2005 TNFa

Regular ArticleEffects of TNF-A and curcumin on the expression of thrombomodulin and endothelial protein C receptor in human endothelial cellsBicheng Nan,Peter Lin,Alan B.Lumsden,Qizhi Yao,Changyi Chen *Molecular Surgeon Research Center ,Division of Vascular Surgery and Endovascular Therapy,Michael E.DeBakey Department of Surgery,Baylor College of Medicine,One Baylor Plaza,Mail stop:NAB-2010,Houston,TX 77030,United StatesReceived 27September 2004;received in revised form 16October 2004;accepted 26October 2004Available online 14November 2004Abstract The objective of this study was to elucidate the effects of tumor necrosis factor-alpha (TNF-a )on the expression of thrombomodulin (TM)and endothelial protein C receptor (EPCR)in human endothelial cells as well as the effect of curcumin,a spice and coloring food compound,as a potential therapeutic agent.Human umbilical vein endothelial cells (HUVECs)treated with TNF-a (2.0ng/ml)showed reduced TM mRNA levels by 80%,97%,94%,and 97%at 3,6,12,and 24h,respectively (P b 0.05),by real-time PCR analysis.Dose-dependent study showed that TM mRNA levels of HUVECs were decreased by 86%,89%,91%,and 94%after treatment of TNF-a (0,0.25,0.5,1,and 2ng/ml)for 6h,respectively (P b 0.05).TM protein levels in HUVECs were significantly reduced by 69%in TNF-a -treated cells as compared to controls (P b 0.05)by Western blot analysis.Secreted protein and activity of TM of HUVEC cultures were also significantly reduced in TNF-a -treated cells.In addition,EPCR mRNA levels of HUVECs were significantly reduced in TNF-a -treated group as compared to controls (P b 0.05).Furthermore,these effects were observed in other types of endothelial cells from human coronary arteries,lung,and skin.Curcumin effectively blocked these effects of TNF-a on downregulation of TM and EPCR.These data demonstrate that TNF-a significantly decreases expression of TM and EPCR at both mRNA and protein levels in several human endothelial cells.0049-3848/$-see front matter D 2004Elsevier Ltd.All rights reserved.doi:10.1016/j.thromres.2004.10.010Abbreviations:TNF-a :tumor necrosis factor-alpha;TM:thrombomodulin;EPCR:endothelial protein C receptor;HUVECs:Human umbilical vein endothelial cells;IL-1:interleukin-1;TF:tissue factor;PAI-1:plasminogen activator inhibitor type-1;tPA:tissue-type plasminogen activator;TAFI:thrombin activatable fibrinolysis inhibitor;APC:activated protein C;ICAM-1:intercellular adhesion molecule-1;IL-8:interleukin-8;HMVECs-L:human lung microvascular endothelial cells;HCAECs:human coronary artery endothelial cells;HMECs:human dermal microvascular endothelial cells;EBM-2:endothelial cell basal medium;hFGF-2:human basic fibroblast growth factor;R3-IGF-1:human recombinant insulin-like growth factor;VEGF:vascular endothelial growth factor;CT:threshold cycle.*Corresponding author .Tel.:+17137984401;fax:+17137986633.E-mail address:jchen@ (C.Chen).KEYWORDSTNF-a ;Curcumin;Thrombomodulin;Endothelial protein C receptor;Endothelial cell;Antioxidant;InflammationThrombosis Research (2005)115,417—426/journals/threCurcumin can effectively block TNF-a-induced endothelial dysfunction.This study suggests a new molecular mechanism of inflammation-induced thrombosis and a new therapeutic strategy to prevent this clinical problem.D2004Elsevier Ltd.All rights reserved.IntroductionEndothelial dysfunction is a key initiating event in the etiology of vascular diseases and has been linked to the pathogenesis of atherosclerosis, stroke,thrombosis,hypertension,and diabetes. Thrombosis has been reported to be a consequence of imbalance of functional phenotypes,such as thrombosis markers of the endothelial cell surface, which is involved in the procoagulant and anti-coagulant pathways.Increasing evidence has shown that inflammatory cytokines,such as interleukin-1 (IL-1)and tumor necrosis factor-a(TNF-a),have a profound effect on endothelial cell injury via regulating endothelial phenotypes,modulating the coagulant system,changing vascular perme-ability,and inducing the expression of adhesion molecules and chemoattractive cytokines.TNF-a is produced by monocytes,macrophages, lymphocytes,and other types of cells during the inflammatory process[1].TNF-a is able to disturb the hemostatic balance on the surface of endothe-lial cells by changing