Pigment Epithelium–Derived Factor Alleviates Tamoxifen-Induced Endometrial Hyperplasia

Published OnlineFirst October 8, 2015; DOI: 10.1158/1535-7163.MCT-15-0523

Cancer Biology and Signal Transduction Molecular Cancer Therapeutics Pigment Epithelium–Derived Factor Alleviates Tamoxifen-Induced Endometrial Hyperplasia Keren Goldberg1, Hadas Bar-Joseph1, Hadas Grossman1, Noa Hasky1, Shiri Uri-Belapolsky1, Salomon M. Stemmer2, Dana Chuderland1, Ruth Shalgi1, and Irit Ben-Aharon2

Abstract

Tamoxifen is a cornerstone component of adjuvant endocrine angiogenic balance favoring VEGF over PEDF. Treatment with therapy for patients with hormone-receptor–positive breast can- recombinant PEDF (rPEDF) abrogated tamoxifen-induced uter- cer. Its signiﬁcant adverse effects include uterine hyperplasia, ine hyperplasia and VEGF elevation, resulting in reduction of polyps, and increased risk of endometrial cancer. However, the blood vessels density. Exploring the molecular mechanism underlying molecular mechanism remains unclear. Excessive revealed that tamoxifen promoted survival and malignant trans- angiogenesis, a hallmark of tumorigenesis, is a result of disrupted formation pathways, whereas rPEDF treatment prevents these balance between pro- and anti-angiogenic factors. VEGF is a pro- changes. Activation of survival pathways was decreased, demon- angiogenic factor shown to be elevated by tamoxifen in the uterus. strated by reduction in AKT phosphorylation concomitant with Pigment epithelium–derived factor (PEDF) is a potent anti-angio- elevation in JNK phosphorylation. Estrogen receptor-a and c-Myc genic factor that suppresses strong pro-angiogenic factors, such as oncoprotein levels were reduced. Our ﬁndings provide novel VEGF. Our aim was to investigate whether angiogenic balance insight into the molecular mechanisms tamoxifen induces in the plays a role in tamoxifen-induced uterine pathologies, elucidate uterus, which may become the precursor events of subsequent the molecular impairment in that network, and explore potential endometrial hyperplasia and cancer. We demonstrate that rPEDF intervention to offset the proposed imbalance elicited by tamox- may serve as a useful intervention to alleviate the risk of tamoxifen. Using in vivo mouse models, we demonstrated that tamoxifen-induced endometrial pathologies. Mol Cancer Ther; 14(12); ifen induced a dose-dependent shift in endogenous uterine 2840–9. 2015 AACR.

Introduction Tamoxifen belongs to a class of selective estrogen receptor modulators (SERM), it binds to estrogen receptor (ER) and elicits Tamoxifen is considered a pivotal component of the hormonal estrogen agonistic or antagonistic responses by recruiting diverse therapy backbone commonly used for treatment of patients with sets of corepressors and coactivators, depending on the target hormone-receptor–positive breast cancer in all settings—adju- tissue (4). Tamoxifen serves as antagonist to ER in breast cancer vant, metastatic, and as a "primary prevention" risk-reducing cells, whereas in the uterus it exerts partial agonistic activity, strategy for high-risk populations (1, 2). The current American resulting in a range of endometrial pathologies including hyper- Society of Clinical Oncology guidelines for adjuvant hormonal plasia, polyps, carcinomas, and sarcomas (5, 6). Several large- therapy by tamoxifen have been recently revised; for premeno- scale studies have reported a 2- to 7-fold increase in the incidence pausal patients for 10 years of adjuvant tamoxifen therapy rather of endometrial cancer in patients treated with tamoxifen (7, 8) than 5 years (3), whereas in postmenopausal patients aromatase compared with control patients. Histopathologic changes in the inhibitors are considered an alternative, though tamoxifen repre- endometrium were observed after tamoxifen administration sents a valid and effective treatment standard. mainly in postmenopausal patients (9), and because endometrial hyperplasia is viewed as a possible precursor for malignant transformation, patients with atypical hyperplasia are closely monitored. 1Department of Cell and Developmental Biology, Sackler Faculty of Tamoxifen is known to exert its effect on the endometrium via 2 Medicine, Tel Aviv University, Tel Aviv, Israel. Institute of Oncology, activation of estrogen-regulated genes (10); however, recent Davidoff Center and Rabin Medical Center, Petah-Tiqva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel. observations indicate that tamoxifen can also activate different Note: Supplementary data for this article are available at Molecular Cancer sets of genes that are not regulated by estrogen (11). As demon- ﬁ Therapeutics Online (http://mct.aacrjournals.org/). strated by gene array pro ling, these genes are related to cellular processes of DNA replication, cell-cycle progression, and cellular D. Chuderland, R. Shalgi, and I. Ben-Aharon contributed equally to this article. organization (12). Yet, the exact molecular mechanism by which This work was performed in partial fulﬁllment of the requirements for a PhD tamoxifen exerts adverse effects on the uterus remains to be degree of K. Goldberg, Sackler Faculty of Medicine, Tel Aviv University, Israel. elucidated. Corresponding Author: Irit Ben-Aharon, Institute of Oncology, Davidoff Center, Angiogenesis, or formation of new blood vessels, is essential for Rabin Medical Center, Sackler Faculty of Medicine, Tel-Aviv University, Kaplan tissue proliferation. Under physiologic conditions, endometrial St., Petach-Tiqva 49100, Israel. E-mail: [email protected] angiogenesis is tightly regulated by pro- and anti-angiogenic doi: 10.1158/1535-7163.MCT-15-0523 factors (13); estrogen plays a pivotal role in establishing new 2015 American Association for Cancer Research. vascular bed and promoting cellular growth and differentiation

