Cushing’s syndrome is caused by excessive adrenocorticotropic hormone (ACTH) secretion derived from pituitary corticotroph tumors (Cushing disease) or from non-pituitary tumors (ectopic Cushing’s syndrome). Hypercortisolemic features of ectopic Cushing’s syndrome are severe, and no definitive treatment for paraneoplastic ACTH excess is available. We aimed to identify subcellular therapeutic targets by elucidating transcriptional regulation of the human ACTH precursor POMC (proopiomelanocortin) and ACTH production in non-pituitary tumor cells and in cell lines derived from patients with ectopic Cushing’s syndrome. We show that ectopic hPOMC transcription proceeds independently of pituitary-specific Tpit/Pitx1 and demonstrate a novel E2F1-mediated transcriptional mechanism regulating hPOMC. We identify an E2F1 cluster binding to the proximal hPOMC promoter region (−42 to +68), with DNA-binding activity determined by the phosphorylation at Ser-337. hPOMC mRNA expression in cancer cells was upregulated (up to 40-fold) by the co-expression of E2F1 and its heterodimer partner DP1. Direct and indirect inhibitors of E2F1 activity suppressed hPOMC gene expression and ACTH by modifying E2F1 DNA-binding activity in ectopic Cushing’s cell lines and primary tumor cells, and also suppressed paraneoplastic ACTH and cortisol levels in xenografted mice. E2F1-mediated hPOMC transcription is a potential target for suppressing ACTH production in ectopic Cushing’s syndrome.
ACTH-dependent Cushing’s syndrome and consequent hypercortisolemia leads to critical metabolic and cardiovascular complications (Chrousos 2009, Nieman et al. 2015, Raff & Carroll 2015). The high ACTH levels that induce adrenal cortisol secretion may derive from pituitary corticotroph adenomas, i.e., Cushing disease, or, less commonly, from non-pituitary tumors, i.e., ectopic or paraneoplastic production (Brown 1928). Ectopic ACTH-secreting tumors can arise from a variety of sites, such as breast, lung (Ilias et al. 2005), colon, pancreas, thyroid, ovary and thymus. Features of hypercortisolemia are usually florid and rapidly progressive, with concomitant severe hypokalemia and higher morbidity and mortality than those observed with pituitary-derived Cushing disease. The source of excess ACTH production may be occult, and pose a challenge to identify, localize and treat. As many such tumors are already metastatic at the time of diagnosis, surgical resection is not usually an option, and failure to control ACTH levels contributes to high morbidity and mortality. Current medical therapies, which are frequently associated with significant side effects and are often poorly tolerated, address the peripheral effects of excess ACTH by blocking adrenal steroidogenesis (Preda et al. 2012, Newell-Price 2014) or by antagonizing cortisol actions (Fleseriu et al. 2012, 2015), but do not target the ACTH source directly (Colao et al. 2012).
Subcellular mechanisms for ectopic Cushing’s syndrome are poorly defined. ACTH produced by post-translational cleavage from the proopiomelanocortin (POMC) precursor is abundantly expressed in pituitary corticotroph cells and in the hypothalamus (Stevens & White 2010). The rodent pituitary Pomc promoter (rPomc), mostly elucidated in AtT20 mouse corticotroph adenoma cells, is regulated by Tpit/Pitx1 (Lamolet et al. 2001, Liu et al. 2001), NeuroD1 (Poulin et al. 1997), nuclear receptors (Maira et al. 1999), TR4 (Du et al. 2013), STAT3 (Bousquet et al. 2000), and cytokines and hormones such as LIF (Bousquet et al. 1999) and CRH (Karalis et al. 2004). However, regulation of the human POMC (hPOMC) gene is not well understood. In addition, whether paraneoplastic hPOMC in ectopic Cushing’s tumors is regulated by the same factors identified in pituitary cells is not known, and pathways controlling POMC/ACTH expression in non-pituitary Cushing’s tumor cells are not well defined (Picon et al. 1995, Ray et al. 1996, Newell-Price 2003). Moreover, investigation of hPOMC gene regulation has been hampered due to the lack of a functional human pituitary cell line.
With a view to identifying a molecular target for controlling ACTH production in patients with ectopic Cushing’s syndrome, we analyzed hPOMC expression using primary tumor cell cultures and human cell lines derived from non-pituitary ACTH-secreting tumors. Here, we show that ectopic hPOMC regulation differs from pituitary regulation and report an E2F1-mediated mechanism for hPOMC transcription. We also show that E2F1 antagonists markedly suppress hPOMC expression and paraneoplastic ACTH secretion in human-derived tumor primary cultures. Our results indicate that E2F1-mediated hPOMC signaling is a potential target for suppressing ACTH production in ectopic Cushing’s syndrome.
Materials and methods
Cell culture and transfection
After obtaining informed consent, ectopic/pituitary ACTH-secreting surgical specimens were processed using Tissue Dissociate Kits (Miltenyi Biotec, Bergisch, Gladbach, Germany) and cultured in low-glucose DMEM 10% FBS for 48 h. Then, the cell viability was monitored by trypan blue staining and culture medium was collected for RIA. DMS79 cells (human small-cell lung carcinoma) and COLO320 cells (human carcinoid-like colorectal carcinoma) were cultured in RPMI-1640 with l-glutamine and 10% FBS. Human samples are described in Supplement 6, see section on supplementary data given at the end of this article. E2F1 siRNA (107660, Thermo Fisher) transfections were performed using a 4D-Nucelofector (Lonza, Basel, Switzerland). Human ACTH and corticosterone levels were analyzed by RIA (MP Biomedicals, Santa Ana, CA, USA).
DMS79 cells (1.5 × 106 cells) were injected subcutaneously to 24 five-week-old male Nu/J mice (Jackson Laboratory). Seven days later, mice were treated by oral gavage with R-roscovitine (200 mg/kg, n = 12) or vehicle (0.5% methylcellulose, 0.5% Tween 80/PBS; 100 μL, n = 12) for 17 days. Animals were then killed, tumors were measured and weighed and blood was collected for ACTH and corticosterone RIA. Pre- and post-treatment ACTH and corticosterone levels are depicted in Supplement 3.
