Critical Role for STAT3 in Murine Pituitary Adrenocorticotropin Hormone Leukemia Inhibitory Factor Signaling*

Leukemia inhibitory factor (LIF) is a pleiotropic neuroimmune cytokine that promotes corticotroph cell differentiation and induces proopiomelanocortin (POMC) mRNA expression and adrenocorticotropin hormone (ACTH) secretion. However, molecular mechanisms for this induction remain elusive. We therefore developed ACTH-secreting AtT20 transformants for wild-type or mutated STAT3, a cytokine signaling molecule, to address whether STAT3 is a determinant of LIF-mediated ACTH regulation. We show that these mutants act in a dominant negative manner by blocking endogenous STAT3 tyrosine phosphorylation or STAT3 DNA binding. Attenuation of STAT3 activity in the dominant negative AtT20 clones prevented LIF from promoting transcriptional activation of the POMC promoter (2.1-fold), whereas this LIF action was enhanced (7.7-fold; p < 0.05) in wild-type STAT3-overexpressing clones in comparison to mock-transfected cells (4.5-fold). However, wild-type or dominant negative STAT3-overexpressing clones showed comparable (4-fold) POMC induction after treatment with cyclic adenosine monophosphate (cAMP), an alternate inducer of POMC transcription, indicating the STAT3 specificity for LIF signaling. Moreover, dominant negative inactivation of STAT3 activity resulted in abrogation of LIF-induced POMC mRNA levels and ACTH secretion, confirming the in vivo role of STAT3 in LIF-mediated corticotroph action. Chemical or molecular blockade of the mitogen-activated protein kinase pathway did not affect LIF-mediated corticotroph function. These results indicate that STAT3 is a critical intrapituitary component of the LIF-mediated neuroimmunoendocrine interface in corticotroph cells.

Leukemia inhibitory factor (LIF) is a pleiotropic neuroimmune cytokine that promotes corticotroph cell differentiation and induces proopiomelanocortin (POMC) mRNA expression and adrenocorticotropin hormone (ACTH) secretion. However, molecular mechanisms for this induction remain elusive. We therefore developed ACTH-secreting AtT20 transformants for wild-type or mutated STAT3, a cytokine signaling molecule, to address whether STAT3 is a determinant of LIF-mediated ACTH regulation. We show that these mutants act in a dominant negative manner by blocking endogenous STAT3 tyrosine phosphorylation or STAT3 DNA binding. Attenuation of STAT3 activity in the dominant negative AtT20 clones prevented LIF from promoting transcriptional activation of the POMC promoter (2.1-fold), whereas this LIF action was enhanced (7.7-fold; p < 0.05) in wild-type STAT3-overexpressing clones in comparison to mock-transfected cells (4.5-fold). However, wildtype or dominant negative STAT3-overexpressing clones showed comparable (4-fold) POMC induction after treatment with cyclic adenosine monophosphate (cAMP), an alternate inducer of POMC transcription, indicating the STAT3 specificity for LIF signaling. Moreover, dominant negative inactivation of STAT3 activity resulted in abrogation of LIF-induced POMC mRNA levels and ACTH secretion, confirming the in vivo role of STAT3 in LIF-mediated corticotroph action. Chemical or molecular blockade of the mitogen-activated protein kinase pathway did not affect LIF-mediated corticotroph function. These results indicate that STAT3 is a critical intrapituitary component of the LIF-mediated neuroimmunoendocrine interface in corticotroph cells.
