Lineage-Specific Restraint of Pituitary Gonadotroph Cell Adenoma Growth

Although pituitary adenomas are usually benign, unique trophic mechanisms restraining cell proliferation are unclear. As GH-secreting adenomas are associated with p53/p21-dependent senescence, we tested mechanisms constraining non-functioning pituitary adenoma growth. Thirty six gonadotroph-derived non-functioning pituitary adenomas all exhibited DNA damage, but undetectable p21 expression. However, these adenomas all expressed p16, and >90% abundantly expressed cytoplasmic clusterin associated with induction of the Cdk inhibitor p15 in 70% of gonadotroph and in 26% of somatotroph lineage adenomas (p = 0.006). Murine LβT2 and αT3 gonadotroph pituitary cells, and αGSU.PTTG transgenic mice with targeted gonadotroph cell adenomas also abundantly expressed clusterin and exhibited features of oncogene-induced senescence as evidenced by C/EBPβ and C/EBPδ induction. In turn, C/EBPs activated the clusterin promoter ∼5 fold, and elevated clusterin subsequently elicited p15 and p16 expression, acting to arrest murine gonadotroph cell proliferation. In contrast, specific clusterin suppression by RNAis enhanced gonadotroph proliferation. FOXL2, a tissue-specific gonadotroph lineage factor, also induced the clusterin promoter ∼3 fold in αT3 pituitary cells. As nine of 12 pituitary carcinomas were devoid of clusterin expression, this protein may limit proliferation of benign adenomatous pituitary cells. These results point to lineage-specific pathways restricting uncontrolled murine and human pituitary gonadotroph adenoma cell growth.


Introduction
Pituitary tumors arise from highly specialized cell types expressing the respective pituitary polypeptide hormones. Thus, tumors derived from somatotrophs secrete growth hormone (GH), lactotrophs, prolactin (PRL), thyrotrophs, thyrotropin (TSH), and corticotrophs, adrenocorticotropin (ACTH). In contrast, nonfunctioning pituitary tumors usually arise from non-secreting cells of gonadotroph origin [1]. Clinically inapparent pituitary tumors are identified in 25% of autopsy specimens with a population prevalence of ,77 cases/10 5 . Pituitary tumors are usually benign neoplasms (adenomas), however, they may also exhibit invasive or recurrent growth. Rarely encountered malignant pituitary carcinomas comprise 0.02% of all pituitary tumors, proliferate rapidly and show extracranial metastases [2,3,4]. Although most aggressive pituitary adenomas persistently exhibit low mitotic activity [3], mechanisms underlying these unique growth properties are largely elusive. We postulate that intrinsic cell-specific trophic properties as well as the lineage-origin of highly differentiated and specialized pituitary cells underlies constrained adenoma proliferation. Cellular senescence is characterized by irreversible proliferative arrest, while cells remain viable and metabolically active. Proliferation arrest may occur as a result of age-related telomere shortening, and also in response to oxidative or genotoxic stress, DNA damage, aneuploidy or chromosomal instability, as well as oncogene activation [5,6]. Thus, oncogenic RAS causes stable proliferative arrest rather than transformation in diploid fibroblasts [7]. BRAF in benign skin nevi elicits an initial increased proliferation followed by DNA stress and cellular senescence [8].
In most human GH-producing pituitary adenomas PTTG overexpression is associated with DNA damage and p21dependent senescence [34], however pathways restraining growth and transformation of the more commonly encountered nonfunctioning pituitary adenomas are not known. We show here that similar to GH-cell adenomas, tumors arising from the gonadotroph lineage exhibit high PTTG levels and DNA damage.
However, unlike GH-cell adenomas, p53/p21 senescence markers are not activated in non-functioning adenomas, which do, however, selectively express abundant cytoplasmic clusterin. High clusterin levels restrain cell proliferation by triggering Cdk inhibitors p15, p16 and p27, while suppression of clusterin expression enhanced pituitary gonadotroph cell proliferation. Thus, we identify a novel role for clusterin in enabling pituitary gonadotroph tumor cell proliferation arrest. FOXL2, a transcription factor specifically expressed in pituitary gonadotroph cells [35] stimulates the clusterin promoter, further highlighting a differential lineage-specific pathway restricting pituitary cell cycle progression, acting to buffer non-functioning pituitary adenomas from unrestrained growth.

