Acute phase mediators modulate thrombin-activable fibrinolysis inhibitor (TAFI) gene expression in HepG2 cells.

Thrombin-activable fibrinolysis inhibitor (TAFI) has recently been identified as a positive acute phase protein in mice, an observation that may have important implications for the interaction of the coagulation, fibrinolytic, and inflammatory systems. Activated TAFI (TAFIa) inhibits fibrinolysis by removing the carboxyl-terminal lysines from partially degraded fibrin that are important for maximally efficient plasminogen activation. In addition, TAFIa has been shown to be capable of removing the carboxyl-terminal arginine residues from the anaphylatoxins and bradykinin, thus implying a role for the TAFI pathway in the vascular responses to inflammation. In the current study, we investigated the ability of acute phase mediators to modulate human TAFI gene expression in cultured human hepatoma (HepG2) cells. Surprisingly, we found that treatment of HepG2 cells with a combination of interleukin (IL)-1 and IL-6 suppressed endogenous TAFI mRNA abundance in HepG2 cells (~60% decrease), while treatment with IL-1 or IL-6 alone had no effect. Treatment with IL-1 and/or IL-6 had no effect on TAFI promoter activity as measured using a luciferase reporter plasmid containing the human TAFI 5'-flanking region, whereas treatment with IL-1 and IL-6 in combination, but not alone, decreased the stability of the endogenous TAFI mRNA. Treatment with the synthetic glucocorticoid dexamethasone resulted in a 2-fold increase of both TAFI mRNA levels and promoter activity. We identified a functional glucocorticoid response element (GRE) in the human TAFI promoter between nucleotides 92 and 78. The GRE was capable of binding the glucocorticoid receptor, as assessed by gel mobility shift assays, and mutation of this element markedly decreased the ability of the TAFI promoter to be activated by dexamethasone.

From the Departments of ‡Biochemistry and §Medicine, Queen's University, Kingston, Ontario K7L 3N6, Canada Thrombin-activable fibrinolysis inhibitor (TAFI) has recently been identified as a positive acute phase protein in mice, an observation that may have important implications for the interaction of the coagulation, fibrinolytic, and inflammatory systems. Activated TAFI (TAFIa) inhibits fibrinolysis by removing the carboxylterminal lysines from partially degraded fibrin that are important for maximally efficient plasminogen activation. In addition, TAFIa has been shown to be capable of removing the carboxyl-terminal arginine residues from the anaphylatoxins and bradykinin, thus implying a role for the TAFI pathway in the vascular responses to inflammation. In the current study, we investigated the ability of acute phase mediators to modulate human TAFI gene expression in cultured human hepatoma (HepG2) cells. Surprisingly, we found that treatment of HepG2 cells with a combination of interleukin (IL)-1␤ and IL-6 suppressed endogenous TAFI mRNA abundance in HepG2 cells (ϳ60% decrease), while treatment with IL-1␤ or IL-6 alone had no effect. Treatment with IL-1␤ and/or IL-6 had no effect on TAFI promoter activity as measured using a luciferase reporter plasmid containing the human TAFI 5-flanking region, whereas treatment with IL-1␤ and IL-6 in combination, but not alone, decreased the stability of the endogenous TAFI mRNA. Treatment with the synthetic glucocorticoid dexamethasone resulted in a 2-fold increase of both TAFI mRNA levels and promoter activity. We identified a functional glucocorticoid response element (GRE) in the human TAFI promoter between nucleotides ؊92 and ؊78. The GRE was capable of binding the glucocorticoid receptor, as assessed by gel mobility shift assays, and mutation of this element markedly decreased the ability of the TAFI promoter to be activated by dexamethasone.
Thrombin activable fibrinolysis inhibitor (TAFI) 1 was first identified in 1989 by two independent groups as a basic carboxypeptidase present in fresh serum that was distinct from the constitutive basic carboxypeptidase N (1,2). By virtue of the intrinsic instability of this enzyme, whose activity disappeared within 2 h upon incubation at 37°C, Hendriks et al. (1) designated the novel activity "unstable" carboxypeptidase or carboxypeptidase U (1). Campbell and Okada (2) determined that the enzyme removed arginine residues from substrates more efficiently than lysines and therefore designated it carboxypeptidase R (2). In 1991, Eaton et al. (3) isolated a cDNA encoding the zymogen form of the enzyme and found that it was highly homologous to pancreatic procarboxypeptidase B. Bajzar et al. (4) independently isolated a protein on the basis of its ability to inhibit fibrinolysis in the setting of sustained activation of the coagulation cascade; on the basis of this property, they named the protein TAFI. Amino acid sequence analysis of TAFI revealed it to be identical to plasma procarboxypeptidase B and procarboxypeptidases U and R. TAFI can be activated by thrombin (4), plasmin (5), and thrombin in complex with thrombomodulin (6), with the last being by far the most efficient activator. Activated TAFI (TAFIa) inhibits fibrin clot lysis by removing the carboxyl-terminal lysine residues from partially degraded fibrin that mediate positive feedback in the fibrinolytic cascade (7). As such, it has been hypothesized that TAFI plays a role in vivo in mediating the balance between coagulation and fibrinolysis.