The coordinated neuroendocrine stress response comprises both classic endocrine as well as neuroimmune regulatory pathways (1). Pituitary corticotroph cell function is controlled by complex hypothalamic (2) and peripheral paracrine signals (3), all ultimately impacting on proopiomelanocortin (POMC) 1 gene expression and adrenocorticotropin hormone (ACTH) secretion. Several cytokines are both peripheral and central modulators of neuroendocrine function and serve as mediators of the immune neuroendocrine system interactions (1,(3)(4)(5). The pituitary is an abundant source of growth factors that regulate the hypothalamo-pituitary-adrenal axis in vivo and pituicyte ACTH production in vitro (6 -9). Leukemia inhibitory factor (LIF) is a pleiotropic neuroimmune cytokine whose gene expression has been demonstrated in human fetal, adult and murine pituitary cells (10,11). LIF is a component of the interleukin-6 (IL-6) cytokine family, which also includes IL-6, IL-11, ciliary neutrophic factor, oncostatin M, and cardiotrophin-1. Signaling initiated by these cytokines is transduced by activation of composite receptors that share a common transmembrane gp130 protein subunit (12)(13)(14)(15)(16). LIF ligand binding is associated with gp130 and LIF receptor subunit heterodimerization and subsequent activation of downstream tyrosine kinases (17)(18). The cytokine receptors lack intrinsic kinase activity. Subsequent to LIF receptor dimerization, JAK2 is activated by tyrosine autophosphorylation (19). These events lead to tyrosine phosphorylation, auto-or hetero-dimerization, and nuclear translocation of STAT3 (signal transducer and activator of transcription) and, to a lesser extent, STAT1␣, which regulates immediate-response gene transcription (20 -22).
In AtT20 murine corticotroph cells, LIF increases POMC mRNA levels and ACTH secretion, and synergizes with corticotropin releasing hormone (CRH) to trigger these effects (20). Recently, we delineated elements of the POMC promoter important for a component of LIF and CRH synergy on POMC transcription (23). However, identification of a specific LIFinduced transcription factor that activates POMC transcription has been elusive. We have recently shown that overexpression of SOCS-3, a cytokine-inducible signaling inhibitor related to JAB, SSI, or CIS (24,25), inhibited LIF activation of POMC gene expression in vitro (26). As SOCS-3 was also potently induced by LIF, SOCS-3 was implicated as a negative intracellular regulator of pituitary LIF signaling. SOCS-3 overexpression in corticotroph cells abrogated LIF-induced gp130 and STAT3 phosphorylation, suggesting STAT3 as a critical regulator of pituitary LIF function.
Although LIF action occurs mainly through the JAK-STAT pathway, other subcellular signaling pathways are activated by IL-6-related cytokines including phospholipase C, phosphoinositol 3-kinase, phosphotyrosine phosphatase D, pp120, the insulin receptor substrate 1, or several components of the mitogen-activated protein kinases cascade pathway, including SHC, GRB2, Raf-1, ERK1, and ERK2 (27)(28)(29)(30)(31)(32). Activation of these widespread signal transducing molecules implies sub-stantial convergence between cytokine-activated pathways and receptor tyrosine kinase-activated pathways. Nevertheless, JAK proteins appear to be key determinants for this convergence as their disruption results in failure of downstream phosphorylation events (31,33). Furthermore, convergence between both pathways may also occur at the level of STAT proteins. Indeed, activation of Raf-1 by interferon ␥ and oncostatin M requires expression of the STAT1 transcription factor (34) and, reciprocally, ERK2 activity is required for stimulation of interferon ␣and ␤-induced gene expression through STAT proteins (35).
To address molecular mechanisms responsible for LIF induction of the neuroendocrine stress response, we tested the involvement of STAT3 or MAPK in this LIF response. Inactivating STAT3 mutants include a phenylalanine substitution at a critical tyrosine carboxyl-terminal residue, Tyr-705 (STAT3F), and a mutant containing two alanine substitutions at positions important for STAT3 DNA binding (Glu-434 and Glu-435) (36). We therefore isolated stable mutant STAT3 AtT20 transformants and show that they behave as dominant negative inhibitors of endogenous corticotroph cell STAT3. Suppression of STAT3 activity is here shown to abrogate LIF action on POMC transcription and ACTH secretion. However, chemical or molecular blockade of the MAPK pathway did not impact LIFinduced POMC promoter activity. These results thus indicate the key role of STAT3 in the LIF-mediated neuroendocrine pituitary stress response.

