Editorial Expression of Concern

The Journal of Pharmacology and Experimental Therapeutics is publishing an Editorial Expression of Concern regarding allegations of figure manipulation or figure duplication in several figures published in JPET. An investigation by an inquiry committee at the University of Perugia, Italy has confirmed or validated findings compatible with alleged hypotheses of electronic duplication and/or figure manipulation. The figure in question in this article is Fig. 6.

************************************************************************

Hepatic fibrosis is a scarring process of the liver that includes components of both increased and altered deposition of extracellular matrix (ECM) and reduced breakdown of ECM components (Mann and Smart, 2002; Friedman, 2003). In chronic liver disease, hepatic stellate cells (HSCs), the major source of ECM in the liver, undergo a process of trans-differentiation from a resting, fat-storing phenotype to a myofibroblast-like phenotype characterized by expression of fibroblastic cell markers such as α-smooth muscle actin (α-SMA) (Mann and Smart, 2002; Friedman, 2003). In addition, there is now considerable evidence that HSCs are a source of both matrix-degrading metalloproteinases (MMPs), including those that degrade type I collagen and the tissue inhibitors of metalloproteinases (TIMPs) (Arthur, 2000). The prototypic member of the TIMP family, TIMP-1, is a broad inhibitor of MMPs, possesses growth factor-like and anti-apoptotic properties (Murphy et al., 2002), and, by promoting survival of HSCs, is mechanistically involved in the development of liver fibrosis (Iredale et al., 1998; Yoshiji et al., 2000, 2002; Smart et al., 2001).

Previous studies have shown that JunD, the predominant Jun family protein expressed in culture-activated rat HSCs, regulates TIMP-1 expression in HSCs (Bahr et al., 1999; Trim et al., 2000; Manning and Davis, 2003). The TIMP-1 promoter contains an activating protein (AP)-1 binding site, and mutation of this site results in greater than 90% reduction of expression of TIMP-1 promoter-reporter constructs in the transient transfection assays (Bahr et al., 1999; Trim et al., 2000; Manning and Davis, 2003).

The farnesoid X receptor (FXR) is a member of the nuclear receptor superfamily of ligand-activated transcription factors that functions as an endogenous sensor for bile acids (Forman et al., 1995; Makishima et al., 1999; Parks et al., 1999). FXR is bound to and activated by bile acid and chenodeoxycholic acid (CDCA) is the natural most active ligand (Forman et al., 1995; Makishima et al., 1999; Parks et al., 1999). Upon binding of a bile acid, FXR alters gene transcription by interacting with the 9-cis-retinoic acid receptor (RXR) (Wang et al., 1999). In liver cells, FXR activation leads to induction of an orphan nuclear receptor termed the short (or small) heterodimer partner (SHP) that mediates inhibition of FXR ligands on a cohort of genes that function to decrease the concentration of bile acids within hepatocytes (Wang et al., 1999; Goodwin et al., 2000; Lu et al., 2000; del Castillo-Olivares and Gil, 2001). FXR is expressed by HSCs, and FXR ligands function as negative regulators of α1(I)collagen synthesis both in vitro and in vivo in rodent models of liver fibrosis (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci et al., 2004), though the mechanism(s) is still not completely understood.

Here, we report that FXR functions as a negative regulator of TIMP-1 gene expression in HSCs. Activation of FXR causes an SHP-dependent inhibition of JunD binding to the AP-1 binding site in the TIMP-1 promoter, reducing TIMP-1 expression/function and increasing the susceptibility of HSCs to apoptogenic stimuli. In addition, we provide evidence that in vivo activation of FXR decreases TIMP-1 expression/activity and promotes resolution of liver fibrosis.

Materials and Methods

Isolation and Culture of HSCs. In vitro studies were performed on primary cultures of rat HSCs; HSC-T6, a rat immortalized cell line; and on LX2 cells, a human HSC line (Fiorucci et al., 2004a). Primary rat HSCs were isolated from control and cirrhotic rats according to techniques described previously (Fiorucci et al., 2004b) and cultured at 37°C in an atmosphere of 5% CO₂ in Dulbecco's modified minimal essential medium (Invitrogen, Carlsbad, CA) containing 10% fetal calf serum, 2 mM l-glutamine, and 5000 IU/ml penicillin/5000 g/ml streptomycin for 1 to 2 days before starting experiments.

