VadasYeesim Khew-Goodall, Carol Wadham, Brian N. Stein, Jennifer R. Gamble and Mathew A.

Human Umbilical Vein Endothelial CellsStat6 Activation Is Essential for Interleukin-4 Induction of P-Selectin Transcription in

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 1999 American Heart Association, Inc. All rights reserved.

Greenville Avenue, Dallas, TX 75231is published by the American Heart Association, 7272Arteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.19.6.14211999;19:1421-1429Arterioscler Thromb Vasc Biol. 

http://atvb.ahajournals.org/content/19/6/1421World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://atvb.ahajournals.org//subscriptions/

at: is onlineArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

  http://www.lww.com/reprints

Information about reprints can be found online at: Reprints: 

document. Question and AnswerPermissions and Rightspage under Services. Further information about this process is available in the

which permission is being requested is located, click Request Permissions in the middle column of the WebCopyright Clearance Center, not the Editorial Office. Once the online version of the published article for

can be obtained via RightsLink, a service of theArteriosclerosis, Thrombosis, and Vascular Biologyin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from by guest on September 26, 2013http://atvb.ahajournals.org/Downloaded from

Stat6 Activation Is Essential for Interleukin-4 Induction ofP-Selectin Transcription in Human Umbilical Vein

Endothelial CellsYeesim Khew-Goodall, Carol Wadham, Brian N. Stein, Jennifer R. Gamble, Mathew A. Vadas

Abstract—Chronic upregulation of P-selectin expression on the surface of the endothelium has been observed in and likelycontributes to a number of chronic inflammatory diseases, including atherosclerosis. Agonists of P-selectin expressionfall into 2 categories: those that induce a very rapid, transient increase, lasting only hours, and those that induceprolonged upregulation lasting days. It is the latter group, which includes interleukin-4 (IL-4), that is likely to be amediator of chronic P-selectin upregulation. The increase in P-selectin expression induced by IL-4 results fromincreased transcriptional activation of the P-selectin gene. The aim of this study was to deduce the postreceptor signalingpathway(s) giving rise to the prolonged increase in P-selectin expression induced by IL-4. We demonstrate the existenceof 2 functional signal transducer and activator of transcription 6 (Stat6) binding sites on the P-selectin promoter andfurther demonstrate, by functional analysis of the P-selectin promoter, that binding of activated Stat6 to at least 1 siteis essential for IL-4-induction of P-selectin transcription. Site 1 (nucleotide[nt]2142) bound Stat6 with a higher affinitythan did site 2 (nt2229), and this difference was reflected functionally as constructs in which only site 1 was functionalshowed full IL-4 inducibility, whereas constructs in which only site 2 was functional showed only 40% of maximal IL-4inducibility. IL-4 also induced prolonged activation of Stat6, which was contingent on the continuous presence of IL-4.The sustained activation of Stat6 induced by IL-4 is likely to be a key factor leading to the prolonged activation of theP-selectin promoter, thereby resulting in prolonged P-selectin upregulation.(Arterioscler Thromb Vasc Biol.1999;19:1421-1429.)

Key Words: P-selectinn interleukin-4n Stat6n inflammationn chronic effects

The role of inflammation in the development of athero-sclerosis is clearly established. The adhesion of circulat-

ing leukocytes, in particular, monocytes, to the endotheliumand their subsequent migration through the endothelial barrieris a requisite step in the formation of fatty streaks, theprecursors to atherosclerotic plaques. Members of the selectinfamily of adhesion receptors, when expressed on the surfaceof activated endothelial cells (ECs) during inflammation,mediate the capture of fast- flowing leukocytes, leading toleukocyte rolling on the vessel wall, a step necessary for thesubsequent firm adhesion and extravasation of leukocytes(reviewed in References 1 and 2).

P-selectin, a member of the selectin family of adhesionreceptors, is a 140-kDa glycoprotein stored in the Weibel-Palade bodies of ECs3,4 and thea-granules of platelets andmegakaryocytes.5 The importance of P-selectin in inflamma-tion is aptly demonstrated by the decrease in numbers ofrolling leukocytes in P-selectin–deficient mice, resulting indelayed and decreased leukocyte extravasation during acuteinflammation.6 Increased expression of P-selectin in theblood vessels at sites of inflammation has been observed in anumber of chronic inflammatory diseases such as rheumatoid

arthritis,7 Graves disease,8 nasal polyps,9 and atherogenicplaques.10 In a recent study of P-selectin–null mice crossedwith low density lipoprotein receptor–null mice (an estab-lished model of familial hypercholesterolemia11), the result-ing double-knockout mice showed significantly smaller andfewer fatty streaks with delayed onset compared with theirlow density lipoprotein receptor–null parents. These resultssuggest that P-selectin plays an essential role in the initiationand perhaps also the progression of atherogenic plaques.Collectively, these observations suggest that a sustainedincrease in endothelial P-selectin expression may play asignificant role in the maintenance of a chronic inflammatorystate that could underpin the development of some chronicinflammatory diseases such as atherosclerosis. Little, how-ever, is known about the mechanism(s) that leads to theprolonged upregulation of P-selectin expression at sites ofchronic inflammation.

Two categories of agonist that bring about cell-surfaceelaboration of P-selectin have been described to date. Mostknown activators result in peak cell-surface expression in 10minutes (eg, thrombin, histamine, and phorbol 12-myristate13-acetate)12,13 or 2 to 4 hours (oxygen radicals14 and tumor

Received September 29, 1998; revision accepted November 25, 1998.From the Hanson Centre for Cancer Research, Division of Human Immunology, Institute of Medical and Veterinary Science, Adelaide, Australia.Correspondence to Yeesim Khew-Goodall, PhD, Hanson Centre for Cancer Research, Division of Human Immunology, Institute of Medical and

Veterinary Science, Frome Road, Adelaide, SA 5000, Australia. E-mail yeesim.khew-goodall@imvs.sa.gov.au© 1999 American Heart Association, Inc.

Arterioscler Thromb Vasc Biol.is available at http://www.atvbaha.org

1421

necrosis factor-a [TNF-a] in the mouse15). The half-life ofthe protein on the cell surface induced by agonists such asthrombin or phorbol 12-myristate 13-acetate is short, with areturn to basal expression levels within several hours, even inthe sustained presence of the agonist. This observationimplies that there is a refractory period during which the cellsare not responsive to further stimulation by the same agonistand also suggests that this class of agonist is more likely to bea mediator of acute inflammation.

