Thrombin-induced Autoinhibitory Factor, Down Syndrome Critical Region-1, Attenuates NFAT-dependent Vascular Cell Adhesion Molecule-1 Expression and Inflammation in the Endothelium*

Activation and dysfunction of the endothelium underlie many vascular disorders including atherosclerosis, tumor growth, and inflammation. We recently reported that thrombin and vascular endothelial growth factor, but not tumor necrosis factor-α, results in dramatic up-regulation of Down syndrome critical region (DSCR)-1 gene in endothelial cells, a negative feedback regulator of calcineurin-NFAT signaling. Constitutive expression of DSCR-1 in activated endothelial cells markedly impaired NFAT nuclear localization, proliferation, tube formation, and tumor growth. The goal of the present study was to elucidate the relative roles of NFAT/DSCR-1 and NF-κB/I-κB in mediating thrombin-responsive gene expression in endothelial cells. DNA microarrays of thrombin-treated human umbilical vein endothelial cells overexpressing DSCR-1 or constitutive active IκBα revealed genes that were dependent on NFAT and/or NF-κB activity. Vascular cell adhesion molecule-1 was inhibited both by DSCR-1 and I-κB at the level of mRNA, protein, promoter activity, and function (monocyte adhesion). Using a combination of transient transfections, electrophoretic mobility shift assays, and chromatin immunoprecipitation, thrombin was shown to induce time-dependent coordinate binding of RelA and NFATc to a tandem NF-κB element in the upstream promoter region of vascular cell adhesion molecule-1. Together, these findings suggest that thrombin-mediated activation of endothelial cells involves an interplay between NFAT and NF-κB signaling pathways and their negative feedback inhibitors, DSCR-1 and I-κB, respectively. As natural brakes in the inflammatory process, DSCR-1 and I-κB may lend themselves to therapeutic manipulation in vasculopathic disease states.

The endothelium is highly malleable cell layer constantly responding to changes within the extracellular environment and responding in ways that are usually beneficial but at times harmful to the organism. Several extracellular mediators activate gene transcription in endothelial cells, resulting in changes in hemostatic balance, increased leukocyte adhesion, permeability, migration, and proliferation. The tight control of these processes is essential for maintaining homeostasis; endothelial cell activation, if excessive, over-sustained, or spatially and temporally misplaced, may lead to vasculopathic disease such as pathological angiogenesis, inflammation, and atherosclerosis.
Under normal conditions, activation signals may be terminated by negative feedback inhibition of downstream transcriptional networks. Such a paradigm is well established for transcription factors, NF-B, Egr-1, and Smad-2/3, which activate the self-inhibitory molecules, IB␣, nerve growth factor induced A-binding protein (NAB), and Smad-7, respectively (1)(2)(3). Moreover, we have recently demonstrated that vascular endothelial growth factor (VEGF) 3 -and thrombin-mediated activation of calcineurin-NFAT signaling in endothelial cells is autoinhibited by the induction of the Down syndrome critical region (DSCR)-1 gene (4).
In humans, the DSCR-1 gene (also known as calcipressin 1, MCIP-1, and Adapt 78) consists of 7 exons, of which exons 1-4 can be alternatively spliced, resulting in a number of different mRNA isoforms (5,6). These isoforms have different expression patterns and are regulated by distinct transcriptional mechanisms. For example, the DSCR-1 exon 4 -7 variant is regulated by a calcineurin-and GATA-dependent pathway (4), whereas the exon 1 variant is under the control of a Notch-and Hes-1-dependent pathway (7).
Thrombin is a multifunctional serine protease that is involved not only in mediating the cleavage of fibrinogen to fibrin in the coagulation cascade but also in activating a variety of cell types, including platelets, vascular smooth muscle, and endothelial cells (8 -10). Thrombin signaling in the endothelium is mediated by a family of seven transmembrane G-protein-coupled receptors, termed protease activated receptors (PARs). Currently, four members of the PAR family have been isolated (PAR-1-4). In endothelial cells PAR-1 is thought to be predominant thrombin receptor. Once activated, PAR-1 is linked to a number of signal intermediates (including mitogenactivated protein kinase, protein kinases C and A, phosphatidylinositol 3-kinase, and calcineurin), transcription factors (such as Egr-1, NF-B, NFAT, SP-1, and AP-1), and downstream target genes (e.g. vascular adhesion molecule (VCAM)-1, intercellular adhesion molecule (ICAM)-1, E-selectin, fractalkine, and monocyte chemoattractant protein-1, endothelial-specific molecule-1, bone morphogenetic protein-2, and HB-EGF) (4,(11)(12)(13).