Additional substrates for TAFIa have been identified that imply a role for the TAFI pathway beyond inhibition of fibrinolysis. TAFIa has been shown to remove the carboxyl-terminal arginines from the anaphylatoxin peptides C3a and C5a (8) as well as from bradykinin (9 -11). As such, TAFIa may modulate inflammatory processes in the vasculature in the setting of activation of the coagulation cascade. Additional evidence for a role for the TAFI pathway in inflammation comes from the recent observation that TAFI is an acute phase protein in mice; injection of the animals with bacterial lipopolysaccharide (LPS) elicited increases in both plasma TAFI concentrations and hepatic TAFI mRNA abundance (12). In order to begin to assess if human TAFI is also an acute phase protein and to determine the molecular mechanisms underlying this potential phenomenon, we have studied the ability of acute phase mediators to alter TAFI gene expression in a cultured human hepatoma cell model.

EXPERIMENTAL PROCEDURES
Materials-Restriction and modification enzymes were from New England Biolabs, Invitrogen, Promega, and Stratagene. [␥-32 P]ATP, [␣-32 P]dATP, and poly(dIdC)⅐(dIdC) was purchased from Amersham Biosciences. TRIzol reagent, minimum essential medium (MEM), Dulbecco's modified Eagle's medium/nutrient mixture F-12 (DMEM/F-12), and penicillin-streptomycin-Fungizone (PSF) were obtained from Invitrogen. Fetal calf serum was purchased from ICN. NucleoTrap mRNA purification kits were from Clontech. Synthetic oligonucleotides were purchased from Cortec DNA Service Laboratories, Inc. (Kingston, Ontario, Canada). IL-1␤, IL-6, dexamethasone, and actinomycin C 1 were purchased from Sigma and were reconstituted as recommended by the manufacturer. Recombinant human glucocorticoid receptor expressed in insect cells and a double-stranded oligonucleotide corresponding to the human tyrosine aminotransferase glucocorticoid response element (TAT-GRE) was obtained from Affinity Bioreagents, Inc. A cDNA clone corresponding to the full-length mRNA encoding the human ␥-chain of fibrinogen was obtained from the American Type Culture Collection.
Cell Culture and RNA Analysis-HepG2 cells (human hepatocellular carcinoma) (13) were grown in MEM containing 10% fetal calf serum and 1% PSF. Cells were maintained in a humidified 37°C incubator under a 95% air, 5% CO 2 atmosphere. Cytokine(s) and/or dexamethasone were added to the growth medium, and the cells were incubated for 24 h prior to harvesting the RNA from the cells. In some experiments, after a 24-h treatment with cytokine(s), actinomycin C 1 was added to the cultures (to 5 g/ml) and incubation continued (in the presence of cytokine(s), where appropriate) for different times up to 8 h prior to harvesting the RNA. RNA was isolated using TRIzol reagent as recommended by the manufacturer. Poly(A) ϩ RNA was prepared using the NucleoTrap mRNA purification kit. For Northern blot analysis, total RNA (20 g/lane) or poly(A) ϩ RNA (ϳ2 g/lane) (in 50% (v/v) formamide, 10 mM MOPS, pH 7.0, 2.2 M formaldehyde) was incubated at 65°C for 15 min, quenched on ice, and then fractionated on a 1% (w/v) agarose gel containing 10 mM MOPS, pH 7.0, 2.2 M formaldehyde. The RNA was blotted onto a nylon membrane (Hybond-XL, Amersham Biosciences) by capillary transfer in 20ϫ SSC (1ϫ SSC is 15 mM trisodium citrate, pH 7, 150 mM NaCl). After ultraviolet cross-linking of the RNA to the membrane, blots were hybridized with radiolabeled probes corresponding to the TAFI cDNA (full open reading frame) (3) or the ␥-chain of human fibrinogen (14). In order to correct for differences in RNA loading and transfer, blots were stripped and hybridized with radiolabeled probes corresponding to either the 36B4 cDNA (human acidic ribosomal phosphoprotein PO) (15) or, in the case of RNA from cells treated with actinomycin C 1 , the glyceraldehyde-6-phosphate dehydrogenase cDNA; in the latter experiments, the amount of TAFI RNA present at each time point was calculated as described by Wilson and Deeley (16), assuming a half-life for the glyceraldehyde-6-phosphate dehydrogenase mRNA of 8 h (17). Probes were prepared using [␣-32 P]dATP and the Prime-It II random primer labeling kit (Stratagene). Hybridization was carried out at 68°C for 1 h in ExpressHyb solution (Clontech). The blots were washed at room temperature in 1ϫ SSC, 0.1% (w/v) SDS, and then at 50°C in 0.2ϫ SSC, 0.1% (w/v) SDS. Blots were exposed to a storage phosphor screen (Kodak), and band intensities were quantitated using a Molecular Imager FX phosphorimager (BioRad).