Recently, another group of agonists were discovered thatinduced cell-surface expression with kinetics very differ-ent from that induced by the agonists described earlier.These are the cytokines interleukin-3 (IL-3), IL-4, andoncostatin-M (OsM), products of activated T cells andmonocytes usually present at chronic inflammatory sites.We showed that IL-3 can upregulate cell-surface P-selectinexpression on cultured human umbilical vein ECs(HUVECs) for a prolonged period, beginning'16 hoursafter stimulation and lasting at least 4 days.16 The increasein cell-surface P-selectin expression was accompanied byan equivalent persistent increase in its mRNA level. IL-4and OsM also caused prolonged expression of bothP-selectin mRNA and cell-surface P-selectin on ECs withsimilar kinetics to IL-3 (Reference 17 and this study).Although Yao et al17 could not confirm the induction ofP-selectin mRNA expression by IL-3, we have found it tobe a consistent observation. Although the increase inmRNA level induced by IL-3 was less than that induced byIL-4, it was similar to that induced by OsM (2-fold;Y.K-G. et al, unpublished observations, 1998).

The unusual kinetics of induction of P-selectin expressionby IL-3, OsM, and IL-4 suggests that they are likely to bemediators of increased P-selectin expression seen in chronicinflammatory diseases and raise intriguing questions regard-ing the nature of the intracellular signaling pathway(s) uti-lized. Binding of IL-4 to its receptor results in activation ofseveral intracellular signaling pathways, including thephosphatidylinositol-3-kinase18 and the Janus kinase (Jak)/signal transducers and activators of transcription (STAT)pathways. Signaling through the Jak/STAT pathway occursby a tyrosine phosphorylation cascade19 whereby Jak tyrosinekinases activated by tyrosine phosphorylation in turn phos-phorylate the STAT family of transcription factors, leading toSTAT dimerization and ultimately, translocation into thenucleus.

IL-4 stimulation results in activation of Stat620 in most celltypes, including HUVECs.21,22 IL-4 suppression of TNF-a–induced E-selectin expression on HUVECs also occursthrough Stat6 activation.23 We therefore hypothesized thatP-selectin gene transcription in response to IL-4 stimulationwould probably occur through activation of Stat6. Theproximal P-selectin promoter contains 2 putative Stat6- bind-ing sites, and in this study, we demonstrate that they are bothfunctional and that the binding of Stat6 to at least 1 Stat6 siteis essential for IL-4 induction of P-selectin gene transcription.We also found that IL-4 not only induced the prolongedexpression of P-selectin but, surprisingly, also induced asimilarly prolonged activation of the transcription factorStat6.

Methods

Cell Culture, Reagents, and Antibodies (Abs)HUVECs were extracted by collagenase treatment according to amodified version of the procedure of Wall et al24 as previouslydescribed.16 Cells were either used 2 to 5 days after establishment ofculture by harvesting with trypsin (0.05%)-EDTA (0.02%) (Gibco-BRL, Life Technologies) for replating or passaged further. Afterreplating, EC growth supplement (Collaborative Research) at 25 mgzmL21 and heparin were included in the culture medium. All reagentsused in the growth, passaging, and stimulation of HUVECs weremade up under endotoxin-free conditions and contained between 10and 100 pg z mL21 endotoxin as ascertained by theLimulusamoebocyte assay. Typically cells between passages 1 and 4 wereused.

Stat6 antiserum was a gift from Dr James Ihle, St Jude Children’sHospital, Memphis Tenn; recombinant human (rh) IL-4 was a gift ofImmunex Corp, Seattle Wash; rhIL-3 was a gift from Dr ChristopherBagley; rhOsM was a gift from Drs Richard Simpson and RobertMoritz of the Ludwig Institute, Melbourne, Australia; and affinity-purified polyclonal P-selectin antibody (Ab) was a gift from DrMichael Berndt, Baker Medical Research Institute, Prahran, Austra-lia. The anti–active mitogen-activated protein kinase (MAPK) poly-clonal Ab was obtained from Promega, and the MAPK polyclonalAb (anti-ERK1 and 2) was from Zymed.

Synthetic Oligonucleotides and Plasmid ConstructsThe following oligonucleotides were used as probes or competitorsin electrophoretic mobility shift assays (EMSAs): (1) site 1 probe,59-agcttTAACTTCATGGGAAGGGCg-39; (2) site 2 probe, 59-agctt-CATCTTCCATGGAAGGGGg-39; (3) Ie, 59-agcttGCAGTTCCCAA-GAACATGg-39; (4) mut, 59-agcttGCAGTTCATGGGCTCATCg-39;and (5) b-cas, 59-agcttAGATTTCTAGGAATTCAAATCCg-39.(The lower- case letters represent the sequences added to generate anHind III restriction site at the 59 end and anSal I site at the 39 end.)Site 1 and 2 sequences were from the P-selectin promoter, and the Iecore Stat6 sequence (italicized) was from the immunoglobuline (Ie)promoter that was previously shown to bind Stat6.25 The mutsequence contained mutations in the site 1 core sequence thatrendered it unable to bind Stat6 (data not shown). Theb-cassequence is the interferon-g activation site–like sequence (GAS)element from theb- casein promoter, which binds Stat526 and acts asa general STAT- binding site (Y.K-G. et al, unpublished observa-tions, 1998, and Reference 27).

The P-selectin promoter fragment from nucleotides (nt’s)2270 to213 was generated by polymerase chain reaction (PCR) amplifica-tion of genomic DNA and specific 59 and 39 primers based on apreviously published sequence28 and subcloned into the chloram-phenicol acetyltransferase (CAT) reporter vector, pBLCAT3, for usein transient transfection assays. Mutations introduced into the Stat6sites were also generated by PCR with oligonucleotides containingthe mutated sequence as primers. All constructs were sequenced toensure that no unintended mutations had been introduced.

Nuclear Extract Preparation and EMSANuclear extracts were prepared from 23106 cells at 4°C accordingto the protocol of Mui et al.29 EMSAs were performed in a totalvolume of 15 mL containing 10 mmol/L Tris-HCl (pH 8.0),100 mmol/L KCl, 5 mmol/L MgCl2, 1 mmol/L DTT, 0.5 mg/mLBSA, 10% glycerol, 1 mmol/L PMSF, 2mg poly(dI-dC), and23105 counts per minute ('0.8 ng) of 32P-labeled probe. Thereactions were initiated by the addition of'9 mg of nuclearextract (in a volume of 2 to 5mL) and allowed to incubate at roomtemperature for 20 minutes before electrophoretic separation on a5% polyacrylamide gel in 0.253 Tris-borate-EDTA buffer. Incompetition assays, unlabeled oligonucleotides were preincubatedwith nuclear extract in the reaction mix 10 minutes before theaddition of labeled probe. Similarly, in Ab supershift assays,anti-Stat6 or preimmune serum was preincubated with nuclearextract in the reaction mix for 1 hour at 4°C before the additionof labeled probe.

