Downregulation of Oestrogen Receptor Associates with Transcriptional

Journal of Pathology J Pathol 2011; 224: 110–120 ORIGINAL PAPER Published online 7 March 2011 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/path.2846 Down-regulation of oestrogen receptor-β associates with transcriptional co-regulator PATZ1 delocalization in human testicular seminomas

Francesco Esposito,1,2# Francesca Boscia,3# Renato Franco,4 Mara Tornincasa,2 Alfredo Fusco,2 Sohei Kitazawa,5 Leendert H Looijenga6 and Paolo Chiefﬁ1,2*

1 Dipartimento di Medicina Sperimentale, II Universita` di Napoli, Naples, Italy 2 IEOS and Dipartimento di Biologia e Patologia, Universita` di Napoli ‘Federico II’, Naples, Italy 3 Dipartimento di Neuroscienze, Universita` di Napoli ‘Federico II’, Naples, Italy 4 Istituto Nazionale dei Tumori ‘Fondazione G. Pascale’, Naples, Italy 5 Department of Molecular Pathology, Kobe University, Japan 6 Department of Pathology, Daniel den Hoed Cancer Centre, JosephineNefkens Institute, Erasmus MC University Medical Centre Rotterdam, The Netherlands

*Correspondence to: Paolo Chiefﬁ, Dipartimento di Medicina Sperimentale, Via Costantinopoli 16, 80138 Naples, Italy. e-mail: Paolo.Chiefﬁ@unina2.it #These authors equally contributed to this study.

Abstract Oestrogen exposure has been linked to a risk for the development of testicular germ cell cancers. The effects of oestrogen are now known to be mediated by oestrogen receptor-α (ERα)andERβ subtypes, but only ERβ has been found in human germ cells of normal testis. However, its expression was markedly diminished in seminomas, embryonal cell carcinomas and mixed germ cell tumours, but remains high in teratomas. PATZ1 is a recently discovered zinc ﬁnger protein that, due to the presence of the POZ domain, acts as a transcriptional repressor affecting the basal activity of different promoters. We have previously described that PATZ1 plays a crucial role in normal male gametogenesis and that its up-regulation and mislocalization could be associated with the development of testicular germ cell tumours. Here we show that ERβ interacts with PATZ1 in normal germ cells, while down-regulation of ERβ associates with transcriptional co-regulator PATZ1 delocalization in human testicular seminomas. In addition, we show that the translocation of PATZ1 from the cytoplasm into the nucleus is regulated by cAMP, which also induces increased expression and nuclear localization of ERβ, while this effect is counteracted by using the anti-oestrogen ICI 182-780. Copyright  2011 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Keywords: ZNF278; PATZ1; MAZR; spermatogenesis; testicular cancer; tumour suppressor; oestrogen receptor-β

Received 12 October 2010; Revised 26 November 2010; Accepted 18 December 2010

No conﬂicts of interest were declared.

Introduction certain hormones (eg oestrogen) at the time of testicular differentiation in utero has long been implicated as a risk factor for developing these neoplasmas [4]. Testicular germ cell tumours (TGCTs), the most Recently, it has been suggested that the carcinogenic common malignancy in males aged 15–34 years, rep- resent a major cause of death attributable to can- effects of oestrogen on testicular cells may involve cer in this age group [1–3]. TGCTs can be subdi- oestrogen receptor-mediated oxidative DNA damage vided into seminoma and non-seminoma germ cell [5]. tumours (NSGCTs), including embryonal cell carci- Oestrogen signalling is mediated by two nuclear α α β β noma, choriocarcinoma, yolk sac tumour and teratoma. receptors, oestrogen receptor- (ER )and (ER ), Neoplasms containing more than one tumour cell com- which are oestrogen-dependent transcription factors. ponents, eg seminoma and embryonal cell carcinoma, ERα is expressed at high levels in human epididymis are referred to as mixed germ cell tumours. Semino- and efferent ductules but not in the testis, whereas mas and NSGCTs not only present distinctive clinical ERβ is expressed in spermatogonia, spermatocytes and features but also show signiﬁcant differences as far as early round spermatids in human testis [6,7]. The ERβ therapy and prognosis are concerned [1,2]. Whereas subtype is the principal mediator of oestrogen action in the aetiology of TGCTs remain undeﬁned, exposure to promoting germ cell survival and development [6–9].

Copyright  2011 Pathological Society of Great Britain and Ireland. J Pathol 2011; 224: 110–120 Published by John Wiley & Sons, Ltd. www.pathsoc.org.uk www.thejournalofpathology.com PATZ1 and ERβ in human testicular seminomas 111