the expression of several thrombosis factors,such as tissue factor(TF), plasminogen activator inhibitor type-1(PAI-1),and tissue-type plasminogen activator(tPA),in a con-certed way and facilitate blood clot formation and lead to thrombosis[2—5].However,the effect of TNF-a on thrombomodulin(TM)and endothelial protein C receptor(EPCR)in different types of human endothelial cells is not known.TM is constitutively expressed on the endothelial cell surface[6,7].The major function of TM is to form a high affinity1:1complex with thrombin. The complex of TM/Thrombin has several impor-tant roles including activation of protein C[8]and thrombin activatable fibrinolysis inhibitor(TAFI) [9],as well as inhibition of inflammation[10]. Thus,TM has a critical effect on anticoagulation and anti-inflammation[10,11].Functional defi-ciency of TM could enhance thrombosis formation.EPCR is a transmembrane glycoprotein[12,13], which effectively augments rate of protein C activation and thereby inhibits the thrombin for-mation from prothrombin[14—16].Formation of activated protein C(APC)depends on the inter-action of at least four proteins,including thrombin, TM,protein C,and EPCR,on the endothelial cell surface.Any change of these factors could reduce the efficiency of APC generation,thereby enhanc-ing the risk of thrombosis[17].Curcumin(diferuloylmethane),a natural and crystalline compound isolated from the plant Curcuma longa,has anti-inflammatory,antioxida-tive,and anticarcinogenic properties[18].It has been reported that curcumin can counteract TNF-a-induced TF expression,intercellular adhesion molecule-1(ICAM-1)expression,and interleukin-8 (IL-8)production[19—23].It is not clear whether curcumin could affect the expression of TNF-a of TM and EPCR in human endothelial cells.The purpose of this study was to determine the effects of TNF-a on TM and EPCR gene expression in human endothelial cells as well as the effect of curcumin as a potential therapeutic agent.Several different types of human endothelial cells were tested.The data from this study will provide a better understanding of the mechanism involved in TNF-a-induced thrombosis as well as provide a new strategy to control inflammation-induced endothe-lial dysfunction.Materials and methodsChemicals and reagentsTNF-a,TRI reagent,thrombin,and monoclonal mouse antihuman h-actin were purchased from Sigma(St.Louis,MO).Chromozym PCa and mono-clonal mouse antihuman TM antibody were obtained from American Diagnostica(Greenwich, CT).Human protein C and antithrombin III were purchased from Calbiochem Novabiochem(La Jolla,CA).Tissue culture plastic wares were purchased from Corning Life Sciences(West Ches-ter,PA).Cell culture media and reagents were ordered from Cambrex(Walkersville,MD)and Invitrogen(Carsbad,CA).Goat polyclonal antihu-man EPCR was purchased from Santa Cruz Biotech-nology(Santa Cruz,CA).ECL plus Western Blotting Detection System and sheep antimouse Ig linked with horseradish peroxidase were purchased from Amersham Pharmacia Biotech(Piscataway,NJ). iScrip cDNA Synthesis kit and iQ SYBR Green Supermix kit were purchased from BioRad Labora-tories(Hercules,CA).B.Nan et al.418Cell cultureHuman umbilical vein endothelial cells(HUVECs), human lung microvascular endothelial cells (HMVECs-L),and human coronary artery endothe-lial cells(HCAECs)were purchased from Cambrex at passage3.Immortalized human dermal micro-vascular endothelial cells(HMECs)were obtained from Dr.Wright S.Caughman,Department of Dermatology,Emory University(Atlanta,GA). HUVECs,HCAECs,and HMECs were routinely main-tained in the endothelial cell basal medium(EBM-2) supplemented with10%fetal bovine serum and a set of EGM-2SingleQuots including0.04%hydro-cortisone,0.4%human basic fibroblast growth factor(hFGF-2),and0.1%each of human recombi-nant epidermal growth factor,human recombinant vascular endothelial growth factor(VEGF),ascorbic acid,heparin,gentamicin sulfate,amphotericin-B (CA-1000),and human recombinant insulin-like growth factor(R3-IGF-1).With the exception of heparin,HMVECs-L were cultured with the same medium.All cells were maintained at378C in a5% CO2humidified environment with the medium changed every2days.Confluent cells were detached with trypsin-EDTA(0.25%,Invitrogen) and reseeded into6-well plates(6.3Â106cells/ well).The cells were then grown in the complete medium for48h to90%confluence,with a change of the medium after24h.The cells were switched to a starvation medium(EBM-2supplemented with 1%fetal bovine serum,0.1%gentamicin sulfate and amphotericin-B,heparin,and ascorbic acid)for24 h.The cells were then stimulated with recombinant human TNF-a(2ng/ml)at378C for6h.Control cells were received only from the fresh starvation medium without TNF-a.All samples were per-formed in triplicate.RNA extraction and real-time PCRThe cells were washed with cold PBS twice and total RNA was isolated by TRI reagent.Briefly,the cells were harvested and treated with1ml TRI reagent for5min at room temperature and then centrifuged at12,000rpm for15min at48C.The aqueous portion containing RNA was transferred to another RNAase free Eppendorf tube,and0.5ml Isopronol was added and incubated for10min at room temperature,followed by precipitating total RNA with centrifugation at12,000rpm at48C for10 min.The pellet was then washed with75%ethanol twice,dried at room temperature on ice for5min, and total RNA pellet was dissolved in20A l H2O. Genomic DNA was digested by DNAase with TURBO DNA free kit(Ambion,Austin,TX).RNA concen-tration was measured at A260with Beckman DU530 Life Science UV/Vis Spectrophotometer.cDNA was obtained from reverse transcription of1A g of total RNA in a20-A l reaction volume by iScrip cDNA Synthesis kit(BioRad Laboratories).The specific primers for TM and EPCR(Table1)were designed with Beacon Designer2.1software(Premier Biosoft International,Palo Alto,CA)and synthesized by Sigma-Genosys(The Woodlands,TX).The sequence for the house keeping gene,h-actin,is as follows: sense:5V CTGGAACGGTGAAGGTGACA3V;antisense: 5V AAGGGACTTCCTGTAACAATGCA3V.Quantitative real-time PCR was performed using the iQ SYBR Green Supermix kit(BioRad Laboratories).Briefly,1 A l reaction product of RT reaction was used in a25-A l real-time PCR reaction.The annealing temper-ature and primer concentration were adapted for each gene.The amplification was initially incu-bated at958C for 1.5min and followed by performing40cycles for1min at608C,1min at 958C,and1min at558C by using iCycler iQ real-time PCR detection system(BioRad Laboratories). Data were analyzed by software version 3.0a (BioRad Laboratories).Data were presented by the relative amount of mRNA with the formula 2(ÀD CT),which stands for the difference of thresh-old cycle(CT)between a gene of interest and the housekeeping gene h-actin.The threshold cycle is the point at which sample fluorescence rises above the background level.Each sample was measured in triplicate.The mRNA(no reverse transcription)or H2O(no DNA samples)were included in the real-time PCR reaction for negative controls.ELISASecreted levels of TM were detected using an IMUBIND thrombomodulin ELISA Kit(AmericanTable1Design of real-time PCR primersGene Accession no.Primer sequenceTM M16552Sense5V TAACGAAGACACAGACTGCGATT3VAntisense5V CTAGCCCACGAGGTCAAGGT3V EPCR BC01445Sense5V AACACCAAAGGGAGCCAAACAAG3VAntisense5V CTACAGCCACACCAGCAATCATG3V Effects of TNF-a and curcumin on the expression of thrombomodulin and EPCR419Diagnostica).Briefly,200A l cell supernatants were added into precoated microtest wells,covered with a sticky plastic lid,and incubated for1h at room temperature.The wells were washed(Â4) with wash buffer.The detection antibody(200A l) was added to each well and incubated for30min at room temperature with lid covering.After the antibody was incubated,the wells were washed with wash buffer.The enzymatic reaction occurred by adding200A l of substrate solution into each well for20min.Blue color indicated a positive reaction.The reaction was terminated by adding100A l of0.5M H2SO4.The absorbance was read in the next30min at the wavelength of450 nm with EL800Universal Microplate Reader(Bio-Tek Instruments).All samples were measured in duplicate.Western blot analysisThe cells were washed three times with cold PBS, lysed with lysis buffer(25mM Tris—HCl,pH7.4, 0.15M NaCl,1%Triton X-100)and complete protease inhibitor cocktail(Roche),and then placed on a shaker at48C for1h.After centrifugation at12,000rpm for10min,the protein concentration of the supernatant was measured by Bio-Rad Protein Assay.Proteins(40 A g)were separated by10%sodium dodecyl