MATERIALS AND METHODS
Cell Culture-AtT20 cells, obtained from the American Type Culture Collection (Rockville, MD), were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml streptomycin, 100 units/ml penicillin, and 0.25 g/ml amphotericin B (Life Technology, Inc.).
Immunoprecipitation and Western Blot Analysis-AtT20 cells were grown in 10-ml dishes to 50% confluence and serum-deprived 16 h before treatment with LIF (R&D Systems, Minneapolis, MN). Cells were then washed once with phosphate-buffered saline (pH 7.0), once with lysis buffer (50 mM Hepes, 150 mM NaCl, 10 mM EDTA, 10 mM Na 4 P 2 O 7 , 100 mM NaF, 2 mM sodium orthovanadate, pH 7.4) and lysed in 500 l of lysis buffer containing 1% Triton X-100 and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 2 M leupeptin). After a 10-min incubation at 4°C, the lysate was collected and centrifuged at 13,000 ϫ g for 10 min at 4°C to remove insoluble material. STAT3 proteins were immunoprecipitated by incubating the solubilized cells for 2 h at 4°C with a monoclonal anti-hemagglutinin antibody, 12CA5 (Roche Molecular Biochemicals) prebound to Sepharose-protein G beads (Sigma) or a polyclonal anti-STAT3 antibody (Santa Cruz) prebound to Sepharose A beads (Sigma). Immune complexes were collected by centrifugation, washed twice with a buffer containing 30 mM Hepes, 30 mM NaCl, and 0.1% Triton X-100, pH 7.4, and boiled for 3 min in 40 l of SDS-sample buffer. Samples were separated on 7.5% SDS-polyacrylamide gel and electroblotted on polyvinylidene difluoride membrane. The membrane was blocked for 16 h at 4°C in 20 mM Tris, pH 7.4, 500 mM NaCl, and 0.1% Tween 20 supplemented with 3% bovine serum albumin, blotted for 3 h at room temperature with polyclonal anti-STAT3 or monoclonal anti-phosphotyrosine antibodies (Santa Cruz). Bound antibodies were detected with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG followed by ECL detection (Amersham Pharmacia Biotech).
Transfection and Luciferase Assay-AtT20 cells were plated in 2-ml dishes (100,000 cells/well) and allowed to adhere for 16 h. Cells were transfected as described (20) with 0.5 g per dish of rat POMC promoter (Ϫ706/ϩ64) fused to a luciferase reporter, and 0.5 g of pSV-lacZ (Promega) expressing the ␤-galactosidase protein as an internal transfection control. In the cotransfection experiments using the dominant negative form of MAPK cDNA, various concentrations of the cDNA were also coexpressed. 24 h after transfection, cells were treated in triplicate for 7 h, washed once in phosphate-buffered saline, pH 7, lysed in reporter lysis buffer, and subjected to assay for luciferase (20) or ␤-galactosidase activity (Promega).
Northern Blot Analysis-Total RNA from AtT20 cells transfected with either STAT3wt, F, or D and treated with or without 0.5 nM LIF for 24 h, was prepared using Trizol reagent (Life Technologies, Inc.). 10 g of RNA/lane was size-fractionated under denaturing conditions using formaldehyde-agarose (1.2%) gel, transferred to nitrocellulose, and hybridized with a 32 P-labeled probe POMC probe (2.5 ϫ 10 6 cpm/2 ml hybridization solution) (Stratagene) according to the manufacturer.
ACTH Assay-ACTH was measured by a radioimmunoassay kit purchased from Diagnostic Products Corp. (Los Angeles, CA).