Reporter Plasmids-The luciferase reporter plasmids TAFI[-2699]luc, TAFI[-1128]-luc, TAFI[-236]-luc, and TAFI[-73]-luc have been described previously (18). The plasmids contain fragments of the 5Јflanking region of the human TAFI gene inserted into the luciferase reporter vector pGL3 Basic (Promega). The numbering refers to the 5Ј-most nucleotide of genomic DNA included in the construct. The 3Ј-boundary of all the reporter plasmids is a HindIII restriction site immediately downstream of the initiator methionine codon such that all possible sites for transcription initiation are present in the construct; the initiator methionine codon was mutated to TTG in all cases. Additional luciferase reporter plasmids representing progressive 5Ј deletions between Ϫ236 and Ϫ73 (TAFI[-120]-luc, TAFI[-100]-luc, TAFI[-90]-luc, and TAFI[-80]-luc) were constructed using PCR in which the 5Ј-most nucleotide of the upstream primer corresponds to the indicated nucleotide in the name of the reporter plasmid.
A mutant variant of TAFI[-1128]-luc (designated TAFI[-1128/ ⌬GRE]-luc) was constructed using PCR according to the method of Nelson and Long (19). The sequence of the mutagenic oligonucleotide was as follows: 5Ј-CACAGGAACAAGAGGGACAGTGCCGTTATATT-TTAACC-3Ј; the underlined nucleotides (positions Ϫ89 to Ϫ87 of the TAFI promoter) are mismatches relative to the wild-type sequence.
Expression Plasmids-Glucocorticoid receptor (GR) expression plasmids were the generous gift of Dr. Robert Haché (University of Ottawa, Ottawa, Canada). pTL2-GR consists of a 2.9-kb BamHI restriction fragment encompassing the full-length open reading frame of the rat GR (20) inserted into pTL2 (a derivative of pSG5, Ref. 21, containing an expanded multiple cloning site) digested with BglII and BamHI. GAL0 -540C encodes a fusion protein containing the Saccharomyces cerevisiae GAL4 DNA-binding domain fused to a carboxyl-terminal fragment of the rat GR (amino acids 540 -795; encompasses the ligand binding domain). rGR-L501P is a mutant variant of the full-length rat GR in which amino acid 501 has been changed from a leucine to a proline. rGR(N525)-pTL2 (encoding the amino-terminal 525 amino acids of the rat GR) was constructed as follows: pEGFP-GRwtN525 (pEGFP-C1 (Clontech) containing a rat GR fragment encompassing amino acids 22-525) was digested with MluI (downstream of the simian virus 40 (SV40) polyadenylation site), the ends were made blunt with T4 DNA polymerase, and the plasmid was then digested with BglII (within the rat GR coding sequence). The resultant fragment was inserted into rGR-pTL2 in which the corresponding sequences of the wild-type receptor cDNA were removed by digestion with BamHI (at which point the ends were made blunt) and BglII. Expression of all rat GR variants in mammalian cells was driven by the SV40 early promoter with the exception of rGR-L501P, whose expression was driven by the Rous sarcoma virus (RSV) long terminal repeat.
Reporter Gene Assays-For luciferase reporter gene assays, HepG2 cells were grown in 6-well plates (Corning) and transfected by the method of calcium phosphate co-precipitation (22). Typically, cells received ϳ1.3 g of luciferase reporter plasmid and 0.6 g of ␤-galactosidase internal control plasmid (RSV-␤gal) (23) (to control for transfection and harvesting efficiency). In some experiments, cells also received 0.6 g of GR expression plasmids. After a 6-h exposure to the precipitate, the cells were washed three times in phosphate-buffered saline and given fresh medium, and incubation was continued for an additional 42 h. In some experiments, hormones or cytokines were added to the culture medium at various times during the 42-h incubation period. The cells were harvested for preparation of cytoplasmic extracts for luciferase and ␤-galactosidase assays as previously described (18). For each sample, the relative luciferase activity was calculated to be the luciferase activity per unit of ␤-galactosidase activity per unit volume of cell extract.