In this study we employed global gene expression profiles to elucidate the relative roles of NFAT-DSCR-1 and NF-B-IB pathways in mediating thrombin signaling in endothelial cells. We identified several thrombin target genes that require both transcription factors for activation. Among these was the proinflammatory gene, VCAM-1, which is reported to play a critical role in firm leukocyte adhesion in diverse inflammatory disease states (14 -17). Additional studies were carried out to determine the interaction between NFAT and NF-B in mediating thrombin induction of VCAM-1. The results suggest that DSCR-1-mediated inhibition of NFAT signaling may be leveraged for therapeutic gain in inflammatory states.
Plasmids and Adenoviruses-Construction of wild type human VCAM-1 luc and Ϫ251NFAT-like-mut-luc was previously described (11). To generate Ϫ216NFAT-like mut-luc and NF-B-mut-luc, a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used according to the manufacturer's instruction with the mutated oligonucleotides shown in supplemental Table I. To generate the mouse VCAM-1-luc plasmid, the mouse 2.5-kilobase promoter was amplified by PCR using the BAC clone (RPCI23 362K11) as template and subcloned into the pGL3-basic vector (Promega, Madison, MI). NFAT expression vector (pCMV-NFATc) was purchased from Invitrogen (IMAGE clone ID 5354603). The RelA expression plasmid (pCMV-RelA) was kindly provided by Dr. Mukesh Jain (Harvard Medical School, Boston, MA). For construction of pTNT-mNFATc, a SalI-NotI fragment containing mNFATc was subcloned into the SalI/NotI-digested pTNT vector (Promega). Human IB␣ cDNA fragment was amplified using PCR from reversetranscribed HUVEC total RNA and subcloned into the pGEM-T easy (Promega). Constitutive active (CA; S32A and S36A) IB␣ was generated using the QuikChange site-directed mutagenesis kit (Stratagene) and mutated primer pairs (supplemental Table I). The EcoRI-digested CA-IB␣ fragment was subcloned into pIRES2-EGFP (Clontech, Mountain View, CA) to generate pIRES2-CA-IB␣-EGFP. The generation of adenovirus (Ad)-Control, Ad-DSCR-1, and Ad-CA-NFAT was previously described (4). For the construction of the Ad-CA-IB␣, pIRES2-CA-IB␣-EGFP was digested with NheI and AflII, and the resulting fragment was subcloned into the pShuttle and Adeno-X DNA (Clontech) using the Adeno-X adenoviral expression system (Clontech). All cloned and subcloned constructs were confirmed by restriction enzyme digestions and automated DNA sequencing.
Microarray Analysis-Two independent lots of HUVEC were infected with either Ad-Control, Ad-DSCR-1, Ad-CA-NFAT, or Ad-CA-IB␣ (multiplicity of infection of 30) for 2 days, serum-starved overnight in medium containing EBM-2 (Clonetics) and 0.5% FBS, and then treated with or without 2 units/ml thrombin for 4 h. RNA was harvested and purified with Trizol (Invitrogen). Preparation of cRNA and hybridization of probe arrays were performed according to the manufacturer's instructions (Affymetrix, Santa Clara, CA). Each array experiment was performed duplicate. File maker software was used for the genes that demonstrated identical patterns in both experiments. Data were analyzed according to the minimum information about a microarray experiment (MIAME) rule. Annotation of the probe numbers and targeted sequences were shown in Affymetrix web page.
Quantitative Real-time PCR-RNA was extracted from endothelial cells with TRIzol reagent (Invitrogen). Two g of total RNA was reverse-transcribed using SuperScript II enzyme using oligo-dT primer as specified by Invitrogen. Real-time PCR including SYBR Green PCR reagent was performed on an instrument according to instructions provided by the manufacturer (Applied Biosystems). Primer pairs are indicated in supplemental Table I.
Plasmid DNA and siRNA Transient Transfections and Luciferase Assay-HUVEC and human embryonic kidney 293 cells were transfected with plasmid DNA using FuGENE 6 reagent (Roche Applied Science), and luciferase activities were calculated using the dual-luciferase assay kit (Promega) as previously described (11). The serum-starved transfected HUVEC were incubated with 50 ng/ml VEGF, 2 units/ml thrombin, and/or 10 ng/ml TNF-␣ for 6 h. For siRNA transfection, HUVEC were incubated with a mixture of 40 nM siRNA and lipofectamine2000 (Invitrogen) for 6 h. Medium was replaced, and cells were cultured with EGM-2-MV for 24 h. After serum-starvation (EBM-2 plus 0.5% FBS) and agonist treatment, cells were processed for real-time PCR, Western blot, or adhesion assays. Targeted siRNA sequences (Invitrogen) are shown in supplemental Table I.
Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extraction and EMSA were carried out as previously described (18). To reduce nonspecific binding, 1 g of poly(dI-dC) (Invitrogen) and 20 fmol of oligonucleotide spanning the flk-1/KDR SP1 motifs (supplemental Table I) were added to the reaction mixture. Double-stranded oligonucleotides were labeled with [␣-32 P]dCTP and Klenow fragment and purified by spin column (Amersham Biosciences). Oligonucleotide sequences used for probes and competitors are shown in supplemental Table I. To test the effect of antibodies on DNA-protein binding, nuclear extracts were preincubated with antibodies against NFATc (Affinity BioReagents, Golden, CO), Ets-1, RelA, and IgG (Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min at room temperature. Recombinant NFATc and RelA were generated using the TNT-coupled wheat germ extract systems (Promega) with pTNT-mNFATc and pCMV-RelA, respectively, according the manufacturer's instruction.

DSCR-1 Negatively Regulates Vascular Inflammation
room temperature. The cells were washed twice with icecold phosphate-buffered saline containing protease inhibitor mixture (Roche Applied Science), then scraped and collected by centrifugation. The cells pellets were resuspended in SDS lysis buffer containing protease inhibitors (Upstate Biotechnology, Lake Placid, NY) and incubated for 10 min. The cross-linked chromatin was subjected to sonication using BioRuptor (Cosmobio, Tokyo, Japan) to obtain DNA fragments of 200 -500 bp. The chromatin complexes were collected by centrifugation at 4°C and diluted 6-fold in ChIP dilution buffer containing protease inhibitor (Upstate). Preclear experiments were employed in the addition of 100 l of salmon sperm DNA/protein A (Upstate) (for RelA) or salmon sperm DNA/protein G (Upstate) (for NFATc1). The samples were immunoprecipitated with 8 g of anti-RelA, anti-Pol II (Santa Cruz), anti-NFATc1 (Affinity Bioreagents), or control IgG (Santa Cruz). Precipitated chromatin complexes were washed and eluted according to the ChIP assay kit instructions (Upstate). PCR and real-time PCR reactions were performed with the primer pairs as shown in supplemental Table I.
Western Blot Analysis-HUVEC were washed with ice-cold phosphate-buffered saline, collected with a cell scraper, and lysed with radioimmune precipitation assay buffer (50 mM Tris-HCl, 150 mM NaCl, 0.5% IGE-PAL CA-630, 1 mM EDTA) containing protease inhibitor mixture tablet (Roche Applied Science). Whole cell lysates (20 g for detection of human VCAM and 10 g for detection of IB␣ and DSCR-1) were separated by SDS-PAGE and transferred to Hybond-P polyvinylidene difluoride membranes (Amersham Biosciences Bioscience) using the semidry blotting method. The membrane was blocked with TBS-T containing 2% skim milk and incubated with primary antibodies against VCAM-1, IB␣, RelA (Santa Cruz), NFATc (Affinity Bioreagents), or DSCR-1 (we generated the monoclonal antibody using the DSCR-1, 94 -197 amino acids fragment). The membrane was washed 3 times with TBS-T and incubated with peroxidase-conjugated secondary antibodies. The signals were detected with WestDura SuperSignal chemiluminescence kit (Pierce). To control for loading, the membranes were stripped for 30 min at 55°C with striping buffer (62.5 mM Tris-HCl, 2% SDS, 100 mM ␤-mercaptoethanol), washed with TBS-T, and re-probed using anti-␤-actin antibody (Sigma).
Monocyte Adhesion Assay-HUVEC were infected with adenovirus and seeded on 24-well tissue culture plates. After grown to confluence, HUVEC were incubated with EBM-2 containing 0.5% FBS for 18 h, then treated in the presence or absence of 50 ng/ml VEGF and 2 units/ml thrombin for 6 h. U937 cells were labeled with red fluorescent dye using PKH-26 staining kit (Sigma), and a total of 3 ϫ 10 5 cells was added to each well of a HUVEC-seeded 24-well plate. 90 min later cells were washed twice with Hanks'-buffered saline (Invitrogen) and examined by fluorescent microscopy. The adhesion intensities were measured with MetaMorph (Molecular Devices Corp., Sunnyvale, CA) and cell image analyzer (Kurabo, Tokyo, Japan).

Identification of NFAT and NF-B-dependent Pathways in
Thrombin-treated Human Primary Endothelial Cells-In a previous study we demonstrated that thrombin and VEGF treatment of HUVEC resulted in marked and rapid up-regulation of the DSCR-1 gene and secondary inhibition of NFAT signaling (4). Other studies have demonstrated an important role for NF-B in thrombin-mediated gene expression (19,20). To determine the relative contributions of these two signaling pathways (NFAT and NF-B) and their respective endogenous inhibitors (DSCR-1 and IB) in thrombin signaling, we carried out DNA microarray analysis of control or thrombin-treated HUVEC overexpressing EGFP (control), DSCR-1, or constitutively active IB␣. High transduction efficiency was confirmed using fluorescent microscopy, real-time PCR, and Western blots ( Fig. 1A and data not shown).
