Human Pathology (2015) 46, 1138–1146
www.elsevier.com/locate/humpath
Original contribution
Mitochondrial estrogen receptor β2 drives antiapoptotic pathways in advanced serous ovarian cancer☆,☆☆ Alessandra Ciucci PhD a , Gian Franco Zannoni MD b , Daniele Travaglia AAS a , Giovanni Scambia MD a , Daniela Gallo PhD a,⁎ a
Department of Obstetrics and Gynecology, Catholic University of the Sacred Heart, 8-00168, Rome, Italy Department of Histopathology, Catholic University of the Sacred Heart, 8-00168, Rome, Italy
b
Received 26 January 2015; revised 19 March 2015; accepted 30 March 2015
Keywords: ERβ2; ERβcx; Ovary; Ovarian carcinomas; Mitochondria
Summary We previously showed an unfavorable prognostic role of the cytoplasmic estrogen receptor β2 (cERβ2) in serous ovarian cancer. Here we aimed to investigate molecular determinants in cell survival function of cERβ2 in this malignant disease. We used immunohistochemistry to evaluate differences in apoptosis (quantified by the expression of cleaved caspase-3) and cell proliferation (quantified by the expression of Ki-67) in 56 advanced serous ovarian cancer cases, stratified according to the absence or presence of estrogen receptor β2 (ERβ2) protein in the cytoplasmic compartment (31 cERβ2− and 25 cERβ2+ cases, respectively). Thereafter, by immunofluorescence, we visualized the subcellular distribution of ERβ2, and by the proximity ligation assays, we characterized in situ its ability to interact with other proteins specifically involved in the apoptosis cascade. Finally, we assessed cytochrome c expression by immunohistochemistry. We demonstrated that, although not affecting tumor proliferation, cytoplasmic ERβ2 expression was indeed associated to a lower apoptotic rate in ovarian cancer cases. Then, we proved that cERβ2 is targeted to mitochondria where it interacts as a binding partner with BAD (B-cell lymphoma [Bcl] 2–associated death promoter). This interaction, precluding the Bcl-xL (B-cell lymphoma extra large)/BAD heterodimer formation, inhibited Bax (Bcl-2–like protein 4) oligomerization, the release of cytochrome c, and ultimately apoptosis. In conclusion, we provide in vivo mechanistic evidence for an antiapoptotic function of mitochondrial ERβ2, a finding supporting the value of its cytoplasmic expression as an unfavorable prognostic biomarker for serous ovarian cancer. © 2015 Elsevier Inc. All rights reserved.
1. Introduction
☆
Competing interests: The authors declare no conflict of interest. Funding/Support: There is no support from any organization for the submitted work. ⁎ Corresponding author. Department of Obstetrics and Gynecology, Catholic University of the Sacred Heart, Largo A. Gemelli, 8-00168, Rome, Italy. E-mail address: [email protected] (D. Gallo). ☆☆
http://dx.doi.org/10.1016/j.humpath.2015.03.016 0046-8177/© 2015 Elsevier Inc. All rights reserved.
In Europe, ovarian cancer accounts for approximately 66 000 new cases and 43 000 deaths each year [1]. Most women have advanced disease, and despite debulking surgery and postsurgical treatment with platinum/taxane-based chemotherapy, most of them will relapse within 2 years, and in most, the disease remains incurable [2,3]. Despite its importance and intensive research, the causes and natural
Mitochondrial ERβ2 in ovarian cancer history of ovarian cancer are among the least understood major human malignancies. The importance of estrogen signaling in the development and progression of ovarian cancer has been assumed to be less significant than in breast or endometrial malignancies, although preclinical studies and clinical data have shown that not only normal ovaries but also many malignant ovarian tumors can be considered as endocrine related and hormone dependent (reviewed by Gallo et al and Simpkins et al [4,5]). Estrogens exert their action through 2 estrogen receptors (ERα and ERβ) that are encoded by separate genes, with at least 5 splice variant isoforms of the ERβ gene product being described (ER β1-ERβ5). The wild-type ERβ (ERβ1) encodes the full-length, 530–amino acid receptor protein, whereas ERβ2 (ERβcx) to ERβ5, which use alternative exons, encode variant receptors with different C-termini [6]. ERβ1 is thought to be the only fully functional isoform that is able to bind ligand, whereas the C-terminally truncated ERβ variants such as ERβ2/cx and ERβ5 are thought to have little activity of their own but may modulate estrogen action when dimerized with ERβ1 or ERα (reviewed by Gallo et al [4]). However, some authors suggest that ERβ2, ERβ4, and ERβ5 might participate in gene regulation, possibly through ligand-independent transcriptional properties [7,8]. In the healthy ovary, the levels of ERβ are high and predominate over ERα, with ERβ1, ERβ2, and ERβ5 being the most represented isoforms [6,9,10]. A high percentage of ovarian tumors also express ERβ, and a progressive decline occurs during development or progression of ovarian cancer [9,11] (reviewed by Gallo et al [4] and Häring et al [12]). This complex scenario is further complicated by the emerging evidence that subcellular ERβ isoforms localization could play different role in the development and progression of some tumor types (reviewed by Gallo et al [4]), and in this context, previous findings from our department revealed that, in advanced serous ovarian cancer, high cytoplasmic ERβ2 expression may define patients with aggressive biology and possibly resistant to chemotherapy [13]. Because insights into the molecular pathways involved in hormone-induced ovarian carcinogenesis may provide the basis for future advances in the treatment of this malignant disease, the present study aimed to investigate molecular determinants in cell survival function of cytoplasmic ERβ2. To this end, we used immunohistochemistry, immunofluorescence, and proximity ligation assays (PLA) technique, demonstrating for the first time ever in serous ovarian cancer samples that ERβ2 localizes in the mitochondrial compartment and interacts with proteins specifically involved in the apoptosis cascade, thus ultimately inhibiting cancer cell death.
2. Materials and methods 2.1. Patients The study included 56 patients with advanced serous ovarian cancer admitted to the Gynecologic Oncology Unit,
1139 Catholic University of Rome, between March 2000 and December 2008. The characteristics of the whole series of patients have already been published [13], and using immunohistochemical data obtained in this previous investigation, cases were stratified according to the absence or presence of ERβ2 in the cytoplasmic compartment (cERβ2− and cERβ2+, respectively). In our institution, a written informed consent is routinely requested from patients for collection of their clinical data as well as paraffin-embedded sections for research use.
2.2. Antibody validation The mouse monoclonal antibody we used in the study (clone 57/3, Serotec Ltd; Oxford, United Kingdom) is raised against the unique COOH-terminal peptides of ERβ2. Antibody specificity was validated by peptide absorption studies with a peptide specific for ERβ2 (CGMKMETLL PEATMEQ) that was used in neutralization studies. Antibody was diluted to its working concentration and incubated with 10 times excess of the peptide overnight at 4°C, before application to tissue.
2.3. Immunohistochemistry Three-micrometer-thick paraffin sections were mounted on Superfrost-coated slides and dried overnight. The immunohistochemical staining for Ki-67 (clone K-2 Prediluted, Ventana Medical Systems; Tucson, AZ) was performed using the Ventana Benchmark XT autostainer. Immunohistochemistry for cleaved caspase-3 (CC3, clone 5A1E, dilution 1:100, Cell Signaling Technology; Leiden, The Netherlands) and cytochrome c (clone 136F3, dilution 1:200, Cell Signaling Technology) was performed using a labeled streptavidin-biotin-peroxidase method. The sections were deparaffinized in xylene and rehydrated in graded solutions of ethanol; antigen retrieval procedure was performed by microwave oven heating in citrate buffer (pH 6). The endogenous peroxidase was blocked with 3% vol/vol H2O2 for 5 minutes. Sections were incubated with 20% vol/vol normal goat serum for 30 minutes at room temperature to reduce nonspecific binding. Cells expressing CC3 or cytochrome c were identified after overnight incubation at 4°C. Sections were incubated with the secondary, antimouse EnVision System-HRP (DakoCytomation, Carpinteria, CA) for 30 minutes, at room temperature. The slides were developed with diaminobenzidine (DAB; DAB Substrate System, DakoCytomation), counterstained with Mayer hematoxylin, dehydrated in ethanol and xylene, and finally mounted. Expression of Ki-67, CC3, and cytochrome c was evaluated by considering the number of cells exhibiting immunoreaction in a minimum of 500 histologically identified neoplastic cells. Data are expressed as percentages of marker-positive cells among total cells. The analysis of all tissue sections was performed by 2 of the authors (G.F.Z. and A.C.) in a blinded fashion. In case
1140 of disagreement, sections were submitted to a joint re-evaluation by a multiheaded microscope.