To investigate the expression of FXR and FXR-regulated genes in HSCs, HSCs (days 0 and 7) and 24-h serum-starved HSC-T6 and LX2 cell lines were incubated for 18 h with the medium alone or with 1 to 50 μM CDCA; 0.1 to 10 μM 6-ECDCA, also indicated as INT-747 (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci et al., 2004a, 2005); or 0.01 to 1 μM GW4064 (Maloney et al., 2000), a nonsteroidal FXR ligand; and FXR, α1(I)collagen, α-SMA, SHP, TIMP-1, TIMP-2, MMP-2, and TGFβ1 mRNA expression investigated by quantitative (q)RT-PCR (Fiorucci et al., 2004a, 2005).

MMP-2 Activity and TIMP-1 ELISA. MMP-2 levels were measured using an ELISA detection kit (Biotrak; Amersham Biosciences, Inc., Piscataway, NJ). This kit measures both the active and total (pro and active) forms of MMP-2. Only data on the activated form are shown. Secretion of TIMP-1 protein by HSCs was measured using a specific ELISA kit (Quantikine rat TIMP-1 immunoassay; R&D Systems, Minneapolis, MN).

Generation of SHP-Overexpressing HSC-T6 (HA-SHP). The SHP coding sequence was cloned from primary rat hepatocytes (Seol et al., 1996; Fiorucci et al., 2004a) induced with 6-ECDCA. For cell transfection, SHP was first cloned into the pCR2.1 vector (TOPO-TA cloning; Invitrogen) and then subcloned in the EcoRI site of the PINCO retroviral vector. 293T modified packaging cell line (ϕNX) were transiently transfected with either the PINCO-SHP chimera or PINCO alone as a negative control. At 48 h post-transfection, the viral supernatant was recovered and used to infect HSC-T6. The PINCO vector contains the emerald green fluorescent protein gene that allows the separation of infected cells (green) from uninfected cells (Fiorucci et al., 2004a). HSC-T6 cells expressing either SHP or empty vector alone were obtained by cell sorting (Beckman Coulter, Milan, Italy).

Silencing of SHP. Selection of siRNAs was based on the characterization of siRNA by Elbashir et al. (2001). A 21-nucleotide RNA with 3′-dAA overhangs was synthesized with Dharmacon 2′ACE technology (Dharmacon Research, Lafayette, CO) in the“ ready-to-use“ option. The primers sequences were 5′-AAGGAGUACGCAUACCUGAAA-3′ and 5′-AACAUCUCCUGUACCCUUCUG-3′ (corresponding to base pairs 498–518 and base pairs 684–704 of the coding sequence of the SHP gene). SHP-siRNAs (15 nM) were introduced into HSC-T6 cell line by transient transfection with Transit TKO transfection reagent (Mirus, Madison, WI). The effect of siRNA was evaluated by measuring the relative expression of SHP mRNA by qRT-PCR (Fiorucci et al., 2004a).

Transactivation Assay and Plasmid Construction. The fragment corresponding to the promoter region of rat TIMP-1 (region between nucleotides–182 and –13) was amplified from genomic DNA obtained from the HSC-T6 cell line (DNAzol; Invitrogen) by polymerase chain reaction using Phusion DNA polymerase (FINNzymes, Espoo, Finland). The fragment was cloned into pCR-Blunt vector (Invitrogen) and then subcloned into KpnI and XhoI sites upstream of the luciferase reporter gene into pGL3 vector (Promega, Madison, WI). The sequences of the primers used were rat TIMP1 (–182) 5′-caggaaggaatgtgcatgacg-3′ and rat TIMP1 (–13) 5′-gttggcgcggaaatcactg-3′. All transfections were performed using the calcium phosphate coprecipitation method in the presence of 25 μM chloroquine. Twenty-four hours before transfection, HSC-T6 were seeded onto six-well plates at a density of 250,000 cells/well. Transient transfections were performed using 500 ng of reporter vector pGL3 vector-Luc, 200 ng of pCMV-βgal as internal control for transfection efficiency, and 50 ng of expression plasmid (pSG5-TIMP-1). The pGEM vector (Promega) was added to normalize the amounts of DNA transfected in each assay (2.5 μg). At 36 to 48 h post-transfection, cells were stimulated with 1 μM 6-ECDCA for 18 h. Control cultures received vehicle (0.1% DMSO) alone. Cells were lysed in 100μ l of diluted reporter lysis buffer (Promega) and 10 μl of cellular lysates were assayed for luciferase activity using the luciferase assay system (Promega). Luminescence was measured using an automated luminometer. Luciferase activities were normalized for transfection efficiencies by dividing the relative light units by β-galactosidase activity expressed from cotransfected pCMV-βgal. Each data point was the average of triplicate assays and repeated three times.