Next, we employed real-time PCR to validate the results of two representative genes from each of the above groups. In group 1, thrombin treatment resulted in significant induction of the cell adhesion molecules, ICAM-1 and E-selectin by 52-and 832-fold, respectively ( Fig. 2 A and B) (a third adhesion molecule belonging to group 1 gene, VCAM-1, is discussed in detail below). Thrombin-mediated induction of ICAM-1 was inhibited 50% by overexpression of DSCR-1 and 87% by CA-IB␣. E-selectin up-regulation was inhibited 62% by the overexpression of DSCR-1 and 84% by CA-IB␣. As examples of group 2 genes, thrombin induced the expression of the calmodulin binding domain containing factor, caldesmon-1 (2.3fold) (21) and the pro-angiogenic factor, Cyr 61 (2.1-fold) (22). The effect of thrombin on caldesmon-1 and Cyr 61 was inhibited by Ad-DSCR-1 (46 and 53%, respectively) but not by Ad-CA-IB␣ (Fig. 2, C and D). In group 3, thrombin stimulated the expression of inhibitors of programmed cell death, baculoviral inhibitor of apoptosis protein (IAP) repeat-containing-3 (40-fold) (23), and A20 (7.0-fold) (24), effects that were attenuated by NF-B blockade (84 and 67%, respectively) but not by DSCR-1 (Fig. 2, E and F ).
Thrombin-mediated Up-regulation of VCAM-1 mRNA and Protein in Human Primary Endothelial Cells Is Dependent on NFAT and NF-B Activation Pathways-According to the DNA microarray data, VCAM-1 is contained within the Group 1 cluster. We have previously shown thrombin-mediated induction of VCAM-1 in endothelial cells involves the coordinate binding of NF-B and GATA to tandem motifs in the upstream promoter region (25). To confirm whether NF-B and NFAT are also involved in thrombin stimulation

DSCR-1 Negatively Regulates Vascular Inflammation
of VCAM-1, we carried out real-time PCR of thrombin-stimulated HUVEC infected with Ad-Control, Ad-DSCR-1, and/or Ad-CA-IB␣. As shown in Fig. 3A, thrombin treatment resulted in 452-fold up-regulation of the VCAM-1 mRNA at 4 h, an effect that was inhibited by Ad-DSCR-1 and Ad-CA-IB␣ (63 and 89%, respectively). Combined overexpression of DSCR-1 and CA-IB␣ resulted in a nearly complete (97%) inhibition of the thrombin effect. Consistent with the effect of Ad-DSCR-1, overexpression of constitutive nuclear-localized NFAT (CA-NFAT) in HUVEC resulted in an 18-fold induction of VCAM-1 expression (Fig. 3B).
Next, we wished to determine the effect of DSCR-1 knockdown on VCAM-1 expression. To that end HUVEC were transfected with siRNA against DSCR-1. As shown in Fig. 3, C and D, two independent siRNAs inhibited thrombin-stimulated DSCR-1 mRNA (87 and 69%, respectively) as well as DSCR-1 protein. Importantly, both siRNAs resulted in a super-induction of VCAM-1 mRNA and protein levels in thrombin-treated HUVEC (Fig. 3, E and F). Similarly, siRNAs against IB␣ accentuated the effect of thrombin on VCAM-1 mRNA expression (supplemental Fig. I). In contrast, siRNA against RelA and NFATc (whose effectiveness in inhibiting their target genes is shown in Fig. 3, G-J) virtually abrogated thrombin-mediated induction of VCAM-1 mRNA (Fig. 3K). Taken together, these data strongly suggest that thrombin stimulation of VCAM-1 is mediated by RelA-and NFATc-dependent signaling pathways and that these pathways are in turn autoinhibited by IB␣ and DSCR-1, respectively.

Thrombin-mediated Up-regulation of VCAM-1 Promoter in Human Primary Endothelial Cells Is Dependent on NFAT and
NF-B Activation Pathways-To confirm the above results at the level of the VCAM-1 promoter, HUVEC were infected with Ad-Control, Ad-DSCR-1, Ad-CA-IB␣, or Ad-CA-NFAT and then transiently transfected with constructs containing the thymidine kinase promoter (control), the human VCAM-1 promoter (1800 bp 5Ј-flanking sequence), or the mouse VCAM-1 promoter (2500-bp 5Ј-flanking sequence) coupled to luciferase cDNA. In contrast to an absence of effect on the thymidine kinase promoter, overexpression of CA-NFAT resulted in a 3-fold induction of mouse and human VCAM-1 promoter activity (Fig. 4A). In co-transfection assays overexpression of NFATc or RelA resulted in a 1.67-and 24.5-fold induction of VCAM-1 promoter activity, respectively. Combined overexpression of NFATc and RelA had a synergistic effect on luciferase activity (42.1-fold induction) (Fig. 4B). Thrombin induced human VCAM-1 promoter activity by 4.5-fold, an effect inhibited by 64 and 92% in cells overexpressing DSCR-1 or CA-IB␣, respectively (Fig. 4C). Taken together with the results of DNA microarrays, real-time PCR, Western blots, and transient transfections, these data support a role for both NF-B/IB␣ and NFAT/DSCR-1 feedback pathways in the temporal regulation of VCAM-1 gene expression.