2840 Mol Cancer Ther; 14(12) December 2015

PEDF Alleviates Tamoxifen-Induced Endometrial Hyperplasia

within the endometrium at each menstrual cycle (13). In ovari- in canola oil. Tamoxifen was administered per os by a ectomized mice, tamoxifen stimulates production of uterine blunt gavage needle. Control mice were administered with VEGF (14), one of the major pro-angiogenic factors in the endo- canola oil. metrium, known to be upregulated in many pathologies and malignancies (15). Ovariectomy Pigment epithelium–derived factor (PEDF) is a 50-kDa glyco- Mice were anesthetized by an intraperitoneal injection of a protein that belongs to the noninhibitory members of the serine mixture of ketamine hydrochloride (Kepro) and xylazine hydro- protease inhibitors (serpin) superfamily. PEDF is one of the most chloride (VMD). The dorsal mid-lumbar area was shaved to potent physiologic inhibitors of angiogenesis (16). The anti- remove hair and prepared for aseptic surgery. A single skin angiogenic effect of PEDF is associated with disruption of endo- incision at the dorsal end of the ribcage was made and muscle thelial proliferation, promotion of endothelial cell apoptosis, and wall was incised, exposing the peri-ovarian fat pad region. The downregulation of pro-angiogenic factors. Increasing amount of ovary was exteriorized by grasping the peri-ovarian fat with tissue evidence indicates that under physiologic conditions, an essential forceps; the oviduct area was ligated and cauterized by extremely balance between PEDF and VEGF exists, and PEDF may counter- hot forceps to avoid bleeding, thus excising the ovary. act the angiogenic potential of VEGF (17). Several studies showed The remaining tissue was returned into the peritoneal cavity. The that PEDF level declines with age, and suggested that this decline procedure was repeated at the other side as well. The muscle may play a key role in the development of angiogenic-related incisions were closed with 4-0 absorbable suture (Ethicon) and pathologies (18). Moreover, it was shown that decrease in PEDF skin incisions were closed with wound clips (Clay-Adams). Mice expression is one of the mechanisms that promote tumor growth were allowed to rest for one week to ensure post-menopausal (19), suggesting PEDF as an anticancer therapy (20, 21). We have state, before experimental onset. recently characterized the expression and regulation of PEDF in human and rodent premenopausal endometrium, indicating that RNA isolation, reverse transcription, and qRT-PCR PEDF is dynamically expressed in mice endometrium throughout RNA was extracted using Trizol reagent (Invitrogen), accord- the estrous cycle, in reciprocity to VEGF, and that the balance ing to the manufacturer's instructions, and quantified with the between the two is regulated by estrogen and progestrone (22, Nano-Drop spectrophotometer (ND-1000; Thermo Scientific). 23). PEDF is a secreted glycoprotein that promotes a variety of Total RNA was reverse transcribed using high capacity cDNA activities upon binding to various receptors (20). In our previous reverse transcription kit (Applied Biosystems; ABI). Changes in studies, we showed that endometrial cells express Patatin-like the level of mRNA expression were detected by SYBR green phospholipase domain-containing protein 2 (PNPLA2; ref. 22), a reagent (SYBR Green PCR Master Mix, Applied Biosystems) PEDF receptor (PEDF-R) that mediates mainly prosurvival activity using 20-ng cDNA and specific primers, on an ABI Prism 7900 (24). Another key protein that regulates the anti-angiogenic Sequence PCR machine (Applied Biosystems). Hypoxanthine activity of PEDF is laminin receptor (LR; ref. 25), which has not phosphoribosyltransferase 1 (HPRT-1) served as an endoge- been characterized in the uterus. We have also demonstrated that nous control. stimulation of ECC-1 endometrial cell line with recombinant PEDF (rPEDF) induces a significant decrease in the level of VEGF expression (22). Finally, we established the role of PEDF as a Western blot analysis potential therapeutic agent for endometriosis and ovarian hyper- Samples were subjected to SDS-PAGE and immunoblotted stimulation syndrome (23, 26). with the appropriate primary antibodies: anti-PEDF (1:200, sc- In the current study, our aim was to explore the mechanism 25594, Santa Cruz Biotechnology), anti-VEGF (1:200, sc-152, accountable for tamoxifen-induced endometrial pathologies in Santa Cruz Biotechnology), anti-phospho-AKT (1:1,000, P- postmenopausal patients, among which these entities represent 4112, Sigma Chemical Company), anti-AKT (1:1,000, #2938, Cell clinical concern. We hypothesized that tamoxifen alters uterine Signaling Technology), anti-phospho-JNK (1:1,000, J-4750, Sig- a angiogenic balance and that the shift in regulatory angiogenic pro- ma), anti-JNK (1:1,000, J-4500, Sigma), anti-ER (1:1,000, ab teins contributes to tamoxifen-induced uterine pathologies. It is 2746, Abcam), anti-c-Myc (1:1,000, sc-40, Santa Cruz Biotechnol- therefore suggested that upon this mechanism, a replacement the- ogy), and anti-actin (1:10,000, MAB1501; Millipore). Blots were rapy using rPEDF may reduce tamoxifen-induced adverse effects. incubated with horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch) and subjected to enhanced fi Materials and Methods chemiluminescence assay (Thermo Scienti c). The intensity of the bands was analyzed by the ImageJ software. Animals Institute for Cancer Research (ICR) female mice (8 weeks old; Histologic evaluation Harlan Laboratories) were housed in temperature- and humidity- Uteri were excised, fixed in 4% paraformaldehyde in PBS, controlled rooms at the animal facilities of the Sackler Faculty of dehydrated by passage through an ascending series of alcohols, fi Medicine, Tel-Aviv University, under arti cial illumination for embedded in paraffin, sectioned (6 mm), and placed on slides ad lib 12 hours daily. Food and water were available . Animal care (Propper). Slides for histologic evaluation were stained with was in accordance with institutional guidelines and approved by hematoxylin and eosin (Bio-Optica). the local authorities. IHC Tamoxifen treatment Sections were deparaffinized by running the slides through Tamoxifen tablets (Teva Pharmaceutical Industries) were xylene (15 minutes) and a descending series of alcohols (5 pulverized to thin powder, dissolved in ethanol, and diluted minutes each) to a total re-hydration in distilled water. The

www.aacrjournals.org Mol Cancer Ther; 14(12) December 2015 2841

Goldberg et al.