2.4. Immunofluorescence of fixed paraffin-embedded ovarian tissue sections Inasmuch as nonspecific tissue staining has traditionally been considered an obstacle in labeling paraffin-embedded sections using immunofluorescence techniques, specific procedures were followed to reduce both nonspecific and background staining [14]. Three-micrometer-thick paraffin sections were mounted on Superfrost coated slides and dried overnight (Thermo Scientific, Menzel GmbH & Co., KG, Braunschweig, Germany). The sections were deparaffinized in xylene, rehydrated in graded solutions of ethanol, and rinsed for 5 minutes in distilled water. Antigen retrieval procedure was performed by microwave oven heating in citrate buffer (pH 6). Sections were incubated with 20% normal goat serum for 30 minutes at room temperature and then incubated at 4°C overnight with primary antibody (ERβ2, clone 57/3, dilution 1:100, Serotec Ltd). The optimal dilution of the primary antibody had been established before by immunoenzymatic staining using conventional techniques (2-stage immunoperoxidase technique, DAB). After overnight incubation, slides were washed in tris-buffered saline and incubated in the dark for 1 hour at room temperature with secondary antibody antimouse Alexa Fluor 488 conjugate (Life Technologies Inc, Milan, Italy). The slides were subsequently incubated with MitoTracker Red (150 nM in tris-buffered saline for 15 minutes, Life Technologies Inc), a widely used probe for mitochondria staining [15]. After extensive washing in phosphate-buffered saline 1% Tween 20, tissues were stained with 4′,6-diamidino-2-phenylindole (DAPI, 1.5 μg/mL) and mounted in Vectashield Mounting Medium (Vector Laboratories; Burlingame, Ontario, Canada). Slides were observed under the fluorescence microscope (Leica; Milan, Italy) using a ×40 or a ×100 oil immersion objective.
2.5. Duolink The Duolink assay (Sigma-Aldrich; St. Louis, MO) based on in situ PLA allows precise detection of protein-protein interactions in fixed cells and tissue samples, in their correct cellular context and at physiologically relevant expression levels [16,17]. Specifically, PLA allows protein complex to be represented as an amplifiable DNA molecule. Recognition is mediated by proximity probes consisting of antibodies coupled with oligonucleotides. Upon dual binding of the proximity probes, the oligonucleotides direct the formation of a circular DNA molecule, which is then amplified by rolling-circle replication. The localized concatemeric product is then detected with fluorescent probes. The assay results in a fluorescent signal in the form of a spot when the 2 proteins of interest are closer than 40 nm; each red dot represents a molecular interaction between the 2 proteins of interest.
A. Ciucci et al. For analysis, 3-μm-thick paraffin sections were mounted on Superfrost coated slides, dried overnight, and subjected to antigen retrieval by microwave cooking in citrate buffer (pH 6.0). The PLA protocol was followed according to the manufacturer’s instructions (Olink Bioscience; Uppsala, Sweden). After blocking (Olink Bioscience), the following primary antibodies were used: mouse monoclonal IgG1 ERβ2 (clone 57/3, Serotec Ltd, dilution 1:100); rabbit polyclonal IgG BAD (B-cell lymphoma [Bcl] 2–associated death promoter, Serotec Ltd, 1:200); mouse monoclonal IgG1 Bcl-xL (B-cell lymphoma-extra large, clone H-5, Santa Cruz, CA, 1:100), rabbit polyclonal IgG Bax (Bcl-2–like protein 4, clone P-19, Santa Cruz, CA, 1:50) with incubation at 4°C overnight. The optimal experimental conditions for primary antibodies had been established before by immunoenzymatic staining using conventional techniques (2-stage immunoperoxidase technique, DAB) and immunofluorescence (Supplementary Fig. S1). On the next day, PLA minus and PLA plus probes (containing the secondary antibodies conjugated with oligonucleotides) were added and incubated 1 hour at 37°C. The ligation reaction was carried out at 37°C for 30 minutes in a humid chamber followed by washing. Slides were then incubated with the amplification mix for 2 hours at 37°C in a darkened humidified chamber. After washing, slides were mounted using the mounting media supplied with the kit.
2.6. Statistical analysis Data were analyzed for homogeneity of variance using an F test. If the group variance appeared homogenous, a Student t test was used. If the variances were heterogeneous, log or reciprocal transformations were made in an attempt to stabilize the variances. If the variances remained heterogeneous, a nonparametric test such as the Mann-Whitney U test was used. Correlations between variables were identified using the Spearman rank correlation. P values are for 2-sided tests; P values ≤ .05 were considered statistically significant. All statistical analyses were performed using the GraphPad Prism5 Software (San Diego, CA).
3. Results 3.1. Antibody validation Specific nuclear and cytoplasmic staining was completely abolished when ERβ2 primary antibody was preabsorbed with its respective blocking peptide (Fig. 1A).
3.2. Apoptosis and cell proliferation rates in cERβ2− and cERβ2+ cases Fifty-six advanced serous ovarian cancer cases were stratified according to the absence or presence of ERβ2 in the cytoplasmic compartment (cERβ2− and cERβ2+, respectively) using immunohistochemical data obtained in our
Mitochondrial ERβ2 in ovarian cancer previous investigation [13]. Fig. 1A shows representative pictures of cERβ2− (n = 31) and cERβ2+ (n = 25) tumors. The antibody specific for active caspases-3 stained the cytoplasm of cells whose nuclear morphology was consistent with apoptosis along with the cytoplasm of morphologically healthy-looking cells, thus suggesting that this antibody recognized activated protein at the early stage of apoptosis. Significantly higher levels of CC3 were detected in cERβ2− when compared with cERβ2+ cases (5.6 ± 1.3 versus 2.6 ± 0.4, respectively, mean ± SEM, P b .05, Fig. 1B). Conversely, tumor proliferation, quantified by the expression of Ki-67, did not appear to be significantly different between the 2 populations (54.2 ± 4.5 versus 57.8 ± 4.4 in cERβ2− and cERβ2+ cases, respectively, mean ± SEM, Fig. 1C). Overall, these findings suggest that cytoplasmic ERβ2 affects
1141 ovarian cancer development by inhibiting pathways that promote cell death rather than directly stimulate cell growth.
3.3. Mitochondrial localization of the ERβ2 protein Determining the subcellular localization of a protein within a cell is often an essential step toward understanding its function, and we used immunofluorescence to detect endogenous ERβ2 protein localization in serous ovarian cancer tissue. Tumor sections were incubated with anti-ERβ2 and Alexa Fluor 488 (antimouse), whereas MitoTracker and DAPI were used for visualization of mitochondrial and nuclear staining. ERβ2 staining in the cytosol presented a punctuate distribution, similar to that of mitochondria. The
Fig. 1 A, Representative pictures for nuclear (cERβ2−) or nuclear/cytoplasmic (cERβ2+) ERβ2 staining in ovarian cancer tissues. Nuclear and cytoplasmic ERβ2 staining was completely abolished after preabsorption with the specific blocking peptide. B, Bar chart showing the distribution of CC3-positive cells in cERβ2− and cERβ2+ cases (Student t test, *P b .05) and representative pictures for CC3 staining showing tumors with high or low levels. C, Bar chart showing the distribution of Ki-67–positive cells in cERβ2− and cERβ2+ cases and representative tumor immunostaining.
1142 overlay of the Alexa dye staining (green) with mitochondrial control MitoTracker Red showed indeed that ERβ2 protein staining coincided precisely with MitoTracker Red staining (Fig. 2). Immunofluorescent staining of 2 different serous ovarian cancer cell lines (ie, Hey and COV-318) showed only in Hey cells both nuclear and mitochondrial localization of ERβ2 protein, whereas COV-318 only displayed nuclear expression (Supplementary Fig. S2).
