FKBPL Regulates Estrogen Receptor Signaling and Determines Response to Endocrine Therapy

Published OnlineFirst January 26, 2010; DOI: 10.1158/0008-5472.CAN-09-2515 Published Online First on January 26, 2010 as 10.1158/0008-5472.CAN-09-2515

Therapeutics, Targets, and Chemical Biology Cancer Research FKBPL Regulates Estrogen Receptor Signaling and Determines Response to Endocrine Therapy

Hayley D. McKeen1, Christopher Byrne1, Puthen V. Jithesh2, Christopher Donley1, Andrea Valentine1, Anita Yakkundi1, Martin O'Rourke1, Charles Swanton3,4, Helen O. McCarthy1, David G. Hirst1, and Tracy Robson1

Abstract The HSP90 chaperone and immunophilin FKBPL is an estrogen-responsive gene that interacts with estogen receptor α (ERα) and regulates its levels. In this study, we explored the effects of FKBPL on breast cancer proliferation. Breast cancer cells stably overexpressing FKBPL became dependent on estrogen for their growth and were dramatically more sensitive to the antiestrogens tamoxifen and fulvestrant, whereas FKBPL knockdown reverses this phenotype. FKBPL knockdown also decreased the levels of the cell cycle inhibitor p21WAF1 and increased ERα phosphorylation on Ser118 in response to 17β-estradiol and tamoxifen. In support of the likelihood that these effects explained FKBPL-mediated cell growth inhibition and sensitivity to endocrine therapies, FKBPL expression was correlated with increased overall survival and distant metastasis-free survival in breast cancer patients. Our findings suggest that FKBPL may have prognostic value based on its impact on tumor proliferative capacity and sensitivity to endocrine therapies, which improve outcome. Cancer Res; 70(3); 1090–100. ©2010 AACR.

Introduction ER-positive tumors do not respond to hormone therapy and 30% to 40% of patients who initially respond positively re- The steroid hormone estrogen has a pivotal role in tumor lapse within 5 years. Therefore, other proteins involved in development (1), with increased levels of circulating estrogen ER signaling need to be assessed as potential targets for ther- being linked with initiation and progression of breast cancer apeutic intervention. (2–4). Estrogen receptors (ER) are members of the nuclear We have previously isolated, and partially characterized, receptor superfamily of ligand-activated transcription factors FKBPL (DIR1/Wisp39). We showed that it belongs to the im- that mediate the biological effects of estrogen (5). Ligand munophilin/FKBP protein family as a consequence of the binding causes the receptor to dimerize and activate or re- presence of COOH-terminal tetratricopeptide repeat do- press transcription of ER target genes through recruitment of mains (TPR) and a partial peptidyl-prolyl cis-trans isomerase coregulators (6). Around two thirds of breast tumors express (PPIase) domain (10, 11). The TPR domains of immunophilin ERα and are therefore dependent on estrogen for their proteins are particularly important for binding to the molec- growth. Anticancer drugs such as tamoxifen exert their anti- ular chaperone Hsp90 (12). In support of this association, tumor activity through binding ERα, blocking the interaction FKBPL has been implicated in Hsp90-dependent stabilization between estrogen and its receptor (7), and altering its confor- of newly synthesized p21 by preventing its proteosomal deg- mation, leading to corepressor recruitment to the complex radation (13); also, overexpression of FKBPL inhibits cell cy- (8, 9). This results in inhibition of ERα-regulated genes that cle progression at the G0-G1 phase in leukemic cells (14). promote tumor growth. However, despite the success of this More recently, a role for FKBPL in Hsp90-associated steroid group of agents in the clinic, almost half of the patients with hormone receptor complexes was identified in which FKBPL regulated glucocorticoid receptor (GR) protein levels, transcriptional activation, and cellular localization in prostate Authors' Affiliations: 1School of Pharmacy, McClay Research Centre and 2Centre for Cancer Research and Cell Biology, Queen's University, cancer cells (15). Belfast, Northern Ireland, United Kingdom; 3Cancer Research UK London In breast cancer, immunophilins and other cochaperones Research Institute, London, United Kingdom; and 4Breast Unit, have been shown to influence ER signaling. To maintain re- Department of Medicine, Royal Marsden Hospital, Sutton, United α Kingdom ceptor stability in the absence of ligand, ER exists in a molecular complex with Hsp90, p23, and the immunophilins Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). FKBP52 or Cyp40 (16). The immunophilins competitively H.D. McKeen and C. Byrne contributed equally to this work. bind Hsp90 through TPR domains (12), and cochaperone recruitment is a dynamic process that controls steroid receptor Corresponding Author: Tracy Robson, School of Pharmacy, McClay Research Centre, Queen's University, 97 Lisburn Road, Belfast, Northern activity in a tissue-specific manner. Cyp40 and FKBP52 are Ireland BT9 7BL, United Kingdom. Phone: 44-028-90972360; Fax: 44- overexpressed in breast tumors (17) and are upregulated 028-90247794; E-mail: [email protected]. transcriptionally and posttranscriptionally by estrogen, doi: 10.1158/0008-5472.CAN-09-2515 whereas exposure to the pure estrogen antagonist ICI ©2010 American Association for Cancer Research. 182,780 (fulvestrant) prevents this estrogen-mediated

