Interleukin-4 Suppression of Tumor Necrosis Factor (cid:97) -stimulated E-selectin Gene Transcription Is Mediated by STAT6 Antagonism of NF- (cid:107) B*

Interleukin-4 (IL-4), an immunoregulatory cytokine secreted from activated T-helper 2 lymphocytes, eosinophils, and mast cells, stimulates the expression of a number of immune system genes via activation of the transcription factor, STAT6. However, IL-4 can concom- itantly suppress the expression of other immune-related gene products, including (cid:107) light chain, Fc (cid:103) RI, IL-8, and E-selectin. We demonstrate that IL-4 activates STAT6 in human vascular endothelial cells and that two STAT6 binding sites are present in the promoter of the E-selec- tin gene. IL-4-induced STAT6 binding does not activate E-selectin transcription but instead suppresses tumor necrosis factor (cid:97) -induced expression of the E-selectin gene. STAT6 was found to compete for binding to a re- gion in the E-selectin gene promoter containing overlap-ping STAT6 and NF- (cid:107) B binding sites, effectively acting as an antagonist of NF- (cid:107) B binding and transcriptional activation. This novel mechanism for IL-4-mediated in- hibition of inflammatory gene expression provides an example of a STAT factor acting as a transcriptional repressor rather than an activator. Competition studies were performed by the addition of molar excess of oligonucleotide the binding reaction. Supershifting experiments were performed by the addition of antibody to the binding reaction for 20 min following the normal 20-min binding reaction. Rabbit polyclonal antibodies to p50 p65 Resultant protein-DNA complexes were resolved on native 6% The gels were prerun 45 and the samples were then electrophoresed 2–3 onto paper vacuum, subjected to autoradiography. Promoter Reporter Assays— HUVECs co-transfected

IL-4 1 is a pleiotropic immunomodulatory cytokine secreted by T-helper 2 (TH2) lymphocytes, eosinophils, and mast cells (1,2). IL-4 promotes the differentiation of premature lymphocytes to the TH2 subset, induces immunoglobulin class switching in B-lymphocytes, and is present at high levels in tissues of patients with chronic inflammatory diseases such as asthma, where it appears to play an important role in disease progression (3)(4)(5)(6). STAT6 (IL-4 STAT) (7), is activated following ligand binding to the IL-4 receptor and has been directly implicated in the transcriptional activation of the low-affinity IgE receptor (FcRMIIb/CD23), the class II major histocompatability complex genes and the constant region of immunoglobulin heavy chain genes (8,9). Mice lacking STAT6 are deficient in TH2 lymphocyte differentiation and immunoglobulin class switching to the IgE phenotype (10,11). STAT proteins (Signal Transducers and Activators of Transcription) are a novel family of transcrip-tion factors which mediate the biological effects of many cytokines (12). Binding of cytokines to cell surface receptors, expressed on a wide variety of cells, activates receptor-associated Janus kinases (JAKS) resulting in tyrosine phosphorylation of cytoplasmic STAT monomers and subsequent homodimerization, nuclear translocation and transactivation of genes containing STAT regulatory elements (13). STAT6 phosphorylation and activation is associated with JAK1 and JAK3 activation (14,15). The STAT6 DNA recognition site has been characterized as a GAS-like element with consensus sequence TTC(N 3-4 )GAA (16). The presence of STAT6 in endothelial cells has not been reported.
Paradoxically, IL-4 also suppresses the expression of many immune-related proteins. For instance, by promoting the maturation of TH2 lymphocytes, IL-4 indirectly inhibits the release of the TH1-derived inflammatory mediators, interferon-␥ and IL-2 (1). IL-4 also directly inhibits the secretion of IL-1␤, tumor necrosis factor (TNF␣), and IL-6 in monocytes (17) and antagonizes many of the stimulatory actions of interferon-␥ and lipopolysaccharide including metalloproteinase biosynthesis in macrophages (18), IgG receptor (Fc␥RI/CD64) expression in monocytes (19), and light chain in murine pre-B cells (20). The immunosuppressing character of IL-4 has heightened interest in its use as a therapeutic agent, and clinical investigations of this cytokine are in progress (2). However, the molecular mechanisms by which IL-4 suppresses gene expression are not currently understood. Such an understanding would not only provide a mechanistic appreciation of the biologic action of this cytokine but might also indicate opportunities for the rational design of pharmacologic agents with enhanced biological specificity.