3.4. Mitochondrial ERβ2 physically interacts with the proapoptotic protein BAD Using a conventional coimmunoprecipitation approach, Zhang et al [18] demonstrated that mitochondrial ERβ physically interacts with the proapoptotic protein BAD, in a ligand-independent manner in lung cancer cells. To examine whether we could detect endogenous complexes between mitochondrial ERβ2 and BAD in serous ovarian cancer cases
A. Ciucci et al. in situ, we performed PLA, a technology capable of detecting protein interactions with high specificity and sensitivity. In cases with only nuclear ERβ2 expression (cERβ2−) there were no significant PLA signals (red dots) for dimers ERβ2/BAD, whereas in tumor samples showing mitochondrial expression, PLA signals could be easily recognized (Fig. 3). Because BAD is a strong heterodimerizing partner of Bcl-xL, to confirm our results, we evaluated the presence of Bcl-xL/BAD heterodimer in the 2 subcohort of cases, showing that the PLA signal Bcl-xL/BAD was only evident in tumors with exclusive nuclear ERβ2 expression, whereas no similar finding was observed in samples showing cytoplasmic/mitochondrial expression (cERβ2+) (Fig. 3). In turn, these latter cases showed Bcl-xL/Bax heterodimers (Fig. 3), a condition that inhibits Bax oligomerization required for mitochondrial outer membrane permeabilization during apoptosis [19]. Single-antibody control reactions did not result in significant PLA signals (not shown).
Fig. 2 Fluorescence microscopy localization of ERβ2 in serous ovarian cancer tissues. Representative pictures of exclusive nuclear (A) and of nuclear/cytoplasmic (B) ERβ2 expression. Nuclei were stained with DAPI (blue), mitochondria with MitoTracker Red (red) and ERβ2 with antimouse Alexa Fluor 488 conjugate (green). Merged images in (B) show the mitochondrial localization of cytoplasmic ERβ2.
Mitochondrial ERβ2 in ovarian cancer
1143
Fig. 3 Mitochondrial ERβ2/BAD (A), Bcl-xL/BAD (B), Bcl-xL/Bax (C), complex formation in ovarian cancer samples, as visualized by the PLA technique. Red dot in merged photomicrographs indicates physical interaction of proteins.
3.5. Higher level of cytochrome c release in cERβ2− cases Because overexpression of the proapoptotic protein Bax has been shown to trigger mitochondrial cytochrome c efflux into the cytosol and cell death [20,21], we examined the expression of cytochrome c and its cellular localization by immunohistochemistry, in our series of ovarian cancer cases. Tumor samples with only nuclear ERβ2 expression (cERβ2−) exhibited higher amounts of cytochrome c in the cytosolic compartment when compared with cERβ2+ cases (P b .05, Fig. 4A), these latter mostly showing a punctuate mitochondrial staining. Because cell-free assays of caspase activation have revealed that cytochrome c is a powerful inducer of procaspase-3 cleavage [22], we assessed the correlation between cytosolic levels of cytochrome c and CC3 expression in our samples, results showing a significant positive correlation (r = 0.388, P b .01, Fig. 4B).
4. Discussion We previously showed that cERβ2 expression is an independent poor prognostic factor in advanced serous ovarian cancer (P = .006), the 5-year survival rate being nearly 28% for patients who did express cERβ2 and 60% in
patients who did not; we also demonstrated that cERβ2 expression is significantly associated with chemoresistance (P = .03) [13]. Notably, similar findings were also reported in familial and sporadic breast cancer, where cytoplasmic ERβ2 staining was associated to the worst survival prognostic [23,24]. Despite these clinical data, the functional nongenomic role of ERβ2 and the mechanisms for its potential influence on cancer biology are far from being elucidated. To address this issue, thus adding further insights in clarifying the discrete role of each specific ERβ isoform, we investigate the molecular determinants in cell survival function of cytoplasmic ERβ2 in ovarian cancer. We firstly demonstrate that, although not affecting the proliferative potential of cancer cells, cytoplasmic ERβ2 is associated to lower apoptotic rates in ovarian cancer cases. Then, we provide strong evidence that ERβ2 is a mitochondrial component in cancer cells that can act as a binding partner of proteins involved in the apoptosis cascade. To examine the in situ existence of endogenous ERβ2 protein complexes in serous ovarian cancer cases, we used the PLA technique on formalin-fixed paraffin sections. Indeed, the PLA assay is a revolutionary technique that enables unprecedented specificity, sensitivity, and accurate visualization of protein-protein interactions [25-27], overcoming some limitations of the conventional coimmunoprecipitation (Co-IP) method, as false-positive detection, or failing to
1144
A. Ciucci et al.
Fig. 4 A, Representative pictures for mitochondrial or cytosolic cytochrome c staining (a detail on the side) in ovarian cancer tissues and bar chart showing the distribution of cytosolic cytochrome c–positive cells in cERβ2− and cERβ2+ cases (Mann-Whitney U test, *P b .05). B, The Spearman rank correlation showing a significant positive correlation between CC3 and cytosolic cytochrome c–positive cells in ovarian cancer samples. **P b .01.
detect weak or transient protein interactions [28]. Notably, by using this innovative technique, we have been able to demonstrate, for the first time ever in serous ovarian cancer samples, that mitochondrial ERβ2 forms protein complexes with BAD at physiologically relevant expression levels and in their correct cellular context. We also show that, in tumor exhibiting ERβ2/BAD interaction, the presence of heterodimers Bcl-xL/Bax is evident, an event that, inhibiting Bax oligomerization, reduces the release of cytochrome c from the mitochondria into the cytoplasm and ultimately inhibits apoptosis. Cytochrome c is indeed a protein that normally is well sequestered in the mitochondrial intermembrane space and the release of cytochrome c into the cytosol can be induced by proapoptotic members of Bcl-2 family (such as Bax, Bad, and Bid) [22,29]. Although cytochrome c is not the sole mitochondrial intermembrane protein capable of activating caspases, the cytochrome c–stimulated pathway of caspase activation is the best characterized, with cell-free assays demonstrating that cytochrome c is a powerful inducer of procaspase-3 cleavage [22]. The reduced levels of CC3 in cERβ2+ tumor samples thus possibly occur as a result of the mitochondrial ERβ2–mediated inhibition of cytochrome c release. Although the difference observed in
the basal apoptotic levels between the 2 populations (cERβ2+ and cERβ2− cases) is not large, it could promote different response rates to drug-induced cytotoxicity and thus a different clinical outcome. Indeed, it is well known that attenuation of apoptosis in ovarian cancer cells contributes to the resistance to subsequent chemotherapy and likely plays an important role in tumor progression [3]. Such a hypothesis is actually supported by our previous results on the association between cERβ2 expression and chemoresistance in patients with serous ovarian cancer [13]. Altogether, our findings provide mechanistic evidence that supports the value of cytoplasmic ERβ2 expression as an unfavorable prognostic biomarker for serous ovarian cancer. Our results are in keeping with previous literature data showing the presence of (total) ERβ in the mitochondria in several cell types, including human non–small cell lung cancer (NSCLC) cells, human breast cancer MCF-7 cells, human osteosarcoma SaOS-2 cells, and liver cancer HepG2 cells [30-32] and with the hypothesis proposed by Zhang et al [18] of a ligand-independent antiapoptotic function of ERβ in lung cancer cells. However, most of these studies did not investigate the contribution of the different ERβ isoforms to this mechanism. Indeed, although our understanding of ERβ
Mitochondrial ERβ2 in ovarian cancer functions has broadened since the discovery of ERβ isoforms [6,33], little is known about their discrete function, expression, and regulation in both normal and tumor tissues as well as about their ability to operate in a ligand-independent or -dependent manner, according to different cellular contexts. On the other hand, there is compelling evidence that supports distinct prognostic significance of ERβ isoforms, at least in some cancer types (reviewed by Gallo et al [4]). Thus, implementation research is certainly needed to uncover the complex biochemical chain of events controlled by the different ERβ isoforms. The availability of robust validated antibodies and of innovative technologies, enabling both measurements of nuclear and cytoplasmic ERβs and exploration of their potential mechanism of action in a relevant cellular context should stimulate further specific investigations to more clearly define their distinct role as outcome factors. In this respect, the potential contribution of cytoplasmic ERβ1 and/or ERβ5 in apoptosis inhibition deserves additional studies, although, in our series, only cytoplasmic ERβ2 was seen to be associated with shorter survival and chemoresistance. In summary, we provide novel insights into the functional nongenomic role of cytoplasmic ERβ2, showing for the first time an antiapoptotic role in advanced serous ovarian cancer. The recognition of molecular features linking cytoplasmic ERβ2 expression to tumor aggressiveness could allow the identification of high-risk patients at the time of the initial diagnosis, thus guiding the use of alternative treatment approaches. Specifically, in these patients, pharmacologic targeting of critical antiapoptotic Bcl-2 proteins might hold great promise for personalized therapeutic interventions.
Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.humpath.2015.03.016.
References [1] Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013;49:1374-403. [2] Jayson GC, Kohn EC, Kitchener HC, Ledermann JA. Ovarian cancer. Lancet 2014;384:1376-88 [Review]. [3] Vasey PA. Resistance to chemotherapy in advanced ovarian cancer: mechanisms and current strategies. Br J Cancer 2003;89(Suppl. 3): S23-8. [4] Gallo D, De Stefano I, Prisco MG, Scambia G, Ferrandina G. Estrogen receptor beta in cancer: an attractive target for therapy. Curr Pharm Des 2012;18:2734-57. [5] Simpkins F, Garcia-Soto A, Slingerland J. New insights on the role of hormonal therapy in ovarian cancer. Steroids 2013;78:530-7. [6] Moore JT, McKee DD, Slentz-Kesler K, et al. Cloning and characterization of human estrogen receptor beta isoforms. Biochem Biophys Res Commun 1998;247:75-8.
1145 [7] Poola I, Abraham J, Baldwin K, Saunders A, Bhatnagar R. Estrogen receptors beta4 and beta5 are full length functionally distinct ERbeta isoforms: cloning from human ovary and functional characterization. Endocrine 2005;27:227-38. [8] Dey P, Jonsson P, Hartman J, Williams C, Ström A, Gustafsson JÅ. Estrogen receptors β1 and β2 have opposing roles in regulating proliferation and bone metastasis genes in the prostate cancer cell line PC3. Mol Endocrinol 2012;26:1991-2003. [9] Suzuki F, Akahira J, Miura I, et al. Loss of estrogen receptor beta isoform expression and its correlation with aberrant DNA methylation of the 5‵-untranslated region in human epithelial ovarian carcinoma. Cancer Sci 2008;99:2365-72. [10] Drummond AE, Fuller PJ. Ovarian actions of estrogen receptor-β: an update. Semin Reprod Med 2012;30:32-8. [11] De Stefano I, Zannoni GF, Prisco MG, et al. Cytoplasmic expression of estrogen receptor beta (ERβ) predicts poor clinical outcome in advanced serous ovarian cancer. Gynecol Oncol 2011;122:573-9. [12] Häring J, Schüler S, Lattrich C, Ortmann O, Treeck O. Role of estrogen receptor β in gynecological cancer. Gynecol Oncol 2012;127:673-6. [13] Ciucci A, Zannoni GF, Travaglia D, Petrillo M, Scambia G, Gallo D. Prognostic significance of the estrogen receptor beta (ERβ) isoforms ERβ1, ERβ2, and ERβ5 in advanced serous ovarian cancer. Gynecol Oncol 2014;132:351-9. [14] Bataille F, Troppmann S, Klebl F, et al. Multiparameter immunofluorescence on paraffin-embedded tissue sections. Appl Immunohistochem Mol Morphol 2006;14:225-8. [15] Murriel CL, Churchill E, Inagaki K, Szweda LI, Mochly-Rosen D. Protein kinase Cdelta activation induces apoptosis in response to cardiac ischemia and reperfusion damage: a mechanism involving BAD and the mitochondria. J Biol Chem 2004;279:47985-91. [16] Söderberg O, Gullberg M, Jarvius M, et al. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat Methods 2006;3:995-1000. [17] Baan B, Pardali E, ten Dijke P, van Dam H. In situ proximity ligation detection of cJun/AP-1 dimers reveals increased levels of cJun/Fra1 complexes in aggressive breast cancer cell lines in vitro and in vivo. Mol Cell Proteomics 2010;9:1982-90. [18] Zhang G, Yanamala N, Lathrop KL, Zhang L, Klein-Seetharaman J, Srinivas H. Ligand-independent antiapoptotic function of estrogen receptor-beta in lung cancer cells. Mol Endocrinol 2010;24:1737-47. [19] Ding J, Mooers BH, Zhang Z, et al. After embedding in membranes antiapoptotic Bcl-XL protein binds both Bcl-2 homology region 3 and helix 1 of proapoptotic Bax protein to inhibit apoptotic mitochondrial permeabilization. J Biol Chem 2014;289:11873-96. [20] Eskes R, Antonsson B, Osen-Sand A, et al. Bax-induced cytochrome C release from mitochondria is independent of the permeability transition pore but highly dependent on Mg2+ ions. J Cell Biol 1998;143: 217-24. [21] Rosse T, Olivier R, Monney L, et al. Bcl-2 prolongs cell survival after Bax-induced release of cytochrome c. Nature 1998;391:496-9. [22] Green D, Kroemer G. The central executioners of apoptosis: caspases or mitochondria? Trends Cell Biol 1998;8:267-71. [23] Shaaban AM, Green AR, Karthik S, et al. Nuclear and cytoplasmic expression of ERbeta1, ERbeta2, and ERbeta5 identifies distinct prognostic outcome for breast cancer patients. Clin Cancer Res 2008; 14:5228-35. [24] Yan M, Rayoo M, Takano EA, kConFab Investigators, Fox SB. Nuclear and cytoplasmic expressions of ERβ1 and ERβ2 are predictive of response to therapy and alters prognosis in familial breast cancers. Breast Cancer Res Treat 2011;126:395-405. [25] Zhu L, Koistinen H, Landegren U, Stenman U-H. Proximity ligation measurement of the complex between prostate specific antigen and a1prostate inhibitor. Clin Chem 2009;55:1665-71. [26] Leuchowius K-J, Weibrecht I, Landegren U, Gedda L, Soderberg O. Flow cytometric in situ proximity ligation analysis of protein interactions and post-translational modification of the epidermal growth factor receptor family. Cytometry A 2009;75A:833-9.
1146 [27] Aubele M, Spears M, Ludyga N, et al. In situ quantification of HER2protein tyrosine kinase 6 (PTK6) protein-protein complexes in paraffin sections from breast cancer tissues. Br J Cancer 2010;103:663-7. [28] Berggård T, Linse S, James P. Methods for the detection and analysis of protein-protein interactions. Proteomics 2007;7:2833-42. [29] Li B, Dou QP. Bax degradation by the ubiquitin/proteasomedependent pathway: involvement in tumor survival and progression. Proc Natl Acad Sci U S A 2000;97:3850-5. [30] Chen JQ, Delannoy M, Cooke C, Yager JD. Mitochondrial localization of ERα and ERβ in human MCF7 cells. Am J Physiol Endocrinol Metab 2004;286:E1011-22.