After activation, these receptors, in association with a approval was given in all instances. The GC-1 cell line myriad of co-activators and repressors, act as nuclear was cultured in Dulbecco’s modified Eagle’s medium transcription factors for targeted genes [10,11]. It has (D-MEM) supplemented with 10% fetal bovine serum ◦ been well documented in literature that ERβ,whichis (FBS; Gibco BRL, Italy) and grown in a 37 C humid- expressed in normal testicular cells, is instead down- ified atmosphere of 5% CO2 [24]. TCam-2 cells were ◦ regulated in seminomas and embryonal cell carcinomas grown at 37 Cina5%CO2 atmosphere in RPMI [12,13]. 1640 (Lonza) supplemented with 10% FBS [25,26]. PATZ1, also named ZNF278 or MAZR, is a recently Spermatogonia were prepared from 7dpp testes, as pre- discovered ubiquitously expressed transcriptional reg- viously described [27]. ulatory factor gene whose product binds to the RING TCam-2 cells were also treated with 8-bromoadeno- finger protein RNF4 that, in turn, associates with a sine-3,5-cyclic monophosphate (8Br-cAMP, 100 µM; variety of transcription regulators [14–17]. By virtue Sigma, Milan, Italy) and with ICI 182–780 (100 µM; of the POZ domain, PATZ1 acts as a transcriptional Zeneca, London, UK). Cells were transfected with repressor on different promoters [18], although it has plasmids contained human PATZ1 and ERβ full-length been also shown to function as a strong activator cDNA using the lipofectamineplus reagent (Invitrogen, of the c-myc promoter [19]. Indeed, PATZ1 is able Milan, Italy), as suggested by the manufacturer. to bind proteins involved in chromatin remodelling, such as HMGA1, that it has been shown to be over- Antibodies expressed in human testicular seminomas [20] and the Antibodies were purchased from the following sources: above mentioned RNF4 [14]. Interestingly, it has been (a) polyclonal mouse antibody anti-PATZ1 (no. found rearranged through a paracentric inversion of ab68646; Abcam, Cambridge, UK); (b) polyclonal rab- 22q12 with the EWS gene, in small round cell sar- bit antibody anti-PATZ1 (no. sc-86 776; Santa Cruz coma, suggesting a potential tumour suppressor role Biotechnology, CA, USA); (c) polyclonal rabbit anti- [17]. Four alternatively spliced transcript variants have body anti-oestrogen receptor-β (ERβ; no. sc-8974, been described for the PATZ1 gene. Our previous pub- Santa Cruz); (d) mouse monoclonal anti-oestrogen lished data demonstrated that only variant 3 of PATZ1 receptor β (clone PPG5/10; code no. M7292, Dako (537 amino acids, about 60 kDa) is expressed in the Cytomation, Denmark); (e) mouse monoclonal anti-β- testis and TGCTs [21]. −/− tubulin (no. T-4026, Sigma, St. Louis, MO, USA); Recently, we have shown that male PATZ1 mice (f) mouse monoclonal anti γ-tubulin (no. T-5326, are unfertile and, although PATZ1 protein expression Sigma); (g) mouse monoclonal anti-Golgin-97 (code was up-regulated in testicular germ cell tumours when no. A21270, Molecular Probes); (h) polyclonal rabbit compared to normal human testis, it was delocalized anti-Hemagglutinin Tag Anitibody (HA) (no. sc-805, in the cytoplasm, thus suggesting an impaired function Santa Cruz) that recognized PATZ1 proteins conju- [21]. In addition, it has been documented, using the gated to HA; (i) polyclonal rabbit anti-SP1 (no. sc- null mice model, that the zinc-finger protein MAZR 14 027, Santa Cruz); (j) mouse monoclonal anti-PKA- is part of the transcription factor network that controls regulatory subunit (PKA-R; code no. 612 242, BD the CD4 versus CD8 lineage fate of double-positive Transduction Laboratories, Franklin Lakes, NJ, USA); thymocytes [22,23]. (k) rabbit polyclonal anti PKA-catalytic subunit (PKA- β Here we show that PATZ1 interacts with ER in C; code no. sc-903, Santa Cruz). normal germ cells. Moreover, in human testicular seminomas, PATZ1 protein delocalization associates Histological analysis and immunohistochemistry with ERβ down-regulation. In addition, we show that the translocation of PATZ1 from the cytoplasm into For light microscopy, tissues were fixed in 10% for- the nucleus is mediated by cAMP, which is also malin and embedded in paraffin by standard proce- responsible for an increased expression and nuclear dures. Sections (4 µm) were stained with haematoxylin localization of ERβ. Interestingly, both PATZ1 and and eosin (H&E) or processed for immunohistochem- ERβ nuclear translocation mediated by cAMP was istry. The classical avidin–biotin peroxidase complex counteracted by using the anti-oestrogen ICI 182–780. (ABC) procedure was used for immunohistochem- Taken together, these results suggest that the failed istry, as described [20,21]. Sections were incubated association between the two proteins might have a role overnight with antibodies against: (a) PATZ1 (diluted in the genesis of the testicular neoplasia. 1 : 200; polyclonal rabbit antibody anti-PATZ1, no. sc- 86 776, Santa Cruz); (b) oestrogen receptor-β (diluted 1 : 200; mouse monoclonal anti-oestrogen receptor-β, β Materials and methods code no. M7292, Dako). For PATZ1 and ER detection additional antibodies were used, which gave sim- ilar results (mouse polyclonal antibody anti-PATZ1, Tissue samples, cell culture, transfection no. ab68646, Abcam; rabbit polyclonal antibody anti- As a source of neoplastic tissues, the tissue bank of oestrogen receptor-β, no. sc-8974, Santa Cruz; not the National Cancer Institute ‘G. Pascale’ provided 21 shown). The following controls were performed: (a) cases of cryopreserved seminomas. Ethical committee omission of the primary antibody; (b) substitution of