1090 Cancer Res; 70(3) February 1, 2010

FKBPL Modulates the Estrogen Receptor Signaling Pathway

increase (18). A recent study has found that the FKBP52 gene Mammalian two-hybrid assay. The CheckMate Mamma- is methylated in ER-negative MDA-MB-231 but not in ER- lian Two-Hybrid System (Promega) was used to assess pro- positive MCF7 cells, suggesting that repression of FKBP52 tein interactions. In-frame cloning of full-length FKBPL into may itself affect ER expression (19). Furthermore, Hsp90 the pBIND-GAL4 binding domain plasmid and Hsp90 into and immunophilin ligands have been shown to target ER the herpes simplex viral protein 16 (pACT-VP16) activation for proteosomal degradation (20). The Hsp90 cochaperone domain plasmid were carried out according to the manu- p23 also plays an important role in ER signal transduction facturer's instructions. MCF-7 cells at 60% to 80% con- (21) and regulates the ER target genes pS2 and cathepsin fluency were transfected with either of these plasmids or D, promoting tumor cell adhesion and invasion (22). the pACT-MyoD and pBIND-Id positive control plasmids Hsp90 and its cochaperones clearly play a role in ER sig- along with the firefly luciferase pG5luc plasmid using naling; however, a role for FKBPL in these ER-associated Lipofectamine Plus according to the manufacturer's instruc- complexes has not yet been described. Understanding the tions. After 24 h, cells were analyzed using the Dual-Glo Lu- possible role of this immunophilin in the ERα-Hsp90 com- ciferase Assay System (Promega). Firefly luciferase activity plex might shed further light on ER signaling. This is espe- was normalized to Renilla luciferase activity and presented cially important because the biological mechanisms as relative light units. Three independent experiments were underlying de novo and acquired tamoxifen resistance have performed with triplicates for each experiment. not been fully elucidated. Previous studies have linked endo- Coimmunoprecipitations. MCF7 or 3.1D2 cells were crine therapy resistance with increased activity of epidermal grown to 90% confluency, washed twice with ice-cold PBS, growth factor receptor (EGFR), human epidermal growth fac- and lysed in buffer [20 mmol/L Tris-HCl (pH 7.4), 1% Igepal, tor receptor 2, and phosphatidylinositol 3-kinase pathways; 12 mmol/L sodium deoxycholate, 0.1% SDS, 10 mmol/L sodi- modification of ER phosphorylation status; deregulation of um molybdate, 1 protease inhibitor tablet] for 30 min on ice, cell cycle proteins (23); and inhibition of CDK10 (24) and then pelleted by centrifugation at 13,000 rpm for 5 min at 4° BRCA1 (25). Here, we show that FKBPL is an estrogen-induc- C. One fifth of the lysate was removed to a fresh tube for use ible gene that acts as a cochaperone in ERα/Hsp90 molecular as whole cell lysate positive control for the Western blots. complexes; furthermore, FKBPL levels may be both a prog- The remaining lysates were precleared by incubating with nostic indicator and determinant of response to endocrine prewashed agarose G beads for 1 h at 4°C with rotation therapy. and then lysates were incubated with Hsp90, FKBPL, ERα, or IgG (negative control) antibody–bound Protein G-Sephar- Materials and Methods ose beads (Cancer Research UK) at 4°C overnight. The beads were washed three times in ice-cold lysis buffer and twice in Cell culture and transfections. The ER-positive cell lines ice-cold PBS then resuspended in 2× Laemmli buffer. West- MCF7 and T47D were obtained from Cancer Research UK ern blot analysis was then carried out. and American Type Culture Collection, respectively, and Western blot analysis. Samples were subjected to SDS- maintained as monolayers in DMEM (Invitrogen) supple- PAGE electrophoresis using the XCell Surelock Mini-Cell Sys- mented with 10% FCS. Cell lines were authenticated by tem (Invitrogen), transferred onto nitrocellulose membranes, short tandem repeat (STR) profiling carried out by the blocked for 1 h at room temperature with 1% skim milk suppliers. The ER status of these cell lines was authenti- blocking solution, and probed with FKBPL rabbit polyclonal cated by Western blot to determine ER expression, and (1:2,000; ProteinTech), Hsp90 (1:1,000; BD Transduction Lab- testing revealed that cells were Mycoplasma-free. All ex- oratories), ERα rabbit monoclonal (1:1,000; Millipore), ca- periments were carried out between passages 1 and 15. thepsin D mouse monoclonal (1:1,000; Abcam), phospho- Transient transfections were carried out using either emp- ERα (Ser118) mouse (1:1,000; Cell Signalling Technologies), ty vector pcDNA3.1 (as a control) or pcDNA3.1/FKBPL plas- p21 mouse monoclonal (1:1,000; Upstate), and glyceralde- mids and transfected using Lipofectamine 2000 (Invitrogen) hyde-3-phosphate dehydrogenase (GAPDH) rabbit mono- according to the manufacturer's instructions. Stable transfec- clonal (1:5,000; Sigma) antibodies. Blots were probed with tants of MCF7 cells were generated by transfecting cells with anti-rabbit or anti-mouse IgG horseradish peroxidase–linked pcDNA3.1/FKBPL plasmid using Lipofectamine Plus accord- whole antibody (secondary; 1:5,000; GE Healthcare). Antibody ing to the manufacturer's instructions. Transfected cells were binding was detected using Supersignal West Pico Chemilu- selected using 750 μg/mL G418 (Sigma) and maintained in minescent Substrate (Pierce) according to the manufac- DMEM with 375 μg/mL G418. In studies requiring estrogen turer's instructions. deprivation, cells were maintained in phenol red–free DMEM Clonogenic assay/growth assays. Cells were plated at a (Invitrogen) containing 10% charcoal-stripped FCS (PAA) density of 500 and 1,000 per well in a six-well plate containing for 48 h before beginning the experiments. For knockdown phenol red–free DMEM + 10% charcoal-stripped FCS or analysis, siCONTROL nontargeting siRNA 1, 5′-UAGCGA- DMEM + 10% FCS and incubated at 37°C for 24 h. The medi- CUAAACACAUCAA-3′ (Dharmacon), and FKBPL siRNA, um was replaced with phenol red–free DMEM + 10% 5′-CGCUUGAGCUGGAAGUAAGtt-3′ (Ambion), were trans- charcoal-stripped FCS containing DMSO as vehicle control − fected using Oligofectamine reagent (Invitrogen) according or 10 8 mol/L 17β-estradiol (Sigma) or DMEM + 10% FCS to the manufacturer's instructions. Cells were assayed 72 h containing cell line IC50 doses of tamoxifen (Sigma) or fulves- posttransfection. trant (Sigma) and incubated under normal conditions for

www.aacrjournals.org Cancer Res; 70(3) February 1, 2010 1091

McKeen et al.