POMC, PTH and GH mRNA levels were analyzed by RT-PCR. Total RNA was isolated using RNeasy mini kit (Qiagen), and cDNA was prepared using a QuantiTect Reverse Transcription Kit with random oligo primers (Applied Biosystems). Quantitative RT-PCR was performed by TaqMan Gene Expression Assays (hPOMC probe; Hs01596743_m1, hGAPDH probe; Hs02758991_g1, hPTH probe; Hs00757710_g1, hGH probe; Hs00792200_g1, Taqman Gene Expression Master mix; 439016, Applied Biosystems). POMC RNA levels were normalized by GAPDH. Each figure depicting RT-PCR results is compiled from at least three independent experiments, and each sample was analyzed in triplicate or quadruplicate.
Immunoblotting was performed using anti-ACTH (ab20358; Abcam), anti-E2F1 (sc-251; Santa Cruz Biotechnology), anti-phospho-Serine 364-E2F1 (ab5391; Abcam), anti-phospho-Serine337-E2F1 (ab135549; Abcam), Lamin A/C (sc-20681; Santa Cruz Biotechnology) and GAPDH (sc-25778; Santa Cruz Biotechnology). Ab binding was detected using Mini PROTEAN electrophoresis protocol (BioRad).
After drug treatment, cells were collected and stained with annexin V and 7-AAD using FITC Annexin V detection kit with 7-AAD (Biolegend, San Diego, CA, USA). Cell viability was analyzed by flow cytomtery using a CyAn flow cytometer (Beckman Coulter).
Mapping of hPOMC RNA start sites
The human POMC transcription start site was determined by 5′ RACE (Frohman et al. 1988) in DMS79 and COLO320 cells, with minor modifications (5). Briefly, cDNA was prepared using an oligo primer (CAGTCAGCTCCCTCTTGAACTCCA) binding to POMC RNA. The G-tail was added to the 5′-end of the resulting cDNA, and 5′-ends were PCR amplified using a poly(C) primer and an antisense POMC primer (ACTCCAGGGGGAAGGCCTCGGCCGA). To concentrate the hPOMC cDNA fragments, a second PCR was performed using the antisense POMC primer (TTGCCCTCGCGCGGGCCCGGCT) and the poly(C) primer. Resulting PCR products were cloned and 5′-ends were determined by DNA sequencing. To avoid mis-mapping due to PCR artifacts, three independent PCR products were analyzed.
Construction of promoters and luciferase assays
The hPOMC promoter (−1200 to +83) was PCR amplified from genomic DNA (Jurkat cells) and cloned. Wild type (Wt) and 5′ deletion mutants were amplified using 5′-primers binding to −1140, −428, −303, −102 and −42, (containing Sal I site at the 5′-end) and 3′-primer binding to +68 (containing Hind III sites at the 3′-end). Resulting amplified fragments were digested with Sal I and Hind III and cloned into Xho I and Hind III sites in pGL3 basic vector. Point mutations in the −428/+68 sequence (Fig. 1C) were introduced by PCR assembly, and mutated fragments were integrated into Xho I and Hind III sites in pGL3 basic vector. DNA sequences of all the inserted fragments were determined to remove defective fragments generated by PCR errors. For minimal promoters, −42 to +68 probes were divided into seven small fragments (Fig. 2B). Sense and antisense oligonucleotides encoding these sequences were synthesized, duplicated and ligated upstream of the minimal promoter in pGL3 as described (Ogawa et al. 2014). Luciferase reporter plasmids containing five direct-repeat inserts were selected by DNA sequencing (Fig. 2B). DNA sequences of inserted fragments were determined to remove the defective fragments generated by PCR errors.
Luciferase reporter assays were performed in 2.5 × 105 DMS79 cells with 0.7 μg luciferase reporter plasmids, and 50 ng pRL-CMV used as an internal control plasmid. Cells were transfected using lipofectamine 2000 (Invitrogen) and cultured in 0.5 mL medium in 24-well plates. Forty-eight hours after transfection, cells were harvested and luciferase activity analyzed by Dual-Luciferase Reporter Assay System (Promega). Luciferase assays were repeated more than three times, and luciferase activities normalized by internal renilla activity. For co-transfection experiments, 0.5 µg luciferase reporter plasmids were co-transfected with 1.5 µg E2F1, E2F3 and/or DP1 expression plasmids constructed using pMF vector carrying the EF-1α promoter. Total DNA (2.0 µg) was adjusted with the empty vector.
Nuclear extracts were prepared as described (Tone et al. 2002). 5 μg nuclear extracts were incubated in 14 μL pre-incubation buffer (5 mM Tris (pH 8.0) and 14% glycerol) for 20 min; 0.5 μg poly (dI-dC), 1.2 μL of 5× EMSA standard buffer (Tris–HCl 4 mM, 12 mM HEPES (pH 7.9), 60 mM KCl, 0.2 mM EDTA, 2 mM DTT and 12% glycerol) and 32P-labeled probes were added for a further 15 min. For competition assays, 100-fold excess unlabeled competitor primer was added to EMSA reaction mixtures. For super-shift assays, nuclear extracts in EMSA reaction buffer were incubated for 15 min with anti-E2F1 (sc-251X; Santa Cruz Biotechnology), anti-E2F2 (sc-633X; Santa Cruz Biotechnology), anti-E2F3 (sc-878X; Santa Cruz Biotechnology), anti-E2F4 (sc-866X; Santa Cruz Biotechnology), anti-Sp1 (07-645; Millipore) or anti-Sp3 (sc-644X; Santa Cruz Biotechnology), and then the probes were added. DNA–protein complexes were analyzed by 4% DNA retardation gel. Probe and competitor sequences (sense strand) used were # 1: CCACCAGGAGAGCTCGGCAAGTATATAAGG, # 2: GTAT ATAAGGACAGAGGAGCGCGGGACCAA, # 3: GACCAA GCGGCGGCGAAGGAGGGGAAG, # 4: AA GGAGGGGAA GAAGAGCCGCGACCGAG, #5: AAGAGCCGCGACCGA GAGAGGCCGCCGAGC, # 6: AGAGGCCGCCGAGCGTC CCCGCCCTCA and # 7: GTCCCCGCCCTCAGAG AGCAGCCTCCC.
EMSA was also performed using CIP-treated nuclear extract from DMS79 cells as described (Tone et al. 2002). Briefly, 5 µg nuclear extract were treated with 0.6 units CIP (New England BioLabs, Ipswich, MA, USA) in EMSA buffer at room temperature for 1 h, and the reaction was terminated by adding phosphatase inhibitor cocktail (Sigma-Aldrich). To prepare control nuclear extracts, phosphatase inhibitor cocktail was added before the addition of CIP and incubated for 1 h.