Figure 1. Immunoprecipitation and western blot analyses with anti-HA–PATZ1 and anti-ERβ antibodies in HEK293 cells transfected with HA–PATZ1 and ERβ constructs or with empty vector. Cell lysates were immunoprecipitated with (A) anti-HA–PATZ1 and (B) anti-ERβ, resolved by SDS–PAGE, transferred to nitrocellulose membranes and challenged with anti-ERβ (rabbit polyclonal) or anti HA–PATZ1 antibodies. The blots are representative of three separate experiments. (C) Western blot analysis of HA–PATZ1 and ERβ expression in HEK293 cells; 40 µg total proteins were resolved on 10% SDS–PAGE, transferred onto nitrocellulose filters and blotted with anti-HA–PATZ1 and ERβ polyclonal antibodies. the primary antiserum with non-immune serum diluted microscope (Carl Zeiss, Thornwood, NY, USA). Sin- 1 : 500 in blocking buffer; (c) addition of the target gle images, taken with an optical thickness of 0.7 µm peptide used to produce the antibody (10−6 M); no and a resolution of 1024 × 1024 pixels, were used for immunostaining was observed after any of the control illustrations. procedures. Protein extraction, immunoprecipitation Immunofluorescence confocal microscopy and western blot analysis Confocal immunofluorescence procedures were performed as described [28]. Sections (4 µm) of cry- Total cell extracts (TCE) were prepared with lysis opreserved seminomas were used. GC1 and TCam-2 buffer (50 mM Tris–HCl, pH 7.5, 5 mM EDTA, cells cultured on glass coverslips coated with 30 mg/ml 300 mM NaCl, 150 mM KCl, 1 mM dithiothreitol, 1% poly-L-lysine (Sigma) were rinsed in 50 mM Tris- Nonidet P40) and a mix of protease inhibitors. Protein buffered saline (TBS), pH 7.4, and fixed in 4% concentration was estimated by a modified Bradford paraformaldehyde (Sigma) for 30 min. The coverslips assay (Bio-Rad). For immunoprecipitations, protein A-Sepharose (3 mg/sample) was incubated with rab- were first blocked in 3% bovine serum albumin (BSA; ◦ Sigma) and 0.05% Triton X-100 (Bio-Rad, Milan, bit anti-ERβ antibodies (5 µl/sample) for 1 h at 4 C Italy) for 1 h and then incubated with the follow- and then washed three times with HNTG (20 mM ing primary antibodies: rabbit polyclonal anti-PATZ1 HEPES, 150 mM NaCl, 0.1% Triton-X-100, 5% glyc- (1 : 1000); mouse monoclonal anti-oestrogen receptor-β erol). 500 µg total protein was added to the anti-ERβ, (diluted 1 : 300); and mouse monoclonal anti-Golgin- anti-PATZ1 and anti HA-PATZ1 antibodies–protein A ◦ ◦ 97 (diluted 1 : 200), at 4 C overnight. Next, the cells mix for 2 h at 4 C. Immunoprecipitates were washed were incubated in a mixture of fluorescent-labelled four times with HNTG (20 mM HEPES, 15 mM NaCl, secondary antibodies Alexa 488-conjugated anti-mouse 0.1% Triton X-100, 5 or 10% glycerol) and prior to and Alexa 594-conjugated anti-rabbit IgGs (diluted this boiled in Laemmli buffer (0.125 M Tris–HCl, pH 1 : 200; Molecular Probes, Eugene, OR, USA) for 6.8, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 1 h. The cell nuclei were stained with Hoechst-33 258 0.002% bromophenol blue) for 5 min. Immunoprecip- (Sigma). Immunofluorescence images were observed itates were subjected to SDS–PAGE (10% polyacry- using a Zeiss LSM 510 META laser-scanning confocal lamide) under reducing conditions.

Figure 2. Co-localization of PATZ1 protein and ERβ in GC-1 and TCam-2 cells. Low-magnification images showing GC-1 (A–D) or TCam-2 (M–P) cells double-labelled for PATZ1 (rabbit polyclonal) and ERβ (mouse monoclonal). Higher-magnification images showing GC-1 (E–H) or TCam-2 (Q–T) cells double-labelled for PATZ1 and ERβ (I–K). A single nucleus of a representative GC-1 cell co-expressing PATZ1 and ERβ (L). Highest magnification of the frame depicted in (K) showing PATZ1 and ERβ immunoreactivities in the GC1 nucleus (U–X). Co.localization of PATZI with Golgin-97 in TCam2-call (A–D). Quantification of PATZ+ cells, ERβ+ cells, PATZ1/ERβ+ double-labelled cells and PATZ1/Golgin-97+ double-labelled cells counted in 10 non-overlapping microscope fields from five coverslips of GC-1 and TCam-2 cells. Data are expressed as percentages of total Hoechst-positive nuclei (a–c) or Golgin-97+ Golgi (d); ∗p < 0.05 versus TCam-2 cells (a–c) or GC-1 cells (d). Scale bars: A–D 25 µm; E–H 10 µm; I–K 5 µm; L–1 µm; M–P 50 µm; Q–X 20 µm.