16 d. For growth assays, 1 × 105 cells were seeded into 35-mm a GAL4-driven luciferase reporter. Following normalization dishes with DMEM + 10% FCS and incubated at 37°C. Cell for transfection efficiency by measurement of Renilla lumi- growth was monitored at 24-h intervals using the trypan blue nescence, luciferase expression in cotransfected cells was in- exclusion assay. duced by 19.5-fold and 4.7-fold compared with FKBPL and Gene expression analysis. Publicly available microarray Hsp90 self-activation controls, respectively, indicating that data sets from National Center for Biotechnology Information FKBPL binds Hsp90 in breast cancer cells (Fig. 1A). To con- Gene Expression Omnibus (26–30) were used for gene expres- firm this interaction, immunoprecipitations were carried out. sion analysis (Table 1). The analysis was performed using Par- FKBPL immunoprecipitates from parental MCF7 cells tek Genomics Suite (Partek, Inc.) and Kaplan-Meier curves showed that endogenous FKBPL coprecipitated with Hsp90 were generated using GraphPad Prism version 5.00 (GraphPad (Fig. 1B). The molar ratio of Hsp90/FKBPL is low, suggesting Software). The raw microarray data were imported into Partek that FKBPL is probably not the dominant immunophilin GS and processed using Robust Multichip Average (RMA) to within Hsp90 chaperone complexes. Furthermore, we were generate the normalized expression values. Expression values able to detect an interaction between FKBPL and ERα in for the probe set 219187_at representing FKBPL were assessed MCF7 parental cells (Fig. 1C) and FKBPL stably overexpres- for normal distribution using Kolmogorov-Smirnov test and sing (3.1D2) cells (Fig. 1D). In estrogen-free conditions, used to stratify samples with low FKBPL expression (values FKBPL immunoprecipates showed that ER did not coprecipi- in the 1st quartile) from the rest. Kaplan-Meier survival curves tate with FKBPL and ER immunoprecipitates showed that were generated based on the stratified data and associated FKBPL coprecipitated very weakly, suggesting that the pres- survival data and compared using Mantel-Cox log-rank test. ence of ligand may be required for this interaction to occur Quantification and statistical analysis. For Western (Supplementary Fig. S1). These results locate FKBPL within blots, image acquisition and densitometric analysis was carried the Hsp90/ER molecular chaperone complex. out using Canon Zoom Browser and one-dimensional ScanEX FKBPL inhibits breast cancer cell growth and sensitizes software. For statistical analysis of all results except the gene cells to estrogen deprivation. Previous studies have sug- expression analysis results, one-way ANOVA was used. gested that FKBPL overexpression can slow the growth of lymphoma cells (14). To assess the relevance of these results Results to breast cancer biology, we assessed cell growth in MCF-7 cells transiently and stably overexpressing FKBPL. Transient FKBPL associates with ERα in the Hsp90 chaperone transfection of MCF-7 cells with pcDNA3.1/FKBPL resulted complex. We hypothesized that FKBPL might bind Hsp90 in a significant decrease in clonogenic survival compared within ER chaperone complexes in breast cancer cells, espe- with cells transfected with an empty vector control (P < cially because we had already identified a role for this protein 0.05;Fig.2A).Inaddition,pcDNA3.1/FKBPLwasstably in glucocorticoid/Hsp90 complexes (15). Using the mamma- transfected into MCF7 cells and clones selected for G418 lian two-hybrid assay, the expression constructs pBIND- antibiotic resistance (showing low and high FKBPL expres- FKBPL and pACT-Hsp90 were cotransfected into ER-positive sion) were analyzed for FKBPL expression by Western blot MCF-7cellsandthebindingabilityofFKBPLwithHsp90 (Fig. 2B). The 3.1D2 and 3.1D3 clones showed a 7-fold and was assessed by measuring luciferase activity derived from 5-fold increase in FKBPL protein levels compared with

Table 1. Microarray data sets

Accession no. No. of samples Treatment ER ratio Microarray FKBPL correlates Reference (−/+) platform with survival

GSE7390 198 No 64/134 Affymetrix U133A Yes Desmedt and P = 0.004 (OS) colleagues (26) P = 0.001 (DMFS) GSE2034 209 No* 0/209 Affymetrix U133A Nonsignificant trend Wang and P = 0.17 colleagues (27) GSE9195 77 Yes 0/77 Affymetrix U133 Nonsignificant trend Loi and colleagues (28) Plus 2.0 P = 0.33 GSE2990 64 Yes 0/64 Affymetrix U133A No Sotiriou and colleagues (29) GSE1378 60 Yes† 0/60 Arcturus 22k oligo No Ma and colleagues (30)

Abbreviations: OS, overall survival; DMFS, distant metastasis-free survival. *Patients did not receive tamoxifen but some patients received radiotherapy treatment. †Patients received radiation before tamoxifen treatment.

1092 Cancer Res; 70(3) February 1, 2010 Cancer Research

FKBPL Modulates the Estrogen Receptor Signaling Pathway

Figure 1. FKBPL is in a complex with Hsp90 and ERα.A, mammalian two-hybrid assay demonstrating an interaction between FKBPL and Hsp90 in MCF7 cells. Representative blots showing protein coimmunoprecipitation in MCF7 whole cell lysates (WCL) immunoabsorbed to protein G–Sepharose beads with antibodies against FKBPL (B) and Hsp90 or FKBPL (C) and ER in DMEM containing 10% FCS. D, representative blot showing protein coimmunoprecipitation in 3.1D2 stable cell line whole cell lysates immunoabsorbed to protein G–Sepharose beads with antibodies against FKBPL and ER in DMEM containing 10% FCS. Anti-rabbit IgG was used as a negative control (NC; n = 3).