Chromatin immunoprecipitation assay
ChIP assay was performed using DMS79 and COLO320 cells as described (Tone et al. 2008). Cells were fixed in 1% formaldehyde, 4.5 mM HEPES pH 8.0, 9 mM NaCl, 0.09 mM EDTA and 0.045 mM EGTA for 10 min at room temperature and sonicated (Bioruptor, Liege, Belgium) in lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris–HCl pH 8.0) with proteinase inhibitor (Sigma-Aldrich P8340). Pre-cleared lysates were incubated overnight at 4°C with anti-E2F1 (sc-251X; Santa Cruz Biotechnology), anti-acetyl histone H4 (06-866; Millipore) and normal rabbit IgG (sc-2027; Santa Cruz Biotechnology) polyclonal antibody. DNA fragments were isolated from immunoprecipitated chromatin and analyzed by RT-PCR analysis with SYBR Green PCR Master Mix (Applied Biosystems). PCR primers used were hPOMC 5′-CCCCTCCTACCCTTGCTGTA, 5′-GGGCACAGCTCCCATAATCA, 5′-CAAAGAGGGCTTC AGAGAACA, 5′-CACTTCAGCATCCACAACATTC, 5′-GAG AGCTCGGCAAGTATAAAG and 5′-TCGGCGCAGAAA GTTTG.
DNA pull-down assay
DNA pull-down assay was performed as described (Ogawa et al. 2014). A 74 bp DNA fragment containing the E2F1-binding site (−24/+50) (Fig. 2B) was amplified by PCR using biotinylated primers. 30 µL nuclear extract (100 µg) were added to 180 µL of 5 mM Tris (pH 8.0)–14% glycerol buffer and pre-incubated on ice. 15 µL poly(dI-dC)(dI-dC) (7.5 µg) and 18 µL of 5× binding buffer (300 mM KCl, 60 mM HEPES (pH 7.9), 20 mM Tris–HCl (pH 8), 0.5 mM EDTA, 25% glycerol and 5 mM DTT) and 6 µL probe (1 µg) were incubated at room temperature for 15 min. DNA–protein complexes were collected with 10 µL Dynabeads M-280 Streptavidin magnetic beads (Life Technologies) and analyzed by immunoblotting.
Differences between vehicle and R-roscovitine groups were analyzed using ANOVA followed by nonparametric t test (Mann–Whitney) or Student t test. Probability of P < 0.05 was considered significant. In bronchial carcinoid tumors, the correlation of POMC and E2F1 expression were assessed by Spearman’s correlation.
Protocols were approved by the Cedars-Sinai Institutional Review Board (protocol #30753). Male and female patients aged more than 18 years with a clinical diagnosis of ectopic or pituitary ACTH-producing tumors and lung carcinoid tumors were screened, and informed consent was obtained. Human samples (age, gender and diagnosis) are described in Supplement 6. Animal protocol (protocol #2603) was approved by the Cedars-Sinai Institutional Animal Care and Use Committee.
hPOMC promoter activity in non-pituitary cells
We used transfection-based luciferase assays to analyze hPOMC promoter activity and investigate transcriptional regulation in non-pituitary human cells. As primary tumor cells are not suitable for these experiments, we used human small-cell lung cancer DMS79 cells that are widely used in studies of ectopic Cushing’s syndrome (Pettengill et al. 1980, White et al. 1989, Picon et al. 1999, Newell-Price et al. 2006). We also found that ACTH-positive human colon carcinoid-like cancer cells COLO320 (Quinn et al. 1979) express POMC mRNA. Therefore, we analyzed the expression of hPOMC mRNA in these two cell lines by RT-PCR. Compared with normal lung BEAS-2B cells (Lechner et al. 1983) and normal colon hNCC cells (Chesnokova et al. 2013), DMS79 and COLO320 cells expressed hPOMC mRNA at high and intermediate levels, respectively (Fig. 1A).
Previously, we showed that POMC gene expression in pituitary cells is suppressed by the cyclin-dependent kinase/cyclin E inhibitor R-roscovitine (Liu et al. 2011, 2015) We therefore analyzed hPOMC gene expression in human tumor cells derived from five patients with ectopic Cushing’s syndrome and in DMS79 and COLO320 cells. Similar to what we found in pituitary cells, hPOMC mRNA expression was suppressed with R-roscovitine treatment (Fig. 1B, C, D), suggesting that hPOMC gene expression is, at least in part, regulated by mechanisms similar to those in human ectopic primary cells. We therefore used these cell lines for subsequent hPOMC gene analysis.
We next considered whether the transcription factors Tpit, Pitx1, and NeuroD1 regulating POMC expression in pituitary corticotrophs are similarly involved in ectopic Cushing’s syndrome. We found the mRNA expression of all three transcription factors in patient-derived ectopic Cushing’s tumors arising from the lung, thymus and liver as well as in DMS79 cells (Fig. 1E), although Pitx1 mRNA was undetectable in the liver tumor cells. We therefore analyzed whether these transcription factors regulate hPOMC promoter activity in non-pituitary cells, as they do in pituitary cells.
To map the hPOMC promoter region, we determined the major transcription start site in non-pituitary tumor cells by 5′-RACE (rapid amplifications of cDNA ends) (Frohman et al. 1988). This start site, which we defined as position +1, is 4-bp downstream from the site previously defined by primer extension (Takahashi et al. 1983) (Fig. 2A). A TATA-like sequence was identified 32-bp upstream of the start site, similar to the previously identified rat promoter (Drouin et al. 1985). We then analyzed hPOMC promoter activity in non-pituitary tumor cells by performing luciferase reporter assays with point mutations within the Tpit/Pitx1 and NeuroD1 sites (Fig. 1F). These sequences identified in the rPomc promoter are conserved in the hPOMC promoter (Supplement 1). When similar point mutations were introduced in the rPomc promoter in AtT20 pituitary cells, promoter activity was attenuated (Fig. 1F), but mutated Tpit/Pitx1 and NeuroD1 sites did not reduce hPOMC promoter activity in ectopic cell lines (Fig. 1G). Next, we performed hPOMC promoter analysis using a series of 5′-deletion mutants. Previously, the hPOMC promoter was analyzed up to −428 bp (Picon et al. 1995); we now extended the analysis to −1160/+68 (Fig. 2A). Similar promoter activity was observed with all deletion mutants. Seventy percent of maximum promoter activity was maintained in the shortest −42/+68 promoter, suggesting this is a key proximal regulatory region.