Western Blot Analysis. Confluent cultures of HSCs, HSC-T6, and LX2 cell lines were serum starved for 24 h and then incubated for 18 h at 37°C with 10 U/ml thrombin alone or in combination with 1 μM 6-ECDCA. Total lysates were prepared by solubilization of cells in sample buffer (62.5 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, and 0.015% bromphenol blue) and separated by polyacrylamide gel electrophoresis. The proteins were then transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA) and probed with primary antibodies to c-Jun, JunD, SHP, FXR, α1(I)collagen, TIMP-1, and MMP-2 (Santa Cruz Biotechnology Inc., Santa Cruz, CA), HA epitope tag (Covance, Berkeley, CA), or α-SMA (Abcam, Cambridge, UK). The anti-immunoglobulin G horseradish peroxidase conjugate (BioRad) was used as the secondary antibody, and specific protein bands were visualized using enhanced chemiluminescence (Amersham Biosciences, Inc.).

Electrophoretic Mobility Shift Assays. The synthesis of [γ-³²P]ATP (Amersham Biosciences, Inc.) radiolabeled probes used for shift analysis (EMSAs) has been described previously (Fiorucci et al., 2004a). For EMSAs, 10 μg of nuclear extract from primary HSCs, HSC-T6, and HA-SHP were incubated with 50,000 cpm of the indicated γ-³²P end-labeled probe in a total volume of 20 μl of binding buffer [50 mM NaCl, 10 mM Tris, pH 7.9, 0.5 mM EDTA, 1 μgof poly(dI-dC), and 10% glycerol] for 20 min at room temperature. For competition assays, excess unlabeled oligonucleotides were preincubated for 15 min before the addition of the radiolabeled probe (50,000 cpm). For antibody-mediated supershift assays, extracts were preincubated with 5 μl of either anti-SHP or anti-JunD antibodies at room temperature for 20 min before the addition of the radiolabeled probe. The reactions were loaded on a 6% polyacrylamide nondenaturing polyacrylamide gel electrophoresis gel in low ionic buffer electrophoresed for 2.5 h at 100 V before drying and exposed to autoradiographic film. The following oligonucleotides were used to construct probes for EMSA experiments: AP-1 consensus (sense), 5′-TATAAAGCA TGAGTCA GACACCTCT-3′; AP-1 consensus (antisense), 5′-AGAGGTGTCTGACTCATGCTTTATA-3′; TIMP1-AP-1 (sense), 5′-TGGGTGGA TGAGTAA TGC-3′; and TIMP1-AP-1 (antisense), 5′-GCATTACTCATCCACCCA-3′.

