Transforming Growth Factor-β (TGF-β) Inhibits the Expression of Factor VII-activating Protease (FSAP) in Hepatocytes*

Deletion of the Habp2 gene encoding Factor VII-activating protease (FSAP) increases liver fibrosis in mice. A single nucleotide polymorphism (G534E) in HABP2 leads to lower enzymatic activity and is associated with enhanced liver fibrosis in humans. Liver fibrosis is associated with a decrease in FSAP expression but, to date, nothing is known about how this might be regulated. Primary mouse hepatocytes or the hepatocyte cell line, AML12, were treated with different factors, and expression of FSAP was determined. Of the various regulatory factors tested, only transforming growth factor-β (TGF-β) demonstrated a concentration- and time-dependent inhibition of FSAP expression at the mRNA and protein level. The TGF-β-Type I receptor (ALK-5) antagonist SB431542 and Smad2 siRNA, but neither SIS3, which inhibits SMAD3, nor siRNA against Smad3 could block this effect. Various regions of the HABP2 promoter region were cloned into reporter constructs, and the promoter activity was determined. Accordingly, the promoter activity, which could phenocopy changes in Habp2 mRNA in response to TGF-β, was found to be located in the 177-bp region upstream of the transcription start site, and this region did not contain any SMAD binding sites. Mutation analysis of the promoter and chromatin immunoprecipitation assays were performed to identify an important role for the ATF3 binding element. Thus, TGF-β is the most likely mediator responsible for the decrease in FSAP expression in liver fibrosis.

Liver fibrosis arises because of varying etiologies such as viral infections, alcohol abuse, fatty diet, and or ingestion/accumulation of toxins (1). Hepatic damage leads to concomitant proinflammatory status leading to an expansion of fibroblast and hepatic stellate cells (HSC). 2 Furthermore, tissue injury recruits inflammatory cells as well as activates hepatic stellate cells (HSC) that further promote extracellular matrix remodeling (2). In patients with hepatic fibrosis there is a very high risk of further progression to hepatocellular carcinoma (HCC) (3). Broad specificity tyrosine kinase receptor inhibitors have shown promising results in reversing liver fibrosis (4).
Factor VII-activating protease (FSAP) is a liver-derived circulating plasma serine protease that is involved in the regulation of coagulation and fibrinolysis (5). Human genetic epidemiological studies have shown that a single nucleotide polymorphism (SNP) in the FSAP-encoding gene (HABP2), rs7080536 or G534E also called the Marburg I (MI) polymorphism is found in about 5% of the European population and it is associated with the severity of liver fibrosis (6). MI-FSAP has about a 5-fold weaker proteolytic activity compared with wildtype (WT) FSAP toward various substrates e.g. platelet-derived growth factor (PDGF)-BB (7). Circulating latent FSAP zymogen is activated by factors that are released from dead/injured or apoptotic cells such as histones and nucleosomes (8,9). This fits well with the previous experimental results that acute application of a hepatotoxin, carbon tetrachloride (CCl 4 ), leads to FSAP activation in mice, presumably through induction of cell death in the liver (10). We have demonstrated that CCl 4 -mediated liver injury in a mouse model is associated with an up-regulation of FSAP expression in the long term (11). To further define the role of FSAP in liver fibrosis, we have measured FSAP mRNA expression in liver from patients with varying progression of disease and found a significant negative correlation between fibrosis/inflammation and FSAP expression. Furthermore, circulating FSAP concentration was lower in patients with advanced liver cirrhosis (12). Experiments with FSAP Ϫ/Ϫ mice also confirmed a protective influence of FSAP on inflammation and liver fibrosis. These studies demonstrate an important role for FSAP in regulating the inflammatory and fibrotic adaptation to liver injury (13).