The endothelium presents a critical barrier between blood and surrounding tissues, thereby serving as a key component of the immune system in regulating leukocyte and macromolecular trafficking and maintaining a nonthrombotic surface (21). Proinflammatory mediators such as lipopolysaccharide, IL-1␤, and TNF␣ stimulate the endothelium to express cell adhesion molecules such as E-selectin and vascular cell adhesion molecule-1 (VCAM-1) and cytokines such as IL-6 and IL-8 (22,23). Together, cell adhesion molecules and cytokines promote the adhesion and migration of leukocytes from the blood to sites of inflammation in surrounding tissues (23). Lipopolysaccharide, IL-1␤, and TNF␣ treatment of endothelium results in the activation of the transcription factor, NF-B, which is a key regulator of cell adhesion molecule and cytokine gene expression (24). The most well studied form of NF-B is a heterodimer composed of p50 and p65 (RelA) subunits, although homodimers and complexes comprised of additional family members are known to exist (25). NF-B is sequestered as an inactive form in the cytoplasm by the presence of an inhibitor protein, IB, which masks the nuclear localization signal present on the p50 and p65 subunits. Activation of NF-B requires * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  both the phosphorylation and proteolytic degradation of IB and results in the translocation of NF-B to the nucleus where it can bind to specific promoter sequence elements and induce gene transcription (26). In vascular endothelium, IL-4 differentially regulates the expression of cell adhesion molecules such as E-selectin and VCAM-1, and proinflammatory cytokines such as IL-6 and IL-8 (16 -18). IL-4 augments the expression of VCAM-1 and IL-6 but concomitantly suppresses the expression of E-selectin and IL-8 (22,27,28), thereby modulating leukocyte recruitment to sites of inflammation (19). Treatment with IL-4 alone increases VCAM-1 protein levels via stabilization of the VCAM-1 mRNA (30). However, the mechanisms by which IL-4 suppresses E-selectin and IL-8 levels is unknown. The studies we describe in this report were designed to explore a possible role for STAT6 in the IL-4 modulation of endothelial E-selectin gene expression.

MATERIALS AND METHODS
Culture of Human Umbilical Vein Endothelial Cells-Human umbilical vein endothelial cells (HUVECs) were isolated from umbilical cords and cultured as described (29). Briefly, HUVECs were plated in tissue culture flasks pretreated with 0.1% gelatin and grown in medium M199 (Life Technologies, Inc.) containing 20% heat-inactivated fetal bovine serum (Hyclone, Logan, UT), 2 mM L-glutamine (Life Technologies, Inc.), 60 g/ml endothelial cell growth supplement (Collaborative Research, Medford, MA), 10 units/ml heparin (Sigma), and 100 units/ml penicillin G with 100 g/ml of streptomycin sulfate (Life Technologies, Inc.). Cells used in experiments were from passages 3 to 5.
Cell Adhesion Molecule ELISA Assays-HUVECs were grown to confluence in 96-well microtitre plates. Cytokine was added as a 10-l addition to the well medium. Treatments were either TNF␣ (300 units/ ml; specific activity, 2 ϫ 10 9 units/mg) or IL-4 (10 ng/ml) (Genzyme, Cambridge, MA) or both as described in the text. At the end of the incubation period, cells were washed once with phosphate-buffered saline (PBS) and incubated with freshly prepared 4% paraformaldehyde solution, pH 7, for 60 min. Plates were then washed once with PBS, blocked overnight at 4°C with 2% bovine serum albumin in PBS, washed once with PBS, and incubated with 1 g/ml primary antibody in 0.1% bovine serum albumin in PBS at 37°C for 2 h. Monoclonal antibody to VCAM-1 (CL40) was from Pharmacia & Upjohn, whereas monoclonal antibody to E-selectin was from R & D Systems (BBA1, Minneapolis, MN). After incubation with primary antibody, the cells were washed three times with 0.05% Tween 20 in PBS, incubated with alkaline phosphatase-conjugated goat anti-mouse IgG (1:500) (Organon Teknika Corp., West Chester, PA) in 0.1% bovine serum albumin in PBS at 37°C for 1 h, washed three times with 0.05% Tween 20 in PBS, and washed once with PBS. The cells were then incubated in chromogenic substrate (1 mg/ml -nitrophenyl phosphate in 1 M diethanolamine, 0.5 mM MgCl 2 , pH 9.8) at 37°C for 10 -30 min, and absorbance was measured at 405 nm using a ThermoMax microplate reader (Molecular Devices, Menlo Park, CA). The results are presented as mean Ϯ standard deviation of quadruplicate samples. Statistical significance was determined using two way analysis of variance.