1422 IL-4–Induced P-Selectin Expression Requires Stat6

Transient Transfections and CAT AssaysHUVECs were transiently transfected using lipofectin (BRL-Gibco,Life Technologies), according to the manufacturer’s recommenda-tions, with 6 mL lipofectin reagent and 1mg plasmid in 1 mLOpti-MEM medium for 105 cells per well in 6-well tissue-culturedishes. Cells were harvested 48 hours after transfection, and the CATassays were then performed. CAT assays were carried out aspreviously described.30 The various acetylated forms of [14C]chlor-amphenicol were detected by using a PhosphorImager (MolecularDynamics) and quantitated by using ImageQuant software (Molec-ular Dynamics).

Flow Cytometry to Detect Cell-SurfaceP-Selectin ExpressionTo determine cell-surface expression of P-selectin, IL-4 stimulationwas carried out in a defined medium (Opti-MEM) containing 2%FCS instead of HUVEC medium containing 20% FCS, which hasbeen used in all other experiments described. Stimulating HUVECsin the presence of a high FCS concentration resulted in a 3- to 4-foldhigher basal P-selectin expression and hence, apparently less IL-4stimulation.

Flow cytometry was used to detect cell-surface P-selectin expres-sion as previously described16 using a rabbit polyclonal Ab againsthuman platelet P-selectin at 10 mg mL21. Before cell fixation, allsolutions used with HUVECs contained 0.2% (wt/vol) NaN3 toprevent reinternalization of P-selectin from the cell surface. Thepolyclonal P-selectin Ab had previously been shown to bind specif-ically to P-selectin and did not cross-react with E-selectin expressedon the surface of Chinese hamster ovary cells transfected with theE-selectin cDNA (data not shown). Cells were washed twice at 4°Cbefore the addition of FITC-conjugated sheep anti-rabbit IgG Fab2

secondary Ab (Silenus, Victoria, Australia) at a 1:80 dilution. Thecells were incubated on ice for 30 minutes; washed once; and thenfixed in 1% formaldehyde, 2% (wt/vol) glucose, and 0.02% (wt/vol)NaN3 in PBS. Flow cytometry was carried out using an EPICSProfile II (Coulter); 10 000 cells per group were analyzed. Thebackground mean fluorescence intensity was determined by usingrabbit nonimmune serum.

Western Blotting to Detect MAPK ActivationUntreated or cytokine- treated HUVECs (23105 cells) were washedin PBS containing 1 mmol/L sodium vanadate and then lysed in lysisbuffer (20 mmol/L Tris-Cl, pH 8.0; 160 mmol/L NaCl; 1 mmol/LCaCl2 ; 1% Triton X-100; 1mmol/L leupeptin; 200mmol/L sodiumvanadate; 1 mmol/L Na F; 2 mmol/L PMSF; 25 mmol/L benzamide;and 5 mmol/Lb-glycerophosphate). The cell lysate was collectedand centrifuged. The supernatant was added to a1⁄3 volume of20 mmol/L Tris-Cl, pH 8.0, containing 20% SDS, 6% sucrose, and10 mmol/L b-mercaptoethanol. The proteins were separated byelectrophoresis on a 7.5% polyacrylamide gel and transferred to apolyvinylidene difluoride membrane. Active MAPK was detectedwith a 1/20 000 dilution of anti–active MAPK Ab. The filters werelater stripped and reprobed with an anti-ERK Ab to detect totalMAPK.

ResultsP-Selectin Promoter Binds Stat6 Induced by IL-4Two putative Stat6- binding sites (Figure 1), fitting theconsensus sequence TTCNNNNGAA,27 were identified inthe proximal promoter of the P-selectin gene28 at nt 2142(denoted as site 1) and2229 (site 2). Scanning 5 kb of theupstream flanking sequence failed to reveal additional Stat6sites. The binding characteristics of these putative Stat6 siteswere investigated by EMSA using either site 1 or site 2oligonucleotide probes with either unstimulated or IL-4–treated HUVEC nuclear extracts (Figure 2). Both putativeStat6 sites bound a factor that was induced by treatment withIL-4 for 15 minutes. Two lines of evidence suggest that thisfactor is Stat6. First, the binding was completely abolished by

the addition of a cold competitor oligonucleotide (denoted Ie)containing the Stat6-binding site of the Ie promoter, whichhad previously been demonstrated to efficiently bind Stat6.25

In contrast, an oligonucleotide probe with a mutation in theStat6 consensus sequence (denoted mut) was unable tocompete for binding to either site 1 or site 2 probes even whenadded at 100-fold molar excess. This finding suggests that theIL-4–inducible factor was specifically binding to both of thecanonical Stat6 sequences of the P-selectin promoter. Second,incubation in the presence of an anti-Stat6–specific Abresulted in a supershift that was not observed when nonim-mune serum was used (Figure 2). This result was againobserved with both site 1 and site 2 probes, confirming thatthe factor binding to both sites was indeed Stat6.

During EMSA with the site 1 and 2 probes, it was notedthat the site 1 probe gave a more intense band than did the site

Figure 1. Location of the 2 Stat6 consensus sequences in theproximal P-selectin promoter. The P-selectin promotersequence from 2270 to 213 is shown with the 2 Stat6 consen-sus sequences at 2142 and 2229 in boldface type; the putativeets sites are underlined; and the GATA site is italicized. Majortranscription start sites28 are denoted by downward arrowheads,and 11 in the sequence is the first nt of the translation initiationcodon ATG.

Figure 2. Both Stat6 consensus sequences specifically bindStat6 induced by IL-4. EMSA with either untreated or IL-4–treated (15 minutes) HUVEC nuclear extract. The 2 cold oligonu-cleotide competitors used were Ie, the Ie promoter Stat6-bindingsequence,25 and mut, the P-selectin promoter site 1 sequencewith a mutation that renders it unable to bind Stat6 (see Meth-ods for actual sequences). Both competitors were used at a100-fold molar excess. The antisera used were S6, a rabbitpolyclonal antiserum specific for Stat6, and NI, rabbit nonim-mune serum. Each gel shown is representative of 3experiments.

Khew-Goodall et al June 1999 1423

2 probe (see Figures 2 and 3), suggesting that site 1 mightbind Stat6 with a higher affinity than site 2. This wasinvestigated further by using either site 1 or site 2 as a probeand unlabeled site 1 or site 2 oligonucleotide as a competitor.The site 2 oligonucleotide was a very poor competitor of thesite 1 probe, reducing Stat6 binding by only 27% at a 50-foldmolar excess (Figure 3a). In contrast, a 5-fold molar excess ofsite 1 oligonucleotide reduced Stat6 binding to the site 1probe by 72%, and a 25-fold molar excess completelyabolished binding. By comparison, a 5-fold molar excess ofeither site 1 or site 2 oligonucleotide abolished binding to thesite 2 probe (Figure 3b). Similarly, site 1 was a more efficientcompetitor than site 2 when the Ie Stat6 sequence was used asa probe (data not shown). These observations are consistentwith site 1’s having a greater affinity for activated Stat6 thanhas site 2.