DNA damage and senescence markers are induced in human pituitary adenomas
Immunoreactive PTTG was induced in all 36 gonadotroph cell adenomas analyzed, but not in normal pituitary tissue ( Figure 1A), confirming previous reports [22,23,24]. Markers of DNA damage and aneuploidy including cH2A.X foci and phopsphorylated kinase mutated in ataxia telangiectasia (pATM) [36,37] were not detected by fluorescent immunohistochemistry in two nontumorous human pituitary specimens. In contrast, all of 12 human pituitary adenomas analyzed (2 GH cell, 10 non-secreting gonadotroph cell) invariably expressed both cH2A.X and pATM, reflecting activated DNA damage signaling ( Figure 1B).
We showed earlier that human GH-secreting adenomas, but not carcinomas, abundantly express intra-nuclear p21, an end-point inhibitor of cell proliferation in the senescence pathway [29,34]. In contrast, gonadotroph cell adenomas did not express p21 ( Figure 2). However, because these tumors also do not (with exceedingly rare exceptions) evolve to malignancy, we analyzed additional pathways restraining pituitary tumor cell proliferation. p16 and p15 of the ARF/INK Cdk inhibitor family act to restrain cellular proliferation in response to activated oncogenes [8,10], and were strongly expressed in gonadotroph adenomas. Seventy percent of gonadotroph adenomas expressed high levels of p15, as compared to 26% of GH-secreting adenomas (p = 0.006, Table 1). Thus, both DNA damage pathways and senescence markers were expressed in gonadotroph cell-derived adenomas ( Figure 2).

aGSU.PTTG mice exhibit features of oncogene-induced pituitary senescence
We recapitulated human gonadotroph tumors in an in vivo transgenic murine model of gonadotroph PTTG expression driven by the aGSU promoter [28]. aGSU.PTTG pituitary glands express up-regulated gonadotroph PTTG with pituitary hyperplasia starting from 4 months of age leading to development of focal pituitary adenomas expressing LH. Other transgenic lines also expressed GH and PRL [38]. In accordance with evidence supporting proto-oncogenic properties of PTTG [26], hyperplastic pre-tumorous pituitary glands derived from transgenic animals were shown to already express markers of increased pituitary proliferation as evidenced by increased BrdU incorporation ( Figure 4A ), and elevated levels of pro-proliferative proteins including PCNA and E2F1 in vivo ( Figure 4B). However, as these animals developed penetrant pituitary tumors only after 10 months, and these invariably remain small, we tested whether pituitary PTTG overexpression also affects anti-proliferative pathways in these transgenic mice.
DNA damage was already evident in pre-tumorous transgenic pituitary glands overexpressing PTTG as evidenced by enhanced pituitary cH2A.X and pATM levels, accompanied by induced DNA damage repair proteins including MSH2, MLH1 and Rad51, as well as tumor suppressors including p19 and p53 ( Figure 4C). The Cdk inhibitor p27, a marker of DNA damage [39], was also induced in pre-tumorous aGSU.PTTG pituitary glands, as were the cell cycle suppressor proteins p15 and p16 ( Figure 4D). Two intracellular forms of pituitary clusterin, a mature glycosylated ,76 kDa secretory form and ,60 kDA presecretory form were up-regulated in the pretumorous transgenic pituitary gland ( Figure 4D).
These results were confirmed by fluorescent immunostaining. Although only modest cytoplasmic clusterin, and intra-nuclear p15 and p16 expression were observed in WT murine pituitary glands, expression of these 3 proteins was enhanced in the transgenic pre-tumorous pituitary, and further induced in aGSU.PTTG pituitary tumors ( Figure 4E). Thus, features of oncogene-induced senescence in the aGSU.PTTG pituitary included induction of the p19/p53/p27 DNA damage pathway, and both p15 and p16 Cdk inhibitors [7,40]. In the pre-tumorous hyperplastic aGSU.PTTG pituitary gland the observed increased SA-b galactosidase activity supported the presence of cellular senescence ( Figure 4F).