Preparation of Nuclear Extracts-HUVECs were treated with TNF␣ (300 units/ml; specific activity, 2 ϫ 10 9 units/mg) or IL-4 (10 ng/ml) or both for 30 min. Cells were suspended by treatment with trypsin and pelleted by centrifugation at 1000 ϫ g for 5 min. Nuclear proteins were isolated as described previously (29).
Western Blotting-Nuclear extracts (125 g) were subjected to electrophoresis on a 10% SDS-polyacrylamide gel, and the fractionated proteins were electrophoretically transferred to Immobilon-P membrane (Millipore, Bedford, MA) using a Multiphor II semi-dry blotting device (Pharmacia Biotech Inc.). The membrane was blocked with 4% nonfat milk powder in PBS with 0.05% Tween 20 (PBS-T), incubated with 0.1 g/ml rabbit polyclonal antibody to STAT6 (Santa Cruz Biotechnology, Santa Cruz, CA) in PBS-T, washed, and then incubated with a polyclonal donkey anti-rabbit IgG antibody (1:2000) conjugated with horseradish peroxidase (Amersham Corp.). After extensive washing with PBS-T, chemiluminescent substrate was added (ECL detection system, Amersham Corp.), and the membrane was subjected to autoradiography with Hyperfilm MP (Amersham Corp.).
Electrophoretic Mobility Shift Assay-Oligonucleotides were end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase as single strands prior to annealing to form double stranded target molecules. The NF-B consensus DNA, 5Ј AGTTGAGGGGACTTTCCCAGGC 3Ј, was pur-chased as a double stranded oligonucleotide from Promega (Madison, WI). The oligonucleotide sequences used for E-selectin and Fc␥RI are given under "Results." For each assay, 10 g of nuclear protein extract was incubated with 35 fmol of 32 P-labeled oligonucleotide probe in binding buffer (4% glycerol, 10 mM Tris.Cl, 50 mM NaCl, 1 mM MgCl, 0.5 mM EDTA, 0.5 mM dithiothreitol, and 50 g/ml poly(dI-dC), pH 7.5) for 30 min at room temperature. Competition studies were performed by the addition of a 50-fold molar excess of unlabeled oligonucleotide to the binding reaction. Supershifting experiments were performed by the addition of antibody to the binding reaction for 20 min following the normal 20-min binding reaction. Rabbit polyclonal antibodies to p50 and p65 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Resultant protein-DNA complexes were resolved on native 6% polyacrylamide gels using a 0.5 ϫ TBE buffer (0.1 M Tris-Cl, 90 mM boric acid, 1 mM EDTA, pH 8.4). The gels were prerun for 45 min at 150 V, and the samples were then electrophoresed at 250 V for 2-3 h at 4°C, dried onto paper under vacuum, and subjected to autoradiography.
Promoter Reporter Assays-HUVECs were co-transfected with pCMV␤ (Clontech, Palo Alto, CA), a mammalian vector containing the ␤-galactosidase reporter driven by the cytomegalovirus promoter, and pE-luc, a mammalian vector containing 861 base pair of the E-selectin promoter (Ϫ9 to Ϫ870 base pairs from translation start site) inserted upstream of the luciferase reporter plasmid pGL2 (Promega, Madison, WI). Site-directed mutagenesis of pE-luc was performed using Pfu DNA polymerase (Stratagene) extension of both plasmid strands from mutated primers. The sense primers were; 5Ј-GCATCGTGGATA(T)(T)C-CCGGGAAAG-3Ј where the Ts in parentheses were changed to Cs to construct two separate point mutants. Clones were sequenced and shown to contain only the introduced mutation. The transfection procedure used Lipofectin reagent as outlined in the manufacturer's instructions (Life Technologies, Inc.). Briefly, cells were seeded into gelatin coated 6-well plates and allowed to grow to 60% confluency. Cells were transfected with 6 l of Lipofectin and 500 ng of each vector in 800 l of serum-free medium for 3.5 h and then incubated in normal medium for 24 -48 h. In experiments where total cytokine treatment was for 5 h, the cytokine was added 48 h after transfection. In experiments involving cytokine pretreatment, the cytokine was added 30 h after transfection. Following cytokine treatment cells were washed once in ice-cold PBS, solubilized by incubation in 200 l of reporter lysis buffer for 15 min (Promega, Madison, WI), transferred to a 96-well plate, and centrifuged to pellet cellular debris, and the supernatant was stored at Ϫ80°C. Luciferase activity was measured using the Luciferase Assay System (Promega, Madison, WI), and ␤-galactosidase activity was determined by the Luminescent ␤-galactosidase Genetic Reporter System (Clontech, Palo Alto, CA). Both activities were quantified in a model ML1000 Luminometer with version 3.993 software (Dynatech, Chantilly, VA). Final E-selectin-luciferase activities were normalized against ␤-galactosidase activity.