Detection of Apoptosis. Apoptosis of HSCs and HSC-T6 was induced by cycloheximide treatment (Murphy et al., 2002). HSC were cultured in 18-well tissue culture plates. Cells were exposed to cycloheximide with and without 1 to 10 μM 6-ECDCA. In the same experiments, recombinant 100 mg/ml TIMP-1 (EMB-Calbiochem, San Diego, CA) was added to the incubation medium. After a 24-h incubation at 37°C, apoptotic cells were detected by flow cytometry after staining with fluorescein isothiocyanate-conjugated Annexin V and propidium iodide (PI) by using a commercially available kit (Annexin V-FITC kit; Immunotech, Marseille, France) as described elsewhere (Fiorucci et al., 2002). Cells were considered apoptotic when they were Annexin V-positive and PI-negative. Staining of cells by PI was an indicator of the loss of plasma membrane integrity. Caspase-3 activity was measured by using a commercial kit (Fiorucci et al., 2002). Briefly, HSCs lysates were incubated with the substrate DEVD-p-nitroanilide (pNA) at 37°C for 60 to 90 min. Results of the reaction were read by a spectrophotometer at 405 nm. A standard calibration curve of pNA was established by a series dilution of pNA solution provided with the kit. Each treatment was performed at least three times. Cytochrome c translocation from mitochondria to cytosol was used as a marker of mitochondrial injury. The preparation of mitochondrial and cytosolic S-100 fractions and Western blot analysis of cytochrome c were performed as described previously (Fiorucci et al., 2002). Protein concentration of the extracts was determined by the Bradford method. Mitochondrial and cytosolic cell lysates (50 mg) were subjected to SDS/polyacrylamide gel electrophoresis. The blot was stained with 0.5% Ponceau S to ensure equal protein loading and transfer. Afterward, the blot was incubated at room temperature with the anti-cytochrome c antibody (BD Biosciences PharMingen, San Diego, CA) diluted 1:1000 in 5% nonfat milk in Tris-buffered saline/Tween 20 for 1 h.

In Vivo Studies. All studies were approved by the Animal Study Committee of the University of Perugia. Male Wistar rats (200–250 g) were obtained from Charles River Laboratories (Monza, Italy) and maintained on standard laboratory rat chow with a 12-h light/dark cycle. Liver fibrosis was induced in rats by intraperitoneal (i.p.) injection of CCl₄, 100μ l/100 g body weight in an equal volume of paraffin oil two times a week for 4 weeks as described previously (Fiorucci et al., 2004a). Control rats were injected i.p. with 100 μl/100 g body weight of paraffin oil alone. Rats were then treated by oral administration of 3 mg/kg 6-ECDCA or 15 mg/kg UDCA in 3% carboxymethyl cellulose (CMC) or 3% CMC alone (control). To investigate the effect of FXR activation on the time-dependent expression of TIMP-1 mRNA, experimental and control rats (4–6 per group) were sacrificed on days 3, 5, and 28 after the first CCl₄ administration. Livers and blood samples were collected for histology, biochemical workup, and RNA extraction.

In a second study, we investigated whether FXR activation promotes fibrosis resolution. For this purpose, we induced fibrosis by 4 weeks administration of CCl₄ (same protocol as described above). Animals were then randomized to receive no further treatment, 3% CMC (control) by gavage or 3 mg/kg 6-ECDCA. Groups of four animals were then sacrificed at 1 (fibrosis peak), 3, 5, and 7 days after the last CCl₄ dose, and livers and blood were collected.

Liver Histology, Immunohistochemistry, and Hydroxyproline Determination. For histological examination, portions of the right and left liver lobes (10–15 mg/each) from each animal were fixed in 10% formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin or Sirius red (Lopez-De Leon and Rojkind, 1985; Fiorucci et al., 2004a). Collagen surface density was quantified using a computerized image analysis system (Image Acquisition System version 005; Delta Sistemi, Rome, Italy) (Fiorucci et al., 2004a). Apoptotic cells were stained by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL). HSCs activation was detected by simultaneous staining of α-SMA, as a marker of activated HSCs, and of PCNA as a marker of S-phase nuclei as described previously (Fiorucci et al., 2004a). HSCs activation was detected by simultaneous staining ofα -SMA, as a marker of activated HSCs, and of PCNA as a marker of S-phase nuclei (Fiorucci et al., 2004a,b).

Apoptosis on nonparenchymal cells in liver section was achieved by TUNEL staining of rat liver sections. The TUNEL staining was performed according to previously published methods (Iredale et al., 1998; Fiorucci et al., 1999). Each liver section was analyzed by a blinded observer who counted the number of nonparenchymal apoptotic figures/bodies present in 10 random high power (40×) fields from each specimen. Apoptotic bodies in a distribution compatible with parenchymal cells (hepatocytes) were not counted. Hepatic and urinary content of hydroxyproline (Fiorucci et al., 2004a) were determined by high-performance liquid chromatography (LC Varian Prostar high-performance liquid chromatograph; Varian, Rome, Italy).