Although hepatocytes are the main source of circulating FSAP, monocytes express low levels of FSAP that is increased by estrogen and progesterone (14) as well as lipopolysaccharide, Interleukin (IL)-6 and IL-1␤ (15). There is also ectopic expression of FSAP in some tumors (16,17). In silico analysis of the HABP2 promoter sequence upstream of the transcription start site indicated the presence of AP-1, SP-1, and HNF1␣ transcription factor binding sites but these have never been experimentally verified (18). To date, no other information is available as to how FSAP expression is regulated in hepatocytes at the molecular level. Thus, we have tested a range of factors on FSAP expression in hepatocytes. The only factor that mediated a robust down-regulation was transforming growth factor-␤1 (TGF-␤). The detailed analysis of how TGF-␤ regulates FSAP expression completes the description of a circuit that further supports the role of FSAP as an effector protease in liver fibrosis.

TGF-␤ Is an Inhibitor of Habp2 Expression in Primary
Hepatocytes and AML12 Cells-To identify factors that regulate the expression of Habp2 in hepatocytes we tested a range of factors on primary mouse hepatocytes and the mouse hepatocyte cell line AML12 cells. The following substances were tested at various concentrations and for various time points and were found not to alter the expression of Habp2 mRNA in either cell type: epidermal growth factor (EGF), PDGF-BB, basic fibroblast growth factor (bFGF), hepatocyte growth factor (HGF), connective tissue growth factor (CTGF), IL-1␤, estrogen (Estr.), and progesterone (Prog.) (Fig. 1A). The only factor found to decrease FSAP mRNA expression was TGF-␤. The effect of TGF-␤ was concentration-and time-dependent (Fig. 1, B and C). The response was evident after 24 h and stronger at 48 h.
Genes that are functionally related to HABP2 and also expressed by hepatocytes, such as Factor XII (F12) and prothrombin (F2), were also down-regulated by TGF-␤ (Fig. 2). As shown before (19) plasminogen activator inhibitor-1 (Serpine1) was up-regulated. Similarly, the neighboring gene at the HABP2 locus, Casp7 was not altered in its expression at 24h but there was some inhibition at 48h (Fig. 2). Expression of Nrap, another neighboring gene, which overlaps with the HABP2 sequence on the opposite strand, was not detected in hepatocytes but in the mouse heart (data not shown). Thus, the downregulation of Habp2 expression is specific response to TGF-␤ treatment.
ALK5 Receptor Is Responsible for the Effect of TGF-␤ on Habp2 Expression-We then investigated the signal transduction pathway responsible for mediating the down-regulation of Habp2 expression. TGF-␤ receptor activation can trigger the canonical pathway including the SMAD2/3 and non-canonical pathways including signaling via p38, JNK, c-Abl, PI3K, and TAK1 (20). Treatment of both cell types with the TGF-␤ receptor (ALK5) inhibitor SB431542 confirmed the role of ALK5 in the TGF-␤-mediated down-regulation of Habp2 mRNA. This was not the case with the SMAD3 inhibitor SIS3 (Fig. 3, A and  B). Western blotting also showed that TGF-␤ could decrease the amount of FSAP protein and that this was reversed by SB431542 but not by SIS3 (Fig. 3C, densitometric analysis in supplemental Fig. S1). To further characterize the signaling FIGURE 1. Inhibition of Habp2 mRNA expression by TGF-␤. A, murine AML12 hepatocyte cell line (black bars) and murine primary hepatocytes (gray bars) were stimulated with epidermal growth factor (EGF, 50 ng/ml), plateletderived growth factor (PDGF-BB, 50 ng/ml), basic fibroblast growth factor (bFGF, 10 ng/ml), hepatocyte growth factor (HGF, 10 ng/ml), connective tissue growth factor (CTGF, 50 ng/ml), transforming growth factor-␤ (TGF-␤, 100 ng/ml), interleukin-1␤ (IL-1␤, 20 ng/ml), estrogen (Estr., 10 g/ml), and progesterone (Prog., 10 g/ml) for 24 h. Relative Habp2 mRNA levels were normalized to Gapdh and represent mean Ϯ S.D. (n ϭ 3). B and C, AML12 cells (square) and primary hepatocytes (circles) were treated with 100 ng/ml of TGF-␤ for increasing time (B), or with increasing concentrations of TGF-␤ for 24 h (C) and Habp2 mRNA levels were quantified. Values are mean Ϯ S.E. (n ϭ 3) of a single experiment and similar results were obtained in three independent experiments. Statistically significant results (p Ͻ 0.05) compared with untreated cells as measured by ANOVA is denoted by *. pathway involved, SMAD3 phosphorylation was investigated in AML-12 cells. SB431542 and SIS3 blocked TGF-␤-mediated phosphorylation of SMAD3 while, total SMAD2/3 levels were unchanged (Fig. 3C, densitometric analysis in supplemental Fig. S1). Immunostaining of primary hepatocytes for FSAP also confirmed the down-regulation of expression by TGF-␤ and the reversal of this effect by SB431542 (Fig. 3D).