Immunofluorescent Staining of Transcription Factors-HUVECs were grown to confluency on gelatin-coated coverslips, treated with TNF␣ (300 units/ml; specific activity, 2 ϫ 10 9 units/mg) and/or IL-4 (10 ng/ml) for 30 min and then fixed with 4% paraformaldehyde for 20 min. Fixed cells were washed in PBS, permeabilized in 100% methanol Ϫ20°C for 6 min, washed again in PBS, and incubated with polyclonal antibody to STAT6 or p65 (Santa Cruz Biotechnology, Santa Cruz, CA) (1 g/ml) in PBS with 0.1% bovine serum albumin for 60 min. Cells were then incubated with goat anti-rabbit IgG antibody conjugated to biotin (Sigma) for 60 min, washed in PBS, and incubated with 20 g/ml fluorescein isothiocyanate-labeled avidin (Sigma) for 20 min in the dark. Coverslips were washed and fixed face down onto microscope slides in the presence of 1 mg/ml o-phenylenediamine dihydrochloride (Sigma) dissolved in PBS-glycerine (1:5, v/v). Cells were viewed at 400ϫ magnification with a Nikon Optiphot-2 microscope equipped with epiillumination and a fluorescein excitation filter and photographed on Kodak Gold 400 print film.
Isolation of RNA and Northern Blotting-Total RNA was isolated from HUVECs using the reagent RNAzol (Tel-Test, Inc., Friendswood, TX) based on the procedure of Chomczynski and Sacchi (30). RNA was resuspended in 15 l of 1 mM EDTA and quantified by absorbance at 260 nm. Total RNA was size fractionated on a 1% agarose 0.6 M formaldehyde gel (31). After electrophoresis the gel was blotted onto Hybond-Nϩ membrane (Amersham Corp.) using the manual capillary blot transfer system (Life Technologies, Inc.), and the RNA was cross-linked to the membrane by UV irradiation in a Stratalinker 2400 (Stratagene, La Jolla, CA). cDNAs of approximately 500 base pairs in length were isolated by reverse transcription-polymerase chain reaction from HU-VECs using the primer pair 5Ј-TGGTCTTACAACACCTC-3Ј and 5Ј-CAGGCTTCCATGCTCAG-3Ј for human E-selectin (32) and 5Ј-TTG-CAGCTTCTCAAGCT-3Ј and 5Ј-ATCCTCAATGACAGGAG-3Ј for human VCAM-1 (33). A glyceraldehyde-3-phosphate dehydrogenase cDNA was isolated from an amplimer kit (Clontech, Palo Alto, CA). cDNA probes were prepared by random prime labeling with [␣-32 P]dCTP (3000 Ci/mmol) using the Prime-It II labeling kit (Stratagene, La Jolla, CA).
Statistical Analysis-Group means and standard deviations were analyzed using one-way analysis of variance.

RESULTS
To establish that IL-4 can differentially modulate cytokineinduced E-selectin and VCAM-1 levels, we examined the cell surface expression of E-selectin and VCAM-1 in HUVECs stimulated with TNF␣ in the presence or the absence of IL-4 ( Fig.  1). Treatment with IL-4 alone for 5 or 23 h had no effect on E-selectin expression (Fig. 1A, lanes 2 and 3). However, consistent with previous reports (28), treatment with IL-4 alone significantly increased VCAM-1 levels above control (Fig. 1B,  lanes 2 and 3). Treatment with TNF␣ resulted in significant increase in E-selectin and VCAM-1 cell surface levels (Fig. 1,  lanes 4). IL-4 treatment in conjunction with TNF␣ resulted in an augmentation of VCAM-1 protein levels and a significant, although not complete, inhibition of E-selectin expression. Of note, although no pretreatment was necessary to observe the IL-4 enhancement of VCAM-1 expression, pretreatment was required for inhibition of E-selectin.