Isolation of Pituicyte Transformants Stably Expressing Dominant Negative Forms of STAT3-
The IL-6 family of cytokines activate the JAK-STAT pathway in several systems, and LIF is a powerful activator of the JAK-STAT pathway, in particular, in AtT20 cells. We therefore isolated pituicyte transformants stably expressing wild-type or dominant negative forms of STAT3 to test whether STAT3 mediates LIF induction of corticotroph function in these cells. Plasmids encoding hemagglutinin peptide (HA)-tagged STAT3F (HA-STAT3F) and HAtagged STAT3D (HA-STAT3D) as well as HA-tagged wild-type STAT3 (HA-STAT3wt) were introduced into AtT20 murine corticotroph pituitary cells. These cells have been shown to be an excellent model for studying the role of cytokine action in the ACTH stress response (20,23). Several different clones selected by G-418 resistance were established; each expressed HA-STAT3F, HA-STAT3D, or HA-STAT3wt. Expression levels of transfected STAT3F or D mutants or STAT3wt, respectively, were determined by immunoprecipitation of HA-STAT3 with anti-HA monoclonal antibody (12CA5) followed by immunoblot analysis with polyclonal anti-STAT3 antibody (Fig. 1). Three independent clones for STAT3wt transformants (clones 1, 2, and 3) and for each dominant negative form of STAT3F (clones 2, 4, and 5) or STAT3D (clones 2, 4, and 6), respectively, were selected. Each clone was shown to express similar levels of HA-STAT3 protein and was used in the ensuing experiments.
To confirm the dominant negative action of HA-STAT3F on endogenous STAT3 function, cell lysates were immunoprecipitated with STAT3 antibody, followed by immunoblotting with anti-phosphotyrosine antibody ( Fig. 2A). This Western blot showed the inability of both endogenous STAT3 and exogenous HA-STAT3F to be phosphorylated after 1 nM LIF treatment for 5 or 10 min in the HA-STAT3F transfected clones. However, in the HA-STAT3wt or HA-STAT3D transfected clones, similar LIF treatments enhanced tyrosine phosphorylation of STAT3, as expected. These results indicated that HA-STAT3F protein behaves in a dominant negative manner by blocking LIF-induced endogenous STAT3 phosphorylation.
To confirm the dominant negative inactivating action of HA-STAT3D, DNA binding was tested by probing with the STAT inducible element (SIEm67) contained within the c-fos pro-moter. This element is readily recognized by either STAT1 or STAT3 (35). After electrophoretic mobility shift assay with extracts derived from mock-transfected AtT20 cells treated for 15 min with 1 nM LIF, three specific complexes were visualized and were effectively competed by 100-fold molar excess of unlabeled SIE probe, but not by an exogenous double-stranded oligonucleotide (Fig. 2B). These three complexes correspond to the previously described SIF (Sis-inducible factor) A (STAT3 homodimer), SIF B (STAT1/STAT3 heterodimer), and SIF C (STAT1 homodimer) (19,21). In the HA-STAT3wt transfected clone, 15 min of 1 nM LIF treatment induced the formation of a single specific complex. STAT3 was shown to be a component of this complex as addition of STAT3 antibody to cell extracts, prior to addition of the SIE probe, inhibited complex formation and induced a faint supershift. However, in the HA-STAT3D transfected clones, this complex was clearly not induced by LIF treatment. This abrogation of DNA binding demonstrated the dominant negative action of HA-STAT3D on endogenous STAT3 action in the HA-STAT3D transfected clones.
Dominant Negative HA-STAT3F and HA-STAT3D Block LIF-induced POMC Promoter Activity-As LIF activates rat POMC expression through the Ϫ706/ϩ64 region of the POMC promoter (20), activity of this promoter was tested in the STAT3 dominant negative clones. Transient transfections of a POMC promoter-driven luciferase reporter, as well as of a vector encoding ␤-galactosidase (internal control), were performed in mock-transfected AtT20, and HA-STAT3wt, F, and D clones, respectively (Fig. 3A). Treatment with 1 nM LIF for 6 h of mock-transfected AtT20 caused a 4.5 Ϯ 0.4-fold (p Ͻ 0.01) luciferase induction. Furthermore, similar LIF treatment of the HA-STAT3wt clones enhanced (p Ͻ 0.05, versus mocktransfected) this induction to 7.7 Ϯ 0.8-fold. However, in the HA-STAT3F and D clones, LIF induction was significantly abrogated (p Ͻ 0.01) as compared with mock-transfected AtT20 or HA-STAT3wt. Indeed, the HA-STAT3F clones only showed 1.81 Ϯ 0.1-fold induction (p Ͻ 0.01), and 2.55 Ϯ 0.2-fold induction (p Ͻ 0.01) was observed in the HA-STAT3D clones. These results indicated the requirement for STAT3 in mediating LIF induction of POMC expression at the transcriptional level.
