Published OnlineFirst April 25, 2016; DOI: 10.1158/0008-5472.CAN-15-3174 Cancer Tumor and Stem Cell Biology Research Molecular Insights of Pathways Resulting from Two Common PIK3CA Mutations in Breast Cancer Poornima Bhat-Nakshatri1, Chirayu P. Goswami2, Sunil Badve3, Luca Magnani4, Mathieu Lupien5, and Harikrishna Nakshatri1,6,7 Abstract The PI3K pathway is activated in approximately 70% of kers of response to PI3K inhibitors. Using a variety of phys- breast cancers. PIK3CA gene mutations or amplifications that iologically relevant model systems with defined natural or affect the PI3K p110a subunit account for activation of this knock-in PIK3CA mutations and/or PI3K hyperactivation, we pathway in 20% to 40% of cases, particularly in estrogen show that PIK3CA-E545K mutations (found in 20% of receptor alpha (ERa)-positive breast cancers. AKT family of PIK3CA-mutant breast cancers), but not PIK3CA-H1047R kinases, AKT1–3, are the downstream targets of PI3K and these mutations (found in 55% of PIK3CA-mutant breast cancers), kinases activate ERa. Although several inhibitors of PI3K have preferentially activate AKT1. Our findings argue that AKT1 been developed, none has proven effective in the clinic, partly signaling is needed to respond to estrogen and PI3K inhibitors due to an incomplete understanding of the selective routing of in breast cancer cells with PIK3CA-E545K mutation, but not in PI3K signaling to specific AKT isoforms. Accordingly, we breast cancer cells with other PIK3CA mutations. This study investigated in this study the contribution of specific AKT offers evidence that personalizing treatment of ER-positive isoforms in connecting PI3K activation to ERa signaling, breast cancers to PI3K inhibitor therapy may benefitfroman and we also assessed the utility of using the components of analysis of PIK3CA–E545K–AKT1–estrogen signaling path- PI3K–AKT isoform–ERa signaling axis as predictive biomar- ways. Cancer Res; 76(13); 3989–4001. Ó2016 AACR. Introduction cancers (3, 4). In addition to E2, multiple cofactors and post- translational modifications control ERa activity (4, 5). These The PI3K pathway is a commonly mutated/amplified pathway include pioneer factors such as FOXA1, GATA3, and AP2g that in cancers (1). Activating mutations of the p110a, the catalytic guide ERa binding to the genome, transcriptional coregulators subunit of PI3K, are common in estrogen receptor alpha (ERa)- that influence transcriptional output from ERa, and ERa phos- positive luminal breast cancers, whereas p110a is amplified phorylation that influences transcriptional activity, stability, and frequently in ERa-negative basal-like breast cancers (2). These subcellular distribution (4). Several growth factor receptor–acti- observations suggest a cross-talk between signaling networks vated kinases including AKT intersect with ERa signaling by emanating from mutant PIK3CA and ERa, which impacts breast phosphorylating ERa and altering transcriptional output (6). cancer initiation and/or progression. AKT family of kinases (AKT1–3) are frequently activated down- ERa is a nuclear receptor activated in response to its ligand stream of PI3K. Published work from multiple groups including estradiol (E2), and plays a significant role in >70% of breast ours has shown significant influence of AKT in ERa phosphory- lation, genome-wide binding, E2-dependent mRNA and miRNA expression, and alternative splicing (7–11). We demonstrated 1Department of Surgery, Indiana University School of Medicine, India- 2 distinct prognostic value of nuclear phospho-AKT in ERa-positive napolis, Indiana. Center for Computational Biology and Bioinformat- fi ics, Indiana University School of Medicine, Indianapolis, Indiana. breast cancers (12). However, there are two signi cant gaps in our 3Department of Pathology & Laboratory Medicine, Indiana University understanding of cross-talk between PI3K and ERa signaling, School of Medicine, Indianapolis, Indiana. 4Division of Cancer, Imperial which this study is designed to address. First, the isoform of AKT College, London, United Kingdom. 5The Princess Margaret Cancer Centre, University Health Network; Ontario Institute for Cancer that preferentially engages PI3K with ERa is unknown. Second, it is Research and Department of Medical Biophysics, University of Tor- unknown whether PIK3CA-E545K mutation, which represents onto, Ontario, Canada. 