sections were placed into preheated antigen unmasking solution Statistical analysis (H-3300; Vector Laboratories Inc.), cooled on ice to 50C, re- Data were expressed as mean SEM. A Student t test (equal heated for three more minutes and cooled on ice to room variance, one-tailed) was used to statistically evaluate the differ- temperature. The slides were then rinsed in PBS and incubated ences between control and experimental results. Differences were for 1 hour with PBSTg (0.2% Tween-20, 0.2% gelatin in PBS), re- considered statistically significant when P < 0.05. rinsed in PBS, blocked for 10 minutes in blocking solution (927B; Cell Marque), and incubated overnight at 4C with anti-CD34 (1:200, CL8927AP; Cedarlane Laboratories) primary antibody. Results Before and after applying the secondary goat anti-rat Alexa Short-term tamoxifen treatment modulates uterine VEGF/PEDF Flour555-conjugated antibody (1:400, Cell Signaling Technolo- balance in vivo gy) and a nuclear marker (Hoechst 3342; Sigma), sections were We have established an in vivo mouse model for examining rinsed in PBSTg (3), in PBS (3), and mounted with mowiol the effect of tamoxifen on postmenopausal uterine tissue (see (Sigma) before being scanned by a laser confocal microscope Supplementary Fig. S1 for the experimental design). To study the (Leica TCS SP5). subacute effect we used the short-term treatment model, which demonstrated a dose-dependent increase of uterine weight rPEDF production (Fig. 1A), as described in the literature (27). The histologic His-tagged human rPEDF (NM_002615.4) was expressed in appearance (Supplementary Fig. S2) showed that tamoxifen Escherichia coli BL21. Bacteria were allowed to grow at 30Cto caused an increase of uterine glands and epithelial height as well OD600 nm of 0.5 to 0.6, induced for 4 to 5 hours by isopropyl-L- as enlargement of the stromal compartment (27). We found that thio-b-D-galactopyranoside (0.5 mmol/L; Sigma), centrifuged, the level of VEGF protein in the uterus was upregulated in a dose- and their pellets lysed. Recombinant protein was purified by ion dependent manner (Fig. 1B and E), correlated with uterine weight. metal affinity chromatography with Ni-NTA His-Bind resin Thereafter, we evaluated the effect of tamoxifen on the expression (Merck KGaA) according to the manufacturer's protocol. Proteins of PEDF protein, and found that PEDF levels were downregulated of the eluted fractions were resolved by SDS-PAGE, dyed by (Fig. 1C and E) as tamoxifen doses were increased, yielding a GelCode (Blue Stain Reagent, Thermo scientific), or Western reciprocal manner of VEGF and PEDF expression (Fig. 1D), blotted with a specific anti-PEDF antibody (57 kDa). Eluates indicating impaired uterine angiogenic balance after tamoxifen with 90% purity were dialyzed against Tris buffer pH 10. administration.

Figure 1. Short-term tamoxifen treatment modulates uterine VEGF/PEDF balance in vivo. Various concentrations of tamoxifen were administered for 7 days to ovariectomized mice. Uteri were excised and lysed. Proteins were resolved by SDS-PAGE and immunoblotted with anti-VEGF or anti-PEDF antibodies. mRNA was extracted using TRIZOL reagent and subjected to RT-PCR. A, uterine weight. B, VEGF protein expression. C, PEDF protein expression. D, VEGF/PEDF protein ratio. E, representative Western blot of VEGF, PEDF, and actin as a loading control. Protein expression was quantiﬁed by ImageJ and calibrated to actin as a loading control. F, PEDF-R mRNA calibrated with HPRT-1. n ¼ 3–4 mice/group. The ratio between each treatment and control is plotted as mean SEM (, P < 0.05; , P < 0.01), statistically signiﬁcantly different from control value, using t test.

2842 Mol Cancer Ther; 14(12) December 2015 Molecular Cancer Therapeutics

PEDF Alleviates Tamoxifen-Induced Endometrial Hyperplasia

Figure 2. rPEDF downregulates both tamoxifen-induced hyperplasia and VEGF elevation, and modifies signaling pathways. Ovariectomized mice were administered for 7 days with tamoxifen (2.5 mg/mouse) or with a combination of tamoxifen and rPEDF (10 mg/kg; days 4–7). Uteri were excised, lysed, proteins resolved by SDS-PAGE, and immunoblotted with anti-VEGF antibody (B and C), anti-p-AKT (D and E), or with anti-p-JNK antibodies (F and G). A, uterine weight. B, VEGF protein expression. C, representative Western blot analyses of VEGF and actin as a loading control. Protein expression was quantified by ImageJ and calibrated to actin as a loading control. Each value is normalized to that of control (tamoxifen-only) level. Tamoxifen, n ¼ 23; tamoxifenþrPEDF, n ¼ 16. D, p-AKT protein expression. E, representative Western blot of p-AKT. Protein expression was quantified by ImageJ and calibrated to gen-AKT as a loading control. F, p-JNK protein expression. G, representative Western blot of p-JNK. Protein expression was quantified by ImageJ and calibrated to gen-JNK as a loading control. Each value is normalized to that of control (tamoxifen-only) level (n ¼ 4 mice/treatment group). Data are plotted as mean SEM (, P < 0.05; , P < 0.01), statistically significantly different from control value, using t test.

We evaluated the expression of PEDF-Rs in the uterus and administration ("rPEDF treatment"). When rPEDF was added found that short-term tamoxifen treatment upregulated the on days 4 to 7 of tamoxifen administration ("rPEDF treatment" expression level of both PNPLA2 and LR (Fig. 1F). regimen), significant decreases in uterine weight (Fig. 2A) and in uterine VEGF level (Fig. 2B and C) were observed, compared rPEDF downregulates both tamoxifen-induced hyperplasia and with the values obtained in mice treated with tamoxifen alone; VEGF elevation, and modifies signaling pathways suggesting that rPEDF had a significant favorable effect on We chose the dose of 2.5 mg/mouse tamoxifen as an optimal tamoxifen-induced uterine hyperplasia. This effect may be dose and added subcutaneous injections of rPEDF (10 mg/kg, attributed, at least in part, to PEDF's ability to resist tamoxi- every third day). We administered rPEDF to mice according to fen-induced VEGF elevation. No effects on uterine weight or two regimens (see Supplementary Fig. S1 for experimental uterine VEGF level were observed in mice treated according to design): (1) starting on day 1 of tamoxifen administration the "rPEDF prevention" regimen (data shown in Supplemen- ("rPEDF prevention"), (2) starting on day 4 of tamoxifen tary Fig. S3).

www.aacrjournals.org Mol Cancer Ther; 14(12) December 2015 2843

Goldberg et al.