Taken together, these results suggest that hPOMC promoter activity is not controlled by the same pituitary-specific factors that regulate promoter activity in Cushing’s disease. Rather, it is mainly controlled by a proximal regulatory region located between −42/+68 bp in non-pituitary cells.
As this short proximal promoter comprises only a 42-bp sequence upstream of the transcription start site, including the TATA (−31 to −25 bp), we could not generate further 5′-deletion mutants. To identify the binding of transcription factors, we constructed a luciferase reporter system with a minimal promoter containing the TATA sequence and transcription start site (Tone et al. 2008) (Fig. 2B). This method was used previously to identify other transcription factors (Tone et al. 2000, 2008, Ogawa et al. 2014). The putative regulatory region (−52 (−42 plus additional 10-bp) to +76 (+68 plus additional 8-bp)) was systematically divided into seven fragments with partially overlapping sequences (Fig. 2B). These short (24–30 bp) DNA fragments (#1 to #7) were integrated upstream of the minimal promoter, and five sequence copies were added as direct repeats (Fig. 2B). When the resulting luciferase reporter plasmids were transfected into DMS79 cells (Fig. 2C), moderate-to-strong luciferase activity was detected with fragments #5 to #7, and mild increments were detected with fragments #1 to #4. We then performed electrophoretic mobility shift assays (EMSA) with fragments #5 to #7 (Fig. 2D). Several bands were detected; the lower bands were consistently observed with #5 and #6 fragments, and the three upper bands (the top thick band consists of two bands) were detected in #6 and #7 fragments. Overlapping regions within #6 and #7 include a CCCCGCCC sequence similar to the Sp1 consensus, and mobility patterns resemble those identified for Sp1/Sp3 binding (Tone et al. 2002). Mutating this sequence resulted in dissipation of the top three bands (Fig. 2D). Sp1/Sp3 binding was also confirmed by super-shift EMSA using anti-Sp1 and anti-Sp3 (Fig. 2E). The lower single band detected with #5 and #6 regions (Fig. 2D, E, arrow) was also seen with the #6 Sp1 mutant probe. We therefore reasoned that a common factor, different from SP1/Sp3, binds this overlapping #5 and #6 sequence region (AGAGAGGCCGCCGAG).
hPOMC expression is regulated by E2F1-mediated transcriptional mechanisms
We introduced a mutation (GCCGCC to AATATT) in the middle of the overlapping #5 and #6 sequence, and, using EMSA, showed partially attenuated binding (about 50%) (Fig. 2F), suggesting a nuclear factor binding to GCCGCC and also indicating the likely presence of another binding site for this probe. Indeed, we observed a similar GCCGC sequence near the 5′-end of the probe (Fig. 2F), with binding reduced to half by mutating GAGCCGC to TAATATT. Using EMSA with additional mutant probes (Fig. 2F), we observed that binding was abolished with M5 and M6 probes, suggesting the presence of two core binding motifs: CGCCGAG in the 3′-site and GAGCCGC (mirror image to the 3′-site core sequence) in the 5′ region of the probe (Fig. 2F). Functions of these two core binding sequences were further confirmed by luciferase reporter assays, which showed that mutating these sequences led to reduced promoter activity of probe #5 (Fig. 2G).
Although no potential transcription factor binding to the identified sequences (GAGCCGC and CGCCGAG) was suggested using databases, we noted that these sequences were similar to the 3′-half of the E2F transcription family consensus motif, i.e., AGTGCCGC (Tao et al. 1997). Using EMSA competition assays with a labeled #5 probe and an unlabeled E2F consensus sequence as competitor, transcription factor binding was abolished by both the #5 probe as well as by E2F consensus competitors, but not by unrelated p53 consensus competitors (Fig. 3A left), suggesting that E2F member(s) bind the #5 probe. Using super-shift EMSA assays with antibodies to E2Fs (E2F1 to E2F4), we detected a super-shifted band with anti-E2F1 (Fig. 3B), but not with other antibodies. E2F1-mediated hPOMC promoter activity was also confirmed by luciferase reporter assay using the #5, #6 and #7 promoter plasmids with co-transfection of either E2F1 or E2F3 expression plasmid. Both promoters #5 and #6 were upregulated by co-expression of E2F1 but not by E2F3, whereas promoter #7 (as a negative control with no E2F1 binding site) was not regulated by either E2F1 or E2F3 co-expression (Fig. 3C).
E2F1 binding to the proximal promoter region was further analyzed (Fig. 2B). Similar sequences of the 3′-half of the E2F consensus (CGGCAA in #1, GCGGGA in #2, GGCGAA in #3, and GAGCCGC in #4) were identified in the −42/+68 hPOMC region, and minimal promoter activities were also moderately upregulated in these fragments (Fig. 2C). We therefore performed EMSA competition assays using the labeled probe #5 with unlabeled competitor (#1 to #4) (Fig. 3A right), and observed that E2F1 binding to probe #5 was blocked by #1 to #4 competitors, suggesting that the E2F1 cluster binds the −42/+68 region.
E2F1 binding to the proximal hPOMC in vivo was confirmed by ChIP assay using DMS79 and COLO320 cells (Fig. 3D, E, F), and histone H4 was highly acetylated in this region, but not in upstream control regions (−1.2 kb or −6.7 kb) identified as an rPomc Tpit palindrome enhancer (Langlais et al. 2011) (Fig. 3E, F).
E2F1 determines hPOMC expression in non-pituitary ACTH-producing tumor cells
To define the functional role of E2F1, we performed luciferase assays and RT-PCR after co-expressing E2F1 and its heterodimer DP1 (Girling et al. 1993). hPOMC promoter activity (−42/+68) and POMC mRNA levels in COLO320 cells were markedly upregulated by co-expression of E2F1 and DP1, but parathyroid hormone, which served as a control gene, was not upregulated (Fig. 4A, B). In DMS79 cells, hPOMC promoter activity and mRNA expression were also upregulated by E2F1 and DP1 co-expression (Fig. 4C, D). Gene-specific POMC knockdown was performed by using Si-RNA E2F1, whereas the control gene (PTH) was unchanged, further suggesting a specific contribution of E2F1 to POMC gene regulation (Fig. 4E).