At Least 1 Stat6 Site on the P-Selectin Promoter IsRequired for IL-4 InductionThe induction of cell-surface P-selectin by IL-4 has previ-ously been shown to be accompanied by an increase intranscriptional activity of the P-selectin gene.17 Having es-tablished that the P-selectin promoter has 2 Stat6- bindingsites, their role in mediating IL-4–induced transcription of theP-selectin gene was investigated. A set of promoter-CATreporter constructs (Figure 4a) was made and transientlytransfected into HUVECs, and the CAT activity in transfec-tants that were either untreated or treated with IL-4 wasdetermined. Transfectants carrying the parent constructp2270-CAT, which contains the proximal P-selectin pro-moter from2270 to 213 and includes both Stat6- bindingsites, showed a 2.560.4-fold (mean6SEM) increase in CATactivity when treated with IL-4 compared with untreatedtransfectants (Figure 4b). The increase in CAT activityobserved in transfectants stimulated with IL-4 suggests thatthe P-selectin promoter from2270 to 213 contains theelements that confer IL-4 responsiveness. In contrast, trans-fectants with reporter-CAT constructs driven by an unrelatedpromoter, such as the thymidine kinase promoter, did notexhibit any increase in CAT activity when stimulated withIL-4 (Figure 4b).

To establish that the Stat6-binding sites are critical formediating IL-4 induction of P-selectin promoter activity,mutations that rendered them unable to bind activated Stat6were introduced into either 1 or both Stat6 sites. Themutations introduced were at the 39 end of the consensussequence, altering it from TTCNNNNGAA to TTC-NNNNCTT. Owing to 2 different problems that arose, 2 setsof constructs containing wild- type Stat6 sites and 2 setscontaining site 1 mutations were made. First, the cloningstrategy used to obtain site 1 and the double mutants resultedin a 2-bp (CG) insertion at nt2120 of the P-selectinpromoter. A control construct (p2270CG-CAT) was there-

Figure 3. Site 1 is a higher- affinity Stat6-binding site than issite 2. a, EMSA using the site 1 probe to detect activated Stat6in HUVEC extract. HUVECs were either untreated or treatedwith IL-4 at 20 ng/mL for the indicated time before being har-vested, and nuclear extracts were made. Assays were per-formed either in the absence of any cold competitor or in thepresence of various molar excesses of either site 1 or site 2cold competitor as indicated. b, EMSA as described in panel a,except that the site 2 probe was used to detect activated Stat6.Data shown are representative of 2 experiments.

Figure 4. IL-4– induced increase inP-selectin promoter activity is dependenton at least 1 Stat6 site. a, Schematic ofP-selectin promoter–CAT constructsused in transient transfection assays inHUVECs. p2270-CAT is the parent con-struct containing the authentic P-selectinpromoter sequence from 2270 to 213.Locations of the 2 Stat6- binding sitesare shown boxed; wt within the boxesindicates the authentic (wild-type) Stat6sequence for the respective site, andmut indicates that the respective Stat6site has been mutated to render the siteunable to bind Stat6. Authentic site 1and site 2 sequences are TTCATGGGAAand TTCCATGGAA, respectively. Mutantsite 1 sequences were either TTCATG-GCTT (39 mutation) or AAGATGGGAA (59

mutation), and the mutant site 2 sequence is TTCCATGCTT (italicized letters are altered nt’s). In the construct in which both sites weremutated [pS6(1/2)-CAT], the site 1 sequence has the 39 mutation. The thick bar indicates the position of major transcription start siteslocated between 2100 and 230.28 The plasmid containing the thymidine kinase promoter (pTK-CAT) is pBLCAT2. b, HUVECs weretransiently transfected with the promoter-reporter constructs shown in panel a and were either treated with IL-4 at 20 ng/mL immedi-ately after transfection for 48 hours or left untreated. The fold induction by IL-4 is the ratio of CAT activity in transfectants treated withIL-4 to the CAT activity in untreated transfectants. At least 3 separate transfections were carried out with each construct, and in somecases as many as 10. *P,0.01 when compared with p2270-CAT (Student’s t test) and #P,0.05 when compared with p2270-CAT orpS6(2)-CAT. The promoterless vector control had a basal activity (measured as % chloramphenicol acetylated) of 0.5560.11(mean6SEM) and was not significantly induced by IL-4 (data not shown). Basal activity for p2270-CAT was 1.9660.40 (P,0.05 whencompared with the promoterless control, Student’s t test); pS6(1)-CAT was 3.8861.27, pS6(2)-CAT was 4.9361.46, and p-S6(1/2)-CATwas 5.4762.77 (P.0.5 for each of the mutated constructs compared with the parental sequence).

1424 IL-4–Induced P-Selectin Expression Requires Stat6

fore generated that contained the 2-bp insertion and wild-type Stat6 sites to ensure that the 2-bp insertion itself did notimpair IL-4 responsiveness. Table 1 shows that the 2-bpinsertion had no significant effect on IL-4 induction ofP-selectin promoter activity. Because introduction of the site1 mutation [in pS6(1)39-CAT] also resulted in mutation of anoverlapping putative ets transcription factor–binding site (seeFigure 1), a second site 1 mutant construct [pS6(1)59-CAT]with the mutations introduced at the 59 end of the Stat6-binding site (AAGNNNNGAA) was made. A comparison ofthe IL-4 inducibility showed no significant difference be-tween the 2 site 1 mutants (Table 2).

In the overall analysis (Figure 4), transfection data ob-tained from the 2 constructs with wild- type Stat6 sites andthe 2 site 1 mutants were pooled and collectively referred toas p2270-CAT and pS6(1)-CAT, respectively. Simulta-neously mutating both Stat6 sites [pS6(1/2)-CAT] resulted incomplete loss of IL-4 induction of P-selectin promoteractivity. Mutation of site 1 alone [pS6(1)-CAT] led to asignificant but partial loss of IL-4 stimulation (1.660.1-fold,P,0.05). This value represents a 60% decrease in IL-4inducibility compared with p2270-CAT, which containedboth authentic Stat6 sites. However, mutation to site 2 alonehad no significant effect on IL-4 stimulation of P-selectinpromoter activity (2.760.4-fold,P50.7). These data suggestthat IL-4 stimulation of P-selectin promoter activity wasdependent on having at least 1 active Stat6 binding site on theP-selectin promoter. They also suggest that site 1 alone issufficient for full stimulation by IL-4, whereas site 2 aloneresults in only partial stimulation.

IL-4 Induces Both Prolonged Cell-SurfaceP-Selectin Expression and Stat6 ActivationThe time course of IL-4 induction of cell-surface P-selectinexpression was examined by flow cytometry using a poly-clonal Ab raised against P-selectin. IL-4 induced a 3-foldincrease in cell-surface P-selectin expression that peaked'24 hours after IL-4 stimulation and was sustained for atleast 72 hours after stimulation (see Figure 6b). Therefore, thekinetics of IL-4–induced Stat6 activation was examined byEMSA using the site 1 oligonucleotide probe from theP-selectin promoter. Figure 5 shows that activation of Stat6by IL-4 was very rapid, occurring within 15 minutes of IL-4addition. Surprisingly, activation of Stat6, like that of cell-

surface P-selectin expression, was also prolonged, lasting atleast 48 hours (Figure 5).