Because SIS3 was ineffective in blocking this response we further used a siRNA approach to re-examine this. Smad2 and Smad3 were separately knocked down in AML-12 cells and its influence on Habp2 expression was determined. Smad3 knockdown did not influence the effect of TGF-␤ but Smad2 knockdown did ( Fig. 4A and supplemental Fig. S2). We also tested whether the, non-canonical pathways, ERK1/2, p38MAPK, or Akt pathways could be involved using specific inhibitors of these pathways. LY294002 (PI3K inhibitor), PD98059 (MEK1 inhibitor) and SB203580 (p38MAPK inhibitor) did not influence the down-regulation of Habp2 mRNA by TGF-␤ (Fig. 4B). The effectiveness of LY294002 and PD98059 was tested in respective Western blotting experiments analyzing ERK1/2 and Akt phosphorylation (supplemental Fig. S3). Thus, a clear involvement of SMAD3 and non-canonical pathways was not evident indicating that ALK5 mediated signaling via SMAD2 was responsible.
The TGF-␤ Response Is Mediated via a 177-bp Promoter Region of the HABP2 Gene that Is Directly Upstream of the Transcription Start Site-We then determined the nature of the transcriptional regulation of the HABP2 gene by TGF-␤.
Deletion mutations representing fragments of the first 2400 bp of the 5Ј region upstream of the transcription start site of the human HABP2 gene were cloned into the luciferase reporter vector pGL4.10[luc2] (Fig. 5A). Basal luciferase activity in untreated AML12 cells was significantly reduced in the case of the sub-fragments Ϫ1/Ϫ177, Ϫ313/Ϫ427, and Ϫ177/ Ϫ313 compared with the 500-bp fragment (Fig. 5B). TGF-␤ induced a decrease in the promoter activity of the Ϫ1/Ϫ2400, Ϫ1/Ϫ1800, Ϫ1/Ϫ1000, Ϫ1/Ϫ500, and the Ϫ1/Ϫ177 bp construct but not of other sub-fragments tested. The effect of TGF-␤ on the FSAP promoter activity was reversed by treatment with SB431542 (Fig. 5C). The direct effect of SB431542 alone on HABP2 expression may be due to the inhibition of the effect of endogenous TGF-␤ produced by the cells. For comparison, we also cloned the previously described Serpine1 (PAI-1) (19) and c-myc promoter (21) since these have been shown to be, respectively, up-or down-regulated in hepatocytes after treatment with TGF-␤ and the expected responses were observed (supplemental Fig. S4) confirming the validity of the test system.
Furthermore, in silico analysis of the human and mouse HABP2 5Ј 177-bp sequence upstream of the transcription start site (Fig. 6) showed a relative high conservation of 69% between the species and a similar pattern of putative binding sites for AP-1, SP-1, and HNF1␣. These results are in agreement with earlier reports regarding the FSAP promoter region (18). Since we determined the activity of the human HABP2 promoter in a mouse hepatocyte cell line AML12 cells we also performed these studies in the human hepatocellular carcinoma-derived hepatocytic cell line HepG2. The activity of the 500-bp HABP2 promoter in human cells was down-regulated by TGF-␤ confirming the inter-species congruence (supplemental Fig. S5).
