Activation of Signal Transducer and Activator of Transcription 1 (STAT1) Is Not Sufficient for the Induction of STAT1-dependent Genes in Endothelial Cells

We compared human endothelial cell (EC) responses to interferon(IFN ) and oncostatin M (OnM), cytokines that utilize Janus kinase/signal transducer and activator of transcription (JAK/STAT) signaling. Both cytokines cause phosphorylation of Tyr residue 701 and Ser residue 727 of STAT1, as shown by immunoblotting. Both activate DNA binding of STAT1 homodimers, shown by electrophoretic mobility shift assay. However, only IFN increases expression of three STAT1-dependent gene products examined, namely transporter associated with antigen processing-1 (TAP1), interferon regulatory factor-1 (IRF1), and class I major histocompatibility complex (MHC) protein, as demonstrated by immunoblotting. Only IFN increases TAP1 transcription assessed by reporter gene assay. OnM pretreatment or co-treatment does not inhibit IFN responses. Interestingly, IFN activation of STAT1 is considerably more long-lived than that produced by OnM. To determine whether duration is functionally significant, we transduced EC with a chimeric receptor containing extracellular domains of platelet-derived growth factor receptor and intracellular regions of gp130, the signaling subunit of the OnM receptor, mutated to prevent binding of the tyrosine phosphatase SHP-2. Addition of platelet-derived growth factor to such transduced cells produces STAT1 activation that is comparable in magnitude and duration to that caused by IFN , but still fails to induce TAP1, IRF1, or class I MHC molecules. OnM also activates STAT1 but not transcription of STAT1-dependent genes in HepG2 cells. Transient transfection of HepG2 cells with a STAT-defective mouse IFN receptor failed to complement the OnM STAT signal. We conclude that STAT1 activation is necessary but not sufficient for induction of transcription of IFN -responsive genes. However, signals provided by IFN other than STAT1 activation cannot be provided in trans to complement the response to OnM.

Vascular endothelial cells (EC) 1 are the principal cellular targets of many pro-and anti-inflammatory cytokines. They are unusually sensitive to activation by interferon-␥ (IFN␥), showing IFN␥-dependent expression of class I and class II major histocompatibility complex (MHC) molecules under basal conditions, when most other cell types are unresponsive (1)(2)(3). A previous study from our laboratory suggested that this sensitivity may be attributed to the fact that EC are relatively deficient in expression of Src homology-containing phosphatase-1 (SHP-1) which normally dampens signaling by IFN␥ (4). The predominant signaling pathway activated by IFN␥ involves Janus kinases (JAKs), namely JAK1 and JAK2, and signal transducer and activator of transcription-1 (STAT1) (5)(6)(7)(8). Binding of IFN␥ to IFN␥ receptor-1 (IFNGR1) in the presence of IFN␥ receptor-2 (IFNGR2) results in receptor clustering followed by rapid phosphorylation of the intracellular region of the IFNGR1 at Tyr residue 440 by receptor-associated JAKs (7). Cytoplasmic STAT1 proteins then bind to this phosphorylated Tyr residue in IFNGR1 and subsequently become phosphorylated themselves at Tyr residue 701 by receptor-associated JAKs. The tyrosine-phosphorylated STAT1 proteins dissociate from the receptor and form a homodimer which translocates to the nucleus and binds to specific ␥-activated sequence (GAS) elements (TTCNNNGAA consensus sequence) in the enhancers of IFN␥-inducible genes, stimulating their transcription (9,10). The ability of STAT1 to activate gene transcription may additionally depend on phosphorylation of Ser residue 727, possibly through the actions of a mitogenactivated protein kinase (MAPK) family member (11)(12)(13).
Many cell types, including EC, respond to IFN␥ by rapidly increasing the expression of TAP1 and IRF1 (14,15). Both responses are STAT1-dependent and involve binding of STAT1 homodimers to GAS elements in the 5Ј-flanking regions of these genes (4,16,17). TAP proteins contribute to the assembly and peptide loading of nascent class I MHC molecules (18,19), while IRF1 mediates the delayed transcription of class I MHC molecules as well as sustained transcription of TAP proteins (15,16,20). The coordinated up-regulation of these proteins by IFN␥ enhances the capacity of these cells to present foreign peptides bound to class I MHC molecules to cytolytic CD8 ϩ T cells (21). In this way, IFN␥ contributes to immune-mediated host defense. Whereas IFN␥ is considered to be pro-inflammatory, interleukin-6 (IL-6)-type cytokines are often immunoregulatory or cytoprotective, reducing inflammation and host injury (22)(23)(24). The members of this cytokine family which include IL-6, IL-11, leukemia-inhibitory factor, ciliary neurotropic factor, cardiotrophin-1, and oncostatin M (OnM), utilize a common cytokine receptor signaling subunit, called gp130, which is expressed on the surface of target cells (25). Typically, IL-6type cytokines associate with a cytokine-binding ␣ chain that, upon ligand binding, heterodimerizes with gp130 and promotes receptor clustering. Receptor-associated JAKs (JAK1, JAK2, and TYK2) then phosphorylate gp130 on several intracellular tyrosine residues, providing docking sites for adapters and transcription factors. Four tyrosine residues in the intracellular region of gp130 (Tyr-767, Tyr-814, Tyr-905, and Tyr-915) contribute to activation of STAT3 and two of these tyrosine residues (Tyr-905 and Tyr-915) also contribute to the activation of STAT1 (26,27). Phosphorylation of Tyr residue 759 leads to recruitment and activation tyrosine phosphatase SHP-2 (28 -32). SHP-2 both down-regulates JAK-STAT signaling through its tyrosine phosphatase activity and acts as a gp130-associated adapter protein, binding Grb2 and initiating the activation of a MAPK pathway independently of its tyrosine phosphatase activity (29,33). The mutation of Tyr residue 759 to Phe in gp130 prevents activation of MAPK (27,29,33). It also causes prolonged STAT3 and STAT1 activation and enhances the induction of STAT3-dependent genes such as ␣ 2 -macroglobulin (␣2M) in human hepatoma HepG2 cells (29 -32, 34).