parental cells, respectively. Stable clone 3.1D9 showed mini- were sufficient to cause a 50% reduction in colony survival, mal overexpression with a 2-fold increase at the protein lev- and in T47D cells, 0.3 μmol/L tamoxifen produced a 50% in- el. These clones displayed inhibition of cell growth across a hibition of clonogenic survival. These doses were used in all 4-day time course (Fig. 2B). Compared with parental con- subsequent experiments. MCF-7 parental and three clones trols, the low-level overexpressing clone 3.1D9 showed inter- that overexpress FKBPL—3.1D2, 3.1D3, and 3.1D9—were mediate growth inhibition, which was significantly different treated with the IC50 dose and clonogenic survival was deter- from parental MCF7 cells at 72 hours (P < 0.05) and 96 hours mined after 16 days. Following treatment with tamoxifen (1 (P < 0.05). The high overexpressing clones 3.1D2 and 3.1D3 μmol/L), clonogenic survival in MCF7 cells was reduced by showed the greatest growth inhibition, which was statistical- 46% compared with untreated controls (Fig. 3A). FKBPL ly significant from parental MCF7 (P < 0.001) and 3.1D9 cells overexpression rendered cells more sensitive to tamoxifen, (<0.05) at 48 hours and highly significantly different from and these responses were not clonal, as independent parental cells at 72 and 96 hours (P < 0.001). This FKBPL- FKBPL-overexpressing clones displayed the same phenotype. mediated inhibition of cell growth was therefore not clonal, In 3.1D2, 3.1D3, and 3.1D9 cell lines, clonogenic survival was as independent FKBPL-overexpressing clones displayed the reduced by 89%, 75%, and 66%, respectively, compared with same phenotype. corresponding untreated controls (Fig. 3A). Furthermore, the FKBPL overexpression clearly affected the growth poten- MCF-7 clone overexpressing FKBPL maximally, 3.1D2, tial of MCF-7 cells under normal growth conditions. Next, showed a dose-dependent increase in sensitivity compared we sought to determine whether FKBPL overexpression with parental controls (Supplementary Fig. S2). The en- could affect growth in estrogen-deprived conditions. In these hanced sensitivity to fulvestrant was even more dramatic experiments, we chose to use the MCF-7 clone overexpres- in FKBPL-overexpressing cell lines (Fig. 3B). Parental MCF7 sing FKBPL maximally, 3.1D2, and compared the effects with cells showed a 54% reduction in clonogenic survival com- the parental MCF-7 cell line. High levels of FKBPL resulted in pared with untreated control, whereas 3.1D2 displayed a re- a 10-fold reduction in clonogenic survival compared with pa- duction of 99% compared with untreated control. 3.1D3 and rental cells (P < 0.001), suggesting a strong dependency on 3.1D9 displayed a 93% and 80% decrease compared with un- estrogen for growth. Furthermore, even in the presence of treated control, respectively (Fig. 3B). In support of a role for physiologic levels of estrogen, FKBPL overexpression inhib- FKBPL in increasing the sensitivity to endocrine therapies, a ited the clonogenic potential by 50% compared with parental targeted knockdown approach was used to determine if cells (P < 0.01; Fig. 2C). these effects could be reversed. Transfection of parental FKBPL sensitizes cells to tamoxifen and fulvestrant. Be- MCF-7 (Fig. 3C) and another ER-positive breast cancer cell cause increased expression of FKBPL sensitized cells to low line, T47D (Fig. 3D), with FKBPL-targeted siRNA increased estrogen levels, we further extended the study to assess the their resistance to tamoxifen compared with the nontarget- effect of high FKBPL levels on response to endocrine therapy. ing siRNA controls. Following treatment with 1 μmol/L ta- Tamoxifen and ICI 182,780 (fulvestrant) IC50 values for MCF7 moxifen, FKBPL siRNA-transfected MCF-7 cells showed a and T47D cells were determined through clonogenic assay. In 30% increase in clonogenic survival compared with nontar- MCF7 cells, 1 μmol/L tamoxifen and 2.5 nmol/L fulvestrant geted siRNA-transfected cells (P < 0.001; Fig. 3C). Following

www.aacrjournals.org Cancer Res; 70(3) February 1, 2010 1093

McKeen et al.

FKBPL knockdown in T47D cells, treatment with 0.3 μmol/L quently affect ERα levels. Using the stable clone 3.1D2, we tamoxifen increased clonogenic survival by 21% compared showed that ERα levels decreased compared with parental with the nontargeted control (P < 0.01; Fig. 3D). In sum- controls, whereas FKBPL knockdown in both parental and mary, the data suggest a correlation between FKBPL levels 3.1D2 cells led to an increase in ERα levels (Fig. 4B). Further- and sensitivity to endocrine therapy in two ER-positive more, FKBPL overexpression caused a decrease in levels of the cell lines. ER-responsive gene, cathepsin D, whereas knockdown in- Finally, we showed that the levels of endogenous FKBPL in creased the levels (Fig. 4C). The regulation of this ER-respon- a range of breast cancer cell lines, including the parental sive gene supports a functional role for FKBPL in physiologic MCF-7, T47D, and ZR-75-1 cells along with two MCF-7 cells ER-mediated signaling. Finally, because FKBPL has already clones stably overexpressing FKBPL, correlated (R2 = 0.76) been identified as essential for the stabilization of newly syn- with sensitivity to tamoxifen (Supplementary Fig. S3A). When thesized p21 (13), we therefore measured p21 levels. Following the ER-negative cell line MDA-231 was included, the correla- FKBPL knockdown, a 5-fold decrease in p21 levels compared tion went down to R2 = 0.6443 (Supplementary Fig. S3B). with nontargeting controls was detected (P < 0.001), whereas FKBPL is an estrogen-inducible gene that inhibits ERα transient overexpression of FKBPL led to a 2.5-fold increase in expression and modifies cathepsin D and p21 expression. p21 protein expression (P = 0.056; Fig. 4D). This is the first time To understand the mechanism relating to the FKBPL-mediat- this has been shown in breast cancer cells. ed regulation of cell growth and sensitivity to tamoxifen, we FKBPL increases ERα phosphorylation on Ser118 follow- first sought to investigate whether FKBPL itself was regulated ing treatment with estrogen or tamoxifen. Loss of p21 ex- by estrogen. A previous study (31) has already identified non– pression, as observed in FKBPL knockdowns, has previously ER-binding estrogen response elements within the FKBPL been shown to cause hyperphosphorylation of ER on Ser118, promoter, and expression of immunophilins and other cocha- leading to an increase in expression of ER-regulated genes perones are modulated by estrogen (18). We therefore investi- and increased resistance to tamoxifen (32). To further exam- gated estrogen-mediated regulation of FKBPL in breast cancer ine the effects of FKBPL modulation on ER signaling, cells − − cells. MCF7 cells were treated with 10 8 mol/L 17β-estradiol were treated with 10 8 mol/L 17β-estradiol (Fig. 5A) or ta- over a 48-hour time period. FKBPL levels increased and moxifen (Fig. 5B) for 30 minutes and ERα phosphorylation reached a maximum by 24 and 48 hours (Fig. 4A). We also de- on Ser118 was evaluated. As expected, phospho-Ser118 ER termined whether modulating FKBPL levels could subse- levels were increased by 98% and 141% in MCF-7 cells