Gel Mobility Shift Assays-Complementary sets of oligonucleotides encompassing the putative GRE in the TAFI promoter were synthesized: sense 5Ј-CAC AGG AAC AAG AGG GAA CAT GCC GTT ATA TTT TAA CC-3Ј; antisense 5Ј-GGT TAA AAT ATA ACG GCA TGT TCC CTC TTG TTC CTG TG-3Ј. Mutant oligonucleotides encompassing the same range, corresponding to the mutation in the GRE (see above), were also synthesized. For radiolabeled TAFI binding site probes for gel mobility shift assays, 5 pmol of sense strand oligonucleotide was endlabeled using [␥-32 P]ATP and T4 polynucleotide kinase. Unincorporated label was removed using a NAP-5 column (Amersham Biosciences). The labeled oligonucleotide was combined with a 5-fold molar excess of cold antisense oligonucleotide, and the two annealed by placing in boiling water and allowing to cool slowly at room temperature. Unlabeled TAFI competitor binding site probes were made by annealing equimolar amounts of sense and antisense oligonucleotides. For the TAT-GRE (sense: 5Ј-CTA GGC TGT ACA GGA TGT TCT GCC TAG-3Ј; antisense: 5Ј-CTA GGC AGA ACA TCC TGT ACA GCC TAG-3Ј), 5 pmol of the double-stranded oligonucleotide was labeled as described above, and then unincorporated label was removed using a NAP-5 column. Unlabeled competitor binding site probes were diluted from the stock TAT-GRE provided by the manufacturer.
Binding reactions were performed in binding buffer (20 mM HEPES pH 7.9, 60 mM KCl, 5 mM MgCl 2 , 5 g/ml bovine serum albumin, 10% (v/v) glycerol, 2 mM dithiothreitol) and contained 1 l of partially purified recombinant human GR expressed in insect cells as well as 2.0 g of poly(dIdC)⅐(dIdC) and 10 fmol of radiolabeled probe (ϳ20,000 cpm). Binding reactions were incubated for 30 min on ice. Some binding reactions contained a 10-, 50-, 100-, or 200-fold molar excess of unlabeled binding site competitor (corresponding either to the wild-type TAFI sequence, the sequence containing the mutant GRE, or the TAT-GRE). Reactions were loaded on a 6% polyacrylamide gel in 0.5ϫ Tris borate-EDTA, 5% (v/v) glycerol that had been pre-electrophoresed at 300 V for 1 h at 4°C. Electrophoresis was continued for a further 2.5 h at this temperature, at which time the gel was fixed, dried, and exposed to film (Kodak BIOMAX MR).

Effect of Acute Phase Mediators on TAFI Gene Expression-It
has been demonstrated that injection of mice with bacterial lipopolysaccharide (LPS) results in an increase in both plasma TAFI concentrations as well as hepatic TAFI mRNA abundance (12). To determine which mediators of the acute phase response regulate TAFI gene expression, we treated human hepatoma (HepG2) cells for 24 h with the cytokines interleukin (IL)-1␤ and/or -6, as well as dexamethasone (a synthetic glucocorticoid hormone analog). We measured the abundance of the endogenous TAFI mRNA by Northern blot analysis, using the acid ribosomal phosphoprotein PO (36B4) mRNA as an internal standard (Fig. 1). While we found that TAFI mRNA abundance was actually decreased by treatment of the cells with a combination of IL-1␤ and IL-6, IL-1␤ alone or IL-6 either alone or in combination with dexamethasone had little or no effect. Treatment with dexamethasone resulted in dosedependent changes in TAFI mRNA abundance: an increase of up to 2-fold was observed that peaked at doses between 0.5 and 2 M hormone.
To verify that our cell culture model was a valid model for the analysis of the effect of acute phase mediators on TAFI gene expression, we measured the abundance of the mRNA for the ␥-chain of fibrinogen (␥-Fgn) under similar conditions. As has been previously reported (24), both IL-6 and dexamethasone induce expression of ␥-Fgn, with the greatest effect occurring when the two mediators are administered in combination. The magnitude of the induction by IL-6 plus dexamethasone was, however, less than has been reported for the mRNA levels of rat ␥-Fgn in primary rat hepatocytes (15-20-fold) (24) or the activity of the rat ␥-Fgn promoter in H35 rat hepatoma cells (10-fold) (25). This finding may result from the fact that HepG2 cells are relatively deficient in glucocorticoid receptor (26). In keeping with published reports (27), IL-1␤ had the effect of blunting the induction of ␥-Fgn mRNA stimulated by IL-6 and IL-6 plus dexamethasone.
Effect of Acute Phase Mediators on TAFI Promoter Activity-To examine whether the acute phase mediators were capable of specifically altering the activity of the TAFI promoter, HepG2 cells were transiently transfected with a luciferase reporter plasmid containing the 5Ј-flanking region of the human TAFI gene. The transfected cells were treated for 24 h with the acute phase mediators described above, and luciferase activity was quantitated in extracts isolated from the cells as a measure of TAFI promoter activity (Fig. 2). Dexamethasone (1.0 M) stimulated TAFI promoter activity more than 2-fold, either alone or in combination with IL-6. None of the other treatments had an appreciable effect on TAFI promoter activity.