Statistical Analysis. Analysis of variance or the Student's t test was used for statistical comparisons. Statistical tests were performed using INSTAT statistical software (GraphPad Software, Inc., San Diego, CA).

Results

FXR Is Expressed in HSCs and Functions as Negative Regulator of TIMP-1. FXR expression was found in both rodent and human HSCs (Fig. 1a). Thus, FXR, mRNA and protein were detected in primary cultures of HSCs, in HSC-T6, and in LX2 cells. FXR expression did not change significantly during the process of trans-differentiation of rats HSCs in primary cultures (data not shown) or after the addition of FXR ligands to the culture medium (Fig. 1a). As shown in Fig. 1, b through g, HSCs in 7-day culture demonstrated a significant increase in α1(I)collagen,α -SMA (not shown), TIMP-1, and MMP-2 compared with day 0 HSCs (n = 4–6; P < 0.01 versus day 0). HSC activation is associated with a significant increase of TIMP-1 release and MMP-2 activity (n = 4–6; P < 0.01 versus day 0). Whereas SHP expression did not change during the process of trans-differentiation, exposure of HSCs to 1 μM 6-ECDCA increased its expression 3-fold (Fig. 1b; n = 4–6; P < 0.01 versus medium) and reducedα 1(I)collagen and TIMP-1 protein expression and secretion by 70 to 80% (Fig. 1, c, d, and f; n = 4–6; P < 0.01 versus medium). Finally, whereas 6-ECDCA had no effect on MMP-2 mRNA (Fig. 1e; n = 4–6; P > 0.05 versus medium), it significantly increased MMP-2 activity (Fig. 1g; n = 4–6; P < 0.01 versus medium). To investigate whether the effect of FXR activation on TIMP-1 expression was maintained in HSCs activated by growth factors, day-7 cultured HSCs were stimulated with 10 U thrombin for 18 h in the presence of natural and synthetic FXR ligands. Exposure to thrombin had no effect on FXR and SHP mRNA, but up-regulatedα 1(I)collagen, α-SMA, TIMP-1, TIMP-2, and MMP-2 gene expression and TIMP-1 release (not shown; n = 6; P < 0.01 versus control cells). Exposure to 1 μM 6-ECDCA caused a 2- to 3-fold induction of SHP (not shown; P < 0.01 versus thrombin) and reduced induction ofα 1(I)collagen, α-SMA, TIMP-1, TIMP-2 mRNA, and TIMP-1 release induced by thrombin by 50 to 80% (not shown; n = 6; P < 0.01 versus thrombin).

SHP Mediates Inhibition of FXR Ligands on TIMP-1. The qualitative and quantitative RT-PCR and Western blot analysis shown in Fig. 2, a and b, demonstrated that transfection of 15 nM SHP siRNA caused an ≈90% reduction of SHP expression (n = 5; P < 0.001 versus untransfected cells). As shown in Fig. 2c, SHP-defective HSC-T6 demonstrated significantly higher levels of TIMP-1 mRNA and released higher amounts of TIMP-1 (not shown) compared with wild-type HSC-T6. Furthermore, although abrogating the expression of SHP had no effect on MMP-2 mRNA expression (not shown), SHP-defective cells have significantly higher MMP-2 activity compared with wild-type HSC-T6 (Fig. 2d; n = 6; P < 0.01 versus wild type).

In contrast to these findings, SHP overexpression inhibits TIMP-1 expression. Thus, TIMP-1 mRNA and TIMP-1 release (not shown) was reduced by 70 to 90% in SHP-overexpressing cells (Fig. 2, e–h; n = 6; P > 0.001 versus wild type). Although SHP overexpression had no effect on MMP-2 mRNA levels (not shown), it significantly increased MMP-2 activity (Fig. 2h; n = 6; P > 0.01 versus HSC-T6).