Figure 2. FKBPL inhibits breast cancer cell line proliferation and sensitizes cells to estrogen deprivation. A, clonogenic assay to assess surviving fraction of MCF7 cells transiently transfected with empty vector or pcDNA3.1/FKBPL expression vector (*, P < 0.05, one-way ANOVA). B, MCF7 cells were transfected with pcDNA3.1/ FKBPL to create stable clones constitutively overexpressing FKBPL at the protein level (shown by Western blot). Proliferation of MCF7 cells and stable FKBPL-overexpressing clones, 3.1D2, 3.1D3, and 3.1D9, was assessed at 24-h intervals over a 4-d time course using trypan blue exclusion assay. C, clonogenic assay to determine surviving fraction in MCF7 and 3.1D2 cells plated in phenol red–free medium containing charcoal-stripped FCS for 24 h, then treated with DMSO vehicle control or 10−8 mol/L 17β-estradiol (E2; **, P < 0.01; ***, P < 0.001, compared with MCF7 cells, one-way ANOVA).

1094 Cancer Res; 70(3) February 1, 2010 Cancer Research

FKBPL Modulates the Estrogen Receptor Signaling Pathway

Figure 3. FKBPL levels determine sensitivity to endocrine therapies. Clonogenic assay to determine surviving fraction in MCF7, 3.1D2, 3.1D3, and 3.1D9 cells treated with (A) 1 μmol/L tamoxifen or (B) 2.5 nmol/L fulvestrant (*, P < 0.05; **, P < 0.01, compared with treated MCF7 cells, one-way ANOVA). MCF7 (C) or T47D (D) cells transfected with nontargeted or FKBPL-specific siRNA and treated with 1 or 0.3 μmol/L tamoxifen (TAM), respectively. (FKBPL knockdown was shown by Western blot; **, P < 0.01; ***, P < 0.001, one-way ANOVA).

transfected with a targeted FKBPL siRNA compared with and Loi and colleagues' data set (GSE9195; tamoxifen trea- nontargeting controls after treatment with estrogen or ta- ted); in ER-positive samples, high FKBPL expression was as- moxifen, respectively. Moreover, in the 3.1D2 FKBPL-overex- sociated with a trend toward increased survival (data not pressing clone, Ser118 phosphorylation was decreased by 55% shown). Furthermore, in the GSE2034 data set, higher p21 le- in DMSO control and 28% with 17β-estradiol treatment com- vels were associated with statistically significant longer time pared with parental control (Fig. 5C). Following tamoxifen to relapse (data not shown). However, in the Desmedt micro- treatment, Ser118 phosphorylation in 3.1D2 was decreased array data set (198 node-negative/134 ER-positive untreated by 50% in untreated control and 30% in treated sample com- patients), high FKBPL expression was highly correlated with pared with parental controls (Fig. 5D). In addition, overex- increased overall survival (hazard ratio, 2.456; 95% confi- pression of FKBPL inhibited ligand-independent Ser118 dence interval, 1.335–4.520; log-rank test P = 0.0039, Fig. phosphorylation of ER, which may also affect endocrine ther- 6A) and distant metastasis-free survival (hazard ratio, 2.647; apy sensitivity. 95% confidence interval, 1.476–4.748; log-rank test P = 0.0011; High FKBPL expression correlates with improved pa- Fig. 6B). These data suggest that FKBPL levels in breast can- tient survival. We have established the downstream molec- cer may correlate with outcome. ular consequences resulting from FKBPL expression that may enhance growth arrest and sensitization to endocrine Discussion agents in breast cancer. We reasoned that FKBPL may carry prognostic power resulting from the effects on ER levels, ER FKBPL has previously been linked with the GR/Hsp90 phosphorylation, and p21 stabilization, which might affect chaperone complex, with implications for GR translocation the tumor proliferative capacity and improve outcome inde- and signaling (15). We have now determined that FKBPL is pendent of ER status. To assess the clinical significance of also involved in the ERα/Hsp90 chaperone complex with our in vitro data, we investigated FKBPL mRNA levels in sev- consequences for downstream estrogenic signaling and re- eral microarray data sets: GSE7390 (26), GSE2034 (27), GSE sponse to endocrine therapy. 9195 (28), GSE2990 (29), and GSE1378 (ref. 30; Table 1). We Strict regulation of ERα levels is essential to maintain ap- saw no correlation in the GSE2990 and GSE1378 data sets. propriate cellular responsiveness to estrogen (33, 34) and the However, a trend toward significance was observed in Wang Hsp90 molecular complex has previously been implicated in and colleagues' data set (GSE2034; no tamoxifen treatment) this process (35, 36). In the present study, we have identified

www.aacrjournals.org Cancer Res; 70(3) February 1, 2010 1095

McKeen et al.