Effect of Acute Phase Mediators on TAFI mRNA Stability-Since the ability of IL-1␤ and IL-6 in combination to decrease TAFI mRNA abundance did not seem to result from a decrease in TAFI promoter activity, we investigated whether these cytokines could influence the stability of the TAFI mRNA transcript. HepG2 cells were treated for 24 h with IL-1␤ and/or IL-6 at which time transcription was arrested by the addition of actinomycin C 1 . RNA was harvested at different times after the addition of actinomycin C 1 and the remaining TAFI mRNA quantified by Northern blot analysis (Fig. 3). The results show that the TAFI mRNA has an intrinsic half-life of about 3 h, identifying the TAFI transcript as a relatively short-lived mRNA species (28); addition of IL-1␤ and IL-6 in combination results in a destabilization of the TAFI mRNA whereas either cytokine administered alone had no effect. From regression of the Northern blot data, the effect of the combined cytokines results in a 22% decrease in the half-life of the TAFI transcript.
Identification of a Glucocorticoid Response Element in the TAFI Promoter-In order to identify sequences in the TAFI promoter that mediate the increase in promoter activity elicited by dexamethasone, we utilized a series of luciferase reporter plasmids containing progressive 5Ј deletions of 5Ј-flanking sequence (Fig. 4A). The plasmids were transiently transfected into HepG2 cells, and the cells were incubated either in the absence or presence of 1.0 M dexamethasone for 42 h prior to harvest and luciferase assay. Full responsiveness to dexamethasone was preserved upon deletion of 5Ј-flanking sequence up to nucleotide Ϫ100 (the numbering refers to the number of nucleotides upstream of one of the transcription start sites in the TAFI promoter) (Fig. 4B). However, deletion right panel); to correct for differences in RNA loading and transfer, the blots were stripped and hybridized to a radiolabeled cDNA probe corresponding to the acid ribosomal phosphoprotein PO (36B4). Band intensities were quantified using a phosphorimager; shown are the corrected intensities of the TAFI or ␥-Fgn bands, with mRNA abundance under each condition presented relative to that in the absence of acute phase mediators (control). The data shown are the mean Ϯ S.E. from three independent blots.

FIG. 2. Effect of acute phase mediators on TAFI promoter activity in HepG2 cells.
HepG2 cells were transiently transfected, by the method of calcium phosphate co-precipitation, with a luciferase reporter plasmid (TAFI[-2699]-luc) containing the 5Ј-flanking region of the human TAFI gene (up to 2699 bp upstream of one of the transcription start sites). Also included in each transfection was the internal control plasmid RSV-␤gal to correct for differences in transfection and harvesting efficiency. After a 6-h exposure to the precipitate, the cells were washed and provided with fresh medium containing the indicated combinations of IL-1␤ (1 ng/ml), IL-6 (10 ng/ml), and dexamethasone (dex) (1.0 M). After a further 42-h incubation, cytoplasmic extracts were prepared for the measurement of luciferase activity as well as ␤-galactosidase activity to allow for correction for differences in transfection and harvesting efficiency. Relative luciferase activities (mean Ϯ S.E. of three independent experiments) are defined as luciferase activity per unit of ␤-galactosidase activity and are shown as a percentage of that observed in the absence of acute phase mediators (control). of sequences between Ϫ100 and Ϫ90 resulted in almost complete elimination of the increase in promoter activity in response to dexamethasone. Deletion of sequences up to Ϫ80 does not result in a further decrease in dexamethasone responsiveness, while deletion of sequences up to Ϫ73 results in a complete loss of TAFI promoter activity, as we have previously reported (18).
Glucocorticoid hormones exert their effects on transcription through binding to and activating the GR, a member of the nuclear receptor superfamily (29). Ligand-bound GR binds as a homodimer to specific sequences, known as GREs, in the promoters of target genes thereby stimulating their transcription. Inspection of the sequence downstream of Ϫ100 of the TAFI promoter revealed the presence of a sequence, between Ϫ92 and Ϫ78, that resembles a consensus GRE (30) in that it is an imperfect inverted repeat with a 3-nucleotide spacing of the 6-nucleotide half-sites (Fig. 5A). Mutations were introduced into this candidate GRE that would be expected to abolish its ability to bind GR (Fig. 5A). HepG2 cells were transfected with a luciferase reporter plasmid encompassing the mutations and including the TAFI 5Ј-flanking sequence up to Ϫ1128 (TAFI-[-1128/⌬GRE]-luc), or the corresponding wild-type reporter plasmid (TAFI[-1128]-luc), and the cells were treated with dexamethasone (Fig. 5B). The mutations greatly decreased the response of the TAFI promoter; interestingly, the decrease in the extent of the response was similar to that observed upon deletion of sequences between Ϫ100 and Ϫ90.