A. Ciucci et al. [31] Solakidi S, Psarra AM, Sekeris CE. Differential subcellular distribution of estrogen receptor isoforms: localization of ERα in the nucleoli and ERβ in the mitochondria of human osteosarcoma SaOS-2 and hepatocarcinoma HepG2 cell lines. Biochim Biophys Acta 1745;2005: 382-92. [32] Zhang G, Liu X, Farkas AM, et al. Estrogen receptor β functions through nongenomic mechanisms in lung cancer cells. Mol Endocrinol 2009;23:146-56. [33] Leung YK, Mak P, Hassan S, Ho SM. Estrogen receptor (ER)-beta isoforms: a key to understanding ER-beta signaling. Proc Natl Acad Sci U S A 2006;103:13162-7.
www.elsevier.com/locate/humpath
Original contribution
Mitochondrial estrogen receptor β2 drives antiapoptotic pathways in advanced serous ovarian cancer☆,☆☆ Alessandra Ciucci PhD a , Gian Franco Zannoni MD b , Daniele Travaglia AAS a , Giovanni Scambia MD a , Daniela Gallo PhD a,⁎ a
Department of Obstetrics and Gynecology, Catholic University of the Sacred Heart, 8-00168, Rome, Italy Department of Histopathology, Catholic University of the Sacred Heart, 8-00168, Rome, Italy
b
Received 26 January 2015; revised 19 March 2015; accepted 30 March 2015
Keywords: ERβ2; ERβcx; Ovary; Ovarian carcinomas; Mitochondria
Summary We previously showed an unfavorable prognostic role of the cytoplasmic estrogen receptor β2 (cERβ2) in serous ovarian cancer. Here we aimed to investigate molecular determinants in cell survival function of cERβ2 in this malignant disease. We used immunohistochemistry to evaluate differences in apoptosis (quantified by the expression of cleaved caspase-3) and cell proliferation (quantified by the expression of Ki-67) in 56 advanced serous ovarian cancer cases, stratified according to the absence or presence of estrogen receptor β2 (ERβ2) protein in the cytoplasmic compartment (31 cERβ2− and 25 cERβ2+ cases, respectively). Thereafter, by immunofluorescence, we visualized the subcellular distribution of ERβ2, and by the proximity ligation assays, we characterized in situ its ability to interact with other proteins specifically involved in the apoptosis cascade. Finally, we assessed cytochrome c expression by immunohistochemistry. We demonstrated that, although not affecting tumor proliferation, cytoplasmic ERβ2 expression was indeed associated to a lower apoptotic rate in ovarian cancer cases. Then, we proved that cERβ2 is targeted to mitochondria where it interacts as a binding partner with BAD (B-cell lymphoma [Bcl] 2–associated death promoter). This interaction, precluding the Bcl-xL (B-cell lymphoma extra large)/BAD heterodimer formation, inhibited Bax (Bcl-2–like protein 4) oligomerization, the release of cytochrome c, and ultimately apoptosis. In conclusion, we provide in vivo mechanistic evidence for an antiapoptotic function of mitochondrial ERβ2, a finding supporting the value of its cytoplasmic expression as an unfavorable prognostic biomarker for serous ovarian cancer. © 2015 Elsevier Inc. All rights reserved.
1. Introduction
☆
Competing interests: The authors declare no conflict of interest. Funding/Support: There is no support from any organization for the submitted work. ⁎ Corresponding author. Department of Obstetrics and Gynecology, Catholic University of the Sacred Heart, Largo A. Gemelli, 8-00168, Rome, Italy. E-mail address: [email protected] (D. Gallo). ☆☆
http://dx.doi.org/10.1016/j.humpath.2015.03.016 0046-8177/© 2015 Elsevier Inc. All rights reserved.
In Europe, ovarian cancer accounts for approximately 66 000 new cases and 43 000 deaths each year [1]. Most women have advanced disease, and despite debulking surgery and postsurgical treatment with platinum/taxane-based chemotherapy, most of them will relapse within 2 years, and in most, the disease remains incurable [2,3]. Despite its importance and intensive research, the causes and natural
Mitochondrial ERβ2 in ovarian cancer history of ovarian cancer are among the least understood major human malignancies. The importance of estrogen signaling in the development and progression of ovarian cancer has been assumed to be less significant than in breast or endometrial malignancies, although preclinical studies and clinical data have shown that not only normal ovaries but also many malignant ovarian tumors can be considered as endocrine related and hormone dependent (reviewed by Gallo et al and Simpkins et al [4,5]). Estrogens exert their action through 2 estrogen receptors (ERα and ERβ) that are encoded by separate genes, with at least 5 splice variant isoforms of the ERβ gene product being described (ER β1-ERβ5). The wild-type ERβ (ERβ1) encodes the full-length, 530–amino acid receptor protein, whereas ERβ2 (ERβcx) to ERβ5, which use alternative exons, encode variant receptors with different C-termini [6]. ERβ1 is thought to be the only fully functional isoform that is able to bind ligand, whereas the C-terminally truncated ERβ variants such as ERβ2/cx and ERβ5 are thought to have little activity of their own but may modulate estrogen action when dimerized with ERβ1 or ERα (reviewed by Gallo et al [4]). However, some authors suggest that ERβ2, ERβ4, and ERβ5 might participate in gene regulation, possibly through ligand-independent transcriptional properties [7,8]. In the healthy ovary, the levels of ERβ are high and predominate over ERα, with ERβ1, ERβ2, and ERβ5 being the most represented isoforms [6,9,10]. A high percentage of ovarian tumors also express ERβ, and a progressive decline occurs during development or progression of ovarian cancer [9,11] (reviewed by Gallo et al [4] and Häring et al [12]). This complex scenario is further complicated by the emerging evidence that subcellular ERβ isoforms localization could play different role in the development and progression of some tumor types (reviewed by Gallo et al [4]), and in this context, previous findings from our department revealed that, in advanced serous ovarian cancer, high cytoplasmic ERβ2 expression may define patients with aggressive biology and possibly resistant to chemotherapy [13]. Because insights into the molecular pathways involved in hormone-induced ovarian carcinogenesis may provide the basis for future advances in the treatment of this malignant disease, the present study aimed to investigate molecular determinants in cell survival function of cytoplasmic ERβ2. To this end, we used immunohistochemistry, immunofluorescence, and proximity ligation assays (PLA) technique, demonstrating for the first time ever in serous ovarian cancer samples that ERβ2 localizes in the mitochondrial compartment and interacts with proteins specifically involved in the apoptosis cascade, thus ultimately inhibiting cancer cell death.
2. Materials and methods 2.1. Patients The study included 56 patients with advanced serous ovarian cancer admitted to the Gynecologic Oncology Unit,
1139 Catholic University of Rome, between March 2000 and December 2008. The characteristics of the whole series of patients have already been published [13], and using immunohistochemical data obtained in this previous investigation, cases were stratified according to the absence or presence of ERβ2 in the cytoplasmic compartment (cERβ2− and cERβ2+, respectively). In our institution, a written informed consent is routinely requested from patients for collection of their clinical data as well as paraffin-embedded sections for research use.
2.2. Antibody validation The mouse monoclonal antibody we used in the study (clone 57/3, Serotec Ltd; Oxford, United Kingdom) is raised against the unique COOH-terminal peptides of ERβ2. Antibody specificity was validated by peptide absorption studies with a peptide specific for ERβ2 (CGMKMETLL PEATMEQ) that was used in neutralization studies. Antibody was diluted to its working concentration and incubated with 10 times excess of the peptide overnight at 4°C, before application to tissue.
2.3. Immunohistochemistry Three-micrometer-thick paraffin sections were mounted on Superfrost-coated slides and dried overnight. The immunohistochemical staining for Ki-67 (clone K-2 Prediluted, Ventana Medical Systems; Tucson, AZ) was performed using the Ventana Benchmark XT autostainer. Immunohistochemistry for cleaved caspase-3 (CC3, clone 5A1E, dilution 1:100, Cell Signaling Technology; Leiden, The Netherlands) and cytochrome c (clone 136F3, dilution 1:200, Cell Signaling Technology) was performed using a labeled streptavidin-biotin-peroxidase method. The sections were deparaffinized in xylene and rehydrated in graded solutions of ethanol; antigen retrieval procedure was performed by microwave oven heating in citrate buffer (pH 6). The endogenous peroxidase was blocked with 3% vol/vol H2O2 for 5 minutes. Sections were incubated with 20% vol/vol normal goat serum for 30 minutes at room temperature to reduce nonspecific binding. Cells expressing CC3 or cytochrome c were identified after overnight incubation at 4°C. Sections were incubated with the secondary, antimouse EnVision System-HRP (DakoCytomation, Carpinteria, CA) for 30 minutes, at room temperature. The slides were developed with diaminobenzidine (DAB; DAB Substrate System, DakoCytomation), counterstained with Mayer hematoxylin, dehydrated in ethanol and xylene, and finally mounted. Expression of Ki-67, CC3, and cytochrome c was evaluated by considering the number of cells exhibiting immunoreaction in a minimum of 500 histologically identified neoplastic cells. Data are expressed as percentages of marker-positive cells among total cells. The analysis of all tissue sections was performed by 2 of the authors (G.F.Z. and A.C.) in a blinded fashion. In case
1140 of disagreement, sections were submitted to a joint re-evaluation by a multiheaded microscope.