The IL-4-dependent increase in VCAM-1 levels is known to result from a stabilization of VCAM-1 mRNA (28). We therefore explored whether treatment with IL-4 also affected E-selectin mRNA stability. IL-4 treatment augmented the levels of VCAM-1 mRNA after 4 h and stabilized VCAM-1 transcripts over the following 12 h as previously reported. In contrast, IL-4 treatment had no effect on the half-life of E-selectin mRNA (t1 ⁄2 ϭ 5.5 h, t1 ⁄2 ϩ IL-4 ϭ 6 h) but did result in a 22% decrease in the levels of steady state E-selectin mRNA at 4 h (data not shown). Therefore, IL-4-dependent stabilization of mRNA is specific for VCAM-1 and was not found to occur with the mRNA for another cell adhesion molecule, namely E-selectin. The observed decrease in E-selectin mRNA levels suggested that the IL-4 effect on E-selectin expression was being mediated at the level of transcription. To demonstrate this further we examined the effect of IL-4 on the cytokine-induced transcriptional activation of the E-selectin promoter, using an E-selectin promoter-luciferase reporter gene construct. HUVEC monolayers were cotransfected with a constitutively expressed ␤-galactosidase reporter gene construct and 861 base pairs of the E-selectin promoter fused to the luciferase reporter gene. After 48 h, cells were treated with either TNF␣ or IL-4 alone, TNF␣ and IL-4 added simultaneously, or IL-4 added as a pretreatment of 20 min prior to TNF␣ (Fig. 2). Cells in growth medium alone (lane In leukocytes, IL-4 has been shown to activate STAT6, a key component of the IL-4-dependent transcription factor, IL-4 STAT (7). We therefore examined whether STAT6 was present in endothelial cells and activated in response to IL-4 treatment. Nuclear extracts from HUVECs treated with TNF␣ or IL-4 were examined for immunoreactive STAT6 protein (Fig. 3A). An immunoreactive species of a molecular weight identical to STAT6 was observed in nuclear extracts from IL-4-treated cells but not in extracts from quiescent or TNF␣-treated cells. Immunofluorescence studies with HUVECs treated with TNF␣ and IL-4, either individually or in combination, were used to examine in situ the activation and nuclear translocation of STAT6 and p65, a subunit of the TNF␣-inducible transcription factor NF-B (35) (Fig. 3B). Quiescent cells exhibited diffuse cytoplasmic staining but no specific nuclear staining for NF-B or STAT6. Treatment with TNF␣ resulted in the nuclear localization of NF-B. STAT6 staining was no different from that observed with untreated cells. In contrast, IL-4 treatment had no effect on NF-B activation but resulted in the nuclear trans-location of STAT6 to the nucleus. Treatment with both TNF␣ and IL-4 resulted in the simultaneous activation and nuclear localization of both NF-B and STAT6. These results demonstrate that STAT6 is present in endothelial cells and is specifically activated by IL-4 and not by TNF␣.
A previous report has shown that IL-4 directly inhibits the activation of NF-B in monocytes but not fibroblasts (36). To investigate this mechanism we treated HUVECs with TNF␣ for 20 min, with and without a 20-min IL-4 pretreatment, and examined nuclear extract DNA binding activity to STAT6 and NF-B oligonucleotides (Fig. 4). STAT6 binding activity was only observed in cell extracts from IL-4-treated cells. We observed a minor decrease in NF-B DNA binding activity from cells treated with TNF␣ and IL-4 compared with cells treated with TNF␣ alone (see legend to Fig. 4). Therefore, IL-4 did not appear to markedly inhibit NF-B activation in HUVECs.
We hypothesized that STAT6 binding sites may be present in the E-selectin gene promoter and that STAT6 binding may mediate the IL-4-induced inhibition of E-selectin expression. Computer analysis of the E-selectin promoter revealed two potential STAT6 recognition sequences (TTCN 3-4 GAA) (16) at positions Ϫ112 to Ϫ121 (E-selectin A) and Ϫ41 to Ϫ50 (Eselectin B) relative to the transcription initiation site (Fig. 5A). Oligonucleotides corresponding to these sites were synthesized, and the ability of nuclear proteins to interact with these two putative sequence elements was examined (Fig. 5B). IL-4-inducible complexes of identical electrophoretic mobility were observed binding to both of the E-selectin oligonucleotides and to an oligonucleotide containing a known STAT6 binding site from the Fc␥RI gene (7). The IL-4-inducible complex formed with the Fc␥RI oligonucleotide was diminished in the presence of STAT6 antibody, confirming that the DNA-protein complex contained STAT6 (Fig. 5B, lane 3). The presence of excess Fc␥RI oligonucleotide resulted in the complete loss of the IL-4-induced E-selectin A and B complexes, demonstrating that the factor interacting with the E-selectin sites was the same as that binding to the Fc␥RI sequence (lanes 6 and 10). Addition of STAT6 antibody to the binding reactions inhibited complex formation similar to that observed with Fc␥RI (lanes 7 and 11). Therefore, STAT6 specifically recognizes two sites within the FIG. 3. STAT6 is activated by IL-4 in endothelial cells. A, immunoblot of STAT6 in HUVEC nuclear extracts. HUVECs were left untreated or treated with TNF␣ (300 units/ml) or IL-4 (10 ng/ml) for 30 min, and nuclear proteins were analyzed for immunoreactivity with a polyclonal STAT6 antibody in a Western blot. B, immunofluorescent staining of STAT6 and p65 in HUVECs. Cells were treated with TNF␣ (300 units/ml) and/or IL-4 (10 ng/ml) for 30 min as labeled, processed as described under "Materials and Methods," and probed with polyclonal antibodies to STAT6 or p65 and visualized with fluorescein isothiocyanate immunofluorescence. E-selectin gene promoter, and STAT6 binding to these sites is induced by IL-4 treatment.