Prolonged Activation of Stat6 and ProlongedP-Selectin Expression Both Require theContinuous Presence of IL-4The prolonged activation of Stat6 after IL-4 stimulationsuggested incomplete receptor downregulation after ligandbinding or a long-lived signal generated after IL-4 binding toits receptor. To differentiate between these 2 possibilities, wetreated HUVECs with IL-4 for 23 hours, after which IL-4 wasthoroughly removed by 3 changes of medium. The cells werethen further incubated in the absence of IL-4 for 1 or 3 hoursbefore nuclear extracts were made. The presence of activatedStat6 was detected with an oligonucleotide probe to site 1.After 23 hours of IL-4 stimulation, there was a significantlevel of activated Stat6 present (Figure 6a). One hour afterremoval of IL-4, there was little change in the level ofactivated Stat6 remaining, but 3 hours after removal of IL-4,no detectable levels of activated Stat6 were observed. In thecontrols to which IL-4 was added again for a further 3 hoursafter the wash, there was no detectable decrease in activatedStat6 (Figure 6a). These results suggest that the continuouspresence of IL-4 is necessary for prolonged activation ofStat6.

A similar approach was followed to assess whether re-moval of IL-4 also results in a shortened duration ofP-selectin expression on the cell surface. Removal of IL-4after 24 hours of stimulation resulted in a 75% loss ofcell-surface P-selectin and a complete return to basal expres-sion 48 hours afterward (Figure 6b). Thus, like Stat6 activa-tion, continuous stimulation by IL-4 for 72 hours or replace-ment of IL-4 into the medium for a further 48 hours afterwashing out the initial stimulant is necessary for maintenanceof prolonged cell-surface expression. However, unlike down-regulation of activated Stat6, a complete return to basal levelswas not observed until at least 24 hours after removal of IL-4.This delay in the kinetics of P-selectin downregulation

Figure 5. IL-4 induces prolonged activation of Stat6. a, EMSAof 1 experiment representative of 3 different experiments eachusing a different HUVEC line. The site 1 oligonucleotide probewas used to detect activated Stat6. HUVECs were treated withIL-4 at 20 ng/mL for the indicated times before being harvested,and nuclear extracts were made. Arrowhead indicates the posi-tion of the Stat6 band. b, Pooled data from 3 separate HUVEClines. The gels were scanned using a PhosphorImager and theStat6 bands quantitated.

TABLE 1. Effect of CG Insertion on IL-4 Inducibility

ConstructFold Induction by IL-4

(Mean6SEM)No. of

Experiments

p2270-CAT 2.6860.44 3

p2270CG-CAT 2.6760.22 3

P50.99 (Student’s t test).

TABLE 2. Comparison of Mutations in the 5* and 3* Ends ofthe Proximal Stat6 Site (Site 1) on IL-4 Responsiveness

ConstructFold Induction by IL-4

(Mean6SEM)No. of

Experiments

p2S6(1)39 CAT 1.5260.05 3

p2S6(1)59 CAT 1.7460.10 3

P50.34 (Student’s t test).

Khew-Goodall et al June 1999 1425

compared with the downregulation of activated Stat6 is mostlikely due to the long half-life of P-selectin mRNA.17

OsM and IL-3 Do Not Activate Stat6Because OsM and IL-3 also induce expression of P-selectinon HUVECs,16,17 we investigated whether these cytokinesutilize the same signaling pathway as IL-4. HUVECs weretreated for 30 minutes with OsM or IL-3, and the activation ofStat transcription factors was analyzed by EMSA with theGAS element of theb- casein promoter. Theb- casein GASelement contains the consensus sequence TTCNNNGAA,which is a general Stat- binding sequence.27 OsM was foundto induce 1 prominent complex (Figure 7a) that was identifiedto be Stat1 by a Stat1 Ab supershift (Figure 7b) and by itsrelative mobility on gels (data not shown). Occasionally, aweaker Stat3 band was observed after OsM stimulation(Figure 7b), but Stat1-Stat3 heterodimers have not beenobserved. Stat1 and Stat5 both bind to a GAS element withthe canonical sequence TTCNNNGAA, of which the Stat6-binding site is a variant. Because there were no sequencemotifs in the P-selectin promoter fitting the canonical se-quence of the Stat1-Stat5 GAS element, we investigatedwhether the 2 Stat6 sites present could bind Stat1 or Stat5.Figure 7a shows that although OsM induced a large activationof Stat1, none of it was able to bind to either of the Stat6 sites.IL-3 was found to activate Stat5, which also did not to bindto either Stat6 site (data not shown). Thus, OsM and IL-3

induce P-selectin expression by a pathway different from thatused by IL-4 and that does not involve Stat6 activation.

In addition to the Jak/STAT pathway, both OsM and IL-3also activate the ras-MAPK pathway in several different celltypes.31,32To investigate whether OsM or IL-3 also activatedMAPK in HUVECs, an Ab specific for active MAPK(phosphorylated p44Erk and p42Erk) was used in Western blotanalysis. Figure 7c shows that both OsM and IL-3 inducedMAPK activation 15 minutes after their addition, with resid-ual active MAPK still detectable 24 hours afterward, althoughthe levels were markedly reduced. In contrast, this was notobserved when cells were stimulated with IL-4.

DiscussionP-selectin on ECs has been shown to be a major determinantof inflammatory responses. Two mechanisms of control ofcell-surface P-selectin expression have been delineated thatappear to function at 2 extremes of the time course: 1 thatresults in rapid upregulation within minutes of stimulation,utilizing only preformed protein stored in Weibel-Paladebodies and lasting only several hours, and the other, takingseveral hours to increase P-selectin expression, through a

Figure 6. Prolonged Stat6 activation and prolonged cell-surfaceP-selectin expression induced by IL-4 requires the continuouspresence of IL-4. a, EMSA using the P-selectin site 1 oligonu-cleotide probe to detect activated Stat6. HUVECs were stimu-lated with 20 ng/mL IL-4 continuously for the indicated times orwere stimulated for 23 hours, and then IL-4 was removed andreplaced with medium containing no IL-4 for a further 1 or 3hours as indicated (wash out). The cells were then harvestedand nuclear extracts made. *Denotes that the replacementmedium contained 20 ng/mL IL-4. b, Cell-surface P-selectinexpression detected by flow cytometry with a polyclonal anti–P-selectin Ab. HUVECs were plated in HUVEC medium containing20% FCS for 17 hours before the cells were washed once andreplaced with Opti-MEM medium containing 2% FCS just beforeIL-4 stimulation. HUVECs were stimulated with IL-4 continuouslyfor the indicated period (open squares) or were stimulated withIL-4 for 24 hours, followed by removal of IL-4 and replacementwith fresh medium containing no IL-4 for a further 24 or 48hours (closed squares). The open triangle denotes that thereplacement medium contained 20 ng/mL IL-4 for a further 48hours. One representative of 3 experiments is shown.