To confirm that the HA-STAT3F and D clones were functionally viable, and yet failed to respond specifically to LIF because of lack of inducible STAT3, cells were also treated with dibutyryl cAMP, which stimulates POMC expression through an AP-1 like element in the first exon of the POMC gene (37), and whose recognition motif is also present within the POMC promoter fragment (Ϫ706/ϩ64) (Fig. 3B). 5 mM dibutyryl cAMP treatment for 6 h similarly stimulated POMC transcription in both mock-transfected AtT20 cells (4.1 Ϯ 0.2), and in the HA-STAT3wt clones (4.3 Ϯ 0.1) as well as in the mutant HA-STAT3F and D clones (3.9 Ϯ 0.2 and 4 Ϯ 0.2, respectively). This A, AtT20 clones transfected with HA-STAT3wt, HA-STAT3F, or HA-STAT3D were serum-starved for 16 h and treated with 1 nM LIF or vehicle (C) for 5 and 10 min. Cells were lysed and STAT3 proteins immunoprecipitated with polyclonal anti-STAT3 antibody, separated on an SDS-7.5% acrylamide electrophoresis gel, and the tyrosine-phosphorylated form of STAT3 was identified with a monoclonal anti-phosphotyrosine antibody (upper panel). To confirm equal protein loading in each lane, the membrane was stripped and reblotted with a polyclonal anti-STAT3 antibody (lower panel). B and C, gel-shift analysis with whole cell extracts (20 g) derived from mock-transfected AtT20 (B) or HA-STAT3wt or HA-STAT3D clones (C) treated with 1 nM LIF or vehicle for 15 and 30 min. 32 P-Labeled SIE oligonucleotide was used as the probe with AtT20 nuclear extracts (B, lanes 1-5), HA-STAT3wt nuclear extract (C, lanes 1-6), or HA-STAT3D nuclear extract (C, lanes 7-12). 100-fold molar excess of unlabeled SIE or unlabeled AP-1 oligonucleotides were used as competitors (B, lanes 4 and 5, respectively; and C, lanes 4 and 10, and lanes 5 and 11, respectively). Anti-STAT3 antibody was added to the nuclear extracts before addition of SIE probe (C, lanes 6 and 12). result indicated that, although the STAT3 pathway was specifically altered in the HA-STAT3F and D clones, cAMP known to induce POMC expression through alternate STAT3-independent transduction signals was still functionally operative in these cells.
Dominant Negative HA-STAT3F and HA-STAT3D Block LIF Induction of Endogenous POMC mRNA Expression-To test the in vivo significance of the preceding results, POMC Northern blots were also performed on mRNA extracted from HA-STAT3wt, F, or D clones stimulated with 1 nM LIF for 24 h (Fig.  4). As expected, LIF treatment for 24 h in HA-STAT3wt clones enhanced POMC mRNA expression (2.24 Ϯ 0.21-fold induction) (Fig. 4C). This induction was observed within 3 h of treatment (Fig. 4A) and was maximal after 24 h. This time point was also utilized to compare LIF effects in wild-type and dominant negative STAT3 clones (Fig. 4, B and C). In HA-STAT3F and D clones, induction of POMC mRNA expression was decreased, respectively, to 1.42-fold Ϯ 0.07 (p Ͻ 0.01) and to 1.59-fold Ϯ 0.16 (p Ͻ 0.02) versus HA-STAT3wt. These results thus confirmed the in vivo importance of STAT3 in transducing neuroendocrine LIF signaling to endogenous POMC.
