A, pie chart depicting the frequency distribution of adjuvant treatments and radiotherapy of the 181 patients from MicMa and ULL cohorts. B, bar graphs depicting the frequency distributions of node status, stage, and p53 biomarkers in the 110 patients who received adjuvant treatment and/or radiotherapy, and the 71 patients who did not. C, Kaplan–Meier plot depicting the breast cancer survival of the 71 patients from the ULL and MicMa cohorts that did not receive any type of adjuvant treatment or radiotherapy and that were stratified into the four groups based on p53 biomarkers. D and E, Kaplan–Meier plots depicting the MFS of the 185 from the ULL and MicMa cohorts (D), and the 68 patients that did not receive any type of adjuvant treatment or radiotherapy (E), both stratified into four groups based on the p53 biomarkers. The P values noted are derived from a log-rank test.

attenuation in breast tumors and breast cancer–derived cell function for mutant p53 that results in more aggressive/metastatic lines, namely TP53 mutation and HDM2 overexpression. cancers (4, 44). The most well-described mechanism involves the Consistent with this, we observed that patients with wt TP53 inhibition of the p53 family member, p63. Speciﬁcally, it is well and low HDMX-S/FL ratios associated with the longest overall known that loss of p63 expression associates with metastatic and MFS times in both cohorts. Together, these data support a phenotypes in a number of tumors, and, indeed, p63 expression model, whereby tumors with wt TP53 and low HDMX-S/FL serves as a marker for noninvasive epithelial tumors (45). Sub- ratios, and therefore lower levels of p53 inhibitors like sequently, it has been shown that mutant p53 can directly inhibit HDM2, will have retained greater p53 pathway-dependent p63, resulting in lower expression levels of p63 target genes, suppression of metastasis, resulting in better outcomes for including antimetastatic genes, such as the miRNA regulator the patients. Dicer, SHARP1, and cyclin G2, which both oppose TGF- Intriguingly, our observations suggest that the p53 attenuation b–mediated metastasis (46–48). It is therefore reasonable to associated with higher HDMX-S/FL ratios could also be relevant to hypothesize that our observation that patients with mutant patients with breast cancer whose tumors contain mutant p53. p53 and low HDMX-S/FL ratios have the worst prognoses is due Indeed, an even larger body of literature has described a gain-of- to the fact that the gain-of-function activity of mutant p53 is less

Figure 2. The HDMX-S/FL ratio and TP53 mutational statuses deﬁne breast cancer survival in a similar manner to microarray-based molecular subtypes. A, schematic representation of the combinations of p53 biomarkers. B, Kaplan–Meier plot depicting the breast cancer survival of 190 patients divided into four groups depending on the TP53 mutational status and the HDMX-S/FL ratios, namely wt TP53 with low HDMX-S/FL ratio (n ¼ 102), wt TP53 with high HDMX-S/FL ratio (n ¼ 27), mutant TP53 with low HDMX-S/FL ratio (n ¼ 26), and mutant TP53 with high HDMX-S/FL ratio (n ¼ 35). C, predicted survival curves for patients with different prognoses associated with TP53 mutational status and different HDMX-S/FL ratios derived from a Cox multivariate regression analysis that was adjusted to the known prognostic factors: pathologic lymph node status and adjuvant therapy treatment. Also noted is the calculated RR. D, schematic representation of the molecular subtypes and their gene expression signatures. E, Kaplan–Meier plot depicting the breast cancer survival of 176 patients divided into four molecular subtypes based on microarray analysis using the PAM50 assay. The P value noted is derived from a log-rank test. F, predicted survival curves for the different molecular subtypes derived from a Cox multivariate regression analysis that was adjusted to the independent prognostic factors: pathologic lymph node status and adjuvant therapy treatment (n ¼ 156).

www.aacrjournals.org Cancer Res; 75(4) February 15, 2015 705

Grawenda et al.