We have previously described the effects of IL-11 on human umbilical vein EC (HUVEC) (35,36). In these experiments, we used OnM as a positive control for activation of STAT1 and STAT3. The levels of STAT1 phosphorylation on Tyr residue 701 induced by treatment with OnM were striking, as assessed by immunoblotting, yet there was no evidence of increased expression of class I MHC molecules. 2 This observation led us to examine more carefully whether OnM can increase the expression of IFN␥-inducible STAT1-dependent proteins in HUVEC. We find that although OnM is a potent activator of STAT1, it cannot activate STAT1-dependent genes. These findings suggest that IFN␥ must have some other, as yet unidentified, activities that are necessary for STAT1-dependent gene expression.

EXPERIMENTAL PROCEDURES
Reagents and Cells-Recombinant human OnM was purchased from R&D Systems (Minneapolis, MN). Recombinant human IFN␥ (0.32 ϫ 10 7 units/mg), recombinant mouse IFN␥, and recombinant human platelet-derived growth factor BB (PDGFBB) were purchased from Upstate Biotechnology (Lake Placid, NY). Fibroblast growth factor-1, commonly called endothelial cell growth supplement, was obtained from Collaborative Research/Becton Dickinson (Bedford, MA) and used in conjunction with porcine intestinal heparin (Sigma). Polybrene, puromycin, poly(dI-dC), and mouse monoclonal antibody (Ab) to ␤-actin were also obtained from Sigma. Rabbit polyclonal Abs to STAT1, to phosphotyrosine STAT1, to phosphotyrosine STAT3, and to phosphothreonine/phosphotyrosine-p42/p44 MAPK were purchased from Cell Signaling Technology, Inc. ( Endothelial cells were isolated from discarded umbilical veins as previously described (37,38), pooled, and cultured on gelatin (J. T. Baker, Phillipsburg, NJ)-coated tissue culture plastic at 37°C in 5% CO 2 -humidified air in Medium 199 containing 20% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin (all from Invitrogen, Grand Island, NY), 50 g/ml fibroblast-growth factor-1, and 100 g/ml porcine intestinal heparin. Confluent cultures were serially passaged and cells were typically used at the second or third subculture. The Phoenix-Ampho packaging cell line for production of high titer amphotropic retroviruses was obtained from Dr. G. Nolan (Stanford University, Stanford, CA) and cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. HepG2 cells were kindly provided by Dr. R. Wells (Yale University School of Medicine, New Haven, CT) and grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Immunoblotting-Confluent HUVEC cultures were washed twice with ice-cold phosphate-buffered saline (PBS) containing 1 mM sodium orthovanadate and 1 mM sodium fluoride and lysed with ice-cold RIPA lysis buffer (PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM pefabloc, 1 g/ml leupeptin, 1 g/ml aprotinin, 1 mg/ml benzamidine, 1 mM sodium orthovanadate, 1 mM sodium fluoride). Cell lysates were clarified by centrifugation at 10,000 ϫ g for 15 min and protein concentrations of supernatant were determined by using a Bio-Rad assay kit (Bio-Rad, Hercules, CA). Lysates were prepared for SDS-PAGE by adding an equal volume of 2 ϫ SDS-PAGE sample buffer (100 mM Tris-Cl, pH 6.8, 200 mM dithiothreitol, 4% SDS, 0.2% bromphenol blue, 20% glycerol) and heating the mixture in a boiling water bath for 3 min. 20 g of protein were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Immobilon P, Millipore, Bedford, MA) by electrophoresis. After blocking with Trisbuffered saline Tween (10 mM Tris-HCl, pH 8.0, 0.150 mM NaCl, 0.05% Tween 20) containing 5% nonfat milk for 1 h at room temperature, the membranes were incubated with blocking solution containing the indicated Ab overnight at 4°C. Membranes were washed and incubated with a horseradish peroxidase-conjugated detecting reagent specific for the primary Ab (Jackson ImmunoResearch, West Grove, PA) and horseradish peroxidase activity was detected using an enhanced chemiluminescence kit according to the manufacturer's instructions (Pierce, Rockford, IL). Exposed films were scanned using a laser densitometer (Fast Scan, Series 300, Molecular Dynamics, Sunnyvale, CA). Nuclear extracts were isolated (see below) and prepared for SDS-PAGE by adding an equal volume of 2 ϫ SDS-PAGE sample buffer. The immunoblotting of nuclear extracts was conducted as described above for total cell lysates.
Plasmids and Transient Transfection-TAP1-growth hormone (TAP1-GH) promoter-reporter gene construct, which contains the 5Јflanking sequence of the TAP1 gene fused to the human GH was kindly provided by Dr. D. R. Johnson (Yale University School of Medicine, New Haven, CT) (16). pGL3␣2M-215-luciferase (␣2M-luciferase) promoterreporter gene construct, which contains the promoter region Ϫ215 to ϩ8 of rat ␣2M gene fused to the firefly luciferase encoding sequence, was kindly provided by Dr. F. Schaper (Institute fü r Biochemie, Aachen, Germany) (28). A constitutively active renilla luciferase expression construct was provided by Dr. S. Ghosh (Yale University School of Medicine, New Haven, CT). Mouse IFNGR2, wild type mouse IFNGR1 (IFNGR1(Y440)), and mutated mouse IFNGR1 (IFNGR1(Y440F)), in which Tyr residue 440 was replaced with Phe to prevent STAT1 binding, were kindly provided by Dr. R. Schreiber (Washington University School of Medicine, St. Louis, MO).
Transient transfection of HUVEC was performed using a DEAEdextran protocol as described previously (39). HepG2 cells were transiently transfected using Targefect F-2 and peptide enhancer reagents according to the manufacturer's instructions (Targeting Systems, Santee, CA).