Transcription Factors Involved in the Regulation of HABP2-Because the Ϫ1/Ϫ177 promoter fragment could phenocopy the changes in Habp2 expression by TGF-␤ we investigated this region in detail. We performed in silico analysis to specifically identify transcription factor binding elements that are unique to this fragment compared with other non-responsive control fragments (Ϫ313/Ϫ427 and Ϫ177/Ϫ313, respectively). This showed that the Ϫ1/Ϫ177 fragment contains binding elements for c-jun, c-fos, AP-1, HNF1␣, HNF3␤, and C/EBP␦ (22)(23)(24). No predicted SMAD binding sites were found. To determine the role of c-fos and c-jun, we investigated the protein/DNA interaction of both transcription factors and this promoter region. Nuclear extracts of AML12 cells were incubated with biotinylated dsDNA fragments of the promoter to capture c-fos and c-jun, which were detected using Western blotting. c-fos binding to the Ϫ1/Ϫ177 promoter fragment was 8-fold and 3-fold higher compared with other control DNA fragments (Ϫ313/Ϫ427 and Ϫ177/Ϫ313, respectively), which showed no TGF-␤ responsiveness in the reporter assays (Fig. 7A). c-jun binding of the Ϫ1/Ϫ177 promoter fragment was 2.5-fold higher compared with the Ϫ313/Ϫ427 fragment and was identical compared with the control DNA fragment Ϫ177/Ϫ313 (Fig. 7A). We then compared chromatin binding of c-fos in AML12 cells with or without TGF-␤ treatment. In TGF-␤treated cells c-fos binding to the Ϫ1/Ϫ177 minimal promoter was reduced by 70% compared with no change in the control  . Activity of human HABP2 promoter reporter constructs. A, indicates details of the HABP2 promoter fragments used. B, basal promoter activity of the human HABP2 promoter fragments of various lengths was tested in AML12 hepatocytes. Significant changes relative to the Ϫ1/Ϫ500 fragment are indicated by *. C, cells were pre-treated with SB 431542 (10 M) or 30 min before stimulated with TGF-␤ (100 ng/ml) for 24h before lysis. Activity is shown as mean Ϯ S.E. (n ϭ 5). Since the basal expression of the different constructs varied, the activities of the treated samples are normalized to the untreated sample for that particular construct. Statistically significant difference (p Ͻ 0.05) between control and TGF-␤ as measured by ANOVA is denoted by *. Similar results were obtained in three independent experiments.
Ϫ313/Ϫ427 fragment (Fig. 7B). Even though no predicted SMAD sites were present we performed ChIP of the Ϫ1/Ϫ177 (TGF-␤-responsive) and the Ϫ313/Ϫ427 (TGF-␤-non-responsive) fragments with SMAD2/3 antibody to rule out the presence of low homology or cryptic binding elements. Although there was pull down of the Ϫ1/Ϫ177 fragment, this was not altered with TGF-␤ treatment indicating that this binding was not relevant for the process (Fig. 7C).
Next, we investigated the role of single transcription factor binding sites in basal expression using ATF3, AP-1, HNF1␣, HNF3␤, and C/EBP␦ deletion mutants of the TGF-␤-responsive Ϫ1/Ϫ177 fragment (Fig. 8A). Basal activity was reduced in all mutants to various degrees except C/EBP␦ (Fig. 8B). However, the deletion of AP-1 and C/EBP␦ had no effect on the TGF-␤ mediated down-regulation of the promoter activity. Deletion of HNF1 and HNF3 showed reduced response to TGF-␤ treatment and a near complete loss of TGF-␤ response was detected after deletion of the ATF3 site. These results were confirmed by determining the activity of short fragments containing only the ATF3 transcription factor binding element (Ϫ120/Ϫ177, Ϫ152/Ϫ177). Both fragments showed increased activity upon SB431542 treatment and decreased activity after TGF-␤ treatment (supplemental Fig. S6). This suggests that ATF3 is an important transcription factor binding element in the TGF-␤-mediated down-regulation of Habp2 expression in hepatocytes.

Discussion
Variations in circulating FSAP levels and activity could have an impact on diverse disease processes ranging from thrombosis, liver fibrosis, thyroid cancer, or stroke (12,16,(25)(26)(27). Circulating FSAP levels show considerable inter-individual variations which may be due to genetic and/or environmental factors (28,29). In a few cases, associations of SNPs with circulating levels of FSAP have been reported but not analyzed at a mechanistic level (16,25,27). Complex formation with inhibitors is also likely to modulate its cellular uptake and also influence the circulating levels (30).