Total proteins were prepared as described [29,30]. (diluted 1 : 1000); (c) anti-β-tubulin (diluted 1 : 1000); Differential extraction of nuclear or cytoplasmic pro- (d) anti-γ-tubulin (diluted 1 : 1000); (e) anti-HA– teins was obtained as previously described [31]. Mem- PATZ1 (diluted 1 : 500); (f) anti-SP1 (diluted 1 : 1000); branes were blocked with 5% non-fat milk proteins (g) anti-PKA-R (diluted 1 : 1000); and (h) anti-PKA-C and incubated with the following primary antibod- (diluted 1 : 1000). Also, for western blot analyses addi- ies: (a) anti-PATZ1 (diluted 1: 500); (b) anti-ERβ tional antibodies were used for PATZ1 and ERβ,which

Figure 3. (A) Western blot analysis of PATZ1, ERβ and PKA on cytoplasmic and nuclear extracts of GC1 and TCam2 cells; 50 µg proteins were resolved on 10% SDS–PAGE, transferred onto nitrocellulose filters and western-blotted with anti-PATZ1 and ERβ rabbit polyclonal antibodies; SP1 and γ-tubulin were used as controls of fractionated proteins. N, nuclear extracts; C, cytoplasmic extracts. (B) Immunoprecipitation and western blot detection in GC1 cells. Cell lysates were immunoprecipitated with ERβ,resolvedbySDS–PAGE, transferred to nitrocellulose membranes and challenged with anti-ERβ, anti-PATZ1 and PKA-C (rabbit polyclonal antibodies) and PKA-R (mouse monoclonal). TCE, total cell extracts. The blots are representative of three separate experiments. (C) Immunoprecipitation and western blot detection in 7-day mice spermatogonia. Cell lysates were immunoprecipitated with ERβ, resolved by SDS–PAGE, transferred to nitrocellulose membranes and challenged with anti-ERβ and anti-PATZ1. TCE, total cell extracts. The blots are representative of three separate experiments. gave the same results (not shown). Bound antibod- in fact, total protein HEK293 cell lysates were ies were detected by horseradish peroxidase-conjugated immunoprecipitated with anti-ERβ antibody, submit- secondary antibodies, followed by enhanced chemilu- ted to SDS–PAGE and then immunoblotted against minescence (Amersham Life Sciences, UK). As a con- HA–PATZ1 antibody (Figure 1A). Conversely, total trol for equal loading of protein lysates, the blotted pro- protein HEK293 cell lysates were immunoprecipi- teins were probed with antibody against anti-γ-tubulin tated with anti-HA–PATZ1 antibody, submitted to protein. SDS–PAGE, and then immunoblotted against ERβ antibody (Figure 1B). Co-immunoprecipitation of Statistical analysis exogenous ERβ and HA–PATZ1 proteins was con- Data are expressed as mean ± SEM of values obtained firmed by western blot analysis of total protein of in three separate experiments. Statistical comparisons HEK293 cell lysates (Figure 1C). β between GC1 cells and TCam-2 cells were performed The coexistence of PATZ1 and ER protein expres- using Student’s t-test. p<0.05 was considered signif- sion has been analysed in the GC1 and TCam2 icant. cell lines, derived from immortalized type B murine spermatogonia and human seminoma, respectively [24–26]. Confocal double immunofluorescence stain- ing indicated that PATZ1 and ERβ immunoreactiv- Results ities were co-expressed in the nuclei of GC1 cells (Figure 2A–D). Quantitative analysis showed that PATZ1 interacts with oestrogen receptor-β (ERβ) 50% of GC1 nuclei expressed PATZ1 or ERβ pro- The PATZ1 gene was first identified by a two- teins, and among them 90% of nuclei co-localized hybrid assay, using the RING finger protein RNF4 (Figure 2A–D). By contrast, in the human-derived as bait [14]. Our data indicate that PATZ1 asso- seminoma TCam2 cells, <10% of nuclei express ciates with ERβ in HEK293 cells transfected with the either PATZ1 or ERβ proteins. In particular, in these HA–PATZ1 construct and the full-length human ERβ cells (90%), PATZ1 expression was confined to the (Figure 1). Immunoprecipitation analysis in HEK293 cytoplasmic compartment of the Golgi, as demon- cells indicated that HA–PATZ1 and ERβ interacted; strated by co-localization experiments with anti-Golgin

Table 1. Immunohistochemical analysis of ERβ and PATZI in human testicular seminomas Case no. Age (years) ERβ PATZ1-n PATZ1-c 127+++ +++ − 232+++++ 344+++++ 434+++ +++ − 533+++++ 644+−+++ 733+++++ 828+++++ 925+−+++ 10 19 +++ +++ − 11 35 +−+++ 12 37 ++ ++ 13 27 +−+++ 14 27 +++ +++ − 15 33 +++++ 16 31 +++ +++ − 17 27 +++++ 18 25 +++++ 19 26 +++ +++ − 20 35 +++++ 21 32 +++++

PATZ1-n, nuclear expression of PATZ1; PATZ1-c, cytoplasmic expression of PATZ1; −, 0–10% of neoplastic expressing cells; +, 11–30% of expressing neoplastic cells; ++, 31–70% of expressing neoplastic cells; +++, >70% of expressing neoplastic cells.