Promoter Reporter Gene Assays-To assess the induction of TAP1-GH promoter-reporter gene construct, culture medium was assayed for GH using a solid phase sandwich radioimmunoassay as described by the manufacturer (Nichols Institute, San Juan Capistrano, CA). Radioactivity was measured in a ␥-counter (20/20 series, Iso-Data Inc., Palatine, IL). Cell lysates were assayed for Renilla luciferase activity using a Promega reporter assay system (Promega, Madison, WI) and measured using a Berthold model LB9501 luminometer (Schwarzwald, Germany). GH values in ng/ml were normalized to luciferase values in relative light units (RLU) to control for transfection efficiency. To assess the induction of ␣2M-luciferase promoter-reporter gene construct, cell lysates were assayed for Renilla and Firefly luciferase activities using a dual Promega reporter assay system (Promega) 2 K. Mahboubi and J. S. Pober, unpublished observations. and measured as described above. Firefly luciferase activity in RLU were normalized to Renilla luciferase values in RLU to control for transfection efficiency.
Nuclear Extractions and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared following a modification of the procedure of Dignam et al. (40). Briefly, cells (3 ϫ 10 6 ) were harvested by scraping into PBS, resuspended into 200 l of buffer A (10 mM HEPES (pH 8.0), 10 mM KCl, 100 M EDTA, 1 mM dithiothreitol) supplemented with the mixture of protease and phosphatase inhibitors used for immunoblotting and incubated 10 min on ice. Cells were lysed by the addition of 15 l of 1% Nonidet P-40 and incubated on ice for 5 min, vortexed for 10 s, and the nuclei pelleted by centrifugation (500 ϫ g, 2 min, 4°C). The nuclear pellets were washed with 100 l of buffer A, resuspended in 50 l of extraction buffer C (20 mM HEPES (pH 8.0), 400 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol) supplemented with protease and phosphatase inhibitors as above, and incubated for 15 min at 4°C on rocking platform, the tubes were then centrifuged (20 min at 11,000 ϫ g, 4°C) to remove debris from the extract. Protein concentrations of nuclear extracts were determined to be 1 to 2 g/l against a BSA standard using a Bio-Rad assay kit (Bio-Rad, Hercules, CA). A final concentration of 10% glycerol was added to the extracts and nuclear extracts were either used immediately or stored at Ϫ80°C.
The following oligonucleotides were used in EMSAs (listed 5Ј to 3Ј: complement sequence are not shown): sis-inducible element of the c-Fos promoter region (m67SIE): GTCGACATTTCCCGTAAATC (41); IRF1-GAS: GTGATTTCCCCGAAATGACG (17). Briefly, complementary oligonucleotides were annealed, labeled with [␥-32 P]ATP (3000 Ci/mmol, Amersham Biosciences, Inc., Arlington Heights, IL) with T4 polynucleotide kinase (New England Biolabs Inc., Beverly, MA), and separated from unincorporated nucleotides over a Sephadex G-25 spin column (Amersham Biosciencss, Inc.). Nuclear extracts (10 l) containing 4 g of protein were incubated at room temperature for 20 min with 10 l of EMSA mixture (2 g of poly(dI-dC), 1 mM dithiothreitol, 5 mg/ml BSA, 20 mM Tris-HCl (pH 7.5), 0.5 M NaCl, 2 M EDTA, 10% glycerol, 1 l of probe). Protein-DNA complexes were separated by electrophoresis on a 4% polyacrylamide gel containing 10% glycerol in a Tris glycine-EDTA buffer (0.8 M glycine, 100 mM Tris base, 4 mM EDTA, pH 8.3) for 4 h. Gels were dried under vacuum and autoradiographed. For immunoanalysis of STAT-DNA complexes, nuclear extracts were incubated with 2 g of rabbit polyclonal antibody to STAT1 (Santa Cruz Biotechnology) or irrelevant rabbit Ab at room temperature for 20 min prior to addition of DNA. The EMSA mixture containing the labeled m67SIEoligonucleotide was then added for an additional 20 min and EMSA was performed. The presence of STAT1 in a complex was assessed by loss of a band in the presence of anti-STAT1 Ab, but not irrelevant Ab.
Construction of Retroviral Vectors Expressing Enhanced Green Fluorescent Protein (EGFP) and PDGFR␤-gp130 -Retroviral expression vectors coding for chimeric receptors containing the extracellular domains of human PDGFR␤ receptor and transmembrane and cytosolic portions of wild type gp130 (gp130(Y759)) or mutant gp130 (gp130(Y759F)), in which Tyr residue 759 was replaced with Phe, were constructed as follows. Human PDGFR␤ cDNA in expression vector pLXSN was obtained from Dr. A. Kazlauskas (Schepens Eye Research Institute, Boston, MA). The DNA fragment coding for the extracellular domains of PDGFR␤ receptor (1593 bp) was isolated by PCR amplification. Human gp130(Y759) cDNA and human gp130(Y759F) in expression vector pSVL was kindly provided by Dr. F. Schaper (Institute fü r Biochemie, Aachen, Germany). The DNA fragments (947 bp) coding for the transmembrane and cytoplasmic domains of gp130(Y759) and gp130(Y759F) were isolated by PCR amplification. The HindIII-EcoRI fragment containing the extracellular domains of PDGFR␤ receptor and EcoRI-NotI fragment containing the transmembrane and cytoplasmic domains of gp130 were subcloned into the LZRSpBMN-Z retroviral vector using 5Ј HindIII and 3Ј NotI cloning sites. The LZRSpBMN-Z retroviral vector and LZRSpBMN-Z retroviral vector containing EGFP was kindly provided from Dr. A. Bothwell (Yale University School of Medicine, New Haven, CT) (42). These retroviral vectors containing EGFP, PDGFR␤-gp130(Y759), and PDGFR␤-gp130(Y759F) were directly transfected into the Phoenix-Amphoteric packaging cell line using LipofectAMINE Plus reagent (Invitrogen). Two days after transfection Phoenix cells were selected in media containing puromycin (1 g/ml) and puromycin-resistant cells were used to condition medium, providing a source of retroviral stock.