Hepatocytes are the major source of circulating FSAP protein and we have made a significant observation that TGF-␤ can downregulate FSAP expression. Furthermore, the molecular mechanism of this inhibition from the ALK5 receptor to the canonical and non-canonical pathways to the 177-bp region of the promoter region was characterized.
Hepatocytes are the major producers of all coagulation factors and certainly interesting was the fact that F2 (prothrombin) and F12 (FXII) genes were also downregulated by TGF-␤. The regulation of these key coagulation factors in hepatocytes by TGF-␤ has never been studied before and unlike Habp2 their regulation has not been observed in comparable microarray analysis (31). Partial hepatectomy in mice causes a proliferative/dedifferentiated phenotype in hepatocytes and this was associated with the down-regulation of the expression of many coagulation factors, including F2 and F12 (32). This points to a coordinated regulation of coagulation factor expression that may be relevant for the interactions between pathophysiological changes in the liver and their consequences for coagulation pathways.
In the canonical TGF-␤ pathway the activation of ALK5 leads to activation of signaling via SMAD2/3-SMAD4 complexes and their interaction with the SMAD binding elements (CAGAC) by association with many cofactors leading to the transcriptional regulation of target genes. SMAD2 and SMAD3 are functionally interchangeable but do have distinct functions. Neither siRNA against SMADS 3 nor a syn-FIGURE 7. c-fos interacts with the minimal Habp2 promoter. A, DNA sequence representing the minimally active and TGF-␤-responsive promoter region (Ϫ1/Ϫ177) and other non-responsive regions (Ϫ313/Ϫ427 and Ϫ177/Ϫ313) were biotinylated and incubated with nuclear extracts from AML12 cells. After pull-down with streptavidin magnetic beads the bound fraction was subjected to Western blotting analysis with an anti-c-fos antibody or anti-c-jun antibody. Band density was quantified, and results represent mean Ϯ S.E. (n ϭ 3). B/C, ChIP was performed on nuclear extracts of AML12 cells with or without TGF-␤ stimulation for 24 h. Pull down was performed with an anti-c-fos (B) or an anti-SMAD 2/3 (C) antibody. Intact extract was used as an input control. PCR was performed to amplify the minimally active and TGF-␤-responsive promoter region (Ϫ1/Ϫ177) and other non-responsive region (Ϫ313/Ϫ427). Values represent mean Ϯ S.E. (n ϭ 3 independent experiments). Statistically significant results (p Ͻ 0.05) compared with untreated cells as measured by ANOVA is denoted by *.  Deletion mutations within the 177-bp HABP2 promoter. A, transcription factor binding sites were deleted from the minimally active and TGF-␤-responsive promoter region (Ϫ1/Ϫ177) as indicated in Fig. 6A. B, basal reporter activity was measured in AML12 cells. C, reporter activity was measured in the absence or presence of TGF-␤ (10 ng/ml) for 24 h. Values represent mean Ϯ S.E. (n ϭ 3). Since the basal expression of the different constructs varied, the activities of the treated samples are normalized to the untreated sample for that particular construct. Statistically significant results (p Ͻ 0.05) compared with TGF-␤-treated cells transfected with the control (Ϫ1/Ϫ177) construct was determined by ANOVA and is denoted by *.
thetic SMAD3 inhibitor could modulate this action of TGF-␤. However, knockdown of SMAD2 significantly influenced the down-regulation of Habp2 by TGF-␤. Of the many non-canonical pathways activated by TGF-␤ (20), we did not find any influence of inhibiting the p38MAPK, ERK1, and Akt pathways. The 177-bp promoter region, that exhibited all the characteristics of the TGF-␤ response, did not have any predicted SMAD binding sites and ChIP analysis with SMAD2/3 antibody also failed to detect any TGF␤ response. Thus further studies are required to complete the understanding of the pathway involved.