Table 3. Higher HDMX-S/FL ratios associate with the p53 pathway alterations in breast cancer tumors and cell lines Breast cancer tumors Breast cancer cell lines Low HDMX-S/FL High HDMX-S/FL Low HDMX-S/FL High HDMX-S/FL ratio (n ¼ 129) ratio (n ¼ 62) ratio (n ¼ 18) ratio (n ¼ 18) Mutant TP53 or wt TP53-high HDM2, % 66 81 78 100 wt TP53-low HDM2, % 34 19 22 0 Fisher exact test P value 0.0359 0.033

attenuated by the associated higher expression of p53 inhibitors, tumors. However, it is likely that p53 pathway biomarkers could like HDM2. Indeed, further support of this hypothesis is provided contribute to better prognostication, and decrease the overtreat- by Terzian and colleagues, who clearly demonstrated that a loss of ment of patients with nonaggressive cancers, such as those that mdm2 in murine models resulted in the stabilization of both retain high levels of p53 signaling. Moreover, both wt and mutant mutant p53 and a gain-of-function metastatic phenotype (49). p53 signaling directly affect malignant progression of tumors, However, we cannot rule out the possibility that the observed therefore prognostic biomarkers in the pathway will not only associations are due to the p53-independent effects of higher serve to foresee prognoses, but also offer potential nodes of levels of HDMX (50). intervention. For example, our observations lend support to the Clearly, further elucidation of the p53 pathway biomarkers is hypothesis that anti-HDM2 agents could serve as metastatic needed in additional breast cancer patient cohorts and model preventative therapies in those patients with breast cancer with systems to fully understand how they interact with each other to shorter MFS times, such as those deﬁned by a wt TP53 gene and better predict and create the aggressive phenotype of primary high HDMX-S/FL ratios.

Figure 4. Breast cancer cell invasion is inhibited by p53 activation. A and B, MCF-10A1, MCF-10AT, and MCF-10CA1a expressing sh ctrl or sh p53 were harvested for total protein extraction and p53 expression was analyzed by Western blot using USP7 as loading control (A), or spheroids were formed and embedded into collagen (B). Invasion was determined by calculating the measured invaded area relative to the area on day 0. Results are expressed as mean SD of at least three spheroids and are representative of three independent experiments; , P < 0.05 compared with sh ctrl of the same cell line on the same day. C and D, MCF-10A1, MCF-10AT, and MCF-10CA1a were mock-treated or treated with 10 mmol/L Nutlin-3 and harvested after 24 hours and analyzed for protein expression by Western blot (C), or spheroids were formed and embedded into collagen for 48 hours (D). Results are expressed as mean SD of at least three spheroids and are representative of three independent experiments; , P < 0.05 compared with mock-treated cells of the same cell line.

706 Cancer Res; 75(4) February 15, 2015 Cancer Research

Biomarkers in the p53 Pathway and Metastatic Breast Cancer

Disclosure of Potential Conﬂicts of Interest Acknowledgments No potential conﬂicts of interest were disclosed. The authors wish to thank Magali Olivier for helpful discussions, Mieke Schutte/John Martens for providing the breast cancer cell lines, Fred Miller for the kind gift of the MCF-10A series of cell lines, Pauline Wijers-Koster for STR Authors' Contributions typing, and Drs. Bjørn Naume, Rolf Karesen, and Anita Langerød for providing Conception and design: A.M. Grawenda, A.-L. Børresen-Dale, A.G. Jochemsen, and curating the clinical data. H. Edvardsen, G.L. Bond Development of methodology: A.M. Grawenda, G.L. Bond Grant Support Acquisition of data (provided animals, acquired and managed patients, This study was supported by funding from the Ludwig Institute for Cancer provided facilities, etc.): A.M. Grawenda, S. Lam, A.-L. Børresen-Dale, Research to G.L. Bond and the Worldwide Cancer Research (formerly known V.N. Kristensen, H. Edvardsen as Association for International Cancer Research) to A.G. Jochemsen (grant #09- Analysis and interpretation of data (e.g., statistical analysis, bio- 0073). Results generated in Oslo were with funding from the K.G. Jebsen Centre statistics, computational analysis): A.M. Grawenda, E.K. Møller, E. Repapi, for Breast Cancer Research; E.K. Møller was a PhD fellow of the Norwegian A.-L.Børresen-Dale,V.N.Kristensen, A.G. Jochemsen, H. Edvardsen Cancer Society, project Nr 71248 - PR-2006-0282 to V.N. Kristensen. Writing, review, and/or revision of the manuscript: A.M. Grawenda, E.K. The costs of publication of this article were defrayed in part by the payment Møller, A.-L. Børresen-Dale, V.N. Kristensen, A.G. Jochemsen, H. Edvardsen, of page charges. This article must therefore be hereby marked advertisement G.L. Bond in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.F.A.S. Teunisse, G.I.G. Alnæs, G.L. Bond Received September 10, 2014; revised November 14, 2014; accepted Study supervision: C.R. Goding, A.G. Jochemsen, G.L. Bond December 2, 2014; published OnlineFirst February 3, 2015.