NFAT Stimulation of the VCAM-1 Promoter Is Mediated by the NF-B Motif-The human VCAM-1 promoter contains NFAT consensus motifs ((A/T)GGAAA) at Ϫ443, Ϫ251, and Ϫ216 relative to the transcriptional start site (Fig. 5). To determine whether one or more of these sites is required for NFATmediated transactivation of the VCAM-1 promoter, human embryonic kidney 293 cells were co-transfected with an NFATc expression plasmid (pCMV-NFATc) and VCAM-1 promoter-  JULY 21, 2006 • VOLUME 281 • NUMBER 29 luciferase constructs containing various mutations/deletions. As shown in Fig. 5A, NFATc expression resulted in a 9-fold induction of the wild type promoter. This effect was unaltered by deletion of the 5Ј-most NFAT element (⌬-289-luc) or mutation of the Ϫ251 or Ϫ216 NFAT sites. In contrast, a promoter containing mutated NF-B sites (NF-B mut-luc) was unresponsive to NFAT (Fig. 5A). In other co-transfection studies, overexpression of RelA resulted in significant induction (58-fold) of the wild type promoter but not NF-B mut-luc (Fig. 5B). Taken together, these results suggest that the NF-B motifs are necessary for NFAT-and NF-B-mediated transactivation of the VCAM-1 promoter.

DSCR-1 Negatively Regulates Vascular Inflammation
Thrombin Induces Binding of NFAT to NF-B Motifs in the VCAM-1 Promoter-EMSAs were carried out to characterize NFAT and NF-B DNA binding interactions. To that end, a radiolabeled probe containing the two adjacent NF-B sites was incubated with nuclear extracts derived from control or thrombintreated HUVEC. As shown in Fig. 6A, thrombin treatment resulted in the appearance and/or induction of three DNA-protein complexes (lanes 2 and 3; open arrow, closed arrow, and arrowhead). These DNA-protein complexes were inhibited by the addition of a 20-fold molar excess of unlabeled self-competitor, confirming their specificity (Fig. 6A, lane 4). The fastest of the three complexes (delineated by the open arrow) was inhibited by the addition of a 20-fold molar excess of unlabeled DNA probe containing the consensus NFAT binding site at the Ϫ220 bp of the DSCR-1 promoter (Fig. 6A, lane 5).
To identify the protein(s) in the thrombin-inducible binding complex, the binding reactions were preincubated with antibodies to RelA or NFATc. Anti-RelA antibody inhibited the formation of the slow and intermediate complexes (delineated by the arrowhead and closed arrow, respectively) and resulted in the appearance of a strong supershifted band (Fig. 6, lane 7, asterisk). Preincubation with anti-NFATc antibody resulted in marked reduction of the slow and fast DNA-protein complexes (delineated by the arrowhead and open arrow, respectively) and the appearance of a faint supershifted band (Fig. 6, lane 8, double asterisk), which was more clearly visualized with longer exposure (data not shown). Preincubation with both anti-RelA and anti-NFATc antibodies resulted in a supershift of all three migrating complexes (Fig. 6, lane 9). In contrast, control IgG had no effect on DNA binding (Fig. 6, lane 10).
To provide further evidence for binding of NFAT to the NF-B sites, EMSA were carried out with nuclear extracts derived from HUVEC overexpressing DSCR-1 or pretreated with the NFAT inhibitor, CsA. As shown in Fig. 6B, thrombin treatment of HUVEC infected with Ad-Control resulted in the induction of three specific DNA-protein complexes (lanes 1 and 2). In contrast, nuclear extracts from thrombin-treated HUVEC overexpressing DSCR-1 or pretreated with CsA demonstrated marked reduction in the fastest and slowest migrating complexes (Fig. 6B, lanes 3 and 4).
Finally, EMSAs were performed using recombinant NFATc and RelA. As predicted from the above studies, NFATc resulted in a fast migrating specific DNA-protein complex and RelA in a more slowly moving complex (Fig. 6C, lanes 2 and 3, open and  closed arrows). When added together, the two proteins yielded an additional higher DNA-protein complex (Fig. 6C, lane 4,  arrowhead). Collectively, these findings suggest that thrombin induces cooperative binding of NFATc and RelA to the VCAM-1 NF-B motifs.