Figure 7. OsM activates Stat1, Stat3, and MAPK. a, EMSAusing either unstimulated HUVEC extracts or extracts after stim-ulation with either OsM (20 ng/mL, 30 minutes) or IL-4 (20ng/mL, 30 minutes). Probes used were either the GAS elementof the b- casein promoter (bcas) or the Stat6 binding sites (site1 or 2) of the P-selectin promoter. b, EMSA using eitherunstimulated HUVEC extract or extract stimulated with OsM.The assays were carried out using the b- casein promoter GASelement as a probe in the presence or absence of an anti-Stat1Ab. c, Western blot of HUVEC lysates after stimulation with theindicated cytokines for the indicated times. The blots were firstdeveloped using an anti–active MAPK Ab and subsequentlystripped and reprobed with an Ab against total ERK.

1426 IL-4–Induced P-Selectin Expression Requires Stat6

sustained increase in gene transcription lasting several days.Recently, we and others have identified a number of cyto-kines (References 16, 17, and 33 and this study) as possiblemediators of the latter mechanism of chronic P-selectinexpression on endothelium by virtue of the following:(1) they are products of activated T cells and monocytes, cellscharacteristic of chronic inflammatory sites and (2) theyinduce prolonged expression of P-selectin on ECs in vitro bystimulating P-selectin gene transcription. One such cytokineis IL-4, a product of activated T cells, mast cells, andeosinophils that is present at sites of chronic and allergicinflammation.

The signal transduction pathway(s) leading to prolongedP-selectin expression is largely unknown. One pathwaycommon to all 3 cytokines that induce prolonged P-selectinexpression is the Jak/STAT pathway. The identification of 2canonical Stat6-binding sequences on the proximal P-selectinpromoter at positions2229 and2142 led us to postulate thatIL-4 stimulation of P-selectin transcription might occurthrough activation of Stat6. We established, by EMSA andwith the use of unlabeled oligonucleotide competitors and ananti-Stat6–specific Ab, that both of these putative Stat6 sitescould indeed specifically bind Stat6 activated by IL-4. Ourdata also demonstrated that Stat6 binding to at least 1 Stat6-binding site was obligatory for inducing transcription of theP-selectin gene after IL-4 stimulation, since simultaneousinactivation of both Stat6 sites led to complete loss of IL-4inducibility.

Northern blot analysis (data not shown) showed a 2- to3-fold increase in P-selectin mRNA as early as 3 hours afterIL-4 addition. This is consistent with the increase inP-selectin promoter activity (Figure 4b) induced by IL-4 andalso with the kinetics of Stat6 activation. The steady-statelevel of P-selectin mRNA continued to increase linearly overtime up to between 16 and 24 hours, when a plateau wasreached. The amount of mRNA at 24 hours varied between 8-and 10-fold above basal levels. Yao et al17 had previouslyinvestigated the kinetics of induction of P-selectin mRNAafter IL-4 stimulation and found a similar response, exceptthat no increase was observed between the 3- to 7-hour timepoint. This result, together with the requirement for proteinsynthesis to upregulate P-selectin mRNA, could be indicativeof an indirect effect of IL-4. However, the long, slow, linearincrease in steady-state mRNA to the plateau level, as wehave observed, is consistent with a small increase in tran-scriptional rate induced by IL-4 coupled with a slow turnoverrate of mRNA. This, together with the data showing that Stat6binding to the P-selectin promoter is essential for IL-4induction of P-selectin transcription, suggests that IL-4–ac-tivated Stat6 is acting directly on the P-selectin promoter toincrease its transcription rate.

Activation of Stat6 occurs through tyrosyl phosphorylationand is thus independent of de novo protein synthesis. Indeed,the presence of cycloheximide for 24 hours did not signifi-cantly alter the amount of Stat6 binding to the site 1 probe(data not shown). Yao et al17 had previously reported and ourdata (not shown) confirm that the increase in mRNA inducedby IL-4 was cycloheximide-sensitive, suggesting that de novoprotein synthesis was also essential for induction of mRNAsynthesis. We therefore postulate that in addition to Stat6

activation, de novo synthesis of a second transcription factoris also required for IL-4 induction of P-selectin transcription.

The 2 Stat6-binding sites on the P-selectin promoter havedifferent properties. They differ by 3 nt’s in the variableregion of the canonical sequence, which results in site 1’shaving a much greater affinity for Stat6 than does site 2. Thissuggests that Stat6 would preferentially occupy site 1 in thispromoter. The difference in Stat6 binding affinity is alsoreflected in the function of the 2 sites: mutations that left onlysite 1 functional showed full IL-4 inducibility compared withthe native promoter, whereas when site 2 only was functional,IL-4 inducibility was reduced to,50%. It is unclear what therole of site 2 is. One possibility is that Stat6 bound to site 2could interact with other transcription factors to affect theiractivity. Alternatively, we hypothesize that it could be abinding site for other Stat proteins and thereby participate intransducing signals from other cytokine receptors to theP-selectin promoter.

IL-316 and OsM17 are 2 other cytokines that induceP-selectin expression and also activate the Jak/STAT path-way. IL-3 activates Stat5 in HUVECs (E.I. Korpelainen, et al,unpublished observations, 1995), whereas OsM predomi-nantly activated Stat1, with lower levels of Stat3 observedoccasionally. Stats 1, 3, and 5 bind a TTCNNNGAA consen-sus sequence that is not present in the P-selectin promoter.We therefore investigated whether they could bind to site 1and 2 sequences and found no complexes binding to eitherStat6 site when HUVECs were stimulated with IL-3 or OsM.The lack of binding sites for Stats1 and 5 on the P-selectinpromoter and the lack of activation of Stat3 by OsM suggeststhat upregulation of P-selectin expression by IL-3 and OsM isunlikely to occur through Stat activation. In addition toactivation of the Jak/STAT pathway, both IL-3 and OsM alsoactivated MAPK in HUVECs. One transcription factor that isa potential downstream target after gp130–ras-MAPK acti-vation is nuclear factor (NF)–IL-6.34 The canonical NF—IL-6binding sequence is TT/GNNGNAAT/G,35 and a sequencefitting this consensus is present at2118 on the P-selectinpromoter. It remains to be seen whether this is the mechanismby which IL-3 and OsM upregulate P-selectin expression.