Examination of signaling pathways known to mediate cellular enous PEDF will more likely bind to PNPLA2 and induce cellular growth and survival (PI3K/AKT/mTOR pathway) as well as apo- survival than promote anti-angiogenic activity through LR. ptosis [c-Jun N-terminal kinases (JNK)] revealed a significant decrease in phosphorylation of AKT (Fig. 2D and E) and an A prolonged combined treatment of tamoxifen and rPEDF increase in phosphorylation of JNK (Fig. 2F and G) in mice treated downregulates tamoxifen-induced hyperplasia, VEGF with the "rPEDF treatment" regimen compared with mice elevation, and blood vessels density treated with tamoxifen alone. Mice of the "rPEDF prevention" We examined the long-term effectiveness of rPEDF treatment cohort showed an increase in phosphorylation of JNK and no administered in vivo according to the prolonged tamoxifen significant change in phosphorylation of AKT compared with (2.5 mg/mouse) treatment model and added rPEDF injections mice treated with tamoxifen alone (data shown in Supplementary (2 mg/kg, every third day). Based on our finding that "rPEDF Fig. S3). treatment" regimen was more effective than "rPEDF prevention" regimen, we started rPEDF injections on day 8 of tamoxifen Prolonged tamoxifen treatment modulates uterine VEGF/PEDF treatment (see Supplementary Fig. S1 for experimental design). balance in vivo Uterine weight did not differ in the group treated with the We established a prolonged-treatment model in which tamox- combination compared with tamoxifen-only (Fig. 4A; P ¼ 0.2). ifen was administered to mice continuously for a period of 1 The expression level of uterine VEGF protein was significantly month. The effect of prolonged tamoxifen treatment was similar decreased in mice of the combined "tamoxifenþrPEDF" mice to that of the short-term treatment; namely, an increase in uterine compared with that of the "tamoxifen-only" group (Fig. 4B and weight (Fig. 3A), upregulation of VEGF protein (Fig. 3B and E), C); suggesting a durable, long-lasting effect of rPEDF on the uterus and downregulation of PEDF protein (Fig. 3C). VEGF/PEDF ratio after 1 month of treatment. was significantly increased (Fig. 3D), suggesting a long-lasting Changes in tissue vasculature were examined by IHC staining effect of tamoxifen. with CD34. Tamoxifen caused a remarkable increase in the Evaluation of the expression of PEDF-Rs after 1 month of density of uterine blood vessels (Fig. 4E and H) compared with tamoxifen treatment (Fig. 3F) revealed that PNPLA2 level was control (Fig. 4D and G), whereas addition of rPEDF (tamoxi- significantly increased, whereas that of LR did not change signif- fenþrPEDF treatment) caused a decrease of CD34 staining (Fig. icantly. This suggests that following tamoxifen treatment endog- 4F and I) compared with tamoxifen-only.

Figure 3. Prolonged tamoxifen treatment modulates uterine VEGF/PEDF balance in vivo. Tamoxifen (2.5 mg/mouse) was administered for 1 month (5 days/week) to ovariectomized mice. Uteri were excised and lysed. Proteins were resolved by SDS-PAGE and immunoblotted with anti-VEGF or anti-PEDF antibodies. mRNA was extracted using TRIZOL reagent and subjected to RT-PCR. A, uterine weight. B, VEGF protein expression. C, PEDF protein expression. D, VEGF/PEDF protein ratio. E, representative Western blot of VEGF and actin as a loading control. F, PEDF-R mRNA calibrated with HPRT-1. n ¼ 6–7 mice/treatment group. Protein expression was quantiﬁed by ImageJ and calibrated to actin as a loading control. The ratio between each treatment and control is plotted as mean SEM (, P < 0.05; , P < 0.01), statistically signiﬁcantly different from control value, using t test.

2844 Mol Cancer Ther; 14(12) December 2015 Molecular Cancer Therapeutics

PEDF Alleviates Tamoxifen-Induced Endometrial Hyperplasia

Figure 4. Prolonged combined treatment of tamoxifen and rPEDF downregulates tamoxifen-induced hyperplasia, VEGF elevation, and blood vessels density. Ovariectomized mice were treated for 1 month with tamoxifen (2.5 mg/mouse) or with a combination of tamoxifen and rPEDF (2 mg/kg), every third day. Uteri were excised, lysed, proteins resolved by SDS-PAGE, and immunoblotted with anti-VEGF antibody. A, uterine weights. B, VEGF protein expression. C, representative Western blot of VEGF and actin as a loading control. D–F, representative CD34 IHC staining (red) of uteri of control (D), tamoxifen (E), and tamoxifenþrPEDF (F) treated mice. G–I, magnification of selected areas (marked by white boxes in D–F). DNA labeling by Hoechst 32242 (blue), n ¼ 6 mice/group. Protein expression was quantified by ImageJ and calibrated to actin as a loading control. Each value was normalized to control (tamoxifen-only) value. Data are presented as mean SEM, , P < 0.05, statistically significantly different from control value, using t test.

Combined treatment of tamoxifen and rPEDF modulates ERa those of ERa mRNA, though with no statistical significance (Fig. and c-Myc expression patterns 5B, P ¼ 0.1). We used RT-PCR to examine the expression pattern of hor- Prolonged tamoxifen treatment upregulated mRNA levels of monal receptors in the uteri of mice treated according to the short- both ERa and ERb, but had no significant effect on PR expression term and prolonged in vivo models; assessing the mRNA levels of (Fig. 5D). The combined treatment of tamoxifenþrPEDF signif- ERa and ERb and progesterone receptor (PR). Short-term tamox- icantly reduced ERa mRNA level compared with tamoxifen. The ifen treatment caused an upregulation of ERa mRNA level, but level of ERa protein expression at the various treatments corre- had no significant effect on that of ERb and PR (Fig. 5A). rPEDF lated with that of ERa mRNA (Fig. 5E). impaired the tamoxifen-induced upregulation of ERa mRNA. The Finally, we tested the effect of rPEDF on c-myc protein. Short- levels of ERa protein along the various treatments resembled term tamoxifen treatment increased the level of c-Myc protein,

www.aacrjournals.org Mol Cancer Ther; 14(12) December 2015 2845

Goldberg et al.