We next confirmed the role of E2F1 in the regulation of hPOMC gene expression using 12 human bronchial carcinoid tumor samples, known to exhibit a propensity for ectopic ACTH production (Aniszewski et al. 2001, Isidori et al. 2006). hPOMC expression was detected in 9 of 12 tumors and, importantly, E2F1 and POMC expression levels correlated strongly (P = 0.025) (Fig. 4F).
Taken together, the results suggest that hPOMC expression levels are determined by E2F1, which mediates transcription through the −42/+68 proximal promoter.
E2F1 binding to hPOMC proximal promoter decreased by R-roscovitine treatment
Given the role of E2F1 in regulating hPOMC expression, we considered whether E2F1 could serve as a target in ectopic Cushing’s tumors. As shown in Fig. 1B, C, D, we used R-roscovitine, known to suppress mPomc in pituitary corticotrophs (Liu et al. 2015), to elucidate the mechanism. Cell viability assays were used to determine the doses for in vitro experiments (Supplement 2). We confirmed the suppression of paraneoplastic hPOMC mRNA and ACTH production in surgically excised ectopic ACTH-producing samples derived from lung, liver and thymus tumors (n = 5) treated with R-roscovitine (Fig. 1B, C). Dose-dependent hPOMC expression was seen in DMS79 and COLO320 cells treated with R-roscovitine for 48 h (Fig. 1D). We also confirmed the in vivo efficacy of this drug in DMS79 cell xenografted mice, showing decreased plasma ACTH and corticosterone levels (Supplement 3).
Next, using minimal promoters and the proximal −42/+68 hPOMC promoter, we showed that R-roscovitine suppressed −42/+68 hPOMC promoter activity dose dependently (Fig. 5A). Marked inhibition of promoter activity by R-roscovitine was observed, especially with fragments #5 and #6 (E2F1 sites), whereas promoter activity was unaltered with fragment # 7 (Sp1/Sp3 site) (Fig. 5B), confirming that R-roscovitine acts on E2F1-binding sites, not via SP1/SP3. To elucidate a mechanism for the observed inhibition, we performed EMSA using the same minimal promoter fragments with or without R-roscovitine treatment. E2F1 binding to probe #1 to #5 (E2F1 cluster) (Fig. 3A) was dissipated by R-roscovitine (Fig. 5C). Reduction of in vivo E2F1 binding to this region was also observed using ChIP assays (Fig. 5D) in DMS79 and COLO320 cells, whereas highly acetylated histone H4 was unaltered by R-roscovitine, suggesting that E2F1 determines hPOMC transcriptional levels without affecting chromatin remodeling. Taken together, these results provide evidence that E2F1 is an appropriate molecular target for suppressing hPOMC.
hPOMC transcription is regulated by E2F1 Ser-337 phosphorylation
As E2F1 DNA binding was modulated by R-roscovitine, we assessed whether E2F1 phosphorylation status affects E2F1 DNA binding. EMSA with nuclear extracts dephosphorylated by CIP (Tone et al. 2002) reduced E2F1 DNA binding to the hPOMC promoter after dephosphorylation (Fig. 6A). To test the site specificity of E2F1 dephosphorylation, we performed a series of immunoblotting experiments in DMS79 cells. Total nuclear E2F1 and phosphorylated Ser-364 E2F1 (pSer364-E2F1) were not altered by R-roscovitine treatment. However, R-roscovitine reduced the nuclear phosphorylated Ser-337 E2F1 (pSer337-E2F1) (Fig. 6B). We further assessed the role of phosphorylated Ser337-E2F1 to DNA binding activity by pull-down assay using excess nuclear extract to saturate the hPOMC probe (−24/+50) (Fig. 6C). Although pSer337-E2F1 input was markedly reduced by R-roscovitine, DNA-bound pSer337-E2F1 was still clearly detected (Fig. 6C, Supplement 4), suggesting highly selective binding of pSer337-E2F1 to the proximal hPOMC region. Point-mutated E2F1 (Serine 337 > Alanine) also decreased hPOMC activity by luciferase assays (Fig. 6D).
Suppression of POMC expression by targeting E2F-binding inhibitor HLM006474
As HLM006474, a direct inhibitor of the E2F family, interrupts DNA binding in melanoma cells (Ma et al. 2008), and also decreases cell proliferation in DMS79 cells (Kurtyka et al. 2014), we investigated the effects of this small molecule on hPOMC. After testing cell viability to determine the appropriate dose for in vitro experiments (Supplement 2), similar to our results with R-roscovitine, we found that HLM006474 strongly suppressed POMC mRNA levels in a dose-dependent manner in DMS79 cells analyzed by RT-PCR (Fig. 7A). mRNA expression of PTH, used as a negative control, was unaltered (Fig. 7B), suggesting that E2F1-mediated antagonists elicit gene-specific hPOMC actions. POMC reduction was also confirmed by immunoblotting, and total E2F1 and pSer337-E2F1 were also both dose-dependently reduced (Fig. 7C). Finally, HLM006474 also suppressed hPOMC mRNA expression in three ACTH-secreting tumor primary cultures derived from liver and lung (Fig. 7D), whereas the expression of PTH remained unaltered (Supplement 5).
Ectopic Cushing’s syndrome is as yet an incurable disease. We elucidated mechanisms underlying hPOMC transcription and ACTH production in non-pituitary tumor cells.
Unlike the rat pituitary Pomc promoter, paraneoplastic hPOMC expression does not seem to be regulated by major pituitary-specific transcription factors (Lamonerie et al. 1996, Poulin et al. 1997, Lamolet et al. 2001, Liu et al. 2001). Rather, we show that hPOMC expression in non-pituitary tumors is regulated by E2F1 and Sp1/Sp3 binding to the proximal hPOMC promoter region (−42 to +68). Since these binding sites are closely located, these two transcription factors may mutually interact (Lin et al. 1996). As hPOMC expression is regulated by E2F1 expression levels and activity, E2F1-mediated POMC transcription is a potential target for ectopic Cushing’s tumors.
In earlier reports, an E2F1 binding site at position −376/−417 in the POMC promoter had been identified by luciferase assays, showing a 1.2-fold to 1.5-fold decrease in activity with deletion of the site (Picon et al. 1995, 1999). We found a similar mild reduction of luciferase activity by deletion of the promoter from −303 to −428 (Fig. 2A). However, we also detected much stronger activity (70% of all activity) at our newly identified proximal E2F1 cluster binding site (Fig. 2A), which was not analyzed in previous studies. We therefore focused on that proximal promoter site as being a major site affecting hPOMC expression in ectopic Cushing’s tumors, rather than the weaker upstream E2F family binding site.