6Department of Biochemistry & Molecular 20% of PIK3CA mutations, and PIK3CA-H1047R mutation, which Biology, Indiana University School of Medicine, Indianapolis, Indiana. represents 55% of PIK3CA mutations in breast cancer (mycancer- 7VA Roudebush Medical Center, Indianapolis, Indiana. genome.org), has similar influence on AKT isoform activation. Note: Supplementary data for this article are available at Cancer Research Exploring these gaps is critical because of recent understanding that Online (http://cancerres.aacrjournals.org/). AKT isoforms are not functionally similar (13, 14). Moreover, Current address for C. P. Goswami: Thomas Jefferson University Hospital, commonly used constitutively active AKT mutants do not discrim- Philadelphia, Pennsylvania. inate the functions of different isoforms of AKT (13). We focused Corresponding Author: Harikrishna Nakshatri, Indiana University School of on AKT1 and AKT2 because AKT3 is relevant for only ERa-negative Medicine, C218C, 980 West Walnut St, Indianapolis, IN 46202. Phone: 317- breast cancers (15, 16). Our results showed that, in general, 278-2238; Fax: 317-274-0396; E-mail: [email protected] PIK3CA-E545K mutation is associated with AKT1 activation, doi: 10.1158/0008-5472.CAN-15-3174 whereas PI3KCA-H1047R mutation with activation of AKT1, AKT2, Ó2016 American Association for Cancer Research. or both. AKT1 is essential for ERa activity, E2 dependency, and www.aacrjournals.org 3989 Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2016 American Association for Cancer Research. Published OnlineFirst April 25, 2016; DOI: 10.1158/0008-5472.CAN-15-3174 Bhat-Nakshatri et al. response to PI3K inhibitors in MCF-7 cells with endogenous noprecipitation (ChIP)-on-chip data and ChIP-seq data from our PIK3CA-E545K mutation. Thus, response of ERa-positive breast previous studies were used to assign ERa-binding sites (8, 22). cancers to PI3K inhibitors may depend on the isoform of AKT activated as a consequence of specific PIK3CA mutation. Analysis of signaling pathways overlap FOXA1-E2 and PBX1-E2 signatures have been described pre- Materials and Methods viously (23). Gene lists for AKT1- and AKT2-dependent E2- induced genes were imported into Oncomine (24). Overlap was Cell lines defined as significant at P value of at least 0.01 and an OR 2. MCF-7, T47-D, LY2, HCC1428, BT-474, SK-BR-3, MDA-MB- Prognostic impact of FOXA1-E2–ERa-AKT1 signature was evalu- 231, MDA-MB-436, MDA-MB-468, UACC812, and ZR-75-1 cells ated using two public databases (25, 26). were purchased from ATCC. 600MPE cell line was a gift from Dr. Paul Spellman (Oregon Health Sciences University, Portland, OR; ref. 17). HCC1428, UACC812, and LY2 cell lines were pur- Results chased within last 1 year and other cell lines have been authen- AKT isoform activity in breast cancer cell lines with endogenous ticated within past 2 years using STR Systems for Cell line iden- PIK3CA aberrations fi ti cation (DNA Diagnosis Center and Genetica DNA Laborato- Recently developed antibodies that recognize activated AKT1 ries,). Human immortalized mammary epithelial cells (HMEC) (AKT1_pS473) and AKT2 (AKT2_pS474) enabled us to reexamine PIK3CA PIK3CA-E545K with targeted replacement of with and whether specific isoforms are activated in response to distinct PIK3CA-H1047R were purchased from Horizon Discovery Limited PI3K aberrations. On the basis of the studies using pan (hTERT-HME1). Drs. Ben Ho Park (Johns Hopkins University, AKT_pS473 antibody, it was suggested that PIK3CA mutation is Baltimore, MD) and Michele Vitolo (University of Maryland, not always associated with AKT activation. For example, MCF-7 PIK3CA College Park, MD) provided MCF10A cells with targeted cells with PIK3CA-E545K mutation were reported to lack consti- PTEN mutants and deletion, respectively (18, 19). Dr. Alex Toker tutive AKT activity (27). In contrast to the results reported using (Harvard Medical School, Boston, MA) provided the parental pan AKT_pS473 antibody, AKT1_pS473 was readily detected in pLKO, AKT1, and AKT2 short hairpin RNA (shRNA) lentivirus MCF-7 cells under serum-deprived and serum-supplemented vectors (20). Supplementary information has additional details of conditions (Fig. 1A). AKT1_pS473 and AKT2_pS474 levels (to lentivirus transfection, siRNAs, and cell proliferation assays. a lesser extent) in MCF-7 cells under serum-treated condition were higher than in 600MPE cell line, which lacks PIK3CA/PTEN Antibodies alterations. AKT2_pS474 was dominant in T47-D cells with Antibodies against AKT1, AKT2, AKT1_pS473, AKT2_pS474, PIK3CA-H1047R mutation as its levels were higher compared and phospho-GSK3a/b (S9/21) were purchased from Cell Sig- with MCF-7 or 600MPE cell lines. Several additional cell lines (12 naling Technology. Antibodies against ERa, GATA3, GSK3a/b, cell lines; LY2 is an anti-estrogen–resistant derivative of MCF-7) FOXA1, cMyc, and cyclin D1 were purchased from Santa Cruz were examined to determine whether PIK3CA mutation, PTEN Biotechnology, whereas AP2g antibody was from Epitomics. mutation and/or HER2 amplification correlates with elevated basal and serum-inducible activated AKT1 or AKT2. All HER2- RNA isolation, microarray,