Figure 5. Combined treatment of tamoxifen and rPEDF modulates ERa and c-Myc expression patterns. Ovariectomized mice were treated for 1 month with tamoxifen (2.5 mg/mouse, 5 days/week) or with a combination of tamoxifen and rPEDF (2 mg/kg). Uteri were excised and lysed. mRNA was extracted using TRIZOL reagent and subjected to RT-PCR. Proteins were resolved by SDS-PAGE and immunoblotted with anti-ERa or anti-c-Myc antibodies. A–C, short-term treatment, 7 days; D–F, prolonged treatment, 1 month. RT-PCR was performed with specific primers and calibrated with HPRT-1. Protein expression was quantified by ImageJ and calibrated to actin as a loading control. Data were normalized to control (tamoxifen-only) level (n ¼ 4 mice/group in short-term experiments; n ¼ 6 mice/group in prolonged experiments). Data are presented as mean SEM (, P < 0.05; , P < 0.01), statistically significantly different from control value, using t test.

whereas rPEDF caused no significant change (Fig. 5C). The level of uterine weight correlates with VEGF increase and PEDF decrease. c-Myc was significantly increased after a prolonged tamoxifen PEDF was shown to be a potent effector of angiogenesis and was treatment as well, and significantly decreased after a combined proposed as a potential therapeutic intervention for other angio- tamoxifenþrPEDF treatment (Fig. 5F). genesis-related pathologies. We suggested using rPEDF as a replacement therapy for reducing tamoxifen-induced adverse effects, and investigated its effect on key proteins involved in Discussion regulation of cells proliferation, survival, and apoptosis. Tamoxifen-induced uterine pathologies present a major clin- Our research is the first to demonstrate the significant impact of ical challenge. To date, the mechanism that accounts for the tamoxifen on the expression of uterine blood vessels, revealed by spectrum of tamoxifen-induced endometrial hyperplasia-meta- enhanced staining of the endothelial marker CD34. We suggest plasia remains unclear. It is postulated that it involves alterations that the increase in blood vessels density may be attributed to the in uterine angiogenesis, though only paucity of supporting evi- impaired balance between pro- and anti-angiogenic factors, dence exists. In the current study, we hypothesized that tamoxifen induced by tamoxifen. Moreover, treatment with rPEDF restored exerts its effects on the uterus by modulating uterine angiogenic the angiogenic balance by resisting tamoxifen-induced uterine balance; we studied the mechanisms using an in vivo mouse VEGF elevation, as seen by the decrease of both uterine weight and model. Because tamoxifen is prescribed in the clinical setting on CD34-marked blood vessels. Former evidence characterizing the a daily basis for an extended period, we established two models changes in uterine vasculature after tamoxifen treatment in post- that represent the subacute and the long-term effects. Our results menopausal patients was not persistent, possibly due to limited were in accordance with previous reports and showed that tamox- sample size and increased heterogeneity (28, 29). ifen upregulates the level of uterine VEGF. We showed, for the first Endometrial hyperplasia, especially atypical hyperplasia, is time, that the level of uterine PEDF is downregulated after generally viewed as a possible precursor for malignant transfor- tamoxifen treatment, resulting in an impaired balance of VEGF mations (30). Expression of VEGF and vascular density, two and PEDF, and demonstrated that tamoxifen-induced increase in important markers of angiogenesis, were found to be increased

2846 Mol Cancer Ther; 14(12) December 2015 Molecular Cancer Therapeutics

PEDF Alleviates Tamoxifen-Induced Endometrial Hyperplasia

Figure 6. Suggested mechanism of action for tamoxifen and rPEDF—schematic representation. Tamoxifen treatment (top) causes increased vascularization by (i) downregulation of PEDF, (ii) upregulation of VEGF, and (iii) downregulation of laminin receptor (LR). In addition, tamoxifen causes increased hyperplasia by (i) upregulation of ERa, (ii) upregulation of c-Myc, and (iii) upregulation of Patatin-like phospholipase domain-containing protein 2 (PNPLA2). These combined processes are suggested to account for the increased risk of endometrial malignant transformation under tamoxifen treatment. PEDF treatment (bottom) causes decreased vascularization by downregulation of VEGF, and decreased hyperplasia by (i) downregulation of ERa, (ii) downregulation of c-Myc, (iii) downregulation of phosphorylated-AKT, and (iv) upregulation of phosphorylated-JNK (iii and iv, during the short-term treatment). PEDF treatment restored the uterine natural angiogenic balance, and decreased pathologic hyperplasia, thus possibly reducing the risk of endometrial malignant transformation. in many premalignant lesions (31) and are known to play a role in alone. Our data indicated that rPEDF treatment reduces activation the spectrum of tissue malignant transformation. Our in vivo of proliferation-associated signaling pathways. models verified that tamoxifen enhanced both hyperplasia and Hormone receptors (ER and PR) expression is a fundamental angiogenic activation in treated mice. Furthermore, administra- principle in tamoxifen treatment for breast cancer. We hypothe- tion of rPEDF reduced the levels of both parameters (vascular sized that due to its SERM traits, tamoxifen-induced effects on the density and VEGF(, implying a reduced risk of endometrial endometrium may involve endometrial ER. The effect of tamox- tumorigenesis following tamoxifen treatment. ifen on the ER may entail recruitment of processes as tissue Based on the ability of PEDF to elicit non-angiogenic cellular proliferation and VEGF regulation, two major pathways relevant pathways as well, we examined its effect on signaling pathways to increased hyperplasia and possibly to malignant transforma- mediating cell proliferation, survival, and apoptosis. Activation of tions. Moreover, a study conducted in Ishikawa endometrial cell the PI3K/AKT/mTOR pathway, a signaling pathway that plays an line suggests that tamoxifen-induced upregulation of VEGF is important role in cellular growth and survival, has recently been mediated mainly through ERa, and not through ERb (38). ERb implicated in endometrial cancer (32). Activated AKT initiates a was shown to modulate, and possibly suppress, ERa-mediated cascade of downstream signaling events in endometrial cancer uterine gene transcription (39); the activity of tamoxifen may cells that promote cellular growth, metabolism, proliferation, therefore be determined by the relative expression of ERa and ERb survival, migration, apoptosis, and angiogenesis. JNKs, key med- in the tissue (40). Our results showed that rPEDF treatment iators of stress and inflammation, were implicated in apoptotic induces a specific decrease in the expression of uterine ERa, responses to DNA damage and to cytotoxic agents (33, 34). without modifying the expression of ERb and PR, implying a Activation of JNK correlates with induction of apoptosis in shift toward higher activity of ERb over ERa. The decrease in ERa endometrial cancer cells (35). PEDF was previously shown to expression confers dual implication with regard to tamoxifen- induce apoptosis by downregulation of phosphorylated-AKT (36) treated uterus: First, the ER-related signaling, known to mediate and upregulation of phosphorylated-JNK (37). Combined hyperplastic response, is diminished, and, second, the uterine rPEDF–tamoxifen treatment causes a significant increase in the response to tamoxifen is reduced because there are less tamoxifen- phosphorylation of the pro-apoptotic signaling molecule JNK, binding receptors per cell. c-Myc protein, known to induce pro- and a decrease in the phosphorylation of the prosurvival signaling liferation and to play a role in the development of endometrial molecule AKT, when compared with treatment with tamoxifen carcinoma (41), is upregulated by tamoxifen (42, 43). In our mice