We previously found that R-roscovitine targets pituitary-specific Tpit/Pitx1-mediated rat Pomc transcription and suppresses Tpit mRNA expression in human pituitary corticotroph tumors (Liu et al. 2015). Pituitary-specific rat Pomc gene expression is also regulated by Tpit/Pitx1 and NeuroD1 (Poulin et al. 1997), binding at −316/−297 and −377/−370 (Poulin et al. 1997). Yet, despite these sequences being conserved in the human POMC promoter (located at −375/−369 and −287/−281), our current data indicate that hPOMC transcription is not altered by deletions or by point mutations of these potential sites in two ectopic tumor cell lines. Therefore, hPOMC promoter activity in ectopic Cushing’s tumors seems not to be similarly regulated as pituitary corticotrophs. Alternatively, hPOMC promoter activity in human pituitary corticotrophs may differ from rodent pituitary expression. Indeed, clustered E2F1-binding sequences in the human promoter region are only partially conserved in the rodent Pomc promoter, even though E2F1 binds rodent Pomc in AtT20 cells (Liu et al. 2015). Epigenetic modifications may also contribute to the species differences and tissue-specific regulation of POMC. For example, CpG sites are more abundant in the hPOMC compared with those in the rPomc promoter, and differences in tissue-specific methylation status were also reported for hPOMC (Newell-Price et al. 2001). However, further elucidation of tissue specificity (i.e., pituitary vs non-pituitary) and species specificity (i.e., human vs rodent) of POMC regulation will be challenging to distinguish because of the lack of functional human pituitary corticotroph cell lines.
We show the functional phenotypes of E2F1 on hPOMC using both over-expression and knock-down studies in DMS79 and in COLO320 cells, and also show a strong positive correlation between E2F1 and POMC expression in human bronchial carcinoid tumors, confirming a role for E2F1 on POMC expression in non-pituitary cells and supporting our in vitro results in human tumors using primary cell cultures. These results, together with our ChIP results showing that E2F1 controls hPOMC transcriptional levels without altering AcH4 levels (chromatin remodeling), further support the rationale for E2F1 as a potential molecular target.
Our observation of ACTH suppression by R-roscovitine in both patient tumors and in mouse models elucidates a novel mechanism that E2F1 DNA binding and POMC expression are regulated by pSer337-E2F1. E2F1 Ser-337 phosphorylation may also directly modify E2F1 structure, including the DNA binding domain, or it may prevent Rb/E2F1 binding (Fagan et al. 1994), and therefore increase E2F1 DNA binding (Robertson et al. 2000). As DMS79 and COLO320 cells also express several Rb family members (e.g., p130 and p107) (Helin et al. 1997), these may also be associated with E2F1 Ser-337 phosphorylation status in ectopic Cushing’s tumors.
Controlling E2F1 activity may be a promising therapeutic strategy for ectopic Cushing’s syndrome, because expression levels of POMC seem to be determined by E2F1 activity in POMC-positive tumor cells. However, E2F1 is a ubiquitously expressed transcription factor, and inhibition of E2F1 binding to the −42/+68 promoter region did not affect the acetylation of histone H4 in this region (Fig. 5D), suggesting E2F1 does not regulate open chromatin status in the proximal promoter region. Induction of POMC may be cell type dependent and regulated by other transcription factors, including STAT3 (Bousquet et al. 2000), through chromatin remodeling and epigenetic modifications such as DNA methylation (Newell-Price et al. 2001). Thus, although we cannot exclude the involvement of other transcription factors, E2F1 activity seems to be a common mechanism controlling POMC expression, and E2F1 over-activity may be integral to the induction of ectopic Cushing’s syndrome. Indeed, E2F1 is highly expressed in small-cell lung cancer (Eymin et al. 2001), which may explain the relatively high occurrence of ectopic Cushing’s syndrome associated with these tumors, and may also underlie suppression of cell proliferation seen with the E2F inhibitor HLM006474 in small-cell lung cancer (Kurtyka et al. 2014).
POMC expression strongly correlates with E2F1 in a variety of POMC-expressing cells. The direct E2F inhibitor HLM006474 robustly suppressed POMC/ACTH levels in DMS79 cells, and also reduced total E2F1 and pSer337-E2F1, suggesting that both E2F1 expression and phosphorylation status are important to hPOMC gene regulation. We recognize that our study used a limited number of human samples to support mechanistic findings derived from POMC-expressing cell lines. Ectopic Cushing’s syndrome is rare and up to 50% are occult (Ilias et al. 2005); therefore, it is often treated based on signs and symptoms rather than definitive localized pathological diagnosis, making tissue samples scarce. It is also likely that such tumors are under-diagnosed, as screening for ACTH oversecretion is typically not performed until fulminant signs of disease develop, further limiting the availability of tissue samples. Confirmation for the role of E2F1 in regulating POMC expression and ACTH secretion in paraneoplastic tumors may best be demonstrated clinically by targeting kinases involved in E2F1 regulation. Our results represent a unique mechanism for E2F1 action on paraneoplastic hPOMC/ACTH hormonal production. Elucidation of subcellular mechanisms for ectopic Cushing’s syndrome and our finding of E2F1 DNA binding as a potential target for excess hPOMC expression provide new insights into this challenging disorder.
This is linked to the online version of the paper at http://dx.doi.org/10.1530/ERC-16-0206.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Financial support was provided by NIH grants DK103198 and T32DK007770 and by the Doris Factor Molecular Endocrinology Laboratory. The funding sources had no role in study design, data collection and analysis, decision to publish, or in the preparation of the manuscript.
Author contribution statement
T A, M T, N L and S M designed the research studies; T A, Y T, D C and R H conducted experiments; T A, N L, M T and S M analyzed the data; and T A, N L, M T and S M wrote the manuscript.
The authors thank Drs Odelia Cooper, Run Yu and Harmik Soukiasian for patient samples, and Norika Chiba, Masato Tsuda, Anat Ben-Shlomo and Shira Berman for technical advice and assistance. Financial support was provided by NIH grants DK103198 and DK007770 to S M and by the Doris Factor Molecular Endocrinology Laboratory.