References 1. Ventura A, Kirsch DG, McLaughlin ME, Tuveson DA, Grimm J, Lintault L, 20. Jones SN, Roe AE, Donehower LA, Bradley A. Rescue of embryonic lethality et al. Restoration of p53 function leads to tumour regression in vivo. Nature in Mdm2-deficient mice by absence of p53. Nature 1995;378:206–8. 2007;445:661–5. 21. Migliorini D, Lazzerini Denchi E, Danovi D, Jochemsen A, Capillo M, 2. Xue W, Zender L, Miething C, Dickins RA, Hernando E, Krizhanovsky V, Gobbi A, et al. Mdm4 (Mdmx) regulates p53-induced growth arrest and et al. Senescence and tumour clearance is triggered by p53 restoration in neuronal cell death during early embryonic mouse development. Molec- murine liver carcinomas. Nature 2007;445:656–60. ular and cellular biology 2002;22:5527–38. 3. Riley T, Sontag E, Chen P, Levine A. Transcriptional control of human p53- 22. Lenos K, Grawenda AM, Lodder K, Kuijjer ML, Teunisse AF, Repapi E, et al. regulated genes. Nat Rev Mol Cell Biol 2008;9:402–12. Alternate splicing of the p53 inhibitor HDMX offers a superior prognostic 4. Muller PA, Vousden KH, Norman JC. p53 and its mutants in tumor cell biomarker than p53 mutation in human cancer. Cancer Res 2012;72: migration and invasion. J Cell Biol 2011;192:209–18. 4074–84. 5. Muller PA, Vousden KH. p53 mutations in cancer. Nat Cell Biol 2013; 23. Malkin D, Li FP, Strong LC, Fraumeni JF Jr, Nelson CE, Kim DH, et al. Germ 15:2–8. line p53 mutations in a familial syndrome of breast cancer, sarcomas, and 6. Hainaut P, Hollstein M. p53 and human cancer: the first ten thousand other neoplasms. Science 1990;250:1233–8. mutations. Adv Cancer Res 2000;77:81–137. 24. Li FP, Fraumeni JF Jr, Mulvihill JJ, Blattner WA, Dreyfus MG, Tucker MA, 7. Hollstein M, Sidransky D, Vogelstein B, Harris CC. p53 mutations in et al. A cancer family syndrome in twenty-four kindreds. Cancer Res human cancers. Science 1991;253:49–53. 1988;48:5358–62. 8. Vousden KH, Lu X. Live or let die: the cell's response to p53. Nat Rev Cancer 25. Bukholm IK, Nesland JM, Karesen R, Jacobsen U, Borresen AL. Relationship 2002;2:594–604. between abnormal p53 protein and failure to express p21 protein in 9. Soussi T, Beroud C. Assessing TP53 status in human tumours to evaluate human breast carcinomas. J Pathol 1997;181:140–5. clinical outcome. Nat Rev Cancer 2001;1:233–40. 26. Wiedswang G, Borgen E, Karesen R, Naume B. Detection of isolated tumor 10. Vazquez A, Bond EE, Levine AJ, Bond GL. The genetics of the p53 pathway, cells in BM from breast-cancer patients: significance of anterior and apoptosis and cancer therapy. Nat Rev Drug Discov 2008;7:979–87. posterior iliac crest aspirations and the number of mononuclear cells 11. Freed-Pastor WA, Prives C. Mutant p53: one name, many proteins. Genes analyzed. Cytotherapy 2003;5:40–5. Dev 2012;26:1268–86. 