97 antibody (Figure 2M–X, a–d). Interestingly, in ERβ down-regulation associates with PATZ1 these cells showing PATZ1 delocalization, no ERβ delocalization in human seminomas immunosignal was observed (Figure 2M–T). In agreement with the immunofluorescence results, Because the transcriptional repressor PATZ1 is ex- western blot analysis of the cytoplasmic and nuclear pressed in spermatogonia and disruption of cell cycle protein fractions from either GC1 or TCam2 cells showed that both PATZ1 and ERβ protein expression control is a hallmark of cancer, we examined samples were abundantly observed in both the nuclear and from 21 patients with seminomas by immunohisto- cytoplasmic fractions of GC1 cells. By contrast, in chemistry, using two different antibodies raised against TCam2 cells, PATZ1 protein was largely detected human PATZ1. In our cases the protein was gener- in the cytoplasmic fraction. In this latter fraction, ally expressed in seminomas (Table 1). In particular, ERβ expression was undetectable (Figure 3A). In the PATZ1 protein was highly expressed in the cytoplasm nuclear fraction of TCam2 cells, both PATZ1 and ERβ in 15 of 21 seminomas examined (Figure 4, Table 1). expression were barely detectable (Figure 3A). These findings are in in line with our previous pub- In addition, immunoprecipitation analysis in GC1 lished results [21]. In addition, an interesting obser- cells clearly indicated that PATZ1 and ERβ inter- vation was that the PATZ1 cytoplasmic localization is acted; in fact, total protein GC1 cell lysates were associated with ERβ down-regulation (Figure 4G,H), immunoprecipitated with anti-ERβ antibody, submitted whereas PATZ1 nuclear localization associated with to SDS–PAGE and then immunoblotted against PATZ1 ERβ high nuclear expression (Figure 4A,B). These antibody. Co-immunoprecipitation of endogenous ERβ observations have been confirmed using immunoflu- and PATZ1 was detected (Figure 3B). Interestingly, we orescence confocal analysis on the same cases. As also observed an interaction of the regulatory subunit shown in Figure 4, in the seminoma (case 4) ERβ of the cAMP-dependent protein kinase (PKA-R) with and PATZ1 co-localized (Figure 4C–F), while they PATZ1, but not with the catalytic subunit of the cAMP- did not in the seminoma (case 2), in which low dependent protein kinase (PKA-C) (Figure 3B). ERβ nuclear and PATZ1 cytoplasmic localization were Because we have previously demonstrated that present (Figure 4I–L). PATZ1, among the normal germ cells, is present exclu- sively in the spermatogonia, we evaluated PATZ1 and These observations have been confirmed by western ERβ protein interaction in 7-day mice spermatogonia blot analysis of tissues obtained from the same patients. [21]. Immunoprecipitation analysis clearly indicated As shown in Figure 5, analysis of the cytoplasmic and that PATZ1 and ERβ interacted; in fact, total pro- nuclear protein fractions from these samples indicated tein mice spermatogonia lysates were immunoprecipi- that PATZ1 protein was more abundant in the cyto- tated with anti-ERβ antibody, submitted to SDS–PAGE plasmic fraction compared to the nuclear fraction in and then immunoblotted against PATZ1 antibody. Co- seminomas (Figure 5) and was associated with low immunoprecipitation of endogenous ERβ and PATZ1 ERβ nuclear expression; these results are in agreement was detected (Figure 3C). with the immunohistochemical and confocal data.

Figure 4. Immunohistochemistry and immunofluorescence analyses of ERβ and PATZ1 protein expression in human testicular seminomas, using anti-PATZ1 (rabbit polyclonal) and ERβ (mouse monoclonal) antibodies. Seminoma (case 4) in which ERβ (A) and PATZ1 (B) intense nuclear positivity was observed by immunohistochemistry; confocal microscopic images displaying both ERβ (C) and PATZ1 (D) immunoreactivities in the nuclei of seminoma (case 4). Scale bars = 10 µm (C–F). Seminoma (case 2) in which an absence of immunopositivity of ERβ (G) was observed in association with an intense and diffuse PATZ1 cytoplasmic immunosignal (H) by immunohistochemistry (magnification, ×40); confocal microscopic images displaying an absence of immunopositivity of ERβ (I) and PATZ1 cytoplasmic (D) immunoreactivities in the nuclei of seminoma (case 2). Scale bar = 10 µm(I–L). cAMP treatment caused the translocation of PATZ1 whether cAMP might be able to promote relocalization from the cytoplasm to the nucleus of PATZ1 into the nucleus in a human seminoma cellu- Recently, it has been shown that the BTB-POZ domain lar line, TCam-2 cells were exposed to the second mes- zinc-finger transcription factor PATZ1 interacts with senger analogue 8Br-cAMP (100 µM), and PATZ1 and the RIa subunit of PKA, and the translocation of ERβ protein expression were analysed by immunoflu- PATZ1 from the cytoplasm into the nucleus is reg- orescence analysis. We observed an increased cyto- ulated by cAMP in normal fibroblasts [32]. To test plasmic expression of PATZ1 after 6 h of 8Br-cAMP