sul-fate-polyacrylamide gel eletrophoresis(SDS-PAGE). The separated proteins were transferred to Hybond TM ECL TM Nitrocellulose Membrane (Amersham Pharmacia Biotech).Immunoblot was blocked with blocking buffer(5%nonfat milk from Bio-Rad,10mmol/l Tris,pH7.5,100mmol/l NaCl, and0.1%T ween-20)at room temperature for1h. TM was detected by1:1000dilution of monoclonal mouse antihuman TM antibody and incubated overnight at48C.Then,the membrane was washed by TBS-T(1ÂTris Buffered Saline,1% T ween20,Sigma)three times for5min.The second antibody(sheep antimouse Ig linked horse-radish)at1:2000dilution was incubated at room temperature for1h.Protein visualization was performed with enhanced chemiluminescence detection kit ECL plus.Densitometric analysis of western blot bands was performed with Alpha image software(Alpha Innotech). Thrombomodulin activity assayEndothelial cell TM activity was assayed by meas-uring the changes in APC as previously described [24].In brief,HUVECs were seeded onto the96-well plate at a density of2Â104and allowed to reattach overnight.Next,the cells were starved for 24h with the starvation medium and treated with TNF-a in a concentration of0.25,0.5,1,or2ng/ml for6h,respectively.After treatment,the cells were washed with a wash buffer(20mM Tris pH7.4, 0.15M NaCl,2.5mM CaCl2,and5mg/ml bovine serum albumin).The cells were incubated with40 A l of reaction buffer at378C for30min(37.5nM thrombin and5A g/ml protein C in the washing buffer).The reaction of protein C activation was terminated by adding40A l of mixture(6IU/ml antithrombin and12IU/ml heparin)at378C for5 min.The enzymatic activity of activated protein C was measured by adding50A l of the peptide substrate of Chromozym PCa at378C(H-d-Lys-Z-Pro-Arg-4-nitroanilidediacetate,0.5mM in20mM Tris,pH7.4,0.15mM NaCl and5mg/ml bovine serum albumin).The absorbance was measured at 405nm with EL800Universal Microplate Reader (Bio-Tek Instruments)within2h.APC(American Diagnostica)was used as a positive control. Statistical analysisAll data were presented as mean F S.E.Where indicated,comparisons between two groups were made by paired Student’s t-test(two tails,Excel program,Microsoft)A P b0.05was considered statistically significant.ResultsEffects of TNF-A on TM mRNA expression in human endothelial cellsTM mRNA levels of HUVECs were determined by real-time PCR.In response to a concentration-dependent treatment of TNF-a for6h(Fig.1A), TM mRNA expression,as normalized with h-actin mRNA,was dramatically decreased.As compared to controls(100%),TM mRNA levels were14%,11%, 9%,and6%with the treatment of TNF-a at0.25, 0.5,1,and2ng/ml,representing reductions by 86%,89%,91%,and94%,respectively(P b0.05).As treated with TNF-a(2ng/ml)for different times (Fig.1B),TM mRNA was decreased by80%,97%, 94%,and97%at3,6,12,and24h,respectively (P b0.05).To determine whether TNF-a could affect TM mRNA expression in other types of endothelial cells which represent different ana-tomic locations,HCAECs,HMVECs-L,and HMECs,as well as HUVECs,were treated with TNF-a(2ng/ml) for6h,and TM mRNA levels were determined by real-time PCR(Fig.1C).As compared to controlsB.Nan et al.420(100%),TM mRNA was reduced to 7%,12%,28%,and 43%of HUVECs,HCAECs,HMVECs-L,and HMECs,representing reductions by 93%,88%,72%,and 57%,respectively (P b 0.05).In control conditions with-out TNF-a treatment,HMECs had much higher TM expression as compared to other types of cells.Thus,TNF-a significantly reduces TM mRNA expres-sion in human endothelial cells.Effects of TNF-A on TM protein levels and activity in HUVECsTo determine whether TNF-a could reduce TM protein levels in HUVECs,the cell culture super-natants were collected and analyzed for soluble TM protein levels by ELISA (Fig.2).With treatment ofTNF-a (2ng/ml)for 6h,the soluble TM antigen level was reduced by 29%as compared to controls (P b 0.05).Cellular TM protein levels were deter-mined by Western blot analysis (Fig.3),showing a 68%reduction in TNF-a (2ng/ml)-treated group as compared to controls (P b 0.05).Additionally,cel-lular TM activity was also determined by measuring the thrombin-catalyzed activation of protein C in HUVECs (Fig.4).After treatment withdifferentFigure 1Effects of TNF-A on TM mRNA expression in human endothelial cells.Total RNA was extracted from endothelial cells exposed to media with TNF-A .TM mRNA levels were measured by quantitative real-time PCR.Data were presented by the relative amount of mRNA normalized by B -actin and the percentage change compared with control.Relative mRNA level was presented as 2[Ct(B actin)ÀCt(gene)].Data are expressed as mean F S.E.of triplicate values.