2.4. Immunofluorescence of fixed paraffin-embedded ovarian tissue sections Inasmuch as nonspecific tissue staining has traditionally been considered an obstacle in labeling paraffin-embedded sections using immunofluorescence techniques, specific procedures were followed to reduce both nonspecific and background staining [14]. Three-micrometer-thick paraffin sections were mounted on Superfrost coated slides and dried overnight (Thermo Scientific, Menzel GmbH & Co., KG, Braunschweig, Germany). The sections were deparaffinized in xylene, rehydrated in graded solutions of ethanol, and rinsed for 5 minutes in distilled water. Antigen retrieval procedure was performed by microwave oven heating in citrate buffer (pH 6). Sections were incubated with 20% normal goat serum for 30 minutes at room temperature and then incubated at 4°C overnight with primary antibody (ERβ2, clone 57/3, dilution 1:100, Serotec Ltd). The optimal dilution of the primary antibody had been established before by immunoenzymatic staining using conventional techniques (2-stage immunoperoxidase technique, DAB). After overnight incubation, slides were washed in tris-buffered saline and incubated in the dark for 1 hour at room temperature with secondary antibody antimouse Alexa Fluor 488 conjugate (Life Technologies Inc, Milan, Italy). The slides were subsequently incubated with MitoTracker Red (150 nM in tris-buffered saline for 15 minutes, Life Technologies Inc), a widely used probe for mitochondria staining [15]. After extensive washing in phosphate-buffered saline 1% Tween 20, tissues were stained with 4′,6-diamidino-2-phenylindole (DAPI, 1.5 μg/mL) and mounted in Vectashield Mounting Medium (Vector Laboratories; Burlingame, Ontario, Canada). Slides were observed under the fluorescence microscope (Leica; Milan, Italy) using a ×40 or a ×100 oil immersion objective.
2.5. Duolink The Duolink assay (Sigma-Aldrich; St. Louis, MO) based on in situ PLA allows precise detection of protein-protein interactions in fixed cells and tissue samples, in their correct cellular context and at physiologically relevant expression levels [16,17]. Specifically, PLA allows protein complex to be represented as an amplifiable DNA molecule. Recognition is mediated by proximity probes consisting of antibodies coupled with oligonucleotides. Upon dual binding of the proximity probes, the oligonucleotides direct the formation of a circular DNA molecule, which is then amplified by rolling-circle replication. The localized concatemeric product is then detected with fluorescent probes. The assay results in a fluorescent signal in the form of a spot when the 2 proteins of interest are closer than 40 nm; each red dot represents a molecular interaction between the 2 proteins of interest.
A. Ciucci et al. For analysis, 3-μm-thick paraffin sections were mounted on Superfrost coated slides, dried overnight, and subjected to antigen retrieval by microwave cooking in citrate buffer (pH 6.0). The PLA protocol was followed according to the manufacturer’s instructions (Olink Bioscience; Uppsala, Sweden). After blocking (Olink Bioscience), the following primary antibodies were used: mouse monoclonal IgG1 ERβ2 (clone 57/3, Serotec Ltd, dilution 1:100); rabbit polyclonal IgG BAD (B-cell lymphoma [Bcl] 2–associated death promoter, Serotec Ltd, 1:200); mouse monoclonal IgG1 Bcl-xL (B-cell lymphoma-extra large, clone H-5, Santa Cruz, CA, 1:100), rabbit polyclonal IgG Bax (Bcl-2–like protein 4, clone P-19, Santa Cruz, CA, 1:50) with incubation at 4°C overnight. The optimal experimental conditions for primary antibodies had been established before by immunoenzymatic staining using conventional techniques (2-stage immunoperoxidase technique, DAB) and immunofluorescence (Supplementary Fig. S1). On the next day, PLA minus and PLA plus probes (containing the secondary antibodies conjugated with oligonucleotides) were added and incubated 1 hour at 37°C. The ligation reaction was carried out at 37°C for 30 minutes in a humid chamber followed by washing. Slides were then incubated with the amplification mix for 2 hours at 37°C in a darkened humidified chamber. After washing, slides were mounted using the mounting media supplied with the kit.
2.6. Statistical analysis Data were analyzed for homogeneity of variance using an F test. If the group variance appeared homogenous, a Student t test was used. If the variances were heterogeneous, log or reciprocal transformations were made in an attempt to stabilize the variances. If the variances remained heterogeneous, a nonparametric test such as the Mann-Whitney U test was used. Correlations between variables were identified using the Spearman rank correlation. P values are for 2-sided tests; P values ≤ .05 were considered statistically significant. All statistical analyses were performed using the GraphPad Prism5 Software (San Diego, CA).
3. Results 3.1. Antibody validation Specific nuclear and cytoplasmic staining was completely abolished when ERβ2 primary antibody was preabsorbed with its respective blocking peptide (Fig. 1A).
3.2. Apoptosis and cell proliferation rates in cERβ2− and cERβ2+ cases Fifty-six advanced serous ovarian cancer cases were stratified according to the absence or presence of ERβ2 in the cytoplasmic compartment (cERβ2− and cERβ2+, respectively) using immunohistochemical data obtained in our
Mitochondrial ERβ2 in ovarian cancer previous investigation [13]. Fig. 1A shows representative pictures of cERβ2− (n = 31) and cERβ2+ (n = 25) tumors. The antibody specific for active caspases-3 stained the cytoplasm of cells whose nuclear morphology was consistent with apoptosis along with the cytoplasm of morphologically healthy-looking cells, thus suggesting that this antibody recognized activated protein at the early stage of apoptosis. Significantly higher levels of CC3 were detected in cERβ2− when compared with cERβ2+ cases (5.6 ± 1.3 versus 2.6 ± 0.4, respectively, mean ± SEM, P b .05, Fig. 1B). Conversely, tumor proliferation, quantified by the expression of Ki-67, did not appear to be significantly different between the 2 populations (54.2 ± 4.5 versus 57.8 ± 4.4 in cERβ2− and cERβ2+ cases, respectively, mean ± SEM, Fig. 1C). Overall, these findings suggest that cytoplasmic ERβ2 affects
1141 ovarian cancer development by inhibiting pathways that promote cell death rather than directly stimulate cell growth.
3.3. Mitochondrial localization of the ERβ2 protein Determining the subcellular localization of a protein within a cell is often an essential step toward understanding its function, and we used immunofluorescence to detect endogenous ERβ2 protein localization in serous ovarian cancer tissue. Tumor sections were incubated with anti-ERβ2 and Alexa Fluor 488 (antimouse), whereas MitoTracker and DAPI were used for visualization of mitochondrial and nuclear staining. ERβ2 staining in the cytosol presented a punctuate distribution, similar to that of mitochondria. The
Fig. 1 A, Representative pictures for nuclear (cERβ2−) or nuclear/cytoplasmic (cERβ2+) ERβ2 staining in ovarian cancer tissues. Nuclear and cytoplasmic ERβ2 staining was completely abolished after preabsorption with the specific blocking peptide. B, Bar chart showing the distribution of CC3-positive cells in cERβ2− and cERβ2+ cases (Student t test, *P b .05) and representative pictures for CC3 staining showing tumors with high or low levels. C, Bar chart showing the distribution of Ki-67–positive cells in cERβ2− and cERβ2+ cases and representative tumor immunostaining.
1142 overlay of the Alexa dye staining (green) with mitochondrial control MitoTracker Red showed indeed that ERβ2 protein staining coincided precisely with MitoTracker Red staining (Fig. 2). Immunofluorescent staining of 2 different serous ovarian cancer cell lines (ie, Hey and COV-318) showed only in Hey cells both nuclear and mitochondrial localization of ERβ2 protein, whereas COV-318 only displayed nuclear expression (Supplementary Fig. S2).