Transduction of HUVEC was accomplished by serial infections over 2 weeks without drug selection as previously described (42). A round of viral infections in the presence of Polybrene (8 g/ml) was performed for 5 h with HUVEC in primary culture. The normal growth medium was replaced and cells were maintained overnight. The infection was re-peated the next day. After the second round of infection, cells were passaged and then the process of double infection was repeated. Using this protocol the percentage of HUVEC expressing transduced genes is routinely Ͼ95%.
Indirect Immunofluorescence and FACS Analysis-For immunostaining, HUVEC were washed with Hanks' buffered saline solution and incubated for 1 min with trypsin/EDTA. Detached cells were collected and washed twice with ice-cold PBS containing 1% BSA, and incubated with specific mouse monoclonal Ab to PDGFR␤ (BD Pharmingen, San Diego, CA) or irrelevant control Ab for 30 min at 4°C. Cells were washed twice with PBS, 1% BSA and then were incubated with a fluorescein isothiocyanate-conjugated polyclonal goat antimouse Ab (Roche Molecular Biochemicals, Indianapolis, IN) for 30 min on ice followed by washing twice with PBS, 1% BSA prior to fixation with 2% paraformaldehyde. After fixation, cells were analyzed by FACS using a FACSort and Lysis II software (Becton Dickinson, San Jose, CA). Surface expression of mouse IFNGR1 on transfected HepG2 cells was detected using the same protocol as above using a hamster monoclonal Ab to mouse IFNGR1 followed by phycoerythrin-conjugated antihamster monoclonal Ab (BD Pharmingen).

RESULTS
OnM Does Not Increase the Expression of IRF1 or TAP1 in HUVEC-We have previously shown that OnM and IFN␥ each stimulate phosphorylation of STAT1 Tyr residue 701 in a dosedependent manner in HUVEC (4,35), as detected by immunoblotting with Ab specific for this tyrosine-phosphorylated form of STAT1 (Tyr-701), although these responses were not compared with each other. However, FACS analysis showed that IFN␥ but not OnM increased expression of class I MHC molecules. 2 We initiated the present study to further examine this discrepancy. IRF1 and TAP1 are two IFN␥-inducible proteins whose rapid induction is mediated via STAT1-dependent mechanisms (4,17). Both are necessary for class I transcription (20) and surface expression (18). We first compared the effects of IFN␥ and OnM on the expression of IRF1 and TAP1 proteins. Six hours after stimulation with IFN␥ there was a significant increase in the expression of IRF1 and TAP1 proteins as detected by immunoblotting (Fig. 1A). However, OnM did not increase either IRF1 or TAP1 protein expression in HUVEC (Fig. 1A). To directly investigate whether OnM increases transcription of STAT1-dependent genes, the 5Ј-flanking sequence of TAP1 gene was subcloned into the promoterless poGH plasmid to yield TAP1-GH construct, in which the transcription of human GH reporter gene is controlled by the IFN-responsive sequences (16), and this TAP1-GH promoter-reporter gene construct was transiently transfected into HUVEC. Cells were then treated with IFN␥ or OnM and the amount of GH was measured in the media 24 h after treatment, as shown in Fig.  1B. There was no activity of TAP1 promoter in unstimulated HUVEC (Fig. 1B, control). TAP1 promoter activity was significantly increased by IFN␥ (Fig. 1B, closed bars) as reported previously (16), whereas poGH plasmid was unresponsive to IFN␥ (Fig. 1B, open bars). However, the TAP1 promoter did not display a transcriptional response to OnM (Fig. 1B, closed  bars). These findings are consistent with the failure of OnM to increase TAP1 protein expression.
Next, we examined whether inhibitory signals generated through gp130 interfere with the transcriptional activity of STAT1 homodimers. To do so, we tested the effects of OnM on IFN␥-mediated STAT1 phosphorylation and on the induction of STAT1-dependent proteins. In these experiments IFN␥ and OnM each stimulate tyrosine phosphorylation of STAT1 and addition of OnM did not reduce the IFN␥ response (Fig. 1C). Six hours and 48 h after addition of IFN␥, there was a significant induction of IRF1 and class I MHC protein heavy chain, respectively (Fig. 1C). OnM did not reduce the IFN␥-mediated induction of IRF1 or class I MHC protein heavy chain (Fig. 1C). We also observed that OnM did not reduce IFN␥-mediated induction of TAP1 promoter activity (Fig. 1B, closed bars).
Similarly, pretreatment of HUVEC with OnM for various times (15 min, 1 h, and 4 h) did not inhibit phosphorylation of STAT1stimulated by IFN␥ (Fig. 1D). IFN␥-mediated induction of IRF1 and class I MHC protein heavy chain also were not inhibited by preincubation with OnM (Fig. 1D). Thus the failure of OnM to induce these proteins despite phosphorylation of STAT1 on Tyr residue 701 does not appear to involve an inhibitory signal.