Pttg over-expression in LbT2 cells results in a senescent phenotype
To recapitulate in vivo effects of pituitary Pttg over-expression, we transiently transfected murine gonadotroph-derived LbT2 cells with a plasmid expressing murine Pttg, and also isolated LbT2 cells stably overexpressing Pttg. As shown in Figure 5, Pttg overexpression lead to induction of clusterin and p15 in both gonadotroph cell transfectants, similar to in vivo patterns observed in the aGSU.PTTG pituitary ( Figure 4D).
Clones of stably transfected LbT2 cells sorted and selected for high PTTG expression, showed lower rates of BrdU incorporation as compared to control vector-expressing cells, reflecting decreased proliferation ( Figure 5C). These transfectants were spread-out, and larger in size with giant aneuploid nuclei, consistent with a senescent phenotype ( Figure 5D). High Pttg expression resulted in decreased apoptosis as detected by TUNEL assay ( Figure 5E), and these cells also exhibited increased SA-bgalactosidase activity ( Figure 5 F,G). Thus, constitutively high gonadotroph cell Pttg expression resulted in premature cellular senescence similar to the in vivo pituitary phenotype observed in aGSU.PTTG mice.   [41,42,43]. In the pre-tumorous hyperplastic aGSU.PTTG pituitary, C/EBPb was induced both in aGSU, and in GH-and PRL-secreting cells ( Figure 6A). In LbT2 cells stably expressing Pttg, several C/EBPb and C/EBPd isoforms [41] ) were also up-regulated (Figure 6 B).
We therefore assessed whether C/EBPs activate the clusterin promoter in LbT2 cells, and also in a murine gonadotroph-derived aT3 cell line. pGL3-mClu-luc reporter plasmid co-transfected with full length murine C/EBPb or with C/EBPd constructs ( Figure 6C) resulted in induced luciferase activity in both cell types. However, the plasmid encoding C/EBPa did not induce the clusterin promoter, indicating the specificity of C/EBPb and d effects. Accordingly, clusterin protein levels were also found to be upregulated in cells transfected with pcDNA3-C/EBPb and d respectively (Figure 6 D,E).
As LbT2 mPttg cells exhibit increased clusterin levels, we treated these cells with siRNAs directed against either C/EBPb or C/ EBPd. However, separate suppression of either of these genes resulted in compensatory increase of the other protein (data not shown), while simultaneous suppression of C/EBPb and d with 3 nM of each RNAi lead to decreased clusterin protein levels in LbT2 mPttg cells after 48 hours. These experiments were conducted with two different siRNA combinations directed against both C/EBPb and d, and a representative Western blot is shown in Figure 6F. The results confirmed that both C/EBP proteins act to regulate gonadotroph cell clusterin expression.