On the other hand, there was a robust pull down of this fragment with a c-fos antibody that was decreased in the presence of TGF-␤. In a corollary experiment this DNA fragment could pull down c-fos protein, but not c-jun, in a specific manner. Further mutation studies suggested a role of c-jun/c-fos, HNF1␣, and HNF3 in the basal expression of the HABP2 gene, since deletion of any of those caused reduced expression in AML12 cells. When we treated the different mutants with TGF-␤ only the ⌬ATF3 mutant showed a significant loss of responsiveness, indicating an important role of ATF3 binding element in this process. ATF3, also known as liver regeneration factor 1 (LRF-1), plays an important role in the stress-response of the liver after partial hepatectomy or CCl 4 treatment (33,34). After CCl 4 treatment ATF3 forms a heterodimer with Jun B and represses gene expression by competing with the c-jun/c-fos complex for the same binding sites (33). c-jun, Jun B and ATF3 are up-regulated in three different human epithelial cell lines upon TGF␤ treatment, while C/EBP␦ is down-regulated. On the basis of these results we hypothesize that upon TGF␤ treatment levels of ATF3 and junB increase, effectively replacing c-jun/c-fos binding to the promoter, which leads to reduced Habp2 expression (35). A similar model has been proposed for the up-regulation of PAI-1 after TGF␤ treatment (36).
Human genetic studies as well as the analysis of FSAP knockout mice indicate an important role for FSAP in the pathophysiology of liver fibrosis, thrombosis, stroke and thyroid cancer. We have for the first time performed a comprehensive analysis of factors that may regulate FSAP expression in hepatocytes and have identified TGF-␤ as a major inhibitory factor. This cytokine is a pleiotropic factor that regulates growth, apoptosis, differentiation, epithelial-mesenchymal transition, deposition of extracellular matrix, and is a strong immune-modulatory factor. It is strongly up-regulated in liver fibrosis and plays a crucial role in the development of liver fibrosis and hepatocellular carcinoma (3). This together with the fact that FSAP can influence the activation of hepatic stellate cells (11), indicates that FSAP could be part of a local paracrine loop in the liver that is regulated by TGF-␤.

Experimental Procedures
Cell Culture-Mouse hepatocyte cell line AML12 (ATCC® CRL-2254 TM ) was cultivated in DMEM (Invitrogen, Darmstadt, Germany) with 10% (v/v) fetal calf serum (FCS), Thermo-Fisher Scientific (Fermentas), Rockford), 10 units/ml penicillin and 10 g/ml streptomycin (Invitrogen) on cell culture-treated plastic (Nunc, Wiesbaden, Germany). Mouse primary hepato-cytes were isolated as described earlier by collagenase perfusion (2) and were cultivated as above. All cells were maintained in a humidified atmosphere of 5% CO 2 at 37°C. All animal studies performed in this study were approved by the local authority for animal welfare; Regierungspraesidium Giessen, Germany and Mattylsynet, Oslo, Norway. The mice were treated in accordance with the recommendation of the guidelines for the care and use of Laboratory animals of the National Institute of Health (USA) and the local authorities.
Western Blotting Analysis-SDS-PAGE was performed, and proteins were transferred to Hybond-P PVDF membrane (GE Healthcare, Frankfurt, Germany). For analysis of Western blotting chemiluminescence plus reagent (GE Healthcare) was used. Polyclonal rabbit antibody against FSAP was kindly provided by Michael Etscheid, Paul Ehrlich Institute, Langen, Germany. SMAD 2/3 was used as a loading control and densitometric analysis was performed to calculate relative expression using the LabImage 1D (Kapelan, Leipzig, Germany) system.
RNA Isolation and qPCR Analysis-Total RNA was extracted using total RNA Miniprep Kit (Sigma Aldrich). Reverse transcription was performed using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). For real-time PCR, SensiMix SYBR Kit (Bioline GmbH, Luckenwalde, Germany) was used, and analysis was performed on a Step One Plus Real-Time PCR System (Applied Biosystems, Darmstad). Melting curve was performed at 60°C for 1 min following 95°C. Amplification (cDNA denaturation 95°C for 15 s, primer hybridization 60°C for 30 s, elongation 72°C for 30 s) plot was monitored over 40 cycles and continuous fluorescence measurement indicated mRNA expression of analyzed genes. Fluorescent threshold cycles (ct) were set and normalized against ct of reference genes Gapdh or Gusb (37). Primer sequences are provided in supplemental Table S1.