Thrombin Induces NFAT and RelA Binding to the VCAM-1 NF-B Region in Chromatin Immunoprecipitation Assays-To further demonstrate a role for NFAT in binding to the VCAM-1 promoter, we carried out sequential ChIP analysis (26) using the genomic VCAM-1 region. HUVEC were treated with

DSCR-1 Negatively Regulates Vascular Inflammation
thrombin for 0, 1, 2, and 4 h and then harvested for formalinfixed genomic DNA. The samples were immunoprecipitated with antibodies to RelA, NFATc1, or Pol II, and the resulting products were used as a template in a PCR reaction containing two sets of primers, one specific for the NF-B region (Fig. 7A, region I) and the other specific for a region immediately down-

DSCR-1 Negatively Regulates Vascular Inflammation
stream of the transcriptional start site (Fig. 7A, region II) of the VCAM-1 promoter. Binding intensities were calculated using real-time PCR and ethidium bromide density. Thrombin induced RelA binding to region I with maximal levels (16-fold) occurring at 1-2 h, whereas RelA binding efficiency on region II was weak (only detected 4 h after the thrombin treatment) (Fig.  7A, left). In contrast, maximal NFATc binding on regions I and II occurred at 1 h (7.8-and 7.9-fold, respectively) (Fig. 7A, middle). Thrombin treatment resulted in a gradual induction of Pol II to chromatin DNA spanning both regions of VCAM-1 (Fig.  7A, right). We also performed ChIP analysis in HUVEC transfected with siRNA against DSCR-1 or control siRNA. Compared with Si-Control, Si-DSCR-1 treatment resulted in a ϳ2.5-fold increase of the NFATc binding levels at 0, 1, and 2 h after the thrombin stimulation (supplemental Fig. IIA). In addition, similar NFATc-chromatin DNA association patterns were detected in ChIP analysis using the human DSCR-1 promoter spanning the multi-NFAT binding regions, whereas RelAchromatin DNA interactions were not detected in that region (supplemental Fig. IIB).
To confirm the temporal pattern of thrombin-inducible NFAT and NF-B binding, we carried out a time course study using EMSA. Consistent with the ChIP results, thrombin-mediated induction of NFAT-DNA binding occurred earlier and was more sustained compared with NF-B binding (Fig. 7B). Taken together, these findings suggest that thrombin results in a dynamic, sequential, and temporally regulated interaction between RelA and NFAT at the level of the VCAM-1 NF-B region in endothelial cells.
Agonist-mediated Induction of VCAM-1 in Human Primary Endothelial Cells-We have previously shown that in HUVEC, NFAT is more strongly activated by VEGF and thrombin compared with TNF-␣, whereas NF-B is more strongly activated by TNF-␣ and thrombin compared with VEGF (4). To validate these findings at the level of the promoter, HUVEC were transfected the VCAM-1 reporter plasmid, treated with various agonist combinations, and then assayed for luciferase activity. As shown in Fig. 8A, 50 ng/ml VEGF, 2 units/ml thrombin, and 10 ng/ml TNF-␣ activated the promoter by 1.9-, 4.7-, and 9.2-fold, respectively. The combined addition of VEGF and thrombin or VEGF and TNF-␣ resulted in synergistic stimulation of promoter activity. Co-incubation with thrombin and TNF-␣ resulted in significantly higher luciferase levels compared with either agonist alone. EMSAs were performed using nuclear extracts from HUVEC treated with various combinations of agonists for 4 h. As shown in Fig. 8B, VEGF, thrombin, or TNF-␣ each induced RelA and NFAT binding (lanes 3-5). The relative effects of these agonists on RelA binding were TNF-␣ Ͼ thrombin Ͼ VEGF and for NFAT binding were VEGF Ͼ thrombin Ͼ TNF-␣. Co-incubation of HUVEC with pairs of agonists resulted in increased intensity of RelA and NFATc binding compared with either agonist alone (Fig. 8B, lanes 4 -6).

DSCR-1 and IB␣ Attenuate VEGF-and Thrombin-mediated Monocyte Adhesion to Human Primary Endothelial Cells
-We next wished to determine the functional role for DSCR-1 and IB␣ autoinhibitory pathways in attenuating endothelial cell activation. To that end, we assayed for adhesion of fluorescent (PKH-26)-labeled U937 monocytes to HUVEC. Treatment of HUVEC with 50 ng/ml VEGF and 2 units/ml thrombin for 4 h resulted in a 75-fold induction in monocyte adhesion compared with untreated controls (Fig. 9, A and B). This effect was almost completely blocked (90%) by antibodies against VCAM-1 but not E-selectin or ICAM-1 (Fig. 9, A and B). As reported previously (27), pretreatment with phosphatidylinositol 3-kinase inhibitor (LY294002) resulted in marked (97%) reduction of agonist-mediated monocyte adhesion. Phosphatidylinositol 3-kinase inhibition blocked the effect of VEGF and thrombin on expression levels of VCAM-1 but not E-selectin and ICAM-1 (supplemental Fig. III). Consistent with the VCAM-1 protein data (Fig. 8C), combined treatment with VEGF and thrombin had a far greater effect on monocyte adhesion compared with either agonist alone (data not shown). Taken together, these findings suggest that the effect of VEGF and thrombin on U937 monocyte adhesion is mediated primarily through the inducible expression of VCAM-1 expression on endothelial cells.