Clusterin restrains pituitary cell proliferation
As high clusterin expression was observed in benign gonadotroph adenomas and in small slow-growing aGSU.PTTG pituitary tumors (Figures 3 and 4E), and also in LbT2 cells overexpressing mPttg ( Figure 5A,B), we analyzed the effects of altering intracellular clusterin levels. Transient transfection with mClu-pIRES2-ZsGreen1 resulted in increased p15, p16, and p27 expression in LbT2, and p16 in the aT3 cell line. In contrast, levels of phosphorylated histone H3 (pH 3), a specific marker for S and M phases were attenuated in both cell types after clusterin transfection ( Figure 7A,B).
After synchronization of aT3 cells stably overexpressing mClu, and adding 10% FBS, cells were pulsed with BrdU, and flow cytometry demonstrated that BrdU incorporation was decreased, reflecting attenuated DNA synthesis ( Figure 7F).
Next we suppressed clusterin expression in LbT2 mPttg stable transfectants, and also tested aT3 cells, where endogenous clusterin levels were relatively high. Both cell lines were transfected with 6 nM of two different siRNAs directed against clusterin, and the pRS shClu-GFP RNA expressing plasmid. Treatments with siClu#2 and transfection with shClu-GFP both depleted clusterin mRNA by 80% as measured by real time PCR (data not shown). Figure 8 (A,B) shows Western blots of representative experiments using siClu#2 where both clusterin protein forms were downregulated, p15 and p16 levels were low, and pH 3 protein levels induced in both cell lines. In LbT2mPttg cells and in aT3 cells, clusterin suppression led to increased numbers of cells incorporating BrdU respectively (p,0.05) ( Figure 8C). Depletion of clusterin by shClu RNA resulted in 30 and 26% increase in BrdU-positive LbT2mPttg cells and aT3 cells respectively (data not shown).
The results show that in pituitary gonadotroph cells, induced clusterin restrains proliferation associated with up-regulated p15 and p16, while clusterin depletion led to decreased p15 and p16, accompanied by increased cell proliferation. Indeed, suppression of p15 and p16 transcription by respective siRNAs markedly increased the number of BrdU-incorporated LbT2mPttg cells 48 hours after transfection, reflective of increased cell proliferation (Figure 8 D,E).

FOXL2 activates clusterin promoter in gonadotroph pituitary cells
Forkhead box gene transcription factor L2 (FOXL2) is a cellspecific factor for pituitary gonadotroph differentiation and triggers aGSU expression [35]. FOXL2 is abundantly expressed in human pituitary gonadotroph and null cell adenomas [45], and in normal pituitary co-localizes with LH, FSH and aGSU [35]. As distribution of FOXL2 appeared to mirror clusterin expression in pituitary adenomas, we tested whether FOXL2 stimulates clusterin in gonadotroph cells. pGL3-mClu-luc reporter plasmid was co-transfected with the murine FOXL2 (pcDNA3-His-Foxl2) in aT3 cells ( Figure 9A), and luciferase activity was induced,3 fold (p,0.01) indicating the stimulatory effect of FOXL2 on the clusterin promoter. Clusterin protein levels were also enhanced in cells transfected with pcDNA3-His-Foxl2 ( Figure 9B). In contrast, clusterin was not induced in ACTH-secreting AtT20 murine corticotroph pituitary cells transfected with pcDNA3-His-Foxl2 (data not shown), indicating cell specificity of FOXL2 action on the clusterin promoter.
We then tested whether FOXL2 is recruited to the endogenous clusterin promoter. Lysates derived from aT3 pituitary gonadotrophs were isolated and chromatin immunoprecipitation (ChiP) assays performed with a polyclonal FOXL2 antibody. FOXL2 was shown to bind the clusterin promoter, spanning 2700-1339 nucleotides upstream from the transcription start site ( Figure 9C), but did not bind the 21468-1865 or +1-700 promoter regions (not shown). Enrichment of specific 2700-1339 clusterin promoter sequences in the precipitate indicated FOXL2 association with the clusterin promoter in vivo.