and delocalization of PATZ1 in human testicular seminomas compared with normal testes. One hypothesis is that the aberrant expression of PATZ1 suggests an important role of PATZ1 in tumorigenesis [21]. More- over, the splice variant PATZ1 has been shown to tran- scriptionally repress the c-myc, CDC6 and galectin-1 promoters [14]. In addition, MAZR, the mouse homo- logue of PATZ1, exhibited transactivation potential from the SV40 and the c-myc promoter [19], thus suggesting that PATZ1 may have transcriptional potential. Moreover, it has been shown that PATZ1 is an androgen receptor (AR) co-regulator that acts by modulating the effect of the AR co-activator RNF4 [37,38], a protein expressed in normal germ cells but not in human testicular tumours [39]. In addition, it is interesting to note that SNURF/RNF4 is a co-regulator of ERβ and both are severely repressed in seminomas [12,13]. In this paper, we report for the first time that PATZ1 interacts with and may function as a novel co-regulator Figure 5. Western blot analysis of PATZ1 and ERβ expression in of ERβ, and interacts also with the regulatory sub- human normal testis and seminoma; 40 µg proteins were resolved on 10% SDS–PAGE, transferred onto nitrocellulose filters and unit of the cAMP-dependent protein kinase (PKA-R). western-blotted with anti-PATZ1 and -ERβ rabbit polyclonal serum For several years the mechanism of steroid receptor on cytoplasmic and nuclear extracts in normal testis (NT) and action has been considered very simple, consisting of seminoma (case 2); SP1 and β-tubulin were used as controls of hormone binding followed by nuclear import, recogni- fractionated proteins. N, nuclear extracts; C, cytoplasmic extracts. tion of specific DNA motifs and direct transcriptional The blots are representative of three separate experiments. activation (or repression). During the last few years, several putative co-regulatory proteins have been iden- treatment (Figure 6E); surprisingly, at the same time- tified for nuclear receptors. Although the physiological point, stimulation with the second messenger analogue role of nuclear receptor co-activators remains to be elu- β 8Br-cAMP caused an up-regulation of ER immunore- cidated, it is becoming clear that these ligand-inducible activity in the cytoplasm (Figure 6F). When the cells transcription factors are capable of associating, proba- were exposed to 8Br-cAMP for 24 h, the translocation bly simultaneously, with multiple target proteins. Iden- β of both PATZ1 and ER proteins from the cytoplasm to tification of an increasing number of proteins with the the nucleus was observed (Figure 6I,J). These effects potential to enhance or repress steroid receptor action were counteracted by using 8Br-cAMP treatment in indicates that the activation process is complex and is combination with the anti-oestrogen ICI 182–780 controlled, at least in part, by the cellular ratio of co- µ (100 M). In fact, when the cells were exposed to 8Br- regulatory proteins. In addition, it appears always more + cAMP ICI 182–780 for 24 h, translocation of both evident that steroid receptor co-factors are important β PATZ1 and ER proteins from the cytoplasm to the in regulating receptor function through activation or nucleus was not observed (Figure 6M,N). repression of various target genes in endocrine cancers [40]. Our data suggest a novel signalling mechanism in Discussion which the sequestration of PATZ1 in the cytoplasm is regulated by cAMP, thus enabling PATZ1 to translo- The down-regulation of ERβ observed in semino- cate into the nucleus and transactivate its target genes mas is in accordance with data from animal mod- upon activation of the cAMP pathway. Therefore, fur- els and human cell culture studies suggesting that ther understanding of the interactions of PATZ1 and ERβ may control and limit cell proliferation during their effects on growth regulation may offer insights breast, prostate, ovarian and colon cancer progression into the mechanisms of cAMP-mediated growth con- [33–36]. These considerations have led us to hypothe- trol. Our results are in accordance with the results of size that exposure to oestrogens or oestrogen mimics, Yang and co-workers, which showed the interaction in some as-yet undefined manner, diminishes ERβ- of the regulatory subunit of the cAMP-dependent pro- mediated growth restraint in spermatagonia, which tein kinase (PKA) with PATZ1 in normal fibroblasts favours unscheduled cell proliferation. The affected [32]. In addition, for the first time we demonstrate that spermatogonia or their descendants may then be able to cAMP induces the expression and nuclear localization escape normal cell cycle regulation and be at a higher of ERβ in TCam-2 cells, in which the ERβ is normally risk of undergoing malignant transformation. down-regulated. The translocation of PATZ1 from the We consistently detected that decreased expression cytoplasm to the nucleus induced by cAMP is counter- of ERβ protein associates with increased expression acted by the anti-oestrogen ICI 182–780, suggesting