To determine whether IL-4-induced activation of STAT6 correlated directly with the observed inhibition of E-selectin expression, we used EMSA and ELISA analyses to compare the dose dependence of IL-4 activation of STAT6 (Fig. 6A) and inhibition of E-selectin cell surface protein levels (Fig. 6B). The dose-dependent concentration range of STAT6 nuclear localization and DNA binding activity to the E-selectin A site was 0.1-10 ng/ml IL-4. Similarly, IL-4 inhibition of TNF␣-induced E-selectin expression was observed at concentrations between 0.1 and 10 ng/ml IL-4, with maximal inhibition at 10 ng/ml. Inspection of the sequences surrounding the E-selectin A site revealed that this STAT6 recognition sequence overlaps two adjacent, well characterized NF-B binding sites critical for TNF␣-induced transcriptional activation (Fig. 7A) (34). This observation suggested that binding of STAT6 to this site might antagonize NF-B binding and account for the observed IL-4 suppression of TNF␣-induced E-selectin expression. To explore this hypothesis, we synthesized an extended oligonucleotide containing both the NF-B sites and the STAT6 site, as shown in Fig. 6A, and examined the binding of TNF␣and IL-4activated nuclear proteins to this sequence (Fig. 7B). Consistent with earlier experiments, nuclear extracts from cells treated with IL-4 induced the formation of a complex that was inhibited by the addition of unlabeled Fc␥RI oligonucleotide or STAT6 antibody (Fig. 7B, lanes 2-4). Nuclear extracts from cells treated with TNF␣ alone induced a distinct complex of different mobility than that identified as containing STAT6 (Fig. 7B, lane 5). Formation of this complex was inhibited by the addition of unlabeled NF-B binding oligonucleotide or antibodies to the NF-B component proteins, p50 and p65 (Fig.  7B, lanes 6 and 7). When nuclear extracts from cells treated with both TNF␣ and IL-4 were analyzed, both the NF-B and STAT6 complexes were present in the same binding reaction (Fig. 7B, lane 8). No novel complexes were observed, suggesting that NF-B and STAT6 cannot bind to the same site simultaneously. Addition of unlabeled Fc␥RI exclusively diminished STAT6 binding, whereas unlabeled NF-B oligonucleotide prevented formation of NF-B complex only (Fig. 7B, lanes 9 -10).
Similarly, addition of STAT6 antibody resulted in a loss of the STAT6 containing complex without affecting the NF-B complex, whereas antibody to p50 and p65 resulted in the loss of the NF-B complex without affecting the STAT6 complex (Fig.  7B, lanes 11-12). Therefore, NF-B and STAT6 appear to bind a shared region of the E-selectin promoter in a manner that is mutually exclusive and consequently antagonistic.