www.aacrjournals.org Mol Cancer Ther; 14(12) December 2015 2847

Goldberg et al.

model tamoxifen treatment causes an increase in the level of short- and long-term treatment models. Tamoxifen-induced vas- c-Myc protein, compared with its level in control mice. Admin- cular proliferation clearly inhibited in mice of the prolonged istration of rPEDF prevents c-Myc elevation, suggesting a rPEDF treatment model. Prolonged rPEDF treatment downregu- decreased risk of malignant development. Our results also indi- lates VEGF and modulates expression patterns of ERa, thus cate that tamoxifen treatment changes the expression patterns of altering uterine response to tamoxifen treatment, diminishing receptors known to mediate PEDF activity. By upregulating the estrogen-related signal activation, and reducing susceptibility to level of PNPLA2, induction of prosurvival pathway is preferred tamoxifen-induced adverse effects. Moreover, the expression of c- over that of anti-angiogenic activity via the LR. Myc oncoprotein is downregulated after rPEDF treatment, possi- Two therapeutic options, Levonorgestrel and Metformin, were bly reducing the risk of endometrial malignant transformation investigated as potential candidates to reduce the risk of devel- (proposed mechanism of action is depicted in Fig. 6). By admin- oping endometrial cancer following tamoxifen treatment. istering PEDF, we restored the altered angiogenic balance of Though Levonorgestrel-releasing intrauterine system (LNG-IUS) uterine tissue to its natural state, and reduced the cascade of was indicated as a suppressor of endometrium proliferation, there events leading to pathologic hyperplasia. is not enough evidence regarding its efficacy in reducing endometrial hyperplasia and cancer (44). Moreover, the use of LNG- Disclosure of Potential Conflicts of Interest IUS in the setting of hormonal therapy for breast cancer, is K. Goldberg, H. Bar-Joseph, D. Chuderland, S.M. Stemmer, R. Shalgi and questionable due to its possible interaction with the antiestro- I. Ben-Aharon have ownership interest in U.S. Provisional Application no. genic therapy (45). When considering the safety of PEDF treat- 61/972,325. No potential conflicts of interest were disclosed by the other ment to patients with breast cancer—PEDF was already shown to authors. inhibit the growth of breast cancer cells in vitro, to exhibit anti- tumor activity in vivo, and to reduce the resistance of tumors to Authors' Contributions tamoxifen (46). Thus, PEDF treatment may have dual efficacy— Conception and design: K. Goldberg, D. Chuderland, R. Shalgi, I. Ben-Aharon treating breast cancer and minimizing the risk of endometrial Development of methodology: K. Goldberg, D. Chuderland, I. Ben-Aharon pathogenesis. The other candidate, the antidiabetic drug Metfor- Acquisition of data (provided animals, acquired and managed patients, min, exerts an antiproliferative effect on the endometrium of provided facilities, etc.): K. Goldberg, H. Bar-Joseph, H. Grossman, N. Hasky, tamoxifen-treated mice via inhibition of mTOR (47). Interesting- S. Uri-Belapolsky ly, Metformin increases PEDF levels in patients with type II Analysis and interpretation of data (e.g., statistical analysis, biostatistics, diabetes (48), and our data demonstrated that rPEDF decreased computational analysis): K. Goldberg, S.M. Stemmer, D. Chuderland, R. Shalgi, I. Ben-Aharon the activity of PI3K/AKT/mTOR pathway. However, inhibition of Writing, review, and/or revision of the manuscript: K. Goldberg, S.M. Stem- PEDF by Metformin was described (49), thus further research is mer, D. Chuderland, R. Shalgi, I. Ben-Aharon needed to fully comprehend the connection among these obser- Administrative, technical, or material support (i.e., reporting or organizing vations. The beneficiary effect of rPEDF in protecting the uterus data, constructing databases): I. Ben-Aharon and the fact that it does not compromise the safety of patients with Study supervision: D. Chuderland, R. Shalgi, I. Ben-Aharon breast cancer, imply that rPEDF therapy could minimize gyneco- The costs of publication of this article were defrayed in part by the payment of logic follow-up of tamoxifen-treated women and reduce medical page charges. This article must therefore be hereby marked advertisement in interventions caused by tamoxifen. accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Our study suggests that angiogenesis-related changes contrib- ute to tamoxifen-induced uterine hyperplasia, and therefore inhi- Received June 22, 2015; revised September 8, 2015; accepted September 21, bition of VEGF can improve tamoxifen's effect on the uterus both 2015; published OnlineFirst October 8, 2015.