27. Parker JS, Mullins M, Cheang MC, Leung S, Voduc D, Vickery T, et al. 12. Hainaut P, Olivier M, Wiman KG. P53 in the Clinics. New York: Springer; Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin 2013. p. 355. Oncol 2009;27:1160–7. 13. Olivier M, Langerod A, Carrieri P, Bergh J, Klaar S, Eyfjord J, et al. The 28. Langerod A, Zhao H, Borgan O, Nesland JM, Bukholm IR, Ikdahl T, et al. clinical value of somatic TP53 gene mutations in 1,794 patients with breast TP53 mutation status and gene expression profiles are powerful prognostic cancer. Clin Cancer Res 2006;12:1157–67. markers of breast cancer. Breast Cancer Res 2007;9:R30. 14. Petitjean A, Mathe E, Kato S, Ishioka C, Tavtigian SV, Hainaut P, et al. 29. Enerly E, Steinfeld I, Kleivi K, Leivonen SK, Aure MR, Russnes HG, et al. Impact of mutant p53 functional properties on TP53 mutation patterns miRNA-mRNA integrated analysis reveals roles for miRNAs in primary and tumor phenotype: lessons from recent developments in the IARC TP53 breast tumors. PLoS ONE 2011;6:e16915. database. Hum Mutat 2007;28:622–9. 30. Elstrodt F, Hollestelle A, Nagel JH, Gorin M, Wasielewski M, van den 15. Toledo F, Wahl GM. Regulating the p53 pathway: in vitro hypotheses, in vivo Ouweland A, et al. BRCA1 mutation analysis of 41 human breast cancer veritas. Nat Rev Cancer 2006;6:909–23. cell lines reveals three new deleterious mutants. Cancer Res 2006;66: 16. Wade M, Wang YV, Wahl GM. The p53 orchestra: Mdm2 and Mdmx set the 41–5. tone. Trends Cell Biol 2010;20:299–309. 31. Wasielewski M, Elstrodt F, Klijn JG, Berns EM, Schutte M. Thirteen new p53 17. Linares LK, Hengstermann A, Ciechanover A, Muller S, Scheffner M. HdmX gene mutants identified among 41 human breast cancer cell lines. Breast stimulates Hdm2-mediated ubiquitination and degradation of p53. Proc Cancer Res Treat 2006;99:97–101. Natl Acad Sci U S A 2003;100:12009–14. 32. Lam S, Lodder K, Teunisse AF, Rabelink MJ, Schutte M, Jochemsen AG. Role 18. Montes de Oca Luna R, Wagner DS, Lozano G. Rescue of early embryonic of Mdm4 in drug sensitivity of breast cancer cells. Oncogene 2010;29: lethality in mdm2-deficient mice by deletion of p53. Nature 1995;378:203–6. 2415–26. 19. Parant J, Chavez-Reyes A, Little NA, Yan W, Reinke V, Jochemsen AG, et al. 33. Wiercinska E, Naber HP, Pardali E, van der Pluijm G, van Dam H, tenDijke Rescue of embryonic lethality in Mdm4-null mice by loss of Trp53 suggests P. The TGF-beta/Smad pathway induces breast cancer cell invasion through a nonoverlapping pathway with MDM2 to regulate p53. Nat Genet 2001; the up-regulation of matrix metalloproteinase 2 and 9 in a spheroid 29:92–5. invasion model system. Breast Cancer Res Treat 2011;128:657–66.