AniszewskiJPYoungWFJrThompsonGBGrantCSvan HeerdenJA2001Cushing syndrome due to ectopic adrenocorticotropic hormone secretion. World Journal of Surgery25934–940. (doi:10.1007/s00268-001-0032-5)
BousquetCSusiniCMelmedS1999Inhibitory roles for SHP-1 and SOCS-3 following pituitary proopiomelanocortin induction by leukemia inhibitory factor. Journal of Clinical Investigation1041277–1285. (doi:10.1172/JCI7924)
BousquetCZatelliMCMelmedS2000Direct regulation of pituitary proopiomelanocortin by STAT3 provides a novel mechanism for immuno-neuroendocrine interfacing. Journal of Clinical Investigation1061417–1425. (doi:10.1172/JCI11182)
ChesnokovaVZhouCBen-ShlomoAZonisSTaniYRenSGMelmedS2013Growth hormone is a cellular senescence target in pituitary and nonpituitary cells. PNAS110E3331–E3339. (doi:10.1073/pnas.1310589110)
ColaoAPetersennSNewell-PriceJFindlingJWGuFMaldonadoMSchoenherrUMillsDSalgadoLRBillerBM2012A 12-month phase 3 study of pasireotide in Cushing’s disease. New England Journal of Medicine366914–924. (doi:10.1056/NEJMoa1105743)
DrouinJChamberlandMCharronJJeannotteLNemerM1985Structure of the rat pro-opiomelanocortin (POMC) gene. FEBS Letters19354–58. (doi:10.1016/0014-5793(85)80078-8)
DuLBergsneiderMMirsadraeiLYoungSHJonkerJWDownesMYongWHEvansRMHeaneyAP2013Evidence for orphan nuclear receptor TR4 in the etiology of Cushing disease. PNAS1108555–8560. (doi:10.1073/pnas.1306182110)
EyminBGazzeriSBrambillaCBrambillaE2001Distinct pattern of E2F1 expression in human lung tumours: E2F1 is upregulated in small cell lung carcinoma. Oncogene201678–1687. (doi:10.1038/sj.onc.1204242)
FaganRFlintKJJonesN1994Phosphorylation of E2F-1 modulates its interaction with the retinoblastoma gene product and the adenoviral E4 19 kDa protein. Cell78799–811. (doi:10.1016/S0092-8674(94)90522-3)
FleseriuMBillerBMFindlingJWMolitchMESchteingartDEGrossC2012Mifepristone, a glucocorticoid receptor antagonist, produces clinical and metabolic benefits in patients with Cushing’s syndrome. Journal of Clinical Endocrinology and Metabolism972039–2049. (doi:10.1210/jc.2011-3350)
FleseriuMPivonelloRYoungJHamrahianAHMolitchMEShimizuCTanakaTShimatsuAWhiteTHilliardA2015Osilodrostat, a potent oral 11beta-hydroxylase inhibitor: 22-week, prospective, Phase II study in Cushing’s disease. Pituitary19138–148. (doi:10.1007/s11102-015-0692-z)
FrohmanMADushMKMartinGR1988Rapid production of full-length cDNAs from rare transcripts: amplification using a single gene-specific oligonucleotide primer. PNAS858998–9002. (doi:10.1073/pnas.85.23.8998)
GirlingRPartridgeJFBandaraLRBurdenNTottyNFHsuanJJLa ThangueNB1993A new component of the transcription factor DRTF1/E2F. Nature36283–87. (doi:10.1038/362083a0)
HelinKHolmKNiebuhrAEibergHTommerupNHougaardSPoulsenHSSpang-ThomsenMNorgaardP1997Loss of the retinoblastoma protein-related p130 protein in small cell lung carcinoma. PNAS946933–6938. (doi:10.1073/pnas.94.13.6933)
IliasITorpyDJPacakKMullenNWesleyRANiemanLK2005Cushing’s syndrome due to ectopic corticotropin secretion: twenty years’ experience at the National Institutes of Health. Journal of Clinical Endocrinology and Metabolism904955–4962. (doi:10.1210/jc.2004-2527)
IsidoriAMKaltsasGAPozzaCFrajeseVNewell-PriceJReznekRHJenkinsPJMonsonJPGrossmanABBesserGM2006The ectopic adrenocorticotropin syndrome: clinical features, diagnosis, management, and long-term follow-up. Journal of Clinical Endocrinology and Metabolism91371–377. (doi:10.1210/jc.2005-1542)
KaralisKPVenihakiMZhaoJvan VlerkenLEChandrasC2004NF-kappaB participates in the corticotropin-releasing, hormone-induced regulation of the pituitary proopiomelanocortin gene. Journal of Biological Chemistry27910837–10840. (doi:10.1074/jbc.M313063200)
KurtykaCAChenLCressWD2014E2F inhibition synergizes with paclitaxel in lung cancer cell lines. PLoS ONE9e96357. (doi:10.1371/journal.pone.0096357)
LamoletBPulichinoAMLamonerieTGauthierYBrueTEnjalbertADrouinJ2001A pituitary cell-restricted T box factor, Tpit, activates POMC transcription in cooperation with Pitx homeoproteins. Cell104849–859. (doi:10.1016/S0092-8674(01)00282-3)
LamonerieTTremblayJJLanctotCTherrienMGauthierYDrouinJ1996Ptx1, a bicoid-related homeo box transcription factor involved in transcription of the pro-opiomelanocortin gene. Genes and Development101284–1295. (doi:10.1101/gad.10.10.1284)
LanglaisDCoutureCSylvain-DroletGDrouinJ2011A pituitary-specific enhancer of the POMC gene with preferential activity in corticotrope cells. Molecular Endocrinology25348–359. (doi:10.1210/me.2010-0422)
LechnerJFMcClendonIALaVeckMAShamsuddinAMHarrisCC1983Differential control by platelet factors of squamous differentiation in normal and malignant human bronchial epithelial cells. Cancer Research435915–5921.