References 1. Osborne CK. Tamoxifen in the treatment of breast cancer. N Engl J Med 9. Cheng WF, Lin HH, Torng PL, Huang SC. Comparison of endometrial 1998;339:1609–18. changes among symptomatic tamoxifen-treated and nontreated premen- 2. Wuttke M, Phillips K-A. Clinical management of women at high risk of opausal and postmenopausal breast cancer patients. Gynecol Oncol breast cancer. Curr Opin Obstet Gynecol 2015;27:6–13. 1997;66:233–7. 3. Burstein HJ, Temin S, Anderson H, Buchholz TA, Davidson NE, Gelmon KE, 10. Shang Y. Molecular mechanisms of oestrogen and SERMs in endometrial et al. Adjuvant endocrine therapy for women with hormone receptor- carcinogenesis. Nat Rev Cancer 2006;6:360–8. positive breast cancer: American Society of Clinical Oncology clinical 11. Wu H, Chen Y, Liang J, Shi B, Wu G, Zhang Y, et al. Hypomethylation- practice guideline focused update. J Clin Oncol 2014;32:2255–69. linked activation of PAX2 mediates tamoxifen-stimulated endometrial 4. Burger HG. Selective oestrogen receptor modulators. Horm Res 2000;53: carcinogenesis. Nature 2005;438:981–7. 25–9. 12. Tamm-Rosenstein K, Simm J, Suhorutshenko M, Salumets A, Metsis M. 5. Cohen I. Endometrial pathologies associated with postmenopausal tamox- Changes in the transcriptome of the human endometrial Ishikawa cancer ifen treatment. Gynecol Oncol 2004;94:256–66. cell line induced by estrogen, progesterone, tamoxifen, and mifepristone 6. Mourits MJ, De Vries EG, Willemse PH, Ten Hoor KA, Hollema H, Van der (RU486) as detected by RNA-sequencing. PloS One 2013;8:e68907. Zee AG. Tamoxifen treatment and gynecologic side effects: a review. Obstet 13. Abulaﬁa O, Sherer DM. Angiogenesis of the endometrium. Obstet Gynecol Gynecol 2001;97:855–66. 1999;94:148–53. 7. Fisher B, Costantino JP, Redmond CK, Fisher ER, Wickerham DL, Cronin 14. Hyder SM, Stancel GM, Chiappetta C, Hyder M, Stancel M, Boettger-tong WM. Endometrial cancer in tamoxifen-treated breast cancer patients: HL, et al. Uterine expression of vascular endothelial growth factor is ﬁndings from the National Surgical Adjuvant Breast and Bowel Project increased by estradiol and tamoxifen. Cancer Res 1996;3954–60. (NSABP) B-14. J Natl Cancer Inst 1994;86:527–37. 15. Roskoski R. Vascular endothelial growth factor (VEGF) signaling in tumor 8. Al-Mubarak M, Tibau A, Templeton AJ, Cescon DW, Ocana A, Seruga B, progression. Crit Rev Oncol Hematol 2007;62:179–213. et al. Extended adjuvant tamoxifen for early breast cancer: a meta-analysis. 16. Nyberg P, Xie L, Kalluri R. Endogenous inhibitors of angiogenesis. Cancer PloS One 2014;9:e88238. Res 2005;3967–79.