OnM Induces Tyrosine Phoshorylation, Serine Phosphorylation, Nuclear Translocation, and DNA Binding Activity of STAT1 Homodimers-Tyrosine phosphorylation of STAT1 on Tyr residue 701 is necessary for dimerization and subsequent translocation of STAT1 homodimers to the nucleus where they bind to specific promoter elements in target genes and increase the rate of transcription. However, it may not be sufficient to cause these responses. We therefore compared the effects of OnM and IFN␥ on STAT1 activation in greater detail, beginning with the dose response and the time course of STAT1 phosphorylation in HUVEC in response to these two cytokines. No tyrosine phosphorylation of STAT1 was detected in unstimulated HUVEC by immunoblot analysis using Ab to phosphotyrosine STAT1 ( Fig. 2A). After 30 min of cytokine treatment, tyrosine phosphorylation of STAT1 was induced by 0.62 ng/ml OnM and increased, in a dose dependent manner, to maximal levels at 10 ng/ml ( Fig. 2A). A dose response to IFN␥ was observed over a similar range ( Fig. 2A), but at all concentrations tested, OnM stimulated a greater level of STAT1 phosphorylation on Tyr residue 701 than did IFN␥. Phosphotyrosine STAT1 levels were significantly increased after 15 min of treatment with IFN␥ and the levels stayed elevated for up to 150 min (Fig. 2B). In contrast, OnM induced rapid tyrosine phosphorylation of STAT1 but phosphotyrosine STAT1 was no longer detectable after 1 h of stimulation with OnM (Fig. 2B).  While phosphorylation on Tyr residue 701 is necessary for STAT1 dimerization, for nuclear translocation, and for DNA binding in response to IFN␥, full transcriptional activation by STAT1 is obtained only when Ser residue 727 in the transcriptional activation domain is also phosphorylated (11)(12)(13). Therefore, we determined whether OnM causes serine phosphorylation of STAT1. HUVEC were stimulated with either IFN␥ or OnM for various times and cell lysates were subjected to immunoblot analysis with an antibody specific to the serine phosphorylated form of STAT1 (Ser-727) (Fig. 2C). There was a small amount of phosphoserine STAT1 in the unstimulated cells (Fig. 2C, control). Increased serine phosphorylation of STAT1 after addition of IFN␥ was seen as early as 30 min and remained elevated for 90 min. OnM also increased serine phosphorylation of STAT1 but this increase largely disappeared by 60 min (Fig. 2C). Thus, similar to tyrsoine phosphorylation of STAT1, serine phosphorylation of STAT1 by OnM was more transient than that caused by IFN␥.
We next examined whether the tyrosine phosphorylation of STAT1 by OnM is associated with nuclear translocation of STAT1 homodimers. HUVEC were stimulated with either IFN␥ or OnM for various times. Nuclear extracts were subjected to immunoblot analysis with an antibody specific to the tyrosine-phosphorylated form of STAT1 (Fig. 3A, upper panel), the serine-phosphorylated form of STAT1 (Fig. 3A, middle panel), or the tyrosine-phosphorylated form of STAT3 (Fig. 3A, lower panel). No phosphorylated STAT1 or STAT3 proteins were detected in the nucleus in the absence of cytokine (Fig. 3A,  control). Strong phosphotyrosine and phosphoserine STAT1 signals were seen at 30, 60, and 120 min after stimulation with IFN␥. IFN␥ did not induce phosphorylation of STAT3 in HU-VEC (Fig. 3A, lower panel). At early times, there was a significant amount of phosphotyrosine and phosphoserine STAT1 and phosphotyrosine STAT3 in the nuclear extracts prepared from OnM-treated cells (Fig. 3A), typically exceeding that seen with IFN␥ treatment. However, phosphorylated STAT1 signal completely disappeared at 60 min after stimulation with OnM and phosphotyrosine STAT3 largely disappeared by 60 min (Fig. 3A), consistent with the kinetics observed with whole cell lysates (Fig. 2).
To assess the activation of DNA binding of STAT1 induced by OnM, the same nuclear extracts examined by immunoblotting were used in EMSA with a labeled sis-inducible element (m67SIE) derived from the c-Fos promoter region, a probe known to bind both STAT1 and STAT3 with high affinity (41). Nuclear extracts collected after 30 min of stimulation with IFN␥ generated a single DNA binding complex (Fig. 3B). This DNA binding activity was maintained through 120 min stimulation with IFN␥ (Fig. 3B). In contrast, three major DNA complexes were formed from nuclear extracts prepared 30 min after addition of OnM and all three complexes were no longer detected in extracts collected after 60 min of treatment (Fig.  3B). These three major complexes, corresponding to SIF-A, -B, -C, have been described previously and shown to be composed of STAT3 homodimers, STAT1/STAT3 heterodimers, and STAT1 homodimers, respectively (26,43). 2 To confirm the presence of STAT1 in the complexes induced by IFN␥ or OnM treatments of HUVEC, experiments were performed with antibodies reactive with STAT1 (Fig. 3B) or control Abs. Preincubation of nuclear extracts with anti-STAT1 Ab (Fig. 3B), but not with anti-STAT3 Ab or irrelevant Ab control (data not shown) resulted in the complete elimination of DNA complexes formed by extracts from IFN␥-treated cells. Addition of STAT1 antibody to the extracts of OnM-treated HUVEC resulted in the complete elimination of SIF-B and SIF-C complexes without affecting the formation of SIF-A (Fig. 3B). Addition of anti-STAT3 antibody resulted in the complete elimination of SIF-A and SIF-B complexes without affecting the formation of SIF-C (data not shown). Nuclear extracts prepared from cells stimulated with IFN␥ or OnM confer binding of STAT1 homodimers to GAS elements of IRF1 (Fig. 3C) and TAP1 (data not shown) promoter region. Preincubation of nuclear extracts with anti-STAT1 Ab resulted in the complete elimination of DNA complexes formed by extracts from either IFN␥-or OnMtreated cells (Fig. 3C). These results confirm the presence of STAT1 homodimers in the DNA binding complexes formed from nuclear extracts of OnM-treated HUVEC, and the overall analysis suggests that OnM differs from IFN␥ in that STAT1 activation in response to OnM is much more transient than that produced by IFN␥.