LinSYBlackARKosticDPajovicSHooverCNAzizkhanJC1996Cell cycle-regulated association of E2F1 and Sp1 is related to their functional interaction. Molecular and Cellular Biology161668–1675. (doi:10.1128/MCB.16.4.1668)
LiuJLinCGleibermanAOhgiKAHermanTHuangHPTsaiMJRosenfeldMG2001Tbx19, a tissue-selective regulator of POMC gene expression. PNAS988674–8679. (doi:10.1073/pnas.141234898)
LiuNAArakiTCuevas-RamosDHongJBen-ShlomoAToneYToneMMelmedS2015Cyclin E – mediated human proopiomelanocortin regulation as a therapeutic target for Cushing disease. Journal of Clinical Endocrinology and Metabolism1002557–2564. (doi:10.1210/jc.2015-1606)
LiuNAJiangHBen-ShlomoAWawrowskyKFanXMLinSMelmedS2011Targeting zebrafish and murine pituitary corticotroph tumors with a cyclin-dependent kinase (CDK) inhibitor. PNAS1088414–8419. (doi:10.1073/pnas.1018091108)
MaYKurtykaCABoyapalleSSungSSLawrenceHGuidaWCressWD2008A small-molecule E2F inhibitor blocks growth in a melanoma culture model. Cancer Research686292–6299. (doi:10.1158/0008-5472.CAN-08-0121)
MairaMMartensCPhilipsADrouinJ1999Heterodimerization between members of the Nur subfamily of orphan nuclear receptors as a novel mechanism for gene activation. Molecular and Cellular Biology197549–7557. (doi:10.1128/MCB.19.11.7549)
Newell-PriceJ2003Proopiomelanocortin gene expression and DNA methylation: implications for Cushing’s syndrome and beyond. Journal of Endocrinology177365–372. (doi:10.1677/joe.0.1770365)
Newell-PriceJ2014Ketoconazole as an adrenal steroidogenesis inhibitor: effectiveness and risks in the treatment of Cushing’s disease. Journal of Clinical Endocrinology and Metabolism991586–1588. (doi:10.1210/jc.2014-1622)
Newell-PriceJKingPClarkAJ2001The CpG island promoter of the human proopiomelanocortin gene is methylated in nonexpressing normal tissue and tumors and represses expression. Molecular Endocrinology15338–348. (doi:10.1210/mend.15.2.0599)
NiemanLKBillerBMFindlingJWMuradMHNewell-PriceJSavageMOTabarinA2015Treatment of Cushing’s syndrome: an endocrine society clinical practice guideline. Journal of Clinical Endocrinology and Metabolism1002807–2831. (doi:10.1210/jc.2015-1818)
OgawaCToneYTsudaMPeterCWaldmannHToneM2014TGF-beta-mediated Foxp3 gene expression is cooperatively regulated by Stat5, Creb, and AP-1 through CNS2. Journal of Immunology192475–483. (doi:10.4049/jimmunol.1301892)
PettengillOSSorensonGDWurster-HillDHCurpheyTJNollWWCateCCMaurerLH1980Isolation and growth characteristics of continuous cell lines from small-cell carcinoma of the lung. Cancer45906–918. (doi:10.1002/1097-0142(19800301)45:5<906::AID-CNCR2820450513>3.0.CO;2-H)
PiconABertagnaXde KeyzerY1999Analysis of the human proopiomelanocortin gene promoter in a small cell lung carcinoma cell line reveals an unusual role for E2F transcription factors. Oncogene182627–2633. (doi:10.1038/sj.onc.1202635)
PiconALeblond-FrancillardMRaffin-SansonMLLenneFBertagnaXde KeyzerY1995Functional analysis of the human pro-opiomelanocortin promoter in the small cell lung carcinoma cell line DMS-79. Journal of Molecular Endocrinology15187–194. (doi:10.1677/jme.0.0150187)
PoulinGTurgeonBDrouinJ1997NeuroD1/beta2 contributes to cell-specific transcription of the proopiomelanocortin gene. Molecular and Cellular Biology176673–6682. (doi:10.1128/MCB.17.11.6673)
PredaVASenJKaravitakiNGrossmanAB2012Etomidate in the management of hypercortisolaemia in Cushing’s syndrome: a review. European Journal of Endocrinology167137–143. (doi:10.1530/EJE-12-0274)
QuinnLAMooreGEMorganRTWoodsLK1979Cell lines from human colon carcinoma with unusual cell products, double minutes, and homogeneously staining regions. Cancer Research394914–4924.
RaffHCarrollT2015Cushing’s syndrome: from physiological principles to diagnosis and clinical care. Journal of Physiology593493–506. (doi:10.1113/jphysiol.2014.282871)
RayDWDavisJRWhiteAClarkAJ1996Glucocorticoid receptor structure and function in glucocorticoid-resistant small cell lung carcinoma cells. Cancer Research563276–3280.
RobertsonKDAit-Si-AliSYokochiTWadePAJonesPLWolffeAP2000DNMT1 forms a complex with Rb, E2F1 and HDAC1 and represses transcription from E2F-responsive promoters. Nature Genetics25338–342. (doi:10.1038/77124)
StevensAWhiteA2010ACTH: cellular peptide hormone synthesis and secretory pathways. Results and Problems in Cell Differentiation5063–84. (doi:10.1007/400_2009_30)
TakahashiHHakamataYWatanabeYKikunoRMiyataTNumaS1983Complete nucleotide sequence of the human corticotropin-beta-lipotropin precursor gene. Nucleic Acids Research116847–6858. (doi:10.1093/nar/11.19.6847)
TaoYKassatlyRFCressWDHorowitzJM1997Subunit composition determines E2F DNA-binding site specificity. Molecular and Cellular Biology176994–7007. (doi:10.1128/MCB.17.12.6994)
ToneMPowellMJToneYThompsonSAWaldmannH2000IL-10 gene expression is controlled by the transcription factors Sp1 and Sp3. Journal of Immunology165286–291. (doi:10.4049/jimmunol.165.1.286)
ToneMToneYBabikJMLinCYWaldmannH2002The role of Sp1 and NF-kappa B in regulating CD40 gene expression. Journal of Biological Chemistry2778890–8897. (doi:10.1074/jbc.M109889200)
ToneYFuruuchiKKojimaYTykocinskiMLGreeneMIToneM2008Smad3 and NFAT cooperate to induce Foxp3 expression through its enhancer. Nature Immunology9194–202. (doi:10.1038/ni1549)
WhiteAStewartMFFarrellWECrosbySRLavenderPMTwentymanPRReesLHClarkAJ1989Pro-opiomelanocortin gene expression and peptide secretion in human small-cell lung cancer cell lines. Journal of Molecular Endocrinology365–70. (doi:10.1677/jme.0.0030065)