To demonstrate this transcription factor antagonism in vivo, we performed site-directed mutagenesis on the E-selectin promoter-reporter construct. Two point mutant constructs were generated that changed the STAT6 binding site but retained the NF-B binding consensus sequence of the upstream site. According to a previous study, mutant-1 should abolish STAT6 binding, and mutant-2 should retain the STAT6 consensus sequence (8). In a promoter-reporter assay (Fig. 8) the control E-selectin luciferase construct showed enhanced luciferase activity in the presence of TNF␣ that was inhibited 50% with an IL-4 pretreatment (lanes 1-3) as shown in an earlier experiment (Fig. 2). Mutant-1 showed an unexpected partial loss of TNF␣ stimulation, suggesting that this point mutation does affect NF-B binding even though the introduced mutation conformed to the theoretical consensus sequence. However, no additional inhibition was observed with IL-4, suggesting that STAT6 binding at this site is responsible for the inhibitory effect observed (Fig. 8, lanes 4 -6). A second mutant-1 isolate exhibited the same activity profile (data not shown). Mutant-2 FIG. 5. STAT6 binds to two sites in the E-selectin gene promoter. A lists the three oligonucleotides used in the EMSA analyses. The specific STAT6 binding sites are underlined, and consensus nucleotides are capitalized. The Fc␥RI oligonucleotide was used as a STAT6 binding control (7). B, EMSA analysis of STAT6 binding to Fc␥RI (lanes 1-3), E-selectin site A (lanes 4 -7), and E-selectin B (lanes 8 -11) oligonucleotide probes. Nuclear extract was prepared from HUVECs treated with medium alone (lanes 1, 4, and 8) or with 10 ng/ml IL-4 for 30 min. Binding was competed with a STAT6 antibody (lanes 3, 7, and 11) or with excess unlabeled Fc␥RI oligonucleotide (lanes 6 and 10). NS, nonspecific binding; Ab, antibody. showed TNF␣ stimulation similar to that observed for the control and was likewise inhibited by IL-4, confirming that this mutation does not abolish STAT6 binding (Fig. 8, lanes 7-9). DISCUSSION IL-4 has been shown in vivo to retard the infiltration of neutrophils and monocytes and to enhance the recruitment of lymphocytes and eosinophils to sites of inflammation (3,37). Selective leukocyte recruitment is a feature of inflammatory diseases such as asthma, which are characterized by an eosinophilic infiltrate and high levels of IL-4 (6). Evidence suggests that the mechanisms whereby IL-4 exerts this effect are diverse but include the augmentation of VCAM-1, an important ligand for eosinophil adhesion but not required for the recruitment of neutrophils (38); the inhibition of IL-8 secretion, a major neutrophil chemoattractant (39); and the suppression of E-selectin, an essential endothelial adhesion molecule for the initial attachment and rolling of neutrophils (40). Intravital microscopic evaluations of leukocyte rolling in E-selectin-deficient mice show changes in initial tethering of leukocytes to the endothelial surface (41). Indeed, monoclonal antibodies to Eselectin inhibit neutrophil attachment to activated endothelium but have little effect on eosinophil adhesion (42). E-selectin expression is activated by proinflammatory mediators such as TNF␣ and IL-1␤ released by activated macrophages and mast cells at sites of inflammation. The molecular pathways by which TNF␣ and IL-1␤ promote the transcriptional activation of E-selectin expression are under intensive investigation but have highlighted the necessity for activation of the transcription factor, NF-B.
We present evidence for a mechanism whereby IL-4 suppresses E-selectin protein levels via the transcription factor STAT6. These experiments demonstrate that NF-B and STAT6 are activated by distinct signal transduction pathways in endothelial cells and that treatment with TNF␣ and IL-4 results in the concurrent nuclear localization of both NF-B and STAT6. EMSA analysis demonstrates that STAT6 can compete for binding to a dual NF-B enhancer element previously shown to be crucial for maximal E-selectin expression (34). The presence of STAT6 in the nucleus corresponds with repressed E-selectin promoter activity and the suppression of E-selectin mRNA and cell surface protein. STAT6 therefore acts as an antagonist of NF-B, blocking its binding and ability to transactivate the E-selectin gene.
The property of STAT6 acting as a transcriptional repressor may not be unique to the E-selectin gene. The antagonistic action of STAT6 that we have identified may also occur in an analogous manner within the Fc␥RI/CD64 gene in monocytes (43). Transcriptional activation of this gene is regulated by the interferon-␥-activated transcription factor, GAF, but is inhibited by treatment with IL-4. STAT6 and GAF can both bind the recognition sequence TTCN 3 GAA, which is present in the Fc␥RI gene promoter. IL-4-activated STAT6 may compete for binding to this site, thereby preventing GAF-mediated transcription of the Fc␥RI gene. Further work is necessary to determine if similar STAT6 antagonistic mechanisms are responsible for the IL-4 suppression of other genes.