Dominant Negative HA-STAT3F and HA-STAT3D Block LIF Induction of ACTH Secretion-The physiologic relevance of STAT3 disruption in the HA-STAT3 clones was also assessed by measuring ACTH levels in conditioned medium after 24 h treatment with 1 nM LIF (Fig. 5). In mock-transfected AtT20 cells as well as in HA-STAT3 wt clones, LIF stimulated ACTH basal levels 2.4 Ϯ 0.2-fold and 2.7 Ϯ 0.1-fold, respectively. However, in HA-STAT3F and D dominant negative clones, this induction was attenuated (respectively, 2 Ϯ 0.08 and 1.9 Ϯ 0.09) (p Ͻ 0.01) in comparison to both mock-transfected AtT20 cells and HA-STAT3wt clones.

Role of MAPK in LIF Induction of POMC Promoter Activity-
The IL-6 cytokine family, which transduces through the gp130 subunit, may also activate Ras-MAPK pathways (27, 29 -32). We therefore tested whether LIF induces MAPK activation in AtT20 cells. Using LIF-activated AtT20 cell extracts in a Western blotting experiment using an anti-tyrosine-phosphorylated form of MAPK antibody, treatment for 10 min with 1 nM LIF was shown to induce MAPK phosphorylation 2-fold (data not shown). To test whether the Ras-MAPK pathway might be a functional component of LIF function in AtT20 cells, we blocked MAPK by either using a chemical inhibitor of mitogenactivated protein kinase kinase (MAPKK), PD 98059, which inhibits MAPK activation and subsequent phosphorylation of MAPK substrates, or, a cDNA encoding a dominant negative form of MAPK (TAYF). Pretreatment of AtT20 cells for 1 h with PD 98059 (5-100 M) did not affect either LIF-induced POMC promoter activity (Table I) or LIF-induced STAT3 binding to the SIE probe (data not shown). Cotransfection of various concentrations of a dominant negative form of MAPK (T183A/ Y185F) with the POMC promoter luciferase construct also did not alter LIF-induced POMC promoter activity (Table II). Thus, MAPK does not appear to be implicated in LIF corticotroph function despite the weak induction of ERK2 tyrosine phosphorylation by LIF in these cells. DISCUSSION We have previously shown that LIF stimulated POMC promoter activity, POMC mRNA expression, and ACTH secretion (20). These results now demonstrate that disruption of STAT3 expression in AtT20 corticotroph cells significantly abrogates these LIF responses. We used two forms of mutated STAT3, which were both expected to behave as signaling inhibitors via different mechanisms (36). Both mutants of STAT3, STAT3F with a phenylalanine substitution at tyrosine 705 and STAT3D mutated at positions important for DNA binding, acted in a dominant negative manner in these cells. Endogenous STAT3 was not tyrosine-phosphorylated in HA-STATF clones, nor did STAT3 bind the SIE probe from the c-Fos promoter in the HA-STAT3D clones. This dominant negative action implies that the mutants compete with endogenous STAT3 protein for recruitment to activated gp130-JAK complex in the HA-STAT3F clones. In the HA-STAT3D clones, endogenous homodimeric STAT3 DNA binding is abrogated. This result is also consistent with previous studies using similar STAT3 mutants and showing STAT3 involvement in IL-6-induced growth arrest and terminal differentiation in M1 cells or in LIF-induced ES cell differentiation (36,38). Furthermore, these HA-STAT3F and D AtT20 transformants were shown, in a separate study, to inhibit LIF-induced SOCS-3 mRNA expression. 2 In the DNA binding assay showing disrupted endogenous STAT3 DNA binding activity in the HA-STAT3D clones, a single complex was consistently observed in the assay using HA-STAT3wt extracts, and was identified as a STAT3 homodimer because addition of STAT3 antibody blocked its formation and also caused a band supershift. This complex was identified as SIF A (Sis-inducible factor) and is composed of a STAT3 homodimer, as distinct from SIF B and C composed, 2 C. J. Auernhammer, C. Bousquet, and S. Melmed, submitted for publication. Luciferase activity was measured as described under "Materials and Methods." The results are representative of three separate experiments using three independent clones, and are expressed as -fold induction above control.