Overexpression of DSCR-1 or CA-IB␣ alone inhibited agonist-mediated monocyte adhesion by 82 and 73%, respectively. Combined expression of both DSCR-1 and CA-IB␣ completely blocked VEGF and thrombin-mediated monocyte adhesion (Fig. 9, C and D). Overexpression of CA-NFAT resulted in increased monocyte adhesion both to unstimulated and agonist-treated HUVEC (Fig. 9, C and D). HUVEC overexpressing DSCR-1, CA-NFAT, and CA-IB␣ did not demonstrate evidence of toxicity (Fig. 9C, upper panels). Compared with VEGF  7 and 9), antibodies against NFATc (lanes 8 and 9), or control IgG (lane 10). The arrows and arrowhead indicate the specific DNA-protein complexes. The asterisks indicate the super-shifted complexes. B, EMSAs were performed with 32 P-labeled probe incubated with nuclear extracts from Ad-Control-infected HUVEC treated in the absence (lanes 1 and 2) 3 and 4). The arrows and arrowhead indicate the specific DNA-protein complexes. The results are representative of at least two independent experiments. and thrombin, TNF-␣ triggered similar levels of monocyte adhesion. However, consistent with the Western blot results, overexpressing DSCR-1 resulted in only a slight reduction (18%) of TNF-␣-mediated monocyte adhesion, whereas overexpression of CA-IB␣ had a far greater effect (78% reduction) (supplemental Fig. IV).

DSCR-1 Negatively Regulates Vascular Inflammation
siRNA-mediated knockdown of DSCR-1 and IB␣ in HUVEC resulted in increased monocyte adhesion (Fig. 9, E and F). In contrast, knockdown of NFATc and RelA markedly reduced agonist-mediated monocyte adhesion (Fig. 9, E and F). Treatment of siRNAs to HUVEC did not demonstrate any morphological abnormalities (supplemental Fig. V). Taken together, these findings suggest that both NF-B and NFAT play critical roles in regulating VEGF and thrombin-mediated cell adhesion to endothelial cells.

DISCUSSION
Activation of endothelial cells by extracellular stimuli, including VEGF, thrombin, and TNF-␣, is a key step underlying many vascular diseases. In a previous study, we demonstrated that VEGF and thrombin resulted in the rapid and pronounced induction of DSCR-1 (4). DSCR-1, in turn, markedly attenuated calcineurin-dependent NFAT signaling, resulting in inhibition of proliferation and angiogenesis. In the present study, we have extended these findings by demonstrating that NFAT, the primary target of DSCR-1, cooperates with NF-B to up-regulate many inflammatory genes and that DSCR-1 and IB function as autoinhibitory factors in these respective pathways.
NF-B family is widely believed to play a critical role in mediating the inflammatory response in endothelium (28). We, and FIGURE 8. Effects of thrombin, VEGF, and TNF-␣ on VCAM-1 expression in HUVEC. A, HUVEC were transiently transfected with VCAM-1-luc plasmid and exposed to 50 ng/ml VEGF, 2 units/ml thrombin, or 10 ng/ml TNF-␣ alone or in combination for 6 h. The results show the mean and S.D. of luciferase light units (relative to the untreated control) obtained in triplicate from five independent experiments. B, EMSA were performed with 32 P-labeled probe incubated in the absence (lanes 1) or presence of nuclear extract from untreated HUVEC (lane 2) or from HUVEC treated with VEGF, thrombin, or TNF-␣ alone (lanes [3][4][5] or in combination (lanes 6 -8) for 4 h. The arrows and arrowhead indicate the specific DNA-protein complexes. C, Western blot analysis of VCAM-1 using 10 g of protein from HUVEC infected with Ad-Control (lanes 1, 2, 4, 6, 8, 10, and 12) or Ad-DSCR-1 (lanes 3, 5, 7, 9, 11, and 13)   others have previously shown thrombin leads to nuclear translocation of RelA and subsequent transactivation of multiple proinflammatory target genes, including VCAM-1, ICAM-1, E-selectin, MCP-1 monocyte chemotactic protein-1, and fractalkine (12,19). In addition, the NF-B family induces the early expression of its inhibitor, IB␣, which serves to dampen further RelA activity (1). IB␣ is susceptible to IKK-mediated phosphorylation and subsequent ubiquitination and degradation. Thus, the non-phosphorylated mutant form of the IB␣ (S32A and S36A; CA-IB␣) functions as a dominant negative factor for the RelA-mediated signaling cascade (29). In our DNA microarrays of HUVEC, overexpression of CA-IB␣ inhibited one-third (54/172) of thrombin-stimulated genes. These data are consistent with an important role for NF-B in thrombin signaling.