Prolonged Activation of STAT1 in HUVEC through PDGFR␤-gp130(Y759F) Chimeric Receptor-These observations raised the possibility that transience of the STAT1 response could account for the failure of OnM to activate STAT1dependent gene expression. The tyrosine phosphatase SHP-2 down-regulates gp130-mediated signaling by association with the Tyr residue 759 of gp130, possibly by dephosphorylating gp130 or its associated JAKs (29 -32). A gp130 molecule mu- Nuclear extracts were analyzed for STAT factor binding by using m67SIE probe (panels B and C) and IRF1-GAS probe (panel C) as described under "Experimental Procedures." Data are from one of two experiments with similar outcome. tated so that it cannot interact with SHP-2 has resulted in more prolonged STAT activation (29 -32). Since the effects of transduction of mutated gp130 into HUVEC could not be readily analyzed due to the presence of endogenous wild type gp130, we created a PDGFR␤-gp130 chimeric receptor, in which the extracellular domains of PDGFR␤ were combined with the transmembrane and intracellular domains of gp130. This chimeric receptor allowed us to turn on the gp130 signaling pathway independently of endogenous gp130 receptors by addition of PDGFBB, taking advantage of the fact that HUVEC lack PDGF receptors. We stably transduced HUVEC with retroviruses encoding either EGFP, PDGFR␤-gp130(Y759) which contains SHP-2 docking site, or PDGFR␤-gp130(Y759F) which contains a mutated nonfunctional SHP-2 docking site. The expression of PDGFR␤ was quantitated by indirect immunofluorescence followed by FACS analysis (Fig. 4A) and by immunoblotting (Fig. 4B). As expected, untransduced HUVEC (Fig. 4A) and EGFP-transduced HUVEC (Fig. 4B) do not express PDGFR␤. Both wild type and mutant chimeras could readily be detected above the background, but two independently transduced cultures indicated that PDGFR␤-gp130(Y759) was consistently three to four times more highly expressed than PDGFR␤-gp130(Y759F).
We next tested the function of these chimeric receptors. Cytokines that signal through gp130 cause threonine/tyrosine phosphorylation of p42 and p44 MAPK in a manner that depends upon docking of SHP-2 (29,30,32). Therefore, the function of the chimeric receptor proteins was assessed by the PDGFBB phosphorylation of p42 and p44 MAPK proteins in PDGFR␤-gp130-transduced cells, as well as by PDGFBB-induced STAT1 phosphorylation. Both PDGFR␤-gp130(Y759)and PDGFR␤-gp130(Y759F)-transduced cells respond to OnM by increasing tyrosine phosphorylation of STAT1 and threonine/tyrosine phosphorylation of p42 and p44 MAPK that was indistinguishable from the control EGFP cells (data not shown). Thus, transduction did not appear to inhibit the endogenous response to OnM. To assess the potential effects of Y759F mutation on signaling, we compared time-dependent changes of STAT1 and p42 and p44 MAPK phosphorylation in PDGFR␤-gp130(Y759) and PDGFR␤-gp130(Y759F) cells in response to PDGFBB. Essentially the same time course of STAT1 (Fig. 5, upper panel) and p42 and p44 MAPK phosphorylation (Fig. 5, middle panel) was detected in PDGFR␤-gp130(Y759) in response to PDGFBB as observed for OnM. However, tyrosine phosphorylation of STAT1 after addition of PDGFBB remained elevated for up to 120 min in PDGFR␤-gp130(Y759F) cells (Fig.  5, upper panel). Moreover, treatment with PDGFBB did not elicit phosphorylation of p42 and p44 MAPK in PDGFR␤-gp130(Y759F) cells (Fig. 5, middle panel). These results are consistent with previous findings that preventing the recruitment of SHP-2 by the Y759F mutation in gp130 both prolongs the activation of STAT proteins and prevents p42 and p44 MAPK activation (29,32).
Finally, we examined the induction of IFN␥-responsive gene products by PDGFBB and IFN␥ in both PDGFR␤-gp130(Y759) and PDGFR␤-gp130(Y759F) cells. After 9 h of cytokine treatment, IRF1 and TAP1 protein expression were induced to the same levels by IFN␥ in both PDGFR␤-gp130(Y759) and PDGFR␤-gp130(Y759F) cells (Fig. 7C). Although the time course of STAT1 phosphorylation induced by IFN␥ and PDGFBB was similar in PDGFR␤-gp130(Y759F) cells, PDGFBB did not stimulate the expression of IRF1, TAP1, or class I MHC protein heavy chain in these cells (Fig. 7C). Thus, we conclude that transient phosphorylation of STAT1 induced by cytokines that signal through gp130 cannot explain why these cytokines do not increase transcription of STAT1-dependent genes in HUVEC.
OnM Activates STAT1 but Does Not Increase the Expression of IRF1 or TAP1 in HepG2 Cells-The STAT3 pathway mediates up-regulation of ␣2M in HepG2 cells by OnM and by other cytokines that signal through gp130. It has not been shown if STAT1 activation, which also occurs in these cells, is functional. To determine whether our results with HUVEC can be generalized to other cell types, we examined the responses of HepG2 cells to OnM and IFN␥. OnM and IFN␥ each increases tyrosine phosphorylation of STAT1 in HepG2 cells (Fig. 8A). As expected, there was a significant increase in IRF1 and TAP1 protein expression in response to IFN␥ in HepG2 cells (Fig.  8A). However, consistent with our HUVEC results, OnM did not increase either IRF1 or TAP1 protein expression in these cells (Fig. 8A). Furthermore, when the TAP1 promoter-reporter construct was transiently transfected into HepG2 cells (Fig.  8B), TAP1 promoter activity was significantly increased by IFN␥ whereas the TAP1 promoter did not display a transcriptional response to OnM (Fig. 8B). As a positive control ␣2M promoter activity was increased in response to OnM in HepG2 cells (Fig. 8C). Therefore, OnM increased the transcription of STAT3-dependent genes, but it did not stimulate the transcription of STAT1-dependent genes in HepG2 cells. We have also observed that IFN␥ but not OnM increases IRF1 expression in coronary artery smooth muscle cells. 2 We conclude that the inability of OnM to activate STAT1 but not induce expression of FIG. 6. Dose-dependent tyrosine phosphorylation of STAT1 by IFN␥ and PDGFBB. HUVEC were stably transduced with retroviruses encoding either PDGFR␤-gp130(Y759) or PDGFR␤-gp130(Y759F) as described under "Experimental Procedures." Cells were washed and incubated with M199 media for 4 h prior to incubating with various doses of IFN␥ or PDGFBB for 30 min, and analyzed for tyrosine phosphorylation of STAT1 (P-STAT1, Tyr-701) and total STAT1 by immunoblotting as described under "Experimental Procedures." Data are from one of two independent sets of transductants with similar outcome.