Discussion
Pituitary lineage specificity determines highly distinct peripheral pituitary trophic hormone actions as exemplified by differentiated GH, PRL, ACTH, TSH or gonadotroph hormone functions. Cell-specific pituitary hormone synthesis is regulated by specific hypothalamic, intrapituitary and peripheral hormone signals [46,47,48,49,50]. Pituitary cells are also sensitive to aneuploidy, DNA damage or oncogene overexpression. In response to these insults, we show here that cell-type specific trophic pathways are activated, with the common end-point of pituitary cell proliferation arrest.
GH secreting tumors exhibit high PTTG, features of aneuplody, chromosomal instability, activation of DNA damage responses, p21-dependent cell proliferation restraint and senescence [34]. Although high PTTG levels are also observed in gonadotroph pituitary adenomas [26], unlike GH-cell adenomas, we now show that gonadotroph adenomas do not express p21, but abundantly express clusterin in a cell specific manner. In contrast, only modest clusterin expression was observed in non-tumorous pituitary glands and in GH/PRL-secreting tumors, while clusterin was undetectable in 76% of the very rarely encountered pituitary carcinomas.
The results in human tumors were validated using in vivo and in vitro models of gonadotroph cell adenomas with high PTTG expression. Reflecting pro-proliferative properties of excess PTTG, aGSU.PTTG pituitary glands exhibit microadenoma formation [38], also evidenced by increased in vivo pituitary BrdU incorporation, and up-regulated proliferation markers including PCNA and E2F1. Similar to human pituitary adenomas, in the aGSU.PTTG pituitary gland, high PTTG levels result in aneuplody and chromosomal instability, also evidenced by DNA damage, pATM induction and activation of p53/p27 pathways known to arrest cell proliferation in the course of continuing DNA damage [39]. Concordantly, both the pre-tumorous aGSU.PTTG The results suggest a biphasic response to transforming effects of excess PTTG in vivo: Abundant PTTG is apparently sufficient to trigger an initial proliferative burst leading to hyperplasia and tumor initiation, however, inability of these pituitary tumors to undergo persistent further growth, is likely due to proliferationrestraining pathways activated by PTTG overexpression. These finding underscore the observed SA-b-galactosidase activation in pre-tumorous aGSU.PTTG glands and in LbT2 cells constitutively expressing Pttg.
High clusterin levels were observed in the aGSU.PTTG pituitary gland and in LbT2 cells stably and transiently transfected with mPttg, suggesting that in human gonadotroph adenomas clusterin might be also induced by high PTTG levels. In gonadotroph cells, PTTG-overexpression is accompanied by C/ EBP induction. C/EBP proteins uniquely regulate cell-type specific growth and differentiation [41,42]. C/EBPb is associated with oncogene-induced senescence [51], while C/EBPd triggers growth arrest and cell differentiation [43,52,53]. Both C/EBPb and C/EBPd are shown here to activate the clusterin promoter, and induced clusterin protein expression was evident in gonadotroph cells and pituitary tissue overexpressing PTTG. Furthermore, a specific gonadotroph cell lineage transcription factor FOXL2, independently activates the clusterin promoter in these cells.
We show that forced clusterin expression in LbT2 and aT3 pituitary gonadotroph cells triggers a linage-specific cytostatic response, inducing p15, p16, or p27; decreased cell proliferation was also evidenced by lower expression of pH 3, similar to observations in prostate cancer cells [54]. Accordingly, when either p15 or p16 gene expression were suppressed, pituitary cell proliferation was enhanced. These results are in accordance with those showing that TGFb-induced p15 decreases proliferation and induces cell cycle arrest in rat GH 3 pituitary cells [55]. Thus, clusterin-triggered p15 and p16 likely restrain pituitary cell proliferation in aGSU.PTTG pituitary tumors and in LbT2 gonadotroph-derived cells. Induced p15 in human gonadotroph adenomas might therefore limit growth of these tumors.
Custerin function in tumorigenesis is unclear. Clusterin expression is enhanced in human prostate cancer, and antisense oligonucleotides targeting clusterin inhibit prostate tumorigenesis  [56]. Clusterin also induces breast cancer cell growth and metastatic progression [57] and is associated with human lung adenocarcinoma cell growth [58]. The nuclear anti-apoptotic form of clusterin is induced in late stage cancers following chemotherapy, hormonal ablation or radiotherapy, thus protecting tumor cells undergoing damaging stress [59]. As pituitary carcinomas are rarely treated with radiation or chemotherapy before surgery, we did not observe clusterin expression in human carcinoma specimens, as expected. Several lines of evidence also point to the role of clusterin as a tumor suppressor protein. Thus, clusterin was down-regulated, and its expression inversely proportional to tumor grade/or metastatic stage [16,18,60]. Patients with clusterin-positive lung cancer have enhanced disease-free survival [61]. Moreover, Clu 2/2 mice are more prone to oncogeneinduced tumorigenesis [17,21]. Although clusterin restrains proliferation of untransformed epithelial cells [21] and acts primarily as a tumor-suppressor during early stages of carcinogenesis [16,21], when re-expressed in advanced cancers, clusterin might promote tumor growth.
Based on the results presented here, we propose that in benign pituitary tumor cells of gonadotroph origin, the role of clusterin is to restrain proliferation. Similar affects were demonstrated for TGFb1, which functions as a tumor suppressor in normal epithelial cells and during early stages of tumor development [62]. In late-stage tumors TGFb1 exhibits features of a tumor promoter, modulating vascular and immune compartments of the tumor stroma [63]. Similarly, in normal fibroblasts [32] and in benign pituitary tumor cells (here and [34]), high PTTG restrains the cell cycle and leads to senescence. In transformed tumor cell lines and in malignant tumors, overexpressed PTTG triggers production of FGF-2 and VEGF-A, cell cycle progression and angiogenesis, and in malignant tumors, high PTTG levels correlate with tumor invasiveness and serve as a marker of poor prognosis (reviewed in [26]). Thus, the cellular environment appears to determine end-point effects of these proteins.
Our hypothesis is outlined in Figure 10. Human gonadotroph tumors express both FOXL2 and PTTG. The results shown here indicate that FOXL2 directly activates the clusterin promoter, while PTTG triggers clusterin via C/EBPs. Both C/EBPs and clusterin are also induced by DNA damage (data not shown and [59]). High clusterin, in turn, provokes p15, p16 and p27 expression in vivo and cell-specifically in vitro thus restraining gonadotroph cell proliferation. These observations point to the existence of intrinsic lineage-specific pathways restricting pituitary cell cycle progression. Activation of these pathways should be considered as a contributing factor underlying the overwhelmingly benign nature of pituitary adenomas, enabling maintenance of vital pituitary homeostatic and metabolic functions, while protecting the hormone-secreting gland from destruction by malignancy.