(A)TM mRNA level in concentration-dependent manner .HUVECs were exposed to media with indicated doses of TNF-A for 12h.TM mRNA was significantly diminished by TNF-A as com-pared to control group (*P b 0.05,n =3).(B)TM mRNA level in time-dependent manner.HUVEC cell was incubated with media with 2ng/ml TNF-a for indicated times.TM mRNA was significantly reduced within 3h and remained low until 24h incubation (*P b 0.05,n =3).(C)TM mRNA level in different endothelial cell lines.Four different endothelial cells were exposed to media with 2ng/ml TNF-a for 6h.Downregulation of TM expression was shown in different cell lines (*P b 0.05,n =3).Data represent the mean F S.E.from three inde-pendent experiments.HUVEC:Human umbilical venous endothelial cell;HCAEC:human coronary arterial endo-thelial cell;HMVEC-L:Human lung microvascular endo-thelial cell;HMEC:Human microvascular endothelial cell;TM:thrombomodulin.Figure 2Effect of TNF-A on soluble TM protein level by ELISA analysis.HUVECs were treated by 2ng/ml TNF-A for 6h,and the supernatant was detected for soluble TM antigen by IMUBIND ELISA kit.TM antigen level was reduced after TNF-A treatment compared to the control (*P b 0.05,n =3).Data shown are the mean F S.E.from three independent experiments.Effects of TNF-a and curcumin on the expression of thrombomodulin and EPCR 421TNF-a(1and2ng/ml)for6h,TM activity was significantly reduced by22%to23%as compared to controls(P b0.05).Thus,TNF-a significantly reduces both TM protein levels and TM activity in HUVECs.Effect of the TNF-A on EPCR mRNA expression in human endothelial cells EPCR is a crucial factor involved in the TM and thrombin-induced protein C activation.To deter-mine whether TNF-a could affect the expression of EPCR,HUVECs were treated with TNF-a similar to the TM experiments described above.As compared to controls(100%),the treatment of TNF-a for6h significantly reduced EPCR mRNA levels in a con-centration-dependent manner(Fig.5A),showing a 35%reduction in the2ng/ml TNF-a treated cells as compared to controls(P b0.05).In addition to HUVECs,the effect of TNF-a on EPCR mRNA expression was also studied in HCAECs,HMVECs-L, Figure3Effect of TNF-A on TM protein levels byWestern blot analysis.(A)HUVEC cell was treated byTNF-A.Protein(40M g)per well was separated by12%SDS-PAGE.TM expression was detected by monoclonal mouseantihuman TM.The density of TM signal was reduced byTNF-A.(B)The ratio of TM/B-actin band density showedreduction in TNF-A treatment group compared to thecontrol(*P b0.05,n=3).42kDa h-actin is used as a loadingcontrol.Data are expressed as mean F S.E.of triplicatevalues.Figure4Effect of TNF-A on TM activity.TM activity inintact HUVEC cell was measured by the change of theactivated protein C level using a two-stage as in theMaterials and methods.TM activity was reduced in aconcentration-dependent manner(*P b0.05,n=3).Valuesrepresent the mean F S.E.of threedeterminations.Figure5Effects of TNF-A on EPCR mRNA expression inhuman endothelial cells.Different endothelial cells wereexposed to TNF-A.EPCR mRNA was determined byquantitative real-time PCR.Data were presented by therelative amount of mRNA normalized by B-actin and thepercentage change compared with control.RelativemRNA level was presented as2[Ct(B actin)ÀCt(gene)].Dataare expressed as mean F S.E.of triplicate values.(A)TNF-A concentration-dependent study in EPCR mRNA levels.