respectively, of STAT1 and 3 heterodimer or of STAT1 homodimer (40). SIF B and C were not observed in the electrophoretic mobility shift assay assay using HA-STAT3wt extracts, whereas both these complexes were evident using mocktransfected AtT20 cell extracts. This may be explained by the strong overexpression of HA-STAT3wt competing with endog-enous STAT1 for SIE binding. Furthermore, SIF C (STAT1 homodimer), which was evident in AtT20 extracts, was not regulated by LIF treatment. This observation also points to the major role of STAT3 for LIF signaling in these cells as compared with other STAT family members.
Using the rat POMC promoter (Ϫ706/ϩ64), we previously FIG. 4. Dominant negative HA-STAT3F and HA-STAT3D block LIF induction of endogenous POMC mRNA transcription. A, 5 g of total mRNA derived from mock-transfected AtT20 cells treated for 1, 3, 6, or 24 h with 1 nM LIF or vehicle were loaded per lane and size-fractionated on a denaturing formaldehyde-agarose (1.2%) gel. After transfer to nitrocellulose membrane, hybridization was performed with radiolabeled POMC cDNA. Equal loading and transfer were verified by reprobing the washed membrane with radiolabeled ␤-actin cDNA. B, 5 g of total mRNA derived from HA-STAT3wt, HA-STAT3F, or HA-STAT3D clones treated for 24 h with 1 nM LIF or vehicle (C) were loaded per lane, size-fractionated on a denaturing formaldehyde-agarose (1.2%) gel, transferred, and probed with radiolabeled POMC cDNA. Equal loading and transfer were verified by reprobing the washed membrane with radiolabeled ␤-actin cDNA. C, scanning quantification of bands corresponding to control (no treatment) and 24-h LIF treatment of HA-STAT3wt, HA-STAT3F, and HA-STAT3D clones. These results are representative of five separate experiments using three independent clones, and are expressed as -fold induction above control.
FIG. 5. Dominant negative HA-STAT3F and HA-STAT3D block LIF induction of ACTH secretion. AtT20 cells, HA-STAT3wt, HA-STAT3F, or HA-STAT3D clones were serum-starved for 48 h before being treated with 1 nM LIF for a subsequent 24 h. Conditioned medium ACTH was measured by radioimmunoassay. These results are representative of three separate experiments using three independent clones, and are expressed as -fold induction above control.
showed that LIF stimulated basal POMC expression and potently enhanced CRH action on rat POMC transcription (20). A common POMC promoter element, which does not contain a STAT-binding site, also mediated a component of the synergy between both ACTH-inducing peptides. Moreover, disruption of this element (Ϫ173/Ϫ160), within the full-length POMC promoter (Ϫ706/ϩ64), only partially blocked LIF and CRH-mediated effects (23). Thus, additional mechanisms may also be involved in this synergistic induction of POMC expression. The effects of STAT3 disruption in AtT20 cells also support this hypothesis, as LIF stimulation of POMC-promoter activity was reduced from 4.5-fold in mock-transfected or 7.7-fold in HA-STAT3wt clones to 1.8-or 2.5-fold in HA-STATF or D clones, respectively. Furthermore, overexpression of STAT3wt in HA-STAT3wt clones induced LIF action in comparison with mocktransfected AtT20 cells (7.7-fold versus 4.5-fold, respectively; p Ͻ 0.05). This observation also further confirmed the STAT3specificity of LIF action, as was the demonstration that, in the STAT3 dominant negative clones, POMC transcription was briskly induced via an alternate pathway. CRH-like receptor binding ligands induce cAMP and activate CREB protein, which also induces POMC transcription (37,(41)(42). We therefore stimulated the STAT3 dominant negative clones, as well as HA-STAT3wt and mock-transfected AtT20 cells with dibutyryl cAMP to increase steady state levels of cAMP, and showed that POMC was effectively induced in all clones.