FKBPL as a member of the Hsp90 complex that binds ER in levels of the CDK inhibitor p21 and an inhibition of pro- thepresenceofestrogen.Similartotheimmunophilins liferation and clonogenicity. This growth delay could be Cyp40 and FKBP52 (18), 17β-estradiol treatment upregulated due to the role of FKBPL in stabilizing newly synthesized FKBPL levels. Moreover, high FKBPL expression decreased p21 (13) and suggests a role for FKBPL in cell cycle pro- protein levels of ERα and the ER-responsive gene, cathepsin gression that has not yet been fully elucidated. More re- D, whereas FKBPL knockdown studies reversed this effect. cently, siRNA targeted knockdown of ERα has been The ubiquitin ligase CHIP has previously been identified as shown to increase p21 transcription and protein expres- a cochaperone of the Hsp90-ERα complex (37). CHIP overex- sion (39), consistent with the effects shown by FKBPL pression decreases ERα levels, whereas CHIP knockdown re- overexpression. Here, we show that higher FKBPL was verses the effect (36)—similar effects to what we have seen linked with better survival outcome and increased distant with FKBPL. More recently, the cochaperone serine/threonine metastasis-free survival in three of five clinical data sets. protein phosphatase 5 has been identified as an estrogen- Data sets GSE7390 and 2034 (untreated patients) show a inducible protein that increases estrogen-dependent tumor significant correlation and a trend toward significance, re- growth (38). spectively, between high FKBPL levels and improved sur- Previously, FKBPL overexpression has been shown to de- vival, suggesting that FKBPL prognosticates for outcome. crease the rate of proliferation of leukemic U937 cells by This fits with our in vitro data in which we show an delaying cells in the G0-G1 phase of the cell cycle (14). In FKBPL-induced growth delay and an increase in p21 levels. breast cancer cells, FKBPL overexpression led to increased The GSE9195 data set from patients who were treated with

Figure 4. FKBPL is an estrogen-inducible gene that affects ER signaling. A, FKBPL levels in MCF7 cells treated with DMSO vehicle control or 10−8 mol/L E2 for the 48-h time course. ERα (B), cathepsin D (C), and p21 (D) protein expression levels in MCF7 cells following FKBPL siRNA knockdown or overexpression (corrected densitometry values are displayed on graphs; **, P < 0.01; ***, P < 0.001, one-way ANOVA).

1096 Cancer Res; 70(3) February 1, 2010 Cancer Research

FKBPL Modulates the Estrogen Receptor Signaling Pathway

Figure 5. FKBPL modulates ER phosphorylation on Ser118.ERα phosphorylation on Ser118 was determined in MCF7 cells following FKBPL siRNA knockdown (A and B) or FKBPL stable overexpression (C and D) and treatment with DMSO vehicle control and 10−8 mol/L E2 or 1 μmol/L tamoxifen for 30 min. FKBPL and ER expression levels were corrected to GAPDH loading control and then Ser118 expression levels were corrected to total ER (corrected densitometry values are displayed on graphs; *, P < 0.05; ***, P < 0.001, one-way ANOVA). tamoxifen showed that high FKBPL correlated with a trend sets showing no correlation were generally smaller in size; toward significance, suggesting that FKBPL might be a po- the number of samples is known to affect statistical signif- tential predictive marker of response, which fits with our icance in this type of study. The assessment of FKBPL ex- in vitro data; high or low FKBPL resulted in sensitivity or pression at the mRNA level provides a clinical context for resistance to tamoxifen, respectively. The final two data our in vitro findings. However, transcriptional effects only sets, GSE1378 and GSE2990 (from tamoxifen-treated pa- partly correlate with protein levels; therefore, further vali- tients), did not show any correlation, but the patient dation in a large randomized trial with assessment of numbers are lower, especially in the GSE1378 data set, FKBPL at the protein level using tissue microarrays would which also used a different microarray platform and mi- be required to fully validate our in vitro findings. However, crodissected tumor cells rather than whole tumor tissues. in support of our hypothesis, a recent study showed that A lack of concordance between data sets is a common fea- the loss of chromosome 6p21.32 region, containing FKBPL, ture in microarray studies (40), but this could also be be- occurred more frequently in patients with cancer recur- cause GSE2034 and GSE1378 patients were treated with rence within 5 years of initial diagnosis (41). radiotherapy and FKBPL has previously been shown to As mentioned, FKBPL seems to be an important determi- be regulated by radiation (10, 11). Furthermore, the data nant of response to endocrine therapies in an in vitro setting.

www.aacrjournals.org Cancer Res; 70(3) February 1, 2010 1097

McKeen et al.

Overexpression of FKBPL increased sensitivity to estrogen diate cytostatic effects and that loss of p21 could be a major deprivation, tamoxifen, and fulvestrant, whereas FKBPL cause of antiestrogen resistance in breast cancer (45, 46). knockdown using a targeted siRNA approach increased resis- Moreover, loss of p21 enhanced the agonistic effects of ta- tance to tamoxifen. Furthermore, FKBPL levels correlated moxifen through hyperphosphorylation of ERα (32) and ele- with the intrinsic tamoxifen sensitivity exhibited in breast vated Ser118 phosphorylation mediated through the EGFR/ cancer cells, with cell lines that had high levels of FKBPL dis- MAPK pathway was identified in tamoxifen-resistant cell playing increased sensitivity to tamoxifen (Supplementary lines (47). However, there is some debate over the prognostic Fig. S3). In accordance with our observations of increased benefit of Ser118 phosphorylation for response to tamoxifen. ERα expression following FKBPL knockdown, high ERα le- It has been linked with ligand-independent activation of ERα vels have been associated with estrogen-independent signal- (48) and patient relapse following tamoxifen treatment (49). ing and endocrine therapy resistance (42). In addition, In contrast, elevated expression of phosphorylated ER has elevated cathepsin D levels in patient samples have been been associated with low tumor grade (49) and improved linked with early relapse (43) and reduced survival (44). outcome in patients treated with hormone therapies (50). FKBPL overexpression inhibited ligand-independent Ser118 Our results support a model summarized in Fig. 6C, in which phosphorylation, suggesting that FKBPL may affect growth repression of FKBPL would lead to low levels of p21 and hy- factor receptor– or mitogen-activated protein kinase perphosphorylation of ERα at Ser118, hence lowering sensitiv- (MAPK)–driven ERα phosphorylation, although this remains ity to tamoxifen and possibly resulting in an increase in to be elucidated. These results suggest that FKBPL may have expression of ER-regulated genes, such as cathepsin D. Con- a direct impact on ER signaling and therefore on response to versely, high levels of FKBPL would convey a more slowly endocrine therapy, which may be partially linked to FKBPL- proliferating tumor phenotype and reduce ERα phosphoryla- mediated effects on p21. A previous study found that p21 tion, abrogating tamoxifen-induced agonist activity and expression was essential for tamoxifen and fulvestrant to me- thereby increasing sensitivity to tamoxifen.