In order to examine the requirement for DNA binding of the GR for dexamethasone-dependent activation of the TAFI promoter, we performed co-transfection experiments with the wild-type or ⌬GRE mutant TAFI promoter reporter plasmid and expression plasmids for rat GR (rGR) variants. The variants used (see Fig. 6A) included the full-length wild-type receptor (rGR-wt), a variant including only the amino-terminal 525 amino acids (i.e. lacking the DNA-and ligand-binding domains; rGR-N525), a variant including only the carboxylterminal 540 amino acids fused to the GAL4 DNA binding domain (i.e. lacking the amino-terminal transactivation domain and the DNA binding domain of the GR; rGR-540C), and a mutant of the full-length GR with a single point mutation that abolishes DNA binding (rGR-L501P).
While ectopic expression of the intact, full-length rGR substantially increased the magnitude of dexamethasone induction of the wild-type TAFI promoter (Fig. 6B), ectopic expression of the rGR variants lacking the ability to bind DNA had little or no influence on the magnitude of induction. Expression HepG2 cells were treated with IL-1␤ (1 ng/ml) and/or IL-6 (10 ng/ml) for 24 h. Actinomycin C 1 was added to 5 g/ml and incubation was continued in the presence of cytokine(s). RNA was harvested at different times after the addition of actinomycin C 1 and the abundance of remaining TAFI mRNA quantitated by Northern blot analysis. Differences in RNA loading and transfer were accounted for by using GAPDH mRNA as an internal standard. The abundance of TAFI mRNA after actinomycin C 1 addition is shown relative to that present immediately before the addition of actinomycin C 1 . The data shown are the mean Ϯ S.E. of three independent experiments. of the N525 variant resulted in a large, dexamethasone-independent induction of the TAFI promoter. This result was not unexpected given that removal of the ligand-binding domain results in a constitutively active GR. Interestingly, while ectopic expression of the intact, full-length rGR did not mediate any dexamethasone-dependent induction of the TAFI[-1128/ ⌬GRE]-luc reporter plasmid, the two rGR variants lacking DNA-binding ability were able to confer an ϳ2-fold induction by this hormone. The constitutively active N525 rGR variant induced this mutant reporter plasmid in a hormone-independent fashion, albeit to a substantially reduced extent compared with the wild-type TAFI[-1128]-luc plasmid.
To demonstrate explicitly that the GR can bind to the TAFI promoter GRE, gel mobility shift assays were performed using radiolabeled double-stranded oligonucleotide probes corre-sponding to the wild-type TAFI GRE and recombinant human GR expressed in insect cells (Fig. 7A). As a positive control, gel mobility shift assays were also performed using the GRE from the TAT gene (Fig. 7B). The autoradiograms show a complex of low mobility is formed using the probe containing the wild-type TAFI-GRE as well as the TAT-GRE. These complexes represent specific binding of the GR because they are competed effectively by an excess of unlabeled oligonucleotides containing the TAT-GRE and the wild-type TAFI-GRE but not effectively by the mutant TAFI-GRE (TAFI-⌬GRE). No specific complexes were observed when radiolabeled oligonucleotides containing the mutant TAFI GRE were utilized (data not shown). DISCUSSION The acute phase reaction is a complex host defense mechanism that, in response to triggers such as trauma, surgery, tissue infarction, or severe infection, aims to counteract the

FIG. 5. Mutagenesis of the GRE in the human TAFI promoter.
Panel A, the consensus GRE is an imperfect inverted repeat of 6-nucleotide half-sites with a 3-nucleotide spacing (30); the underlined nucleotides in the consensus are present in Ͼ80% of GREs (30). The TAFI promoter contains an imperfect inverted repeat resembling this consensus between Ϫ92 and Ϫ78. Also shown is the GRE from the rat ␣ 1 -acid glycoprotein gene, which contains a stretch of ten consecutive nucleotides that are identical in the TAFI gene. A mutation of the putative TAFI promoter GRE (⌬GRE), indicated below the TAFI promoter sequence, was introduced into the luciferase reporter plasmid TAFI- underlying challenge while restoring homeostasis (reviewed in Refs. 31 and 32). Among the features of the acute phase response are systemic changes such as fever, increases in neutrophil production, changes in lipid and amino acid metabolism and activation of the coagulation and complement cascades as well as changes in the expression of a panel of liver-expressed plasma proteins. These acute-phase proteins are classified as either positive or negative acute phase proteins depending on whether their expression is induced or repressed, respectively, in the acute phase. Among the acute phase proteins are C-reactive protein, serum amyloid A, ␣ 1 -acid glycoprotein, components of the complement cascade, proteins of the coagulation and fibrinolytic cascades, protease inhibitors, and proteins involved in transport and inflammatory functions. The magnitude of the changes in expression can be small (such as a 50% increase in the expression of ceruloplasmin) to vast (such as a greater than 1000-fold increase in the expression of C-reactive protein and serum amyloid A).