Figure 6. Effect of 8Br-cAMP on cytoplasmic and nuclear expression of PATZ1 and ERβ in TCam-2 cells. Co-localization of PATZ1 and ERβ proteins in TCam-2 cells following 6 or 24 h of 8Br-cAMP (10−9 M) exposure or 24 h of 8Br-cAMP + anti-oestrogen ICI 182–780 exposure (10−9 M). Confocal microscopic images displaying both PATZ1 and ERβ immunoreactivities in control cells (A–D), in cells exposed to 8Br-cAMP (10−9 M) for 6 h (E–H) or 24 h (I–L), and in cells exposed to 8Br-cAMP (10−9 M) plus the anti-oestrogen ICI 182–780 (10−9 M) for 24 h (M–P). Nuclei were counterstained with Hoechst. Scale bars = 20 µm(A–P). that it is mediated by ERβ. Yang and co-workers sug- thecMycpromoterandthepresenceofcAMPandco- gest that in addition to inhibiting the C subunit kinase expression with RIa modulates its transactivation [32]. activity, the RIa subunit may have novel function by Taken together, our results demonstrate the interac- interacting with and sequestering PATZ1 in the cyto- tion between PATZ1 and ERβ in normal germ cells; plasm, thereby regulating its transcriptional activity and in addition, we show that PATZ1 translocation from function in cell growth control in response to cAMP. the cytoplasm into the nucleus is mediated by cAMP, The effects of cAMP in the cell are predominantly which also induces increased ERβ expression and mediated by the cAMP-dependent PKA, which is com- nuclear localization. Consistently, the PATZ1 delocal- posed of two genetically distinct subunits, catalytic ization associated with ERβ down-regulation in human (C) and regulatory (R), forming a tetrameric holoen- testicular seminomas could be ascribed to impaired zyme R2C2. It has been shown that over-expression cAMP-mediated signalling to generate the testicular of RIa, but not the C subunit kinase, is associated neoplasia. with neoplastic transformation [32]. In addition, it has also been demonstrated that mutation in the RIa but not the C subunit is associated with increased resis- Acknowledgment tance to the DNA-damaging anticancer drug cisplatin, thus suggesting that the RIa subunit of PKA may have This work was supported by grants to Paolo Chiefﬁ functions independent of the kinase [41,42]. The cyto- from the Italian Ministry of Education (Grant No. plasmic–nuclear translocation is inducible by cAMP. MIUR-PRIN 2007). We thank Fabrizio Fiorbianco It has been shown that C-terminus deletion abolishes (www.studiociotola.it) for skilful technical assistance PATZ1 interaction with RIa [32]. PATZ1 transactivates with the artwork.

Author contributions interplay between protein binding zinc fingers. JBiolChem2000; 275: 26144–26149. 17. Mastrangelo T, Modena P, Tornielli S, et al. A novel zinc finger FE carried out experiments (transfection, western blot- gene is fused to EWS in small round cell tumor. Oncogene 2000; ting and immunoprecipitation analyses); FB conceived 19: 3799–3804. and carried out experiments (immunofluorescence con- 18. Bilic I, Koesters C, Unger B, et al. Negative regulation of CD8 focal microscopy analysis) and analysed data; RF per- expression via Cd8 enhancer-mediated recruitment of the zinc formed histological analysis and immunohistochem- finger protein MAZR. Nat Immunol 2006; 7: 392–400. istry; MT performed experiments (western blotting and 19. Kobayashi A, Yamagiwa H, Hoshino H, et al. A combinatorial immunoprecipitation analyses); AF, SK and LL con- code for gene expression generated by transcription factor Bach2 tributed cell line; PC designed the study, conceived and and MAZR (MAZ-related factor) through the BTB/POZ domain. MolCellBiol 2000; 20: 1733–1746. carried out experiments, contributed cell line, analysed 20. Franco R, Esposito F, Fedele M, et al. Detection of high mobility data, wrote the paper and PC provide financial support. group proteins A1 and A2 represents a valid diagnostic marker All authors had final approval of the manuscript. in post-puberal testicular germ cell tumours. J Pathol 2008; 214: 58–64. 21. Fedele M, Franco R, Salvatore G., et al. PATZ1 gene has a critical References role in the spermatogenesis and testicular tumours. J Pathol 2008; 215: 39–47. 1. Chieffi P. Molecular targets for the treatment of testicular germ 22. Sakaguchi S, Hombauer M, Bilic I, et al. The zinc-finger protein cell tumors. Mini-Rev Med Chem 2007; 7: 755–759. MAZR is part of the transcription factor network that controls the 2. Chieffi P, Franco R, Portella G. Molecular and cell biology of CD4 versus CD8 lineage fate of double-positive thymocytes. Nat testicular tumors. Int Rev Cell Mol Biol 2009; 278: 277–308. Immunol 2010; 11: 442–448. 3. Oosterhuis JW, Looijenga LH. Testicular germ-cell tumors in a 23. Takada K, Takahama Y. Another zinc finger in the pie of broader perspective. Nat Rev Cancer 2005; 5: 210–222. CD4 –CD8 lineage choice. Nature Immunol 2010; 11: 370–371. 4. Henderson BE, Benton B, Jing J, et al. Risk factors for cancer of 24. Hofmann MC, Narisava S, Hess RA, et al. Immortalization of the testis in young men. Int J Cancer 1979; 23: 598–602. germ cells and somatic testicular cells using the SV40 large T 5. Wellejus A, Loft S. Receptor-mediated ethinylestradiol induced antigen. Exp Cell Res 1992; 201: 417–435. oxidative DNA damage in rat testicular cells. FASEB J 2002; 16: 25. Mizuno Y, Gotoh A, Kamidono S, et al. Establishment and char- 195–201. acterization of a new human testicular germ cell tumor cell line 6. Saunders PT, Millar MR, Macpherson S, et al.ERβ1andthe TCam-2. Nippon Hinyokika Gakkai Zasshi 1993; 84: 1211–1218. ERβ2 splice variant (ERβcx/β2) are expressed in distinct cell 26. de Jong J, Stoop H, Gillis AJ, et al. Further characterization of the populations in the adult human testis. J Clin Endocrinol Metab first seminoma cell line TCam-2. Genes Chrom Cancer 2008; 47: 2002; 87: 2706–2715. 185–196. 7. Vicini E, Loiarro M, Di Agostino S, et al.17β-estradiol elicits 27. Rossi P, Dolci S, Albanesi C, et al. Follicle-stimulating hormone genomic and non-genomic responses in mouse male germ cells. induction of steel factor (SLF) mRNA in mouse Sertoli cells, and J Cell Physiol 2006; 206: 238–245. stimulation of DNA synthesis in spermatogonia by soluble SLF. 8. Kuiper GG, Enmark E, Pelto-Huikko M, et al. Cloning of a novel Dev Biol 1993; 155: 68–74. receptor expressed in rat prostate and ovary. Proc Natl Acad Sci 28. Boscia F, Gala R, Pannaccione A, et al. NCX1 expression and USA 1996; 93: 5925–5930. functional activity increase in microglia invading the infarct core. 9. Pentikainen V, Erkkila K, Suomalainen L, et al. Estradiol acts as Stroke 2009; 40: 3608–3617. a germ cell survival factor in the human testis in vitro. J Clin 29. Chieffi P, Battista S, Barchi M, et al. HMGA1 and HMGA2 Endocrinol Metab 2000; 85: 2057–2067. protein expression in mouse spermatogenesis. Oncogene 2002; 21: 10. Katzenellenbogen BS, Montano MM, Ediger TR, et al. Estrogen 3644–3650. receptors: selective ligands, partners, and distinctive pharmacology. 30. Esposito F, Libertini S, Franco R, et al. Aurora B expression in Recent Prog Horm Res 2000; 55: 163–193. post-puberal testicular germ cell tumours. J Cell Physiol 2009; 11. Pettersson K, Gustafsson JA.˚ Role of estrogen receptor-β in estro- 221: 435–439. gen action. Ann Rev Physiol 2001; 63: 165–192. 31. Di Vizio D, Cito L, Boccia A, et al. Loss of tumour suppressor 12. Pais V, Leav I, Lau KM, et al. Estrogen receptor-β expression in gene PTEN marks the transition from intratubular germ cell human testicular germ cell tumours. Clin Cancer Res 2003; 9: neoplasias (ITGCN) to invasive germ cell tumors. Oncogene 2005; 4475–4482. 24: 1882–1894. 13. Hirvonen-Santti SJ, Rannikko A, Santti H, et al. Down- 32. Yang Wl, Ravtn R, Kudoh K, et al. Interaction of the regula- regulation of estrogen receptor-β and transcriptional coregulator tory subunit of the cAMP-dependent protein kinase with PATZ1 SNURF/RNF4 in testicular germ cell cancer. Eur Urol 2003; 44: (ZNF278). Biochem Biophys Res Commum 2010; 391: 1318–1323. 742–747. 33. Roger P, Sahla ME, Makela S, et al. Decreased expression of 14. Fedele M, Benvenuto G, Pero R, et al. A novel member of the estrogen receptor-β protein in proliferative preinvasive mammary BTB/POZ family, PATZ, associates with the RNF4 RING finger tumors. Cancer Res 2001; 61: 2537–2541. protein and acts as a transcriptional repressor. JBiolChem2000; 34. Pasquali D, Rossi V, Esposito D, et al. Loss of estrogen receptor 275: 7894–7901. beta expression in malignant human prostate cells in primary 15. Galili N, Nayak S, Epstein JA, et al. Rnf4, a RING protein cultures and in prostate cancer tissues. J Clin Endocrinol Metab expressed in the developing nervous and reproductive systems, 2001; 86: 2051–2055. interacts with Gsc1, a gene within the DiGeorge critical region. 35. Weyant MJ, Carothers AM, Mahmoud NN, et al. Reciprocal Dev Dyn 2000; 218: 102–111. expression of ERα and ERβ is associated with estrogen-mediated 16. Lingsø C, Bouiteiller G, Damgaard CK, et al. Interaction between modulation of intestinal tumorigenesis. Cancer Res 2001; 61: the transcription factor SPBP and the positive cofactor RNF4. An 2547–2551.