In addition, our study strongly supports a role for NFAT in mediating thrombin-responsive gene expression. First, overexpression of NFAT in HUVEC resulted in up-regulation (Ͼ4fold) of 51 genes, including many that were categorized as mediators in cell-growth, inflammation, and signaling (data not

DSCR-1 Negatively Regulates Vascular Inflammation
shown). Second, overexpression of DSCR-1 resulted in downregulation of approximately one-third (52/172) of thrombinstimulated genes in HUVEC. Half of these transcripts (including VCAM-1) were also inhibited by CA-IB␣, suggesting that many thrombin-responsive genes are under the dual regulation of NFAT and NF-B.
VCAM-1 is believed to play a major role in mediating leukocyte adhesion in diverse disease states such as acute sepsis, early step of the atherosclerosis and chronic rheumatoid arthritis (14 -17). We recently reported that thrombin increases the expression of VCAM-1 in endothelial cells by a mechanism that involves the coordinate binding of RelA and GATA to neighboring tandem consensus motifs in the upstream promoter (25). The current study provides compelling evidence for the role of NFAT-DSCR-1 and NF-B-I-B autoinhibitory pathways in mediating thrombin induction VCAM-1 mRNA, protein, and promoter activity. Specifically, the thrombin response was inhibited by overexpression of DSCR-1 and CA-IB␣ and by siRNA against RelA or NFATc, super-induced by siRNA against DSCR-1 or IB␣, and mimicked by overexpression of NFATc and RelA. Importantly, the role for NFAT-DSCR-1 and NF-B-I-B in mediating the effect of thrombin (and VEGF) on VCAM-1 expression was confirmed in functional assays of monocyte adhesion.
Previous studies have shown that agonist-mediated Ca 2ϩcalcineurin-dependent dephosphorylation of NFAT results in the translocation of the transcription factor into the nucleus and secondary induction of multiple target genes. In many cases, NFAT has been shown to cooperate with other transacting factors to promote target gene expression, including GATA, AP-1, Maf, and Mef-2 (30 -33). In addition to binding to classic NFAT consensus elements, NFAT has been shown to bind to an NF-B motif in the interleukin-8 or human immunodeficiency virus promoters (34 -36). Using a combination of transient transfection, EMSA, and ChIP assays, we have shown that NFAT similarly binds to the tandem NF-B motif in the VCAM-1 promoter and positively regulates its expression. However, rather than replacing or competing with RelA, NFAT appears to associate with NF-B function, as evidenced by the DNA complexes in EMSA, sequential ChIP, and the additive/ synergistic effects of Ad-CA-IB␣ and Ad-DSCR-1 on VCAM-1 expression and monocyte adhesion. Further studies using in vivo (co-immunoprecipitation) and in vitro (GST pulldown) assays will be required to identify and characterize the precise nature of the dynamical regulated protein-protein and protein-DNA interaction.
Calcineurin/NFAT signaling has been implicated in cell growth, differentiation and contractile state neuronal cells, cells of the immune system, cardiac or smooth muscle cells, and endothelial cells. CsA and FK506 are calcineurin-specific inhibitors that are approved for clinical use to prevent graft rejection after organ transplantation (37,38). However, widespread use of these drugs is limited by their neurotoxic and nephrotoxic effects (39,40). In endothelial cells, high concentrations of CsA lead to toxic effects, including increased lactate dehydrogenase release (41). We found that whereas 50 nM CsA inhibited thrombin-mediated induction of VCAM-1 but not E-selectin and ICAM-1, treatment with 10 M CsA actually increased basal levels of E-selectin (supplemental Fig. VI). In contrast to CsA, DSCR-1 is an endogenous inhibitor of calcineurin, with a K i in the low nanomolar range (42). Chan et al. (43) recently reported the C-terminal 57 residues of DSCR-1 encoded by exon 7 binds calcineurin with high affinity and inhibits its activity with potency similar to full-length DSCR-1 (43). A 31-amino acid stretch similar to serine-proline repeats of NFAT also inhibits calcineurin activity by acting as a competitive inhibitor (44). Thus, DSCR-1 (and its fragments) appears to hold promise as a more selective and less toxic calcineurin inhibitor for therapy in inflammatory conductions, including vascular disease.
Vascular diseases, including atherosclerosis, thrombosis, and tumor angiogenesis represent complex pathological phenomena. Disease-associated changes in the extracellular environment may have profound effects on endothelial cells, with net signal inputs governing cellular phenotype in a time-dependent manner. Here, we show that thrombin, VEGF, and TNF-␣ serve to induce VCAM-1 expressions via overlapping yet distinct signaling pathways. An important goal is to further understand the relative role of intracellular signaling pathways and transcriptional networks in mediating endothelial cell activation and to use this information for tailoring anti-inflammatory therapies according to the nature of the signal input and the degree of desired attenuation.