(B)EPCR mRNA expression in different cell lines.EPCRexpression in TNF-A-treated endothelial cells was down-regulated except for HMECs(*P b0.05,n=3).B.Nan et al. 422and HMECs (Fig.5B).In the control condition without TNF-a treatment,HUVECs had a much higher EPCR mRNA expression than other types of cells.Following TNF-a (2ng/ml)treatment for 6h,EPCR mRNA levels were significantly reduced by 36%,39%,and 46%in HUVECs,HCAECs,and HMVECs-L,respectively,as compared to controls (P b 0.05).However ,TNF-a increased EPCR mRNA by 27%in HMECs as compared to controls.Effects of curcumin on TNF-A -induced downregulation of TM and EPCR in HUVECsSince curcumin has an anti-inflammatory effect,we attempted to determine whether curcumin could block TNF-A -induced downregulation of TM and EPCR expression.Three different concentra-tions of curcumin (5,10,and 25M M)with TNF-A (2ng/ml)were used to treat HUVECs for 6h,and both TM and EPCR mRNA levels were determined by real-time PCR.Curcumin (5,10,and 25M M)significantly increased TM mRNA levels by 127%,286%,and 229%,respectively,as compared to the TNF-A -treated group (P b 0.05.Fig.6A).Furthermore,curcumin (5,10,and 25A M)also significantly increased EPCR mRNA levels by 57%,66%,and 112%,respectively,as compared to the TNF-a treated group (P b 0.05,Fig.6B).Thus,curcumin effectively blocks TNF-a -induced downregulation of TM and EPCR in HUVECs in a concentration-dependent manner.DiscussionIn this study,we show that TNF-a significantly decreased TM expression in a concentration-or time-dependent manner in human endothelial cells,including HUVECs,HCAECs,HMVEC-L,and HMECs.Both mRNA and protein levels,as well as biological activity of TM,were decreased in TNF-a -treated cells.Soluble TM in the culture supernatant of TNF-a -treated cells was also decreased,respec-tively.EPCR mRNA levels were significantly reduced in the TNF-a -treated cells,including HUVECs,HCAECs,and HMVEC-L,but not HMECs.Further-more,curcumin concentration-dependently blocked TNF-a -induced TM and EPCR downregua-tion.The data from this study suggests a new molecular mechanism of TNF-a -associated throm-bosis and a new strategy to prevent this problem.TM on the endothelial cell surface functions as a receptor for thrombin-mediated activation of pro-tein C,providing an anticoagulant property.Down-regulation of TM impairs normal endothelial function and promotes thrombosis formation.Our data demonstrate that TNF-a has potent effects on the downregulation of TM mRNA expression in several human endothelial cells,including HUVECs,HCAECs,HMVECs-L,and HMECs.Protein levels and activity of TM were also significantly decreased after TNF-a treatment in HUVECs.Our findings are consistent with previous reports [25—27].Muruge-san et al.[28]reported that TM mRNA reduction by TNF-a was greater (~À50—80%)in microvascular and pulmonary artery EC than in aortic EC (~30%).However,our data showed additional potent effects of TNF-a on TM downregulation.Further-more,our study used more cell types.TNT-a treatment (2ng/ml for 6h)reduced TM mRNA by 93%,88%,72%,and 57%in HUVECs,HCAECs,HMVECs-L,and HMECs,respectively.HMECs had much higher basal TM expression than other types of cells.It is not clear whether the downregulation of TM mRNA occurs at the transcriptional level orasFigure 6Effect of curcumin on TNF-A induced dowre-gualtion of TM and EPCR.HUVEC cell was incubated with various concentrations of curcumin for 30min before adding TNF-A .The mRNA expression was detected by quantitative real-time PCR.Results are expressed as the percentage of the control value after normalization to B -actin mRNA levels and represent the mean from three independent experiments.Data are expressed as mean F S.E.of triplicate values.(A)Curcumin partially reversed the effect of TNF-A on TM expression (*P b 0.05,n =3).(B)EPCR mRNA was increased significantly by pretreatment with curcumin (*P b 0.05,n =3).Effects of TNF-a and curcumin on the expression of thrombomodulin and EPCR 423a function of the mRNA degradation rate.It has been reported that TNF-a primarily inhibits the TM gene transcription[25,29].However,Scarpati and Sadler[30]reported that TNF-a reduced TM activity approximately80%with no change in TM mRNA levels in HUVECs.This discrepancy may be due to the different culture conditions and assay methods used in the different investigations.TNF-a could promote the endocytosis and degradation of TM[30,31].The protein C anticoagulant pathway is critical for controlling microvascular thrombosis and is initiated when thrombin binds to TM on the surface of endothelial cells.Protein C activation is potentially augmented by an EPCR,which is shed from the endothelium by inflammatory mediators and thrombin[32].Impairment of the APC path-way enhances thrombin formation from prothrom-bin,thereby promoting thrombosis.The role of TNF-a and other inflammatory mediators