FXR Activation Inhibits AP-1 Binding to the TIMP-1-Promoter. Because SHP overexpression inhibits TIMP-1 mRNA expression and TIMP-1 release is induced by thrombin, a known AP-1 inducer, we then investigated whether SHP interacts with AP-1 and modulates its binding to the TIMP-1 promoter. For this purpose, lysates from wild-type and SHP-overexpressing HSC-T6 were immunoprecipitated with a monoclonal anti-HA or polyclonal anti-JunD antibodies and analyzed by Western blotting using the reverse antibody used for immunoprecipitation. As shown in Fig. 3A, exposure of HA-SHP to thrombin induces JunD expression. In coimmunoprecipitation experiments (Fig. 3A) HA-SHP immunoreactivities were detected in anti-JunD immunoprecipitates. Similarly, JunD immunoreactivities were found in HA-SHP immunoprecipitates (Fig. 3B), indicating that SHP directly interacts with AP-1. We next analyzed whether JunD-SHP interaction prevents binding of JunD to the AP-1 binding site in the TIMP-1 promoter. For this purpose, an EMSA was performed using nuclear extracts from wild-type and SHP-overexpressing HSC-T6 cells and a probe with the AP-1 response element obtained from the TIMP-1 promoter. As shown in Fig. 3C, AP-1 binding was found in wild-type HSCs infected with the empty vector and in HA-SHP cells left untreated. SHP overexpression, however, was sufficient to reduce the binding of AP-1 to the TIMP-1 promoter. Addition of thrombin enhances AP-1 binding in wild-type cells, but not in the HA-SHP cells. Exposure to 6-ECDCA, 1 μM, reduces AP-1 binding in HSC-T6, but not in HA-SHP cells likely due to the inability of these cells to further augment the level of SHP protein in response to FXR ligands. This was confirmed by real-time polymerase chain reaction of SHP mRNA (data not shown).

We have thus investigated whether FXR ligands directly modulate TIMP-1 gene transcription by transactivation assay using a sequence containing the AP-1 consensus from the TIMP-1 promoter cloned upstream to a luciferase reporter gene in the pGL3 vector. As shown in Fig. 3D, exposure to thrombin causes the transactivation of the TIMP-1 promoter in HSC-T6 cells transfected with pGL3-TIMP-1 chimera (black columns). In this experimental setting, exposure to 1 μM 6E-CDCA inhibits TIMP-1 transactivation (black columns). Similarly to the exposure to the FXR ligand, transactivation of TIMP-1 promoter induced by thrombin in HA-SHP cells was inhibited by SHP overexpression (gray columns). Again, addition of 6-ECDCA to HA-SHP cells failed to further increase the inhibition caused by SHP overexpression.

FXR Ligands Cause a TIMP-1-Dependent Increase of HSCs Apoptosis. As seen in Fig. 4a, although exposure to 1 μM 6-ECDCA had no effect on the rate of HSC-T6 apoptosis, coincubating 6-ECDCA with 50 μM cycloheximide increased the rate of apoptosis induced by the protein synthesis inhibitor by ≈90% (from 23.5± 3.6 to 43.2 ± 4.2%; n = 6; P < 0.01). Similar to 6-ECDCA, 20 μM CDCA and 100 nM GW4064 increased the rate of apoptosis induced by cycloheximide (Fig. 4b; n = 6; P < 0.01 versus cycloheximide alone). Exposure of HSC-T6 to 100 ng/ml rTIMP-1 effectively reduced the rate of apoptosis induced by cycloheximide alone or in combination with 6-ECDCA (Fig. 4c; n = 6; P < 0.01). Furthermore, whereas cycloheximide increased caspase 3 activity (Fig. 4d), and this effect was significantly enhanced by coincubating the cells with 6-ECDCA (n = 6; P < 0.05 versus cycloheximide), exposure of HSC-T6 to 100 ng/ml rTIMP-1 significantly reduced caspase 3 activity (Fig. 4d; n = 6; P < 0.05). Translocation of cytochrome c from the mitochondria to the cytosol correlates with loss of potential of the mitochondrial membrane (Fiorucci et al., 2002) and is considered a marker of mitochondrial injury. Although exposure to cycloheximide, alone or in combination with 6-ECDCA, correlated with relocation of cytochrome c in the cytosol, this event was prevented by incubating cells with rTIMP-1 (Fig. 4e; n = 4).

SHP-overexpressing HSC-T6 were significantly more sensitive to apoptosis induced by cycloheximide than wild-type cells (Fig. 4, f and g; P< 0.05). In contrast, SHP-deficient HSC-T6 cells, generated by siRNA, were protected against apoptogenic stimuli as assessed by the measurement of Annexin V+/PI– cells and caspase 3 activity (P < 0.05 versus wild-type cells).