Human tissue samples
Pituitary tumors were freshly collected at transsphenoidal surgery according to an approved Cedars Sinai and Mayo Clinic Institutional Review Board protocols. Written informed consent were obtained from all participants. Samples were formalin-fixed and paraffin-embedded for immunohistochemistry. Diagnosis of individual tumors was established on the basis of clinical features, histology, and pituitary hormone immunohistochemistry. Nonfunctioning tumors exhibiting gonadotroph-cell markers, including alpha-glycoprotein subunit (aGSU), LH or FSH were selected for study. Normal anterior pituitary tissue controls were freshly obtained at surgery.

Animals
Experiments were approved by the Cedars Sinai Institutional Animal Care and Use Committee (protocol # 2683). Mice in a B6C3 genetic background harboring the aGSU-PTTG1-IRES-eGFP (aGSU.PTTG) transgene were previously described [28]. To obtain WT and aGSU.PTTG mice from the same breeding, we crossbred aGSU.PTTG +/2 males and females, and genotyped by PCR.

BrdU incorporation in vivo
Mice were injected i.p. with BrdU (50 mg/g BW, Sigma-Aldrich, St Louis, MO) three times at 3 hour intervals, sacrificed 24 h after the first injection and pituitary sections stained (5-BrdU Labeling and Detection Kit, Roche, Palo Alto, CA). Three randomly chosen visual fields (1000 cells/per field) were counted, and three sections/per animal were derived from three of each genotype analyzed.

SA-b-galactosidase activity
Senescence-associated (SA)-b-galactosidase enzymatic activity was detected in pituitary cryosections (10 mm) using a bgalactosidase staining kit (Senescence Cell Staining Kit, Sigma-Aldrich). Only senescent cells stain at pH 6.0. SA-b-galactosidase activity in vitro was assayed in 6-well plates in triplicate [29]. Three randomly chosen visual fields/per well were identified, and 1000 cells/per field counted.
Human pituitary adenoma PTTG was detected by immunohistochemistry with the same antibodies as for Western blot, and using an avidin-biotin-peroxidase kit (Vector Laboratories, Burlingame, CA).