in EPCR expression in human endothelial cells is not clear. With quantitative real-time PCR,our data showed that TNF-a downregulated the expression of EPCR mRNA in a concentration-dependent manner.How-ever,the extent of EPCR mRNA decrease was less than that of TM.In addition,EPCR mRNA down-regulation was delayed as compared to TM mRNA in response to TNF-a.Fukudome and Esmon[12] reported that EPCR mRNA was detected at high levels only in HUVECs among the tested cell lines. Our results demonstrated that four different types of human endothelial cells had EPCR mRNA expression,and HUVECs had higher expression than the others.TNF-a treatment reduced EPCR mRNA levels in HUVECs,HCAECs,and HMVEC-L, but not HMECs.In fact,TNF-a increased EPCR mRNA by27%in HMECs.Thus,TNF-a has differ-ential effects on TM and EPCR expression in human endothelial cells.Curcumin is a major constituent of the spice turmeric from the plant Curcuma longa and is a potent agent with diverse functions,including antioxidation,antithrombosis,and anti-inflamma-tion[21].Curcumin effectively inhibited endothe-lial cell proliferation and the tube formation in vivo [33,34].Recently,several studies have shown that curcumin can antagonize actions of certain cyto-kines such as verotoxin-(VT-1)and TNF-a in human endothelial cells through inactivation of transcrip-tional factors NF n B and AP-1[19—21,35].Curcumin also can block leukocyte recruitment by inhibition of TNF-a-induced monocytes adhering to endothe-lial cells as well as the cell surface expression of ICAM-1,VACM-1,and ELAM-1in HUVEC[23].For the first time,our data elucidates that curcumin inhibited the TNF-a-induced TM and EPCR mRNA downregulation in a concentration-dependent man-ner.Our study and other published data suggest that curcumin is a very promising nonsteroidal anti-inflammatory drug,which may have potent appli-cations in the treatment of vascular diseases.Endothelial cells derived from different ana-tomical locations may have different biological properties and responses to inflammation.Indeed, human aorta endothelial cells have much higher basal ICAM-1levels than endothelial cells from human vena cava and dermal microvessels[36]. HUVECs were mainly used in this study to test the effect of TNF-a and curcumin on TM and EPCR.In order to verify the finding in HUVECs,other types of human endothelial cells were also included in the present study.HCAEC and HMVEC-L expressed slightly lower TM and significant lower EPCR mRNA levels than HUVECs at the same culture condi-tions.TNF-a reduced TM and EPCR expression in these cell types.However,HMECs showed differ-ent properties and responses to TNF-a treatment. HMECs expressed much higher TM and significant lower EPCR mRNA levels than HUVECs.TNF-a significantly reduced TM expression but slightly increased EPCR levels in HMECs.However,clinical implications of the diversity of human endothelial cells have not been investigated.It is possible that different types or locations of endothelial cells involve different mechanisms in vascular lesion formation.Specifically,it is not clear whether this difference between HUVECs and other types of cells is due to the different origin (venous cells)or age.To address this question,the study should include more control cells such as adult and pediatric artery and vein endothelial cells.This could be an interesting investigation in the future.HMEC is an SV-40T transfected human micro-vascular endothelial cell line that maintains many endothelial characteristics,including constitutive expression of von Willebrand factor(vWF)and TM [37,38].Tissue factor(TF)can be induced in response to stimulation with TNF-a,interleukin-1, and phorbol12-myristate13-acetate(PMA)[38]. This cell line has been extensively used as a model to investigate coagulation/fibrinolytic properties and gene expression of microvascular endothelium [39—41].Because HMECs are immortalized cells,it is not clear whether these results generated from HMECs are consistent with those from normal endothelial cells from human dermal microvessels. This could be a limit of these data.Direct comparison between the HMEC line and primary cells is warranted in future study.In summary,our data showed that TNF-a signifi-cantly decreased the expression of TM and EPCR inB.Nan et al.424。