In Vivo FXR Activation Prevents Fibrosis Development and TIMP-1 Induction. CCl₄ administration up-regulated liver expression of a number of genes involved in liver fibrosis development (Fig. 5, a–e) and caused a rapid induction of TIMP-1 and TIMP-2 mRNA (Fig. 5, f and g; n = 4; P < 0.01 versus control). The expression of TIMP-1 mRNA increased by ≈30 fold in response to a single administration of CCl₄ and declined at later time points (n = 4; P< 0.01 versus control rats). Consistent with these findings TIMP-1 protein, as measured by ELISA in liver homogenates, was found maximally elevated in liver samples collected 3 to 5 days after the first CCl₄ administration and declined in a later phases. In contrast,α 1(I)collagen and α-SMA mRNA expression increased over time reaching the peak at 4 weeks after the first dose of CCl₄ administration. Administration of rats rendered cirrhotic by CCL₄ with 3 mg/kg 6-ECDCA caused a 60 to 80% reduction of TIMP-1 and TIMP-2 mRNA expression at days 3 and 5 (Fig. 5; P < 0.05 versus CCl₄ alone), whereas it had no effect on MMP-2 gene expression. However, MMP-2 activity at days 3, 5, and 28 increased significantly in response to the FXR ligand (Fig. 5; n = 4; P < 0.01 versus CCL₄). Consistent with these findings, 4-week administration of 6-ECDCA reduced α1(I)collagen and α-SMA by ≈70%. In contrast to 6-ECDCA, 15 mg/kg UDCA had no effect onα 1(I)collagen, α-SMA, TIMP-1, and TIMP-2 mRNA expression, nor it did modulate the activity of either TIMP-1 or MMP-2 (Fig. 5).

The morphometric analysis (Fig. 6, a–f) of liver samples obtained from rats administered CCl₄ after 28 days revealed a significant increase in the fibrotic surface compared with control rats (Fig. 6b), resulting in a continuous meshwork of connective tissue infiltrating the hepatic parenchyma with central-central, central-portal, and portal-portal bridging. In addition, CCl₄ increased liver hydroxyproline content 5-fold (P < 0.01 versus control rats) and urinary excretion of hydroxyproline by 3-fold (Fig. 6, g and h; P< 0.01 versus control rats). Administering rats with 6-ECDCA protected against liver fibrosis development. Upon morphometric analysis the area of liver parenchyma occupied by fibrotic tissues was reduced by ≈70% after 6-ECDCA treatment (P < 0.01 versus CCL₄) and as shown in Fig. 6c, bridging was significantly attenuated by 6-ECDCA. 6-ECDCA administration also reduced liver and urinary hydroxyproline content (Fig. 6, a–h) and significantly attenuated the increase in body/liver weight ratio caused by CCL₄ administration (Table 1). In contrast to 6-ECDCA, UDCA failed to reduce liver fibrosis (Fig. 6). Attenuation of liver fibrosis in animals treated with the FXR ligand was not due to a reduced liver toxicity of CCL₄. Consistent with this view, analysis of aspartate aminotransferase and alanine aminotransferase plasma levels measured on days 3, 5 (not shown), and 28 demonstrates a similar liver injury in rats treated with CCL₄ alone or in combination with 6-ECDCA or UDCA.

These findings were further confirmed by analysis of TIMP-1 and MMP-2 expression in lysates obtained from HSCs isolated from CCL₄-treated rats. Administering rats with 3 mg/kg 6-ECDCA significantly reducedα -SMA and TIMP-1 expression caused by CCL₄ (Fig. 6i), whereas it had no effect on MMP-2 expression. In vivo administration of 6-ECDCA increased SHP and FXR expression in HSCs (Fig. 6i). In contrast, no effect on α-SMA, TIMP-1, MMP-2, SHP, and FXR was detected in HSCs obtained from animals treated with UDCA.