STAT1-dependent genes is not restricted to HUVEC.
IFN␥ Does Not Complement OnM STAT1 Activation-So far, we have demonstrated that STAT1 activation via gp130 is not sufficient for the induction of STAT1-dependent genes. We next tested whether additional signals might be provided by IFN␥ to complement the actions of OnM. We transiently transfected HepG2 cells with either mouse IFNGR2 plus wild type mouse IFNGR1(Y440) or mouse IFNGR2 plus mutant IFNG-R1(Y440F) containing the mutation of the IFNGR1 chain that prevents STAT1 binding (Fig. 9). Surface expression of IFNG-R1(Y440) and mutant IFNGR1(Y440F) was quantitated by indirect immunofluorescence followed by FACS analysis and indicated that both IFNGR1(Y440) and IFNGR1(Y440F) were expressed to a similar level (data not shown). Like untransfected cells, HepG2 cells transfected with IFNGR2 plus IFNG-R1(Y440) or IFNGR1(Y440F) show responsiveness to human IFN␥ but not to OnM, assessed by TAP1 promoter-reporter gene assay (Fig. 9A) and show responsiveness to OnM, assessed using a STAT3-dependent ␣2M-luciferase promoter-reporter gene assay (Fig. 9B). As predicted, HepG2 cells transfected with IFNGR2 plus wild type IFNGR1 acquire responsiveness to mouse IFN␥ (measured by the TAP1-GH promoter-reporter assay) whereas cells transfected with IFNGR2 plus mutated IFNGR1(Y440F) do not (Fig. 9A). Significantly, mouse IFN␥ treatment of cells expressing mutated receptor did not complement the OnM response (Fig. 9A). In these experiments, we did observe that OnM partially reduced mouse IFN␥-mediated induction of TAP1 promoter activity in HepG2 cells transfected with IFNGR2 plus IFNGR1(Y440) (Fig. 9A). OnM also partially reduced human IFN␥-mediated induction of TAP1 promoter activity in HepG2 cells. However, OnM did not diminish human IFN␥-mediated induction of IRF1 and TAP1 protein in HepG2 cells 2 and we conclude that the observed inhibition is restricted to promoter-reporter genes and unlikely to be of biological significance. DISCUSSION In the present study we examined whether activation of the STAT1 pathway in vascular EC by OnM could result in the induction of STAT1-dependent gene products similar to those seen in response to IFN␥. We have found that while OnM induces serine and tyrosine phosphorylation of STAT1, as well as nuclear translocation and DNA binding by STAT1 homodimers at levels comparable with those induced by IFN␥, OnM does not lead to the induction of IFN␥-responsive STAT1dependent gene products such as IRF1, TAP1, and class I MHC in HUVEC. We noted that STAT1 phosphorylation in response to OnM is more transient than that caused by IFN␥. However, transient phosphorylation of STAT1 is not the reason why OnM does not induce the expression of IFN␥-responsive gene products. This conclusion is based on use of a chimeric PDGFR␤-gp130 receptor mutated to prevent SHP-2 binding. In HUVEC transduced with such a receptor, PDGFBB leads to prolonged tyrosine and serine phosphorylation of STAT1 comparable in duration to IFN␥ responses, yet still does not cause the induction of IRF1, TAP1, and class I MHC protein heavy chain. Since OnM does not inhibit IFN␥ responses, we conclude that the activation of STAT1 is not sufficient for the induction of IFN␥responsive STAT1-dependent gene products in HUVEC.
IFN␥ acts on many cell types, including EC, to up-regulate the transcription of class I MHC molecules and related genes such as TAP1 and IRF1 (14,15,17). STAT1 is required for these IFN␥dependent responses as first shown by use of STAT1 negative mutant cell lines (17,44,45). The biological importance of STAT1 in mediating these IFN␥ responses has since been confirmed in the studies of mice that lack an intact STAT1 gene (5,6). Specifically, targeted disruption of the mouse STAT1 gene results in compromised innate immunity to viral disease and absence of induction of immunomodulatory proteins, such as class I MHC and class II MHC molecules (5,6). Although, these studies indicated that STAT1 is required for the induction of many IFN␥-dependent genes, they do not establish that STAT1 is sufficient. Very recently, it has been observed that some IFN␥ responses are retained in the absence of STAT1, implying that IFN␥ can activate additional signal transduction pathways (46,47). For example, it was observed that in macrophages derived from STAT1-deficient mice, IFN␥ can induce changes in the expression of a large number of genes and that this STAT-independent response still requires the presence of IFN␥ receptor and JAK1 (46). The existence of a STAT1-independent signal(s) raises the possibility that such signals may be necessary, albeit not sufficient, for induction of STAT-dependent genes.
The TAP1 promoter contains B, SP1, IFN-consensus se-FIG. 8. OnM does not induce the expression of TAP1 and IRF1 in HepG2 cells. A, HepG2 cells were either untreated (control), treated with IFN␥ (40 ng/ml), or OnM (10 ng/ml) for 15 min and 6 h. Samples harvested after 15 min of stimulation were analyzed for tyrosine phosphorylation of STAT1 (P-STAT1, Tyr-701) and total STAT1 using immunoblotting as described under "Experimental Procedures." Samples harvested after 6 h stimulation were analyzed for TAP1, IRF1, and ␤-actin protein expression using immunoblotting as described under "Experimental Procedures." B, HepG2 were transiently transfected with TAP1-GH promoter-reporter gene construct and a renilla luciferase expression construct. HepG2 cells were stimulated with cytokines as described in panel A for 24 h. GH activity was expressed as ng/ml normalized to renilla luciferase activity. C, HepG2 cells were transiently transfected with a ␣2M-luciferase promoter-reporter gene construct and a renilla expression construct and were stimulated with cytokines as described in panel A for 24 h. Cell lysates were assayed for luciferase activities and Firefly luciferase activity in RLU were normalized to renilla values in RLU to control transfection efficiency. Data are from one of three experiments with similar outcome.