Construction and Mutations of the HABP2 Promoter-Various genomic fragments from the human HABP2 gene were cloned into the luciferase reporter vector pGL4.10[luc2] (Promega, Mannheim, Germany). These constructs served as templates for smaller constructs, deletion mutants and for sitedirected mutagenesis. These constructs included up to 2400 bp of the 5Ј region of the human HABP2 gene directly upstream of the transcription start site as defined earlier (18). Site-directed mutagenesis was performed using QuickChange Site-directed Mutagenesis Kit (Stratagene, La Jolla, CA). Details of the con-structs are shown in the results section. DNA sequencing was performed to check for the correct sequence of the resulting clones, deletion, and other mutants.
Transfection and Reporter Assays-AML12 or HepG2 cells were transfected with empty vector as well as the various constructs (500 ng) using Lipofectamine2000 (Invitrogen). Adjustment for the relative transfection efficiency was performed by co-transfecting the Renilla vector (pGL4.75[hRluc/CMV]) (Promega). Cells were then switched to serum-free medium with or without TGF-␤ for 24 or 48 h. After lysis, determination of the firefly and Renilla luciferase was performed using a Dual luciferase kit from Promega. Normalized Renilla luciferase activity to control firefly activity was calculated and expressed as fold change compared with empty vector.
Protein/DNA Interaction Pull-down Assay-Confluent AML12 cells were untreated or treated with 100 ng/ml TGF-␤ for 48 h in serum-free medium or regular growth medium. Lysis buffer containing 10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM DDT was added to the cells to induce swelling and bursting of cells for 20 min on ice. After centrifugation in a microfuge the pellet was resuspended in a buffer containing 50 mM HEPES pH 7,8, 3 mM MgCl 2 , 300 mM NaCl, 1 mM DTT, 0.1 mM PMSF as added to the pellet and subjected to sonification. Streptavidin-conjugated Dynabeads (Thermo Fisher, 11205D) were prepared and used according to the manufacturer's protocol. The immobilization of nucleic acids was perform using biotinylated DNA baits which were amplified from plasmid DNA containing the FSAP promoter region. The same primer sequences as in the ChIP were used (supplemental Table S1).
Chromatin Preparation by Sonication and Immunoprecipitation Assays (ChIP)-The chromatin preparation by sonication has been performed as described before; 10 7 cells were disrupted by sonication to prepare fragmented chromatin. These were incubated overnight at 4°C with antibody (anti-cjun SC-1694 and anti-cfos SC-52, both Santa Cruz) pre-coupled for 2 h to magnetic Dynabeads Protein G (Invitrogen) as described before (38).
Immunofluorescence Analysis-Primary hepatocytes were fixed with 4% (w/v) paraformaldehyde for 10 min followed by permeabilization with 0.2% (w/v) Triton X-100. After blocking with 3% (w/v) BSA the cells were stained with the primary antibody followed by an appropriate secondary antibody. Nuclei were stained with (4Ј,6Ј-diamidino-2-phenylindole) (DAPI), and the slides were mounted in Vectashield TM mounting medium (Biozol Diagnostica). Controls were performed by including an isotype-matched negative control primary antibody. Images were taken on a Leica DMR HC fluorescence microscope (Leica Microsystems, Wetzlar, Germany), and image analysis was performed using Metamorph software, Meta Series 7.0 (Visitron Systems, Puchheim, Germany).
Statistical Analysis of the Data-Statistical significance was tested using Analysis of Variance (ANOVA) and the individual comparisons were made using Bonferroni test using Graphpad Prism5 (version 5.02) (GraphPad Software, Inc, La Jolla, CA). In some cases a Student's t test was used. A value of p Ͻ 0.05 was considered to be statistically significant.
Author Contributions-S. L. performed most of the experiments and analyzed the data and edited the manuscript. S. S. conducted experiments deleting TFBS in the promoter and measured luciferase activity under various conditions, ChIP experiments and edited the manuscript. A. M. P. performed the expression experiments and edited the manuscript. L. M. performed immunofluorescence analysis and edited the manuscript. S. M. K. designed the study, analyzed the data and wrote the manuscript together with S. L.