Three different observations are particularly relevant in thesustained increase in P-selectin expression induced by IL-4.First, we show here that IL-4 induced prolonged activation ofStat6 itself, with elevated levels of Stat6 detectable by EMSAas long as 48 hours after IL-4 stimulation. This was somewhatsurprising because the activation of STATs has often beenobserved to be rapid and transient. The consequence ofsustained activation of Stat6 after IL-4 stimulation would bea prolonged duration of increased synthesis of P-selectinmRNA. Second, P-selectin mRNA itself has an intrinsicallylong half-life.17 Third, our studies also showed that bothprolonged Stat6 activation and prolonged P-selectin expres-sion are contingent on the continuous presence of IL-4.Removal of IL-4 after stimulation led to a rapid loss ofactivated Stat6 followed by a slower loss of cell-surfaceP-selectin expression. The requirement for a continuouspresence of IL-4 to sustain Stat6 activation suggests thatcontinuous receptor-mediated events were necessary. It alsosuggests that removal of IL-4 or its source may be aneffective way of downmodulating P-selectin expression atsites of chronic inflammation.

Khew-Goodall et al June 1999 1427

Finally, IL-4 appears to have different immunomodulatoryeffects on the endothelium that are mediated through differentsignaling pathways and mechanisms. IL-4 upregulates theexpression of P-selectin and augments TNF-a–induced vas-cular cell adhesion molecule-1 (VCAM-1) expression36 butinhibits TNF-a induction of E-selectin expression.37 IL-4upregulates P-selectin expression by increasing transcriptionthrough the Jak/Stat6 pathway (this study), whereas theVCAM-1 promoter does not contain any Stat6-binding sites.VCAM-1 upregulation occurs by stabilization of its mRNA36

through an as-yet-unidentified signaling pathway. On theother hand, IL-4 also inhibits TNF-a induction of E-selectinexpression through a Jak/Stat6 pathway,23 but this is effectedthrough competition with NF-kB binding. The Stat6-bindingsite on the E-selectin promoter overlaps the NF-kB site,which mediates TNF-a induction, such that Stat6 bindingresults in competition with NF-kB binding. In the E-selectinpromoter, Stat6 binding after IL-4 stimulation of HUVECsdoes not result in transactivation of promoter activity. Thereason for this is unknown, but 1 suggestion is that thenecessary component for Stat6 to interact with the basaltranscriptional machinery is not present. It is presentlyunclear whether transactivation is a prerequisite for IL-4induction of P-selectin gene transcription or whether theincrease in transcription occurs because Stat6 binding iscompeting away a repressor.

The reasons for the different immunomodulatory effects ofIL-4 are not entirely understood, but they may serve todetermine the leukocyte subtypes that extravasate at differentstages of the inflammatory response. In the acute stages ofinflammation, when TNF-a is present, upregulation ofE-selectin expression and IL-8 production may facilitateneutrophil extravasation. As inflammation progresses to thetransition stages when T-cell infiltration occurs, IL-4 isproduced, switches the selectin elaboration on the endotheli-um from E- to P-selectin, and also enhances and prolongsVCAM-1 expression. The combination of P-selectin andVCAM-1 would facilitate monocytic infiltration over a moreprolonged period. At later stages, if the source of IL-4persists, then there exists the possibility for chronic inflam-mation to develop. However, sites of inflammation are oftencomplex, with a plethora of cytokines and immunomodula-tory molecules being expressed, which then establishes acomplex signaling network. Unraveling the signaling net-works and how cross-talk between different pathways caninfluence the final outcome will be an important step towardthe next generation of therapeutics.

AcknowledgmentsThis work was supported by a program grant (No. 973204) from theNational Health and Medical Research Council of Australia. Wethank the staff of the delivery wards of the Women’s and Children’sHospital and Burnside Memorial Hospital, Adelaide, for the collec-tion of umbilical cords. We also thank Immunex Corp, Seattle,Wash, for the generous gift of rhIL-4, Drs Richard Simpson andRobert Moritz for rhOsM, Dr Christopher Bagley for rhIL-3, DrJames Ihle for the antiserum to Stat6, Dr Michael Berndt for thepolyclonal Ab to P-selectin, and Jenny Drew for preparationof HUVECs.

References1. Butcher EC. Leukocyte-endothelial cell recognition: three (or more) steps

to specificity and diversity.Cell. 1991;67:1033–1036.

2. Lawrence MB, Springer TA. Leukocytes roll on a selectin at physiologicflow rates: distinction from and prerequisite for adhesion throughintegrins.Cell. 1991;65:859–873.

3. McEver RP, Beckstead JH, Moore KL, Marshall-Carlson L, Bainton DF.GMP-140, a plateleta-granule membrane protein, is also synthesized byvascular endothelial cells and is localized in Weibel-Palade bodies.J ClinInvest. 1989;84:92–99.

4. Bonfanti R, Furie BC, Furie B, Wagner DD. PADGEM (GMP140) is acomponent of Weibel-Palade bodies of human endothelial cells.Blood.1989;73:1109–1112.

5. Stenberg PE, McEver RP, Shuman MA, Jacques YV, Bainton DF. Aplatelet a-granule membrane protein (GMP-140) is expressed on theplasma membrane after activation.J Cell Biol. 1985;101:880–886.

6. Mayadas TN, Johnson RC, Rayburn H, Hynes RO, Wagner DD. Leu-kocyte rolling and extravasation are severely compromised in P selectin-deficient mice.Cell. 1993;74:541–554.

7. Grober JS, Bowen BL, Ebling H, Athey B, Thompson CB, Fox DA,Stoolman LM. Monocyte-endothelial adhesion in chronic rheumatoidarthritis: in situ detection of selectin and integrin-dependent interactions[see comments].J Clin Invest.1993;91:2609–2619.

8. Miyazaki A, Mirakian R, Bottazzo GF. Adhesion molecule expression inGraves’ thyroid glands; potential relevance of granule membrane protein(GMP-140) and intercellular adhesion molecule-1 (ICAM-1) in thehoming and antigen presentation processes.Clin Exp Immunol. 1992;89:52–57.

9. Symon FA, Walsh GM, Watson SR, Wardlaw AJ. Eosinophil adhesion tonasal polyp endothelium is P-selectin-dependent.J Exp Med. 1994;180:371–376.

10. Johnson-Tidey RR, McGregor JL, Taylor PR, Poston RN. Increase in theadhesion molecule P-selectin in endothelium overlying atheroscleroticplaques: coexpression with intercellular adhesion molecule-1.Am JPathol. 1994;144:952–961.

11. Johnson RC, Chapman SM, Dong ZM, Ordovas JM, Mayadas TN, HerzJ, Hynes RO, Schaefer EJ, Wagner DD. Absence of P-selectin delaysfatty streak formation in mice.J Clin Invest. 1997;99:1037–1043.

12. Hattori R, Hamilton KK, Fugate RD, McEver RP, Sims PJ. Stimulatedsecretion of endothelial von Willebrand factor is accompanied by rapidredistribution to the cell surface of the intracellular granule membraneprotein GMP-140.J Biol Chem. 1989;264:7768–7771.