FIG. 9. Signals provided by IFN␥ do not complement the actions of OnM. HepG2 cells were transiently transfected with a renilla luciferase expression construct, a mouse IFNGR2 expression construct plus either a mouse IFNGR1(Y440) expression construct, or a IFNGR1(Y440F) expression construct as indicated, and a TAP1-GH promoter-reporter gene construct (panel A) or a ␣2M-luciferase promoter-reporter gene construct (panel B). HepG2 cells were treated with human IFN␥ (40 ng/ml), mouse IFN␥ (10 ng/ml), OnM (10 ng/ml), or mouse IFN␥ plus OnM for 24 h. A, GH activity was expressed as nanograms/milliliter normalized to luciferase activity. B, cell lysates were assayed for luciferase activities as described in Fig. 8C. Data are from one of three experiments with similar outcome. quence, GAS, and IFN-stimulated response elements. IFNconsensus sequence, IFN-stimulated response element, and GAS bind to STAT1 homodimers in response to IFN␥. Min et al. (4) demonstrated that mutation of either GAS or the IFNstimulated response element partially reduced the IFN␥ response and that both elements must be mutated to abolish to IFN␥ response in EC. In the experiments using TAP1-GH reporter construct reported here, we demonstrated that the TAP1 promoter did not display a transcriptional response to OnM. Therefore, we concluded that STAT1 activation by OnM is not sufficient to activate the transcription of TAP1 gene.
A possible explanation for our findings is that STAT1 activation by IFN␥ is qualitatively different from that induced by OnM. A recent study by Mowen et al. (57) showed that in addition to phosphorylation of the Tyr and Ser residues, arginine methylation of STAT1 is also required for IFN␣/␤-induced transcription of STAT1-dependent genes. It is not known if IFN␥ similarly causes arginine methylation. However, it is formally possible that the lack of response to OnM is due to the inability of this cytokine to induce arginine methylation of STAT1 in HUVEC. This possibility will require further investigation.
Transient transfection of HepG2 cells with a STAT-defective mouse IFNGR1 failed to complement the OnM STAT signal. This finding suggested that signals provided by IFN␥ other than STAT1 activation cannot be provided in trans to complement the response to OnM. An alternative explanation is that STAT-defective mouse IFNGR1 is not only unable to activate STAT1, but is also not able to activate signals provided by IFN␥ other than STAT1 activation. In other words, Tyr-440 of IF-NGR1 may mediate signals other than STAT1 activation and our experiments do not formally exclude this possibility.
Several lines of evidence have indicated that EC may be an important cellular target of OnM. In vitro, OnM has been reported to stimulate polymorphonuclear leukocyte transmigration through confluent monolayers of HUVEC (49), release of IL-6 (50) and endothelin-1 (51), and expression of P-selectin (49,52), E-selectin (49), intracellular adhesion molecule-1 (49), and vascular cell adhesion molecule-1 (49). However, despite these reports, the activities of OnM in vivo are primarily antiinflammatory and there is no evidence of adhesion molecule induction on EC in vivo (24). Human EC have been shown to express a high affinity cell-surface receptor for OnM (50), presumably OnMR, and it has been reported that Ab to gp130 inhibits the induction of E-selectin induced by OnM (49). We have recently found that HUVEC can respond to human leukemia-inhibitory factor by activation of STAT and MAPK pathways, 2 raising the possibility that OnM utilizes both gp130-LIFR and gp130-OnMR complexes in this cell type. There is also evidence that OnM uses a tyrosine phosphorylation signal transduction pathway in human EC involving the activation of the p62 yes tyrosine kinase, and that this tyrosine kinase may lead to the induction of IL-6 (53). Finally, although OnM is not a mitogen for EC, 2 OnM is the major autocrine growth factor for Kaposi's sarcoma cells that are thought to be of EC origin (54,55).
The functional role of protein-tyrosine phosphatase SHP-2 in signal transduction in hepatoma cells has been assessed indirectly by preventing recruitment of SHP-2 to gp130 or by overexpressing enzymatically inactive SHP-2 mutants (29,31,32). The data indicated that activation of SHP-2 via gp130 is not required for the induction of STAT3-dependent genes but it is required for the MAPK activation (29 -32). Our observations are concordant with the conclusion that the binding of SHP-2 to tyrosine 759 on the cytoplasmic tail of gp130 negatively regulates STAT1 as well as STAT3 and positively regulates MAPK activation. They also suggest that the p42 and p44 MAPK are not required and probably not involved in gp130-induced STAT1 serine phosphorylation in HUVEC, extending previous observations indicating the p42 and p44 MAPK-dependent and -independent pathways involved in serine phosphorylation of STAT3 in response to gp130 activation (56). The contribution of other MAPKs to the serine phosphorylation of STAT1 in response to gp130 activation as well as identity of the serine kinase activated by IFN␥ remain to be determined.
Recently, Kerr and co-workers (48) demonstrated that a chimeric receptor comprising the external region of erythropoietin receptor, the transmembrane and JAK-binding domains of gp130, and a 7-amino acid STAT1 recruitment motif (Y440) from IFNGR1, efficiently mediates an IFN␥-like response. However, the receptor comprising the external region of erythropoietin receptor, the transmembrane and full-length intracellular domains of gp130 did not mediate such a response (48). These results, together with those reported here, seem to show that the specific STAT1 binding sequence of the IFNGR1 is essential for STAT1 signaling, albeit not for STAT1 phosphorylation. This raises the probability that additional covalent modifications of this transcription factor beyond tyrosine and serine phosphorylation are needed for transcriptional activities.
In summary, our analysis of the response to OnM has led to the conclusion that sustained STAT1 activation is insufficient to activate STAT1-dependent genes in HUVEC. It is as yet unclear what additional signals might be provided by IFN␥ but not OnM that are necessary for STAT1-dependent responses.