Complete inhibition of E-selectin expression by IL-4 was not observed, probably because a third NF-B site, located downstream from those present in the E-selectin A site, has been shown to promote up to 50% of the total NF-B-mediated transcriptional activation of the E-selectin gene (34). This site is not in close proximity to either of the STAT6 sites identified in the study. The function of the second STAT6 site (E-selectin B), which is located downstream of the A site, is at this time not clear. This site does not appear to overlap any binding sites for known transcription factors. The presence of STAT6 alone is not sufficient to induce transcription of the E-selectin gene,  1, 4, and 7), treated with TNF␣ for 5 h (lanes 2, 5, and 8), or pretreated with IL-4 for 20 min followed by TNF␣ for 5 h (lanes 3, 6, and  9). The values represent the means Ϯ S.D. of six replicates and are representative of three independent experiments. #, p Ͻ 0.001; *, p Ͻ 0.005 compared with TNF␣-treated cells. Ab, antibody.
presumably because it is unable to promote the formation of an effective transcriptional complex. This is interesting because a STAT6-containing transcription factor has been shown to activate the FcRMIIb/CD23, class II major histocompatability complex, and mouse C domain of immunoglobulin heavy chain genes in leukocytes (8,9). It will be important to compare the composition of STAT6 transcription factor complexes in endothelial cells and monocytes. It will also be valuable to identify genes that are transcriptionally activated by STAT6 in endothelial cells so as to define additional nuclear factors required for STAT6-mediated transcription.
Activation and translocation of STAT6 by IL-4 is rapid, leading to STAT6 DNA binding activity within 30 min of exposure to cytokine. Although some inhibition of TNF␣-induced E-selectin promoter activity was observed when TNF␣ and IL-4 were added simultaneously, the inhibitory effect of IL-4 was more pronounced when HUVECs were incubated with IL-4 for 20 min prior to TNF␣ treatment. This suggests that either the pathway of NF-B activation is more rapid than that of STAT6 or that protein-DNA binding kinetics at the E-selectin A site in vivo favor binding of NF-B so that elevated levels of STAT6 in the nucleus are required to effectively compete with NF-B for binding. Our observation that the E-selectin A site is completely integrated within two adjacent NF-B sites presented the possibility that conservation of the STAT6 recognition sequence might be a secondary consequence of conservation of the NF-B sites. Although the decameric NF-B consensus sequence varies considerably in many genes shown to be transcriptionally activated by NF-B (26), we were unable to destroy STAT6 activity without affecting NF-B activity. Therefore, conservation of the STAT6 site within two potentially variable NF-B sites could represent the selective retention of an important biological role for IL-4 on expression of the Eselectin gene. In this regard, Schindler and Baichwal (34) have shown that nucleotide changes to the key residues comprising the STAT6 binding site result in significant inhibition of NF-B-mediated promoter activity. Therefore, the conservation and interaction of these three sites appears closely linked.
The mechanisms whereby IL-4 modulates the expression of E-selectin and VCAM-1 in endothelium highlights the fact that an individual cytokine may activate different intracellular signal transduction pathways. Although IL-4 treatment resulted in STAT6 activation and the consequent suppression of Eselectin gene transcription, we found no STAT6 binding sites in the VCAM-1 promoter. IL-4 augmentation of VCAM-1 expression is instead mediated via a poorly characterized pathway that results in the stabilization of VCAM-1 mRNA. This mechanism appears highly selective because it had no effect on E-selectin mRNA or glyceraldehyde-3-phosphate dehydrogenase mRNA stability (28). In addition, IL-4 has been shown to result in the phosphorylation of a 170-kDa substrate called 4PS (44). This protein is related by homology to insulin receptor substrate-1 and interacts with phosphatidylinositol-3 kinase. Therefore, IL-4 may activate at least three signal transduction pathways in a single cell type. Drug discovery efforts that focus upon exploiting such discrete signaling pathways may yield therapies with enhanced biological specificity and reduced side effects.
The purpose of this study was to examine the molecular mechanism whereby an immunomodulatory cytokine, IL-4, may act on E-selectin expression and thereby alter the inflammatory profile of diseases such as asthma. That E-selectin suppression is mediated by an antagonism of NF-B reiterates the importance of this transcription factor in inflammatory disease. NF-B has been shown to be activated in conditions as diverse as allergic airway inflammation (45), atherosclerosis (46), endotoxic shock (47), ischaemia-repurfusion injury (48), rheumatoid arthritis (49), restenosis (50), and sunburn (51). Furthermore, it is notable that many agents that exhibit antiinflammatory properties have been shown to inhibit NF-B action including glucocorticoids (52,53), antioxidants (54), salicylates (55), gliotoxin (56), flavinoids (57), as well as endogenous mediators such as nitric oxide (58) and the immunomodulatory cytokine, IL-10 (59). We now show that IL-4 also exerts some of its immunomodulatory action by inhibiting the actions of NF-B.