Figure 6. High FKBPL expression correlates with improved patient survival. FKBPL expression was analyzed using microarray data from publicly available data set (GSE7390). Kaplan-Meier survival curves of breast cancer patients (n = 198 node negative; 134 ER positive) that have low FKBPL expression (defined as the lowest quartile expression) had significantly reduced (A) overall survival and (B) distant metastasis-free survival. C, schematic representation of proposed FKBPL role in ERα phosphorylation.

1098 Cancer Res; 70(3) February 1, 2010 Cancer Research

FKBPL Modulates the Estrogen Receptor Signaling Pathway

In summary, we provide evidence that FKBPL is involved Grant Support in ER stability and signaling with subsequent implications for the growth of ER-responsive tumors and their sensitivity to Breast Cancer Campaign, United Kingdom, and Action Cancer Northern endocrine therapies. Ireland. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in Disclosure of Potential Conflicts of Interest accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received 7/6/09; revised 11/3/09; accepted 11/18/09; published OnlineFirst No potential conflicts of interest were disclosed. 1/26/10.

References 1. Pietras RJ. Biologic basis of sequential and combination therapies 21. Knoblauch R, Garabedian MJ. Role for Hsp90-associated cochaper- for hormone-responsive breast cancer. Oncologist 2006;11:704–17. one p23 in estrogen receptor signal transduction. Mol Cell Biol 1999; 2. Sommer S, Fuqua SA. Estrogen receptor and breast cancer. Semin 19:3748–59. Cancer Biol 2001;11:339–52. 22. Oxelmark E, Roth JM, Brooks PC, Braunstein SE, Schneider RJ, 3. Cummings FJ. Evolving uses of hormonal agents for breast cancer Garabedian MJ. The cochaperone p23 differentially regulates estro- therapy. Clin Ther 2002;24 Suppl C:C3–25. gen receptor target genes and promotes tumor cell adhesion and 4. MacGregor JI, Jordan VC. Basic guide to the mechanisms of anties- invasion. Mol Cell Biol 2006;26:5205–13. trogen action. Pharmacol Rev 1998;50:151–96. 23. Riggins RB, Schrecengost RS, Guerrero MS, Bouton AH. Pathways 5. Cheskis BJ, Greger JG, Nagpal S, Freedman LP. Signaling by estro- to tamoxifen resistance. Cancer Lett 2007;256:1–24. gens. J Cell Physiol 2007;213:610–7. 24. Iorns E, Turner NC, Elliott R, et al. Identification of CDK10 as an im- 6. Hall JM, McDonnell DP. Coregulators in nuclear estrogen receptor portant determinant of resistance to endocrine therapy for breast action: from concept to therapeutic targeting. Mol Interv 2005;5: cancer. Cancer Cell 2008;13:91–104. 343–57. 25. Wen J, Li R, Lu Y, Shupnik MA, et al. Decreased BRCA1 confers ta- 7. Osborne CK. Tamoxifen in the treatment of breast cancer. N Engl J moxifen resistance in breast cancer cells by altering estrogen recep- Med 1998;339:1609–18. tor-coregulator interactions. Oncogene 2009;28:575–86. 8. Oesterreich S, Zhang Q, Hopp T, et al. Tamoxifen-bound estrogen 26. Desmedt C, Piette F, Loi S, et al. Strong time dependence of the 76- receptor (ER) strongly interacts with the nuclear matrix protein gene prognostic signature for node-negative breast cancer patients HET/SAF-B, a novel inhibitor of ER-mediated transactivation. Mol in the TRANSBIG multicenter independent validation series. Clin Endocrinol 2000;14:369–81. Cancer Res 2007;13:3207–14. 9. Wu L, Wu Y, Gathings B, et al. Smad4 as a transcription corepressor 27. Wang Y, Klijn JG, Zhang Y, et al. Gene-expression profiles to predict for estrogen receptor α. J Biol Chem 2003;278:15192–200. distant metastasis of lymph-node-negative primary breast cancer. 10. Robson T, Joiner MC, Wilson GD, et al. A novel human stress re- Lancet 2005;365:671–9. sponse-related gene with a potential role in induced radioresistance. 28. Loi S, Haibe-Kains B, Desmedt C, et al. Predicting prognosis using Radiat Res 1999;152:451–61. molecular profiling in estrogen receptor-positive breast cancer trea- 11. Robson T, Price ME, Moore ML, et al. Increased repair and cell ted with tamoxifen. BMC Genomics 2008;9:239. survival in cells treated with DIR1 antisense oligonucleotides: im- 29. Sotiriou C, Wirapati P, Loi S, et al. Gene expression profiling in breast plications for induced radioresistance. Int J Radiat Biol 2000;76: cancer: understanding the molecular basis of histologic grade to im- 617–23. prove prognosis. J Natl Cancer Inst 2006;98:262–72. 12. Ratajczak T, Carrello A. Cyclophilin 40 (CyP-40), mapping of its 30. Ma XJ, Wang Z, Ryan PD, et al. A two-gene expression ratio predicts hsp90 binding domain and evidence that FKBP52 competes with clinical outcome in breast cancer patients treated with tamoxifen. CyP-40 for hsp90 binding. J Biol Chem 1996;271:2961–5. Cancer Cell 2004;5:607–16. 13. Jascur T, Brickner H, Salles-Passador I, et al. Regulation of p21 31. Vega VB, Lin CY, Lai KS, et al. Multiplatform genome-wide identifi- (WAF1/CIP1) stability by WISp39, a Hsp90 binding TPR protein. cation and modeling of functional human estrogen receptor binding Mol Cell 2005;17:237–49. sites. Genome Biol 2006;7:R82. 14. Li YY, Liu LQ, Yang J, et al. Effect of WISp39 on proliferation, cell 32. Abukhdeir AM, Vitolo MI, Argani P, et al. Tamoxifen-stimulated cycle and apoptosis of U937 cells. Zhongguo Shi Yan Xue Ye Xue growth of breast cancer due to p21 loss. Proc Natl Acad Sci U S A Za Zhi 2007;15:733–7. 2008;105:288–93. 15. McKeen HD, McAlpine K, Valentine A, et al. A novel FK506-like bind- 33. Eckert RL, Mullick A, Rorke EA, Katzenellenbogen BS. Estrogen re- ing protein interacts with the glucocorticoid receptor and regulates ceptor synthesis and turnover in MCF-7 breast cancer cells mea- steroid receptor signaling. Endocrinology 2008;149:5724–34. sured by a density shift technique. Endocrinology 1984;114:629–37. 16. Pratt WB, Toft DO. Steroid receptor interactions with heat shock 34. Nardulli AM, Katzenellenbogen BS. Dynamics of estrogen receptor protein. Endocr Rev 1997;18:306–60. turnover in uterine cells in vitro and in uteri in vivo. Endocrinology 17. Ward BK, Mark PJ, Ingram DM, Minchin RF, Ratajczak T. Expression 1986;119:2038–46. of the estrogen receptor-associated immunophilins, cyclophilin 40 35. Tateishi Y, Kawabe Y, Chiba T, et al. Ligand-dependent switching of and FKBP52, in breast cancer. Breast Cancer Res Treat 1999;58: ubiquitin-proteasome pathways for estrogen receptor. EMBO J 267–80. 2004;23:4813–23. 18. Kumar P, Mark PJ, Ward BK, Minchin RF, Ratajczak T. Estradiol- 36. Fan M, Park A, Nephew KP. CHIP (carboxyl terminus of Hsc70- regulated expression of the immunophilins cyclophilin 40 and interacting protein) promotes basal and geldanamycin-induced deg- FKBP52 in MCF-7 breast cancer cells. Biochem Biophys Res Com- radation of estrogen receptor-α. Mol Endocrinol 2005;19:2901–14. mun 2001;284:219–25. 37. Ballinger CA, Connell P, Wu Y, et al. Identification of CHIP, a novel 19. Ostrow KL, Park HL, Hoque MO, et al. Pharmacologic unmasking of tetratricopeptide repeat-containing protein that interacts with heat epigenetically silenced genes in breast cancer. Clin Cancer Res shock proteins and negatively regulates chaperone functions. Mol 2009;15:1184–91. Cell Biol 1999;19:4535–45. 20. Gougelet A, Bouclier C, Marsaud V, et al. Estrogen receptor α and β 38. Golden T, Aragon IV, Zhou G, Cooper SR, Dean NM, Honkanen RE. subtype expression and transactivation capacity are differentially af- Constitutive over expression of serine/threonine protein phospha- fected by receptor-, hsp90- and immunophilin-ligands in human tase 5 (PP5) augments estrogen-dependent tumor growth in mice. breast cancer cells. J Steroid Biochem Mol Biol 2005;94:71–81. Cancer Lett 2004;215:95–100.