The abrogation of LIF effects by STAT3 disruption differed quantitatively on POMC promoter activity, and on POMC mRNA expression or ACTH secretion; LIF effects on endogenous POMC expression and ACTH secretion were not as marked as those observed in the POMC promoter studies. This could be explained by the sensitivity of the luciferase assay for studying POMC promoter activity, especially in the mocktransfected AtT20 and HA-STAT3wt cells, than the in vivo assays. We utilized a 770-base pair fragment of the POMC promoter to study LIF action on luciferase reporter expression. However, the POMC promoter containing potential LIF-regulated elements may be longer in vivo, and thus may require transcription factors other than STAT3 (20).
Disrupting STAT3 signal transduction by either blocking tyrosine phosphorylation (HA-STAT3F clones) or preventing DNA binding (HA-STAT3D) resulted in decreasing LIF effects. These results also imply that the DNA-binding event of STAT3 is indispensable for LIF action, and thus render it unlikely that STAT3 tyrosine phosphorylation serves as the sole link between the JAK-STAT and known or unknown pathways (e.g. CREB) involved in POMC transcriptional regulation.
Furthermore, we examined whether other LIF-induced pathways might be implicated in LIF corticotroph function. LIF weakly induced MAPK phosphorylation (2-fold) (data not shown). The magnitude of LIF activation of MAPK differs in various cell lines but is generally low, and less efficient than oncostatin M (29). The MAPK pathway might be a component of LIF signaling for several reasons. Our results show that LIF weakly induces MAPK phosphorylation in AtT20 cells; JAK2 has been shown to activate both STAT and Ras-dependent or -independent MAPK pathways (33), and STAT3 contains a serine phosphorylation site corresponding to a consensus site for MAPK serine phosphorylation activity (35). Nevertheless, neither coexpression of a dominant negative form of MAPK nor pretreatment with a specific inhibitor of MAPKK abrogated POMC promoter activity. We therefore conclude that, even if LIF weakly stimulates MAPK phosphorylation, it is not a primary mechanism by which LIF activates corticotroph function.
STAT3 involvement in LIF promotion of the stress response in AtT20 cells was not previously considered likely, mainly because no apparent STAT3 element has been identified in the POMC promoter. The Ϫ173/Ϫ160 element of POMC promoter shown to mediate LIF and CRH transcriptional synergy does not contain an evident STAT3 binding motif. Nevertheless, we now show that STAT3 disruption significantly blocks LIF transcriptional activation of POMC and ACTH secretion. Direct interactions between STAT3 and POMC promoter remain, however, to be elucidated. STAT3 might interact directly with the POMC promoter by binding to a non-classical STAT binding site. In HepG2 cells, STAT3 and a CRE-like transcription factor interaction were stimulated by interleukin-6 and shown to bind to a JRE-IL6 response element, which differed from a classical high affinity STAT binding site but which contained a low affinity STAT-binding site overlapping with an Ets-like and a CRE-like site (43). Interactions between STAT3 and CRE-like binding proteins on the POMC promoter will be important to explore. Indeed, CRH stimulates POMC transcriptional activation by activating a CRE binding protein (37). Furthermore, interaction between STAT and other transcription factors have been reported elsewhere (44,45).
The dominant negative forms STAT3 transfectants are useful for studying molecular mechanisms by which LIF specifically determines corticotroph cell activation in the hypothalamo-pituitary-adrenal axis. The use of a pituitary cell line lacking endogenous STAT3 was elusive mainly because the POMC gene is specifically and abundantly expressed in the pituitary corticotroph cell, which contains tissue-specific POMC promoter elements (39). Thus, STAT3 appears to be a critical mediator for the LIF-mediated neuroimmunoendocrine interface in corticotroph cells.