www.aacrjournals.org Cancer Res; 75(4) February 15, 2015 707

Grawenda et al.

34. Naume B, Zhao X, Synnestvedt M, Borgen E, Russnes HG, Lingjaerde OC, 42. Sankpal NV, Willman MW, Fleming TP, Mayfield JD, Gillanders WE. et al. Presence of bone marrow micrometastasis is associated with different Transcriptional repression of epithelial cell adhesion molecule contributes recurrence risk within molecular subtypes of breast cancer. Mol Oncol to p53 control of breast cancer invasion. Cancer Res 2009;69:753–7. 2007;1:160–71. 43. Kim NH, Kim HS, Li XY, Lee I, Choi HS, Kang SE, et al. A p53/miRNA-34 35. Millar EK, Graham PH, McNeil CM, Browne L, O'Toole SA, Boulghourjian axis regulates Snail1-dependent cancer cell epithelial-mesenchymal tran- A, et al. Prediction of outcome of early ERþ breast cancer is improved using sition. J Cell Biol 2011;195:417–33. a biomarker panel, which includes Ki-67 and p53. Br J Cancer 2011;105: 44. De Craene B, Berx G. Regulatory networks defining EMT during cancer 272–80. initiation and progression. Nat Rev Cancer 2013;13:97–110. 36. Reis-Filho JS, Pusztai L. Gene expression profiling in breast cancer: clas- 45. Melino G. p63 is a suppressor of tumorigenesis and metastasis interacting sification, prognostication, and prediction. Lancet 2011;378:1812–23. with mutant p53. Cell Death Differ 2011;18:1487–99. 37. Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, et al. 46. Adorno M, Cordenonsi M, Montagner M, Dupont S, Wong C, Hann B, et al. Molecular portraits of human breast tumours. Nature 2000;406: A mutant-p53/Smad complex opposes p63 to empower TGFbeta-induced 747–52. metastasis. Cell 2009;137:87–98. 38. Royds JA, Iacopetta B. p53 and disease: when the guardian angel fails. Cell 47. Neilsen PM, Noll JE, Suetani RJ, Schulz RB, Al-Ejeh F, Evdokiou A, et al. Death Differ 2006;13:1017–26. Mutant p53 uses p63 as a molecular chaperone to alter gene expression and 39. Vassilev LT, Vu BT, Graves B, Carvajal D, Podlaski F, Filipovic Z, et al. In vivo induce a pro-invasive secretome. Oncotarget 2011;2:1203–17. activation of the p53 pathway by small-molecule antagonists of MDM2. 48. Su X, Chakravarti D, Cho MS, Liu L, Gi YJ, Lin YL, et al. TAp63 suppresses Science 2004;303:844–48. metastasis through coordinate regulation of Dicer and miRNAs. Nature 40. Senzer N, Nemunaitis J, Nemunaitis M, Lamont J, Gore M, Gabra H, et al. 2010;467:986–90. p53 therapy in a patient with Li-Fraumeni syndrome. Mol Cancer Ther 49. Terzian T, Suh YA, Iwakuma T, Post SM, Neumann M, Lang GA, et al. The 2007;6:1478–82. inherent instability of mutant p53 is alleviated by Mdm2 or p16INK4a loss. 41. Wang SP, Wang WL, Chang YL, Wu CT, Chao YC, Kao SH, et al. p53 controls Genes Dev 2008;22:1337–44. cancer cell invasion by inducing the MDM2-mediated degradation of Slug. 50. Li Q, Lozano G. Molecular pathways: targeting Mdm2 and Mdm4 in cancer Nat Cell Biol 2009;11:694–704. therapy. Clin Cancer Res 2013;19:34–41.

708 Cancer Res; 75(4) February 15, 2015 Cancer Research

Interaction between p53 Mutation and a Somatic HDMX Biomarker Better Defines Metastatic Potential in Breast Cancer

Anna M. Grawenda, Elen K. Møller, Suzanne Lam, et al.

Cancer Res 2015;75:698-708. Published OnlineFirst February 3, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-14-2637

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2021/03/31/0008-5472.CAN-14-2637.DC1

Cited articles This article cites 48 articles, 16 of which you can access for free at: http://cancerres.aacrjournals.org/content/75/4/698.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/75/4/698.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/75/4/698. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.