Regulation of the expression of acute phase proteins is most often at the level of their transcription in liver and is largely a function of the action of certain inflammatory cytokines (reviewed in Ref. 33). Acute phase proteins are divided into two broad categories based on the cytokines that regulate their expression: class I proteins, including serum amyloid A, C-reactive protein, complement factor C3, and ␣ 1 -acid glycoprotein, are induced by IL-1-type cytokines in concert with IL-6-type cytokines; class II proteins, including fibrinogen, haptoglobin, and ␣ 2 -macroglobulin, are induced by IL-6-type cytokines with the participation of glucocorticoid hormones.
A study in which mice were injected intraperitoneally with a lethal dose of bacterial lipopolysaccharide, a maneuver that would be expected to provoke a robust acute phase response, resulted in increases in concentrations of TAFI in plasma as well as in hepatic TAFI mRNA abundance (12). This study, therefore, identified TAFI as a positive acute phase protein. A role for the TAFI pathway in the acute phase response is reasonable to expect since this pathway may impact both on hemostatic as well as inflammatory functions. Activation of TAFI, with the attendant inhibition of fibrinolysis (4,7), may stabilize clots formed in response to tissue damage or as a means to isolate regions of severe infection. On the other hand, the TAFI pathway may influence the vascular responses to inflammatory stress: the ability of TAFIa to remove the carboxyl-terminal arginine residues from C3a and C5a (the anaphylatoxins) and from bradykinin could have effects on vascular tone and permeability (8 -11).
The potential increase in plasma TAFI concentrations in the acute phase may reflect a requirement for enhanced activity of the TAFI pathway during host defense. Alternatively, the activation of the coagulation and fibrinolytic systems that occurs during the acute phase may result in consumption of the existing pool of plasma TAFI that could be compensated for by an increase in hepatic TAFI expression. Further analysis of the TAFI pathway during the acute phase will likely yield valuable insights into the role of this pathway in regulating the balance between coagulation and fibrinolysis and in mediating interactions between the coagulation and inflammatory systems.
We have utilized a cultured human hepatoma cell model to assess if TAFI gene transcription is induced by acute phase mediators and to investigate the molecular mechanisms underlying these effects. HepG2 cells represent a well characterized model system for the study of transcriptional regulation in the acute phase, as they retain many of the characteristics of hepatocytes, endogenously express many liver-specific genes, and contain cell surface receptors for the relevant inflammatory cytokines (13,33). Surprisingly, we found that treatment of HepG2 cells with IL-1␤ and IL-6 in combination reduced expression of the endogenous TAFI gene, as assessed by Northern blot analysis (Fig. 1). However, we did not observe an effect of this combination of cytokines on TAFI promoter activity measured by transient transfection of luciferase reporter plasmids (Fig. 2). Indeed, we found that these two cytokines, when administered in combination but not alone, were capable of decreasing the stability of the TAFI mRNA, which likely accounts for the ability of these cytokines to decrease TAFI mRNA abundance. We hypothesize that IL-1␤ and IL-6 together modulate the expression of a factor or factors that influences TAFI mRNA decay. IL-6 or IL-1␤ alone had no effect on TAFI mRNA abundance in HepG2 cells or TAFI promoter activity (Figs. 1 and 2). However, dexamethasone treatment, either in the presence or absence of IL-6, increased TAFI promoter activity ϳ2-fold; dexamethasone alone increased TAFI mRNA abundance almost 2-fold but had no effect in the presence of IL-6.
We identified a functional glucocorticoid response element in the TAFI promoter between Ϫ92 and Ϫ78; mutation of this GRE greatly decreased the ability of the TAFI promoter to be activated by dexamethasone (Fig. 5B) and of the TAFI-GRE to bind to the GR (Fig. 7). Of note, the unlabeled TAFI-GRE was a less effective competitor for binding of GR to either probe than unlabeled positive control TAT-GRE (Fig. 7), suggesting that the latter GRE possesses a higher affinity for GR. Indeed, the TAFI GRE contains numerous key substitutions in the downstream half-site sequence (Fig. 5A), relative to the consensus GRE (30) and the TAT-GRE, which differs from the consensus at only one position in the upstream half-site (TG-TACAGGATGTTCT). In addition, a small but detectable extent of competition by the unlabeled TAFI-⌬GRE was observed with both the TAFI-GRE and TAT-GRE probes (Fig. 7), suggesting that the TAFI-⌬GRE retains a weak affinity for the GR. Indeed, the TAFI promoter containing the mutant GRE retains a small extent of dexamethasone inducibility (Fig. 5B) and exhibits dexamethasone-independent induction by ectopic expression of the constitutively active N525 GR variant, albeit to a reduced extent relative to the wild-type promoter (Fig. 6B). On the other hand, ectopically expressed full-length GR abolishes the small extent of dexamethasone inducibility of the mutant TAFI promoter (Fig. 6B). An explanation for this finding might be that the overexpressed full-length receptor consumed factors required for dexamethasone-dependent transactivation of the mutant TAFI promoter. That the TAFI-GRE is not an optimal GR binding sequence is perhaps consistent with the relatively modest extent to which dexamethasone induces transcription of the TAFI promoter, compared with the TAT promoter (e.g. 8 -10-fold increase in TAT mRNA in primary rat hepatocytes induced by 1 M dexamethasone, Ref. 34).