13. Geng JG, Bevilacqua MP, Moore KL, McIntyre TM, Prescott SM, KimJM, Bliss GA, Zimmerman GA, McEver RP. Rapid neutrophil adhesionto activated endothelium mediated by GMP-140.Nature. 1990;343:757–760.

14. Patel KD, Zimmerman GA, Prescott SM, McEver RP, McIntyre TM.Oxygen radicals induce human endothelial cells to express GMP-140 andbind neutrophils.J Cell Biol. 1991;112:749–759.

15. Weller A, Isenmann S, Vestweber D. Cloning of the mouse endothelialselectins: expression of both E- and P-selectin is inducible by tumornecrosis factor-a. J Biol Chem. 1992;267:15176–15183.

16. Khew-Goodall Y, Butcher CM, Litwin MS, Newlands S, Korpelainen EI,Noack LM, Berndt MC, Lopez AF, Gamble JR, Vadas MA. Chronicexpression of P-selectin on endothelial cells stimulated by the T-cellcytokine, interleukin-3.Blood. 1996;87:1432–1438.

17. Yao L, Pan J, Setiadi H, Patel KD, McEver RP. Interleukin 4 oroncostatin M induces a prolonged increase in P-selectin mRNA andprotein in human endothelial cells.J Exp Med. 1996;184:81–92.

18. Imani F, Rager KJ, Catipovic B, Marsh DG. Interleukin-4 (IL-4) inducesphosphatidylinositol 3-kinase (p85) dephosphorylation: implications forthe role of shp-1 in the il-4-induced signals in human b cells.J Biol Chem.1997;272:7927–7931.

19. Schindler C, Darnell JE Jr. Transcriptional responses to polypeptideligands: the JAK-STAT pathway.Annu Rev Biochem. 1995;64:621–651.

20. Mikita T, Campbell D, Wu P, Williamson K, Schindler U. Requirementsfor interleukin-4-induced gene expression and functional characterizationof Stat6.Mol Cell Biol. 1996;16:5811–5820.

21. Palmer-Crocker RL, Hughes CC, Pober JS. IL-4 and IL-13 activate theJAK2 tyrosine kinase and Stat6 in cultured human vascular endothelialcells through a common pathway that does not involve thegc chain.J Clin Invest. 1996;98:604–609.

22. Lugli SM, Feng N, Heim MH, Adam M, Schnyder B, Etter H, Yamage M,Eugster HP, Lutz RA, Zurawski G, Moser R. Tumor necrosis factor-aenhances the expression of the interleukin (IL)-4 receptora-chain onendothelial cells increasing IL-4 or IL-13-induced Stat6 activation.J BiolChem. 1997;272:5487–5494.

23. Bennett BL, Cruz R, Lacson RG, Manning AM. Interleukin-4 suppressionof tumor necrosis factora-stimulated E-selectin gene transcription is

1428 IL-4–Induced P-Selectin Expression Requires Stat6

mediated by STAT6 antagonism of NF-kB. J Biol Chem. 1997;272:10212–10219.

24. Wall RT, Harker LA, Quadracci LJ, Striker GE. Factors influencingendothelial cell proliferation in vitro.J Cell Physiol. 1978;96:203–213.

25. Fenghao X, Saxon A, Nguyen A, Ke Z, Diaz-Sanchez D, Nel A. Inter-leukin 4 activates a signal transducer and activator of transcription (Stat)protein which interacts with an interferon-g activation site-like sequenceupstream of the I epsilon exon in a human B cell line: evidence for theinvolvement of Janus kinase 3 and interleukin-4 Stat.J Clin Invest.1995;96:907–914.

26. Gouilleux F, Pallard C, Dusanter-Fourt I, Wakao H, Haldosen LA,Norstedt G, Levy D, Groner B. Prolactin, growth hormone, erythropoietinand granulocyte-macrophage colony stimulating factor induceMGF-Stat5 DNA binding activity.EMBO J. 1995;14:2005–2013.

27. Seidel HM, Milocco LH, Lamb P, Darnell JE Jr, Stein RB, Rosen J.Spacing of palindromic half sites as a determinant of selective STAT(signal transducers and activators of transcription) DNA binding andtranscriptional activity.Proc Natl Acad Sci U S A. 1995;92:3041–3045.

28. Pan J, McEver RP. Characterization of the promoter for the humanP-selectin gene.J Biol Chem. 1993;268:22600–22608.

29. Mui AL, Wakao H, O’Farrell AM, Harada N, Miyajima A. Interleukin-3,granulocyte-macrophage colony stimulating factor and interleukin-5transduce signals through two STAT5 homologs.EMBO J. 1995;14:1166–1175.

30. Khew-Goodall Y, Mayer RE, Maurer F, Stone SR, Hemmings BA.Structure and transcriptional regulation of protein phosphatase 2A cat-

alytic subunit genes [published erratum appears inBiochemistry.1991;30:5328].Biochemistry. 1991;30:89–97.

31. Faris M, Ensoli B, Stahl N, Yancopoulos G, Nguyen A, Wang S, Nel AE.Differential activation of the extracellular signal-regulated kinase, Junkinase and Janus kinase-Stat pathways by oncostatin M and basicfibroblast growth factor in AIDS-derived Kaposi’s sarcoma cells.AIDS.1996;10:369–378.

32. Stancato LF, Sakatsume M, David M, Dent P, Dong F, Petricoin EF,Krolewski JJ, Silvennoinen O, Saharinen P, Pierce J, Marshall CJ, SturgillT, Finbloom DS, Larner AC.b-Interferon and oncostatin M activateRaf-1 and mitogen-activated protein kinase through a JAK1-dependentpathway.Mol Cell Biol. 1997;17:3833–3840.

33. Modur V, Feldhaus MJ, Weyrich AS, Jicha DL, Prescott SM,Zimmerman GA, McIntyre TM. Oncostatin M is a proinflammatorymediator: in vivo effects correlate with endothelial cell expression ofinflammatory cytokines and adhesion molecules.J Clin Invest1997;100:158–168.

34. Akira S. IL-6-regulated transcription factors.Int J Biochem Cell Biol.1997;29:1401–1418.

35. Faisst S, Meyer S. Compilation of vertebrate-encoded transcriptionfactors.Nucleic Acids Res. 1992;20:3–26.

36. Iademarco MF, Barks JL, Dean DC. Regulation of vascular cell adhesionmolecule-1 expression by IL-4 and TNF-a in cultured endothelial cells.J Clin Invest. 1995;95:264–271.

37. Thornhill MH, Haskard DO. IL-4 regulates endothelial cell activation byIL-1, tumor necrosis factor, or IFN-g. J Immunol. 1990;145:865–872

Khew-Goodall et al June 1999 1429