2848 Mol Cancer Ther; 14(12) December 2015 Molecular Cancer Therapeutics

PEDF Alleviates Tamoxifen-Induced Endometrial Hyperplasia

17. Ohno-Matsui K, Morita I, Tombran-Tink J, Mrazek D, Onodera M, Uetama 34. Sunayama J, Tsuruta F, Masuyama N, Gotoh Y. JNK antagonizes Akt- T, et al. Novel mechanism for age-related macular degeneration: an mediated survival signals by phosphorylating 14-3-3. J Cell Biol 2005;170: equilibrium shift between the angiogenesis factors VEGF and PEDF. J Cell 295–304. Physiol 2001;189:323–33. 35. Reno EM, Haughian JM, Jackson TA, Thorne AM, Bradford AP. c-Jun N- 18. Palmieri D, Watson JM, Rinehart CA. Age-related expression of PEDF/EPC- terminal kinase regulates apoptosis in endometrial cancer cells. Apoptosis 1 in human endometrial stromal fibroblasts: implications for interactive 2009;14:809–20. senescence. Exp Cell Res 1999;247:142–7. 36. Haribalaganesh R, Sheikpranbabu S, Elayappan B, Venkataraman D, 19. Cai J, Parr C, Watkins G, Jiang WG, Boulton M. Decreased pigment Gurunathan S. Pigment epithelium–derived factor down regulates hyper- epithelium-derived factor expression in human breast cancer progression. glycemia-induced apoptosis via PI3K/Akt activation in goat retinal peri- Clin Cancer Res 2006;12:3510–7. cytes. Angiogenesis 2009;12:381–9. 20. Manalo KB, Choong PFM, Becerra SP, Dass CR. Pigment epithelium- 37. Konson A, Pradeep S, D'Acunto CW, Seger R. Pigment epithelium-derived derived factor as an anticancer drug and new treatment methods following factor and its phosphomimetic mutant induce JNK-dependent apoptosis the discovery of its receptors: a patent perspective. Expert Opin Ther Pat and p38-mediated migration arrest. J Biol Chem 2011;286:3540–51. 2011;21:121–30. 38. Mueller MD, Pritts EA, Zaloudek CJ, Dreher E, Taylor RN. Regulation of 21. Becerra SP, Notario V. The effects of PEDF on cancer biology: mechanisms vascular endothelial growth factor by tamoxifen in vitro and in vivo. of action and therapeutic potential. Nat Rev Cancer 2013;13:258–71. Gynecol Obstet Invest 2003;55:119–24. 22. Chuderland D, Ben-Ami I, Friedler S, Hasky N, Ninio-Many L, Goldberg 39. Hall JM, McDonnell DP. The estrogen receptor beta-isoform (ERbeta) of K, et al. Hormonal regulation of pigment epithelium-derived factor the human estrogen receptor modulates ERalpha transcriptional activity (PEDF) expression in the endometrium. Mol Cell Endocrinol 2014;390: and is a key regulator of the cellular response to estrogens and antiestro- 85–92. gens. Endocrinology 1999;140:5566–78. 23. Chuderland D, Hasky N, Ben-Ami I, Kaplan-Kraicer R, Grossman H, Shalgi 40. Weihua Z, Saji S, M€akinen S, Cheng G, Jensen E V, Warner M, et al. Estrogen R. A physiological approach for treating endometriosis by recombinant receptor (ER) beta, a modulator of ERalpha in the uterus. Proc Natl Acad Sci pigment epithelium-derived factor (PEDF). Hum Reprod 2013;0:1–9. U S A 2000;97:5936–41. 24. Subramanian P, Notario PM, Becerra SP. Pigment epithelium-derived 41. Bircan S, Ensari A, Ozturk S, Erdogan N, Dundar I, Ortac F. Immunohis- factor receptor (PEDF-R): a plasma membrane-linked phospholipase with tochemical analysis of c-myc, c-jun and estrogen receptor in normal, PEDF binding affinity. Adv Exp Med Biol 2010;664:29–37. hyperplastic and neoplastic endometrium. Pathol Oncol Res 2005;11: 25. Matsui T, Higashimoto Y, Yamagishi S. Laminin receptor mediates anti- 32–9. inflammatory and anti-thrombogenic effects of pigment epithelium- 42. Gielen SCJP, Kuhne€ LCM, Ewing PC, Blok LJ, Burger CW. Tamoxifen derived factor in myeloma cells. Biochem Biophys Res Commun 2014;443: treatment for breast cancer enforces a distinct gene-expression profile on 847–51. the human endometrium: an exploratory study. Endocr Relat Cancer 26. Chuderland D, Ben-Ami I, Kaplan-Kraicer R, Grossman H, Ron-El R, Shalgi 2005;12:1037–49. R. The role of pigment epithelium-derived factor in the pathophysiology 43. Bai J-X, Yan B, Zhao Z-N, Xiao X, Qin W-W, Zhang R, et al. Tamoxifen and treatment of ovarian hyperstimulation syndrome in mice. J Clin represses miR-200 microRNAs and promotes epithelial-to-mesenchymal Endocrinol Metab 2013;98:E258–66. transition by up-regulating c-Myc in endometrial carcinoma cell lines. 27. Fong CJ, Burgoon LD, Williams KJ, Forgacs AL, Zacharewski TR. Compar- Endocrinology 2013;154:635–45. ative temporal and dose-dependent morphological and transcriptional 44. Chin J, Konje JC, Hickey M. Levonorgestrel intrauterine system for endo- uterine effects elicited by tamoxifen and ethynylestradiol in immature, metrial protection in women with breast cancer on adjuvant tamoxifen. ovariectomized mice. BMC Genomics 2007;8:151. Cochrane Database Syst Rev 2009;CD007245. 28. Hague S, Manek S, Oehler MK, MacKenzie IZ, Bicknell R, Rees MCP. 45. Trinh XB, Tjalma WAA, Makar AP, Buytaert G, Weyler J, van Dam PA. Use of Tamoxifen induction of angiogenic factor expression in endometrium. Br the levonorgestrel-releasing intrauterine system in breast cancer patients. J Cancer 2002;86:761–7. Fertil Steril 2008;90:17–22. 29. Miranda SP, Traiman P, Cândido EB, Lages EL, Freitas GF, Lamaita RM, 46. Jan R, Huang M, Lewis-Wambi J. Loss of pigment epithelium-derived et al. Expression of p53, Ki-67, and CD31 proteins in endometrial polyps of factor: a novel mechanism for the development of endocrine resistance postmenopausal women treated with tamoxifen. Int J Gynecol Cancer in breast cancer. Breast Cancer Res 2012;14:R146. 2010;20:1525–30. 47. Erdemoglu E, Guney€ M, Giray SG, Take G, Mungan T. Effects of metformin 30. Lacey J V, Chia VM. Endometrial hyperplasia and the risk of progression to on mammalian target of rapamycin in a mouse model of endometrial carcinoma. Maturitas 2009;63:39–44. hyperplasia. Eur J Obstet Gynecol Reprod Biol 2009;145:195–9. 31. Raica M, Cimpean AM, Ribatti D. Angiogenesis in pre-malignant condi- 48. Akn S, Aksoy DY, Cnar N, Aydn K, Karaagao glu E, Aryurek€ M, et al. Pigment tions. Eur J Cancer 2009;45:1924–34. epithelium-derived factor increases in type 2 diabetes after treatment with 32. Slomovitz BM, Coleman RL. The PI3K/AKT/mTOR pathway as a thera- metformin. Clin Endocrinol 2012;77:852–6. peutic target in endometrial cancer. Clin Cancer Res 2012;18:5856–64. 49. Yang S, Lv Q, Luo T, Liu L, Gao R, Chen S, et al. Metformin inhibits 33. Weston CR, Davis RJ. The JNK signal transduction pathway. Curr Opin expression and secretion of PEDF in adipocyte and hepatocyte via pro- Genet Dev 2002;12:14–21. moting AMPK phosphorylation. Mediators Inflamm 2013;2013:429207.

www.aacrjournals.org Mol Cancer Ther; 14(12) December 2015 2849

Pigment Epithelium−Derived Factor Alleviates Tamoxifen-Induced Endometrial Hyperplasia

Keren Goldberg, Hadas Bar-Joseph, Hadas Grossman, et al.

Mol Cancer Ther 2015;14:2840-2849. Published OnlineFirst October 8, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-15-0523

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2015/10/08/1535-7163.MCT-15-0523.DC1

Cited articles This article cites 46 articles, 7 of which you can access for free at: http://mct.aacrjournals.org/content/14/12/2840.full#ref-list-1

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://mct.aacrjournals.org/content/14/12/2840. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.