In the study in mice that identified TAFI as a positive acute phase protein (12), the magnitude of the increase in hepatic TAFI mRNA abundance was not carefully measured; as such, it is difficult to directly compare it to our in vitro data. Nonetheless, it is perhaps surprising that the acute phase mediators we examined either increased TAFI gene expression by a relatively modest amount (ϳ2-fold increase mediated by dexamethasone) or even decreased TAFI gene expression (ϳ60% decrease mediated by IL-1␤ ϩ IL-6). However, it is important to stress that even small changes in plasma concentrations of TAFI can be expected to impact on the potential of the TAFI pathway to influence fibrinolysis. The range of plasma concentrations of TAFI (the upper limit of which has been reported to exceed 200 -400 nM; Refs. 35-37) is well below the K m for activation of TAFI by thrombin or thrombin-thrombomodulin (ϳ1 M; Ref. 6); as such, a change in concentration of TAFI in plasma would result in a corresponding change in the rate of TAFI activation by thrombin or thrombin-thrombomodulin. Indeed, a strong correlation between plasma TAFI concentrations and in vitro clot lysis times has been observed (35,37).
No data currently exist that explicitly address potential changes in TAFI plasma concentrations in the acute phase in humans, although associations between TAFI concentrations and those of fibrinogen (38, 39) 2 and C-reactive protein 2 (38) have been reported, suggesting that TAFI gene expression could be positively regulated by inflammatory mediators. Intracellular signals elicited by IL-1 ultimately result in the activation of certain transcription factors, specifically AP-1, NF-B, and C/EBP␤ (33). IL-6 signaling results in the activation of the transcription factor STAT3 (33). Functional cooperation between GR and all of these transcription factors has been reported (40 -44). In addition to the functional GRE, we have identified a functional C/EBP binding site between Ϫ52 and Ϫ40 of the TAFI promoter (45). Examination of the TAFI gene 5Ј-flanking sequence for consensus transcription factor binding sites using Matinspector (46) revealed potential binding sites for AP-1 but not STAT3 or NF-B (data not shown). Alternatively, it cannot be ruled out that in humans, TAFI is either not an acute phase protein or is a negative acute phase protein. Different acute phase proteins have distinct temporal patterns of induction and subsequent return to baseline (31); the temporal pattern of TAFI expression during the acute phase could reflect the changing balance between cytokine and glucocorticoid signaling pathways over time.
The association of TAFI concentrations with inflammatory markers and fibrinogen suggests a role for glucocorticoid hormones (or other inflammatory mediators) in regulating plasma TAFI levels in the setting of chronic inflammation. For example, one study found that compared with preoperative plasma TAFI levels in patients requiring coronary artery bypass grafting (which were higher than in healthy controls), postoperative plasma TAFI levels fell 17% by day 3, then rose again 14% by day 6 (38). On the other hand, glucocorticoid hormones are important clinically as anti-inflammatory drugs. Interestingly, one study compared the distribution of plasma TAFI concentrations in a sample of patients with rheumatoid arthritis with that in a healthy control population: plasma TAFI concentrations are clearly higher in patients with rheumatoid arthritis, although the use of glucocorticoids in the patients was not accounted for (36).
In conclusion, we have documented the effect of acute phase mediators on endogenous TAFI gene expression in HepG2 cells by Northern blot analysis as well as on TAFI promoter activity by transient transfection into HepG2 cells of luciferase reporter plasmids harboring the TAFI 5Ј-flanking region. We found that when administered in combination, IL-1␤ and IL-6 decreased TAFI mRNA abundance by 60%, an effect that was associated with a destabilization of the TAFI mRNA transcript. Dexamethasone resulted in a 2-fold increase in both TAFI mRNA abundance and promoter activity, and we were able to identify a functional GRE in the TAFI promoter. Further studies will be required to fully elucidate the significance of these observations with respect to regulation of TAFI gene expression in the acute phase and other inflammatory states.