Cells, constructs, plasmids and transfections
Mouse pituitary gonadotroph LbT2 and aT3 cell lines were generously provided by Dr. Pamela Mellon (UC San Diego). These cells, immortalized with SV40 T-antigen [64], are the only functional gonadotroph cell lines available.
Murine testis mRNA was used as a source for Pttg1. Primers were designed as follows: forward, GGAATTCCATGGC-TACTCTTATCTT, reverse CGGGATCCCCGAATATCTG-CATCGT. The Expand High Fidelity PCR system (Roche Diagnostics) was used for amplification reactions. PCR products were double digested (EcoR1/BamH1), purified and ligated (DNA ligation Kit, Takara Bio, Japan) into pIRES2-ZsGreen1 vector (Clontech, Mountain View, CA) to generate cells that co-express Pttg and a ZsGreen tag.
Mouse clusterin promoter fragment (21855 to +169) was amplified from mouse genomic DNA using TaKaRa LA Taq, inserted into the pGL3-Basic luciferase reporter vector (Promega, San Luis Obispo, CA) and the following primers used for PCR: forward 59 GGG GTA CCA CAT TCC TCC AAG TTT CTG 39, and reverse 59 CGG GAT CCA TGG GCT CTA GTC ACC TC 39.
LbT2 and aT3 cells were stably transfected with mPttg-pIRES2-ZsGreen1 or with mClu-pIRES2-ZsGreen1 to create LbT2mPttg cells or aT3mClu cells respectively, or with pZsGreen1-N1 alone (vector). Cells were grown in the presence of 400 mg/ml geneticin (Invitrogen, Carlsbad, CA). Fourth-and fifth-generation enriched stable LbT2mPttg or aT3mClu or vector cells were used for experiments.
pcDNA3-His-mFoxl2 overexpressing plasmid was a generous gift from Dr. Wei-Hsiung Yang (Mercer University School of Medicine, Savannah, GA) with the permission of Dr. Buffy S. Ellsworth (Southern Illinois University School of Medicine, Carbondale,IL).

Luciferase Assays
LbT2 and aT3 cells were transfected with 800 ng pGL3-luc basic vector or pGL3-luc-mClu reporter plasmid and co-transfected with pcDNA3 or pcDNA3 encoding murine C/EBPb or d in 12-well plates. Twenty four later cells were harvested and monitored for luciferase activity (Promega, Madison, WI). Light emission was evaluated by luminometer, and normalized to a bgalactosidase luciferase reference plasmid.

Chromatin Immunoprecipitation Assay (ChiP)
ChiP was performed (Chip-IT Express kit, Active Motif, Carlsbad, CA) using aT3 pituitary gonadotroph cells which exhibit abundant clusterin expression. Cells were cross-linked with formaldehyde, harvested, sonicated, nuclear fraction isolated, and chromatin immunoprecipitation performed with polyclonal FOXL2 antibody (Abcam) as well as non-specific mouse IgG. DNA released from precipitated complexes was amplified by PCR with 3 pairs of specific primers spanning 1855 nucleotides upstream from the murine clusterin transcription start site. Primer

Apoptosis
The rate of apoptosis was assessed in vitro using the In Situ Cell Death Detection Kit, AP (TUNEL assay, Roche Diagnostics, Indianapolis, IN). Three randomly chosen visual fields, each containing 1000 cells were counted.

Cell proliferation assay
Asynchronized cells were pulsed with 10 mM BrdU (Sigma, St Louis, Missouri) in phosphate buffered saline for 30 min at 37uC. Cells were washed, harvested, fixed in 75% ethanol/50 mM glycin at pH 2.0 and analyzed by FACScan (Becton Dickinson, Mountain View, California). Experiments were performed in triplicate. In separate experiments aT3mClu or aT3vector cells were synchronized in 0.1% FBS for 18 hours, and then stimulated by addition of 10% FBS. At the indicated times, duplicate samples were pulsed with BrdU for 30 min, analyzed by flow cytometry, and S-phase cells identified by staining with BrdU antibodies.

Statistics
Clusterin and p15 protein expression were compared between human GH/PRL-and gonadotroph cell pituitary tumors using Wilcoxon Rank Sum Test. Cellular labeling indices were analyzed using ANOVA followed by non-parametric t-test (Mann-Whitney) or Student t-test. Probability of p,0.05 was considered significant.