www.aacrjournals.org Cancer Res; 70(3) February 1, 2010 1099

McKeen et al.

39. Liu W, Konduri SD, Bansal S, et al. Estrogen receptor-α binds p53 45. Cariou S, Donovan JC, Flanagan WM, Milic A, Bhattacharya N, tumor suppressor protein directly and represses its function. J Biol Slingerland JM. Down-regulation of p21WAF1/CIP1 or p27Kip1 ab- Chem 2006;281:9837–40. rogates antiestrogen-mediated cell cycle arrest in human breast 40. Ein-Dor L, Zuk O, Domany E. Thousands of samples are needed to cancer cells. Proc Natl Acad Sci U S A 2000;97:9042–6. generate a robust gene list for predicting outcome in cancer. Proc 46. Mukherjee S, Conrad SE. c-myc suppresses p21WAF1/CIP1 expres- Natl Acad Sci U S A 2006;103:5923–5928. sion during estrogen signaling and antiestrogen resistance in human 41. Han W, Han MR, Kang JJ, et al. Genomic alterations identified by breast cancer cells. J Biol Chem 2005;280:17617–25. array comparative genomic hybridization as prognostic markers in 47. Britton DJ, Hutcheson IR, Knowlden JM, et al. Bidirectional cross talk tamoxifen-treated estrogen receptor-positive breast cancer. BMC between ERα and EGFR signalling pathways regulates tamoxifen- Cancer 2006;6:92. resistant growth. Breast Cancer Res Treat 2006;96:131–46. 42. Kuske B, Naughton C, Moore K, et al. Endocrine therapy resistance 48. Bunone G, Briand PA, Miksicek RJ, Picard D. Activation of the unli- can be associated with high estrogen receptor α (ERα) expression ganded estrogen receptor by EGF involves the MAP kinase pathway and reduced ERα phosphorylation in breast cancer models. Endocr and direct phosphorylation. EMBO J 1996;15:2174–83. Relat Cancer 2006;13:1121–33. 49. Sarwar N, Kim JS, Jiang J, et al. Phosphorylation of ERα at serine 43. Gaci Z, Bouin-Pineau MH, Gaci M, Daban A, Ingrand P, Metaye T. 118 in primary breast cancer and in tamoxifen-resistant tumours is Prognostic impact of cathepsin D and c-erbB-2 oncoprotein in a indicative of a complex role for ERα phosphorylation in breast cancer subgroup of node-negative breast cancer patients with low histolog- progression. Endocr Relat Cancer 2006;13:851–61. ical grade tumors. Int J Oncol 2001;18:793–800. 50. Murphy LC, Niu Y, Snell L, Watson P. Phospho-serine-118 estro- 44. Rodriguez J, Vazquez J, Corte MD, et al. Clinical significance of ca- gen receptor-α expression is associated with better disease out- thepsin D concentration in tumor cytosol of primary breast cancer. come in women treated with tamoxifen. Clin Cancer Res 2004;10: Int J Biol Markers 2005;20:103–11. 5902–6.

1100 Cancer Res; 70(3) February 1, 2010 Cancer Research

FKBPL Regulates Estrogen Receptor Signaling and Determines Response to Endocrine Therapy

Hayley D. McKeen, Christopher Byrne, Puthen V. Jithesh, et al.

Cancer Res Published OnlineFirst January 26, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-09-2515

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2010/01/27/0008-5472.CAN-09-2515.DC1

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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2010/01/26/0008-5472.CAN-09-2515. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.