36. Staibano S, Franco R, Mezza E, et al. Oestrogen receptor-β, 40. Barnes CJ, Vadlamudi RK, Kumar R. Novel estrogen receptor PCNA, p53, and DNA ploidy: are they markers of biological behav- coregulators and signaling molecules in human diseases. Cell Mol ior in ovarian granulosa cell tumours? Histopathology 2003; 43: Life Sci 2004; 61: 281–291. 254–262. 41. Cvijic ME, Chin KV. Regulation of P-glycoprotein expression in 37. Pero R, Lembo F, Palmieri EA, et al. PATZ attenuates the RNF4- cyclic AMP-dependent protein kinase mutants. Cell Growth Differ mediated enhancement of androgen receptor-dependent transcrip- 1997; 8: 1243–1247. tion. JBiolChem2002; 277: 3280–3285. 42. Liu B, Cvijic ME, Jetzt A, et al. Cisplatin resistance and regula- 38. Pero R, Lembo F, Chiefﬁ P, et al. Translational regulation of a tion of DNA repair in cAMP-dependent protein kinase mutants. novel testis-speciﬁc RNF4 transcript. Mol Reprod Dev 2003; 66: Cell Growth Differ 1996; 7: 1105–1112. 1–7. 39. Pero R, Lembo F, Di Vizio D, et al. RNF4 is a growth inhibitor expressed in germ cells and lost in human testicular tumours. Am J Pathol 2001; 159: 1225–1230.

Copyright  2011 Pathological Society of Great Britain and Ireland. J Pathol 2011; 224: 110–120 Published by John Wiley & Sons, Ltd. www.pathsoc.org.uk www.thejournalofpathology.com