3.4. Mitochondrial ERβ2 physically interacts with the proapoptotic protein BAD Using a conventional coimmunoprecipitation approach, Zhang et al [18] demonstrated that mitochondrial ERβ physically interacts with the proapoptotic protein BAD, in a ligand-independent manner in lung cancer cells. To examine whether we could detect endogenous complexes between mitochondrial ERβ2 and BAD in serous ovarian cancer cases
A. Ciucci et al. in situ, we performed PLA, a technology capable of detecting protein interactions with high specificity and sensitivity. In cases with only nuclear ERβ2 expression (cERβ2−) there were no significant PLA signals (red dots) for dimers ERβ2/BAD, whereas in tumor samples showing mitochondrial expression, PLA signals could be easily recognized (Fig. 3). Because BAD is a strong heterodimerizing partner of Bcl-xL, to confirm our results, we evaluated the presence of Bcl-xL/BAD heterodimer in the 2 subcohort of cases, showing that the PLA signal Bcl-xL/BAD was only evident in tumors with exclusive nuclear ERβ2 expression, whereas no similar finding was observed in samples showing cytoplasmic/mitochondrial expression (cERβ2+) (Fig. 3). In turn, these latter cases showed Bcl-xL/Bax heterodimers (Fig. 3), a condition that inhibits Bax oligomerization required for mitochondrial outer membrane permeabilization during apoptosis [19]. Single-antibody control reactions did not result in significant PLA signals (not shown).
Fig. 2 Fluorescence microscopy localization of ERβ2 in serous ovarian cancer tissues. Representative pictures of exclusive nuclear (A) and of nuclear/cytoplasmic (B) ERβ2 expression. Nuclei were stained with DAPI (blue), mitochondria with MitoTracker Red (red) and ERβ2 with antimouse Alexa Fluor 488 conjugate (green). Merged images in (B) show the mitochondrial localization of cytoplasmic ERβ2.
Mitochondrial ERβ2 in ovarian cancer
1143
Fig. 3 Mitochondrial ERβ2/BAD (A), Bcl-xL/BAD (B), Bcl-xL/Bax (C), complex formation in ovarian cancer samples, as visualized by the PLA technique. Red dot in merged photomicrographs indicates physical interaction of proteins.
3.5. Higher level of cytochrome c release in cERβ2− cases Because overexpression of the proapoptotic protein Bax has been shown to trigger mitochondrial cytochrome c efflux into the cytosol and cell death [20,21], we examined the expression of cytochrome c and its cellular localization by immunohistochemistry, in our series of ovarian cancer cases. Tumor samples with only nuclear ERβ2 expression (cERβ2−) exhibited higher amounts of cytochrome c in the cytosolic compartment when compared with cERβ2+ cases (P b .05, Fig. 4A), these latter mostly showing a punctuate mitochondrial staining. Because cell-free assays of caspase activation have revealed that cytochrome c is a powerful inducer of procaspase-3 cleavage [22], we assessed the correlation between cytosolic levels of cytochrome c and CC3 expression in our samples, results showing a significant positive correlation (r = 0.388, P b .01, Fig. 4B).
4. Discussion We previously showed that cERβ2 expression is an independent poor prognostic factor in advanced serous ovarian cancer (P = .006), the 5-year survival rate being nearly 28% for patients who did express cERβ2 and 60% in
patients who did not; we also demonstrated that cERβ2 expression is significantly associated with chemoresistance (P = .03) [13]. Notably, similar findings were also reported in familial and sporadic breast cancer, where cytoplasmic ERβ2 staining was associated to the worst survival prognostic [23,24]. Despite these clinical data, the functional nongenomic role of ERβ2 and the mechanisms for its potential influence on cancer biology are far from being elucidated. To address this issue, thus adding further insights in clarifying the discrete role of each specific ERβ isoform, we investigate the molecular determinants in cell survival function of cytoplasmic ERβ2 in ovarian cancer. We firstly demonstrate that, although not affecting the proliferative potential of cancer cells, cytoplasmic ERβ2 is associated to lower apoptotic rates in ovarian cancer cases. Then, we provide strong evidence that ERβ2 is a mitochondrial component in cancer cells that can act as a binding partner of proteins involved in the apoptosis cascade. To examine the in situ existence of endogenous ERβ2 protein complexes in serous ovarian cancer cases, we used the PLA technique on formalin-fixed paraffin sections. Indeed, the PLA assay is a revolutionary technique that enables unprecedented specificity, sensitivity, and accurate visualization of protein-protein interactions [25-27], overcoming some limitations of the conventional coimmunoprecipitation (Co-IP) method, as false-positive detection, or failing to
1144
A. Ciucci et al.
Fig. 4 A, Representative pictures for mitochondrial or cytosolic cytochrome c staining (a detail on the side) in ovarian cancer tissues and bar chart showing the distribution of cytosolic cytochrome c–positive cells in cERβ2− and cERβ2+ cases (Mann-Whitney U test, *P b .05). B, The Spearman rank correlation showing a significant positive correlation between CC3 and cytosolic cytochrome c–positive cells in ovarian cancer samples. **P b .01.
detect weak or transient protein interactions [28]. Notably, by using this innovative technique, we have been able to demonstrate, for the first time ever in serous ovarian cancer samples, that mitochondrial ERβ2 forms protein complexes with BAD at physiologically relevant expression levels and in their correct cellular context. We also show that, in tumor exhibiting ERβ2/BAD interaction, the presence of heterodimers Bcl-xL/Bax is evident, an event that, inhibiting Bax oligomerization, reduces the release of cytochrome c from the mitochondria into the cytoplasm and ultimately inhibits apoptosis. Cytochrome c is indeed a protein that normally is well sequestered in the mitochondrial intermembrane space and the release of cytochrome c into the cytosol can be induced by proapoptotic members of Bcl-2 family (such as Bax, Bad, and Bid) [22,29]. Although cytochrome c is not the sole mitochondrial intermembrane protein capable of activating caspases, the cytochrome c–stimulated pathway of caspase activation is the best characterized, with cell-free assays demonstrating that cytochrome c is a powerful inducer of procaspase-3 cleavage [22]. The reduced levels of CC3 in cERβ2+ tumor samples thus possibly occur as a result of the mitochondrial ERβ2–mediated inhibition of cytochrome c release. Although the difference observed in
the basal apoptotic levels between the 2 populations (cERβ2+ and cERβ2− cases) is not large, it could promote different response rates to drug-induced cytotoxicity and thus a different clinical outcome. Indeed, it is well known that attenuation of apoptosis in ovarian cancer cells contributes to the resistance to subsequent chemotherapy and likely plays an important role in tumor progression [3]. Such a hypothesis is actually supported by our previous results on the association between cERβ2 expression and chemoresistance in patients with serous ovarian cancer [13]. Altogether, our findings provide mechanistic evidence that supports the value of cytoplasmic ERβ2 expression as an unfavorable prognostic biomarker for serous ovarian cancer. Our results are in keeping with previous literature data showing the presence of (total) ERβ in the mitochondria in several cell types, including human non–small cell lung cancer (NSCLC) cells, human breast cancer MCF-7 cells, human osteosarcoma SaOS-2 cells, and liver cancer HepG2 cells [30-32] and with the hypothesis proposed by Zhang et al [18] of a ligand-independent antiapoptotic function of ERβ in lung cancer cells. However, most of these studies did not investigate the contribution of the different ERβ isoforms to this mechanism. Indeed, although our understanding of ERβ
Mitochondrial ERβ2 in ovarian cancer functions has broadened since the discovery of ERβ isoforms [6,33], little is known about their discrete function, expression, and regulation in both normal and tumor tissues as well as about their ability to operate in a ligand-independent or -dependent manner, according to different cellular contexts. On the other hand, there is compelling evidence that supports distinct prognostic significance of ERβ isoforms, at least in some cancer types (reviewed by Gallo et al [4]). Thus, implementation research is certainly needed to uncover the complex biochemical chain of events controlled by the different ERβ isoforms. The availability of robust validated antibodies and of innovative technologies, enabling both measurements of nuclear and cytoplasmic ERβs and exploration of their potential mechanism of action in a relevant cellular context should stimulate further specific investigations to more clearly define their distinct role as outcome factors. In this respect, the potential contribution of cytoplasmic ERβ1 and/or ERβ5 in apoptosis inhibition deserves additional studies, although, in our series, only cytoplasmic ERβ2 was seen to be associated with shorter survival and chemoresistance. In summary, we provide novel insights into the functional nongenomic role of cytoplasmic ERβ2, showing for the first time an antiapoptotic role in advanced serous ovarian cancer. The recognition of molecular features linking cytoplasmic ERβ2 expression to tumor aggressiveness could allow the identification of high-risk patients at the time of the initial diagnosis, thus guiding the use of alternative treatment approaches. Specifically, in these patients, pharmacologic targeting of critical antiapoptotic Bcl-2 proteins might hold great promise for personalized therapeutic interventions.
Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.humpath.2015.03.016.
References [1] Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013;49:1374-403. [2] Jayson GC, Kohn EC, Kitchener HC, Ledermann JA. Ovarian cancer. Lancet 2014;384:1376-88 [Review]. [3] Vasey PA. Resistance to chemotherapy in advanced ovarian cancer: mechanisms and current strategies. Br J Cancer 2003;89(Suppl. 3): S23-8. [4] Gallo D, De Stefano I, Prisco MG, Scambia G, Ferrandina G. Estrogen receptor beta in cancer: an attractive target for therapy. Curr Pharm Des 2012;18:2734-57. [5] Simpkins F, Garcia-Soto A, Slingerland J. New insights on the role of hormonal therapy in ovarian cancer. Steroids 2013;78:530-7. [6] Moore JT, McKee DD, Slentz-Kesler K, et al. Cloning and characterization of human estrogen receptor beta isoforms. Biochem Biophys Res Commun 1998;247:75-8.
1145 [7] Poola I, Abraham J, Baldwin K, Saunders A, Bhatnagar R. Estrogen receptors beta4 and beta5 are full length functionally distinct ERbeta isoforms: cloning from human ovary and functional characterization. Endocrine 2005;27:227-38. [8] Dey P, Jonsson P, Hartman J, Williams C, Ström A, Gustafsson JÅ. Estrogen receptors β1 and β2 have opposing roles in regulating proliferation and bone metastasis genes in the prostate cancer cell line PC3. Mol Endocrinol 2012;26:1991-2003. [9] Suzuki F, Akahira J, Miura I, et al. Loss of estrogen receptor beta isoform expression and its correlation with aberrant DNA methylation of the 5‵-untranslated region in human epithelial ovarian carcinoma. Cancer Sci 2008;99:2365-72. [10] Drummond AE, Fuller PJ. Ovarian actions of estrogen receptor-β: an update. Semin Reprod Med 2012;30:32-8. [11] De Stefano I, Zannoni GF, Prisco MG, et al. Cytoplasmic expression of estrogen receptor beta (ERβ) predicts poor clinical outcome in advanced serous ovarian cancer. Gynecol Oncol 2011;122:573-9. [12] Häring J, Schüler S, Lattrich C, Ortmann O, Treeck O. Role of estrogen receptor β in gynecological cancer. Gynecol Oncol 2012;127:673-6. [13] Ciucci A, Zannoni GF, Travaglia D, Petrillo M, Scambia G, Gallo D. Prognostic significance of the estrogen receptor beta (ERβ) isoforms ERβ1, ERβ2, and ERβ5 in advanced serous ovarian cancer. Gynecol Oncol 2014;132:351-9. [14] Bataille F, Troppmann S, Klebl F, et al. Multiparameter immunofluorescence on paraffin-embedded tissue sections. Appl Immunohistochem Mol Morphol 2006;14:225-8. [15] Murriel CL, Churchill E, Inagaki K, Szweda LI, Mochly-Rosen D. Protein kinase Cdelta activation induces apoptosis in response to cardiac ischemia and reperfusion damage: a mechanism involving BAD and the mitochondria. J Biol Chem 2004;279:47985-91. [16] Söderberg O, Gullberg M, Jarvius M, et al. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat Methods 2006;3:995-1000. [17] Baan B, Pardali E, ten Dijke P, van Dam H. In situ proximity ligation detection of cJun/AP-1 dimers reveals increased levels of cJun/Fra1 complexes in aggressive breast cancer cell lines in vitro and in vivo. Mol Cell Proteomics 2010;9:1982-90. [18] Zhang G, Yanamala N, Lathrop KL, Zhang L, Klein-Seetharaman J, Srinivas H. Ligand-independent antiapoptotic function of estrogen receptor-beta in lung cancer cells. Mol Endocrinol 2010;24:1737-47. [19] Ding J, Mooers BH, Zhang Z, et al. After embedding in membranes antiapoptotic Bcl-XL protein binds both Bcl-2 homology region 3 and helix 1 of proapoptotic Bax protein to inhibit apoptotic mitochondrial permeabilization. J Biol Chem 2014;289:11873-96. [20] Eskes R, Antonsson B, Osen-Sand A, et al. Bax-induced cytochrome C release from mitochondria is independent of the permeability transition pore but highly dependent on Mg2+ ions. J Cell Biol 1998;143: 217-24. [21] Rosse T, Olivier R, Monney L, et al. Bcl-2 prolongs cell survival after Bax-induced release of cytochrome c. Nature 1998;391:496-9. [22] Green D, Kroemer G. The central executioners of apoptosis: caspases or mitochondria? Trends Cell Biol 1998;8:267-71. [23] Shaaban AM, Green AR, Karthik S, et al. Nuclear and cytoplasmic expression of ERbeta1, ERbeta2, and ERbeta5 identifies distinct prognostic outcome for breast cancer patients. Clin Cancer Res 2008; 14:5228-35. [24] Yan M, Rayoo M, Takano EA, kConFab Investigators, Fox SB. Nuclear and cytoplasmic expressions of ERβ1 and ERβ2 are predictive of response to therapy and alters prognosis in familial breast cancers. Breast Cancer Res Treat 2011;126:395-405. [25] Zhu L, Koistinen H, Landegren U, Stenman U-H. Proximity ligation measurement of the complex between prostate specific antigen and a1prostate inhibitor. Clin Chem 2009;55:1665-71. [26] Leuchowius K-J, Weibrecht I, Landegren U, Gedda L, Soderberg O. Flow cytometric in situ proximity ligation analysis of protein interactions and post-translational modification of the epidermal growth factor receptor family. Cytometry A 2009;75A:833-9.
1146 [27] Aubele M, Spears M, Ludyga N, et al. In situ quantification of HER2protein tyrosine kinase 6 (PTK6) protein-protein complexes in paraffin sections from breast cancer tissues. Br J Cancer 2010;103:663-7. [28] Berggård T, Linse S, James P. Methods for the detection and analysis of protein-protein interactions. Proteomics 2007;7:2833-42. [29] Li B, Dou QP. Bax degradation by the ubiquitin/proteasomedependent pathway: involvement in tumor survival and progression. Proc Natl Acad Sci U S A 2000;97:3850-5. [30] Chen JQ, Delannoy M, Cooke C, Yager JD. Mitochondrial localization of ERα and ERβ in human MCF7 cells. Am J Physiol Endocrinol Metab 2004;286:E1011-22.
A. Ciucci et al. [31] Solakidi S, Psarra AM, Sekeris CE. Differential subcellular distribution of estrogen receptor isoforms: localization of ERα in the nucleoli and ERβ in the mitochondria of human osteosarcoma SaOS-2 and hepatocarcinoma HepG2 cell lines. Biochim Biophys Acta 1745;2005: 382-92. [32] Zhang G, Liu X, Farkas AM, et al. Estrogen receptor β functions through nongenomic mechanisms in lung cancer cells. Mol Endocrinol 2009;23:146-56. [33] Leung YK, Mak P, Hassan S, Ho SM. Estrogen receptor (ER)-beta isoforms: a key to understanding ER-beta signaling. Proc Natl Acad Sci U S A 2006;103:13162-7.