FXR Ligand Promotes Liver Fibrosis Resolution. In rats administered CCl₄ for 4 weeks, resolution of liver fibrosis become evident as early as 5 to 7 days after the last dose of CCl₄. The qRT-PCR analysis shown in Fig. 7 demonstrates that a significant reduction in α1(I)collagen andα -SMA expression was detectable 5 to 7 days after the last dose of CCl₄ (n = 4; P < 0.05 versus peak). In contrast to these changes, neither TIMP-1 nor MMP-2 mRNA or protein changed in the first postadministration week (Fig. 7, f and g). Similarly, we found no increase in the number of nonparenchyma apoptotic cells during this time (Fig. 7i). Administration of 6-ECDCA caused a notable increase in the velocity of fibrosis resolution. At 3 to 5 days of administration of 3 mg/kg 6-ECDCA, there was a 50 to 60% reduction of liver hydroxyproline content, α1(I)collagen, α-SMA, TIMP-1, and TIMP-2 mRNA expression compared with fibrosis peak (Figure 7, a–e). Furthermore, 6-ECDCA decreased TIMP-1 protein content and although it had no effect on MMP-2 mRNA expression, it significantly increased the MMP-2 activity (Fig. 7h). Finally, administration of rats with 6-ECDCA led to an increase in the number of apoptotic figures in nonparenchyma cells (Fig. 7). TUNEL-positive cells had similar distribution to that of α-SMA-positive cells, although we failed to achieve good double immunostaining with α-SMA and TUNEL.

Discussion

The FXR is a ligand-regulated transcription factor that functions as an endogenous sensor for bile acids (Makishima et al., 1999; Parks et al., 1999; Wang et al., 1999; Goodwin et al., 2000; del Castillo-Olivares and Gil, 2001; Lu et al., 2003). We have reported previously that modification of CDCA, the most potent FXR ligand, by addition of an ethyl group in position 6 results in a semisynthetic bile acid, namely, 6-ECDCA, which has potent FXR agonist activity (Mi et al., 2003). The resolution of the crystal structure of the FXR ligand binding domain complexed with 6-ECDCA and the coactivator peptide Src-1 (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci et al., 2004) demonstrates that in the presence of 6-ECDCA, helix 12, the activation function 2 of the receptor, adopts the agonist conformation and stabilizes the binding of coactivator peptides. This would provide a mechanism for enhanced binding of coactivators through intermolecular contacts between their LXXLL sequences, explaining the enhanced potency of 6-ECDCA compared with the parent compound (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci et al., 2004). 6-ECDCA has an EC₅₀ value of 90 nM in the coactivator recruitment assay (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci al., 2004) and is ≈100-fold more potent than CDCA in activating FXR in transactivation assays (Pellicciari et al., 2002; Mi et al., 2003; Fiorucci et al., 2004). Similarly to GW4064 (Maloney et al., 2000), 6-ECDCA protects the liver in a rodent model of cholestasis (Lu et al., 2003).

In the present study, we have shown that FXR activation in HSCs leads to SHP-dependent inhibition of TIMP-1 expression/activity and increases the susceptibility of HSCs to apoptogenic stimuli. SHP is an atypical member of the nuclear receptor superfamily since it lacks a DNA-binding domain (Lee et al., 2000). It is mostly expressed in the liver where it binds to and inhibits the function of other nuclear receptors. SHP activation by FXR ligands in hepatocytes leads to the repression of cholesterol 7-α-hydroxylase expression, the rate-limiting enzyme in bile acid production from cholesterol (Goodwin et al., 2000). SHP is also known to repress the activity of many nuclear hormone receptors in vitro, including estrogen receptors, androgen receptor, hepatocyte nuclear factor 4, constitutive androstane receptor, RXR, liver X receptor α, and liver receptor homologous protein-1 (Lee et al., 2000; Lai et al., 2003). Here, we demonstrated that 6-ECDCA induces SHP in HSCs and that SHP induction is essential for FXR ligands to suppress TIMP-1 expression. Support for this concept comes from the observation that exposure of HSCs to FXR ligands increases SHP mRNA and that FXR ligands are unable to suppress TIMP-1 expression in SHP-deficient HSCs. Furthermore, we found that SHP overexpression is sufficient to cause a 70 to 80% reduction of TIMP-1 mRNA in resting HSCs and abrogates the up-regulation of TIMP-1 induced by thrombin. Together, these data indicate that SHP is an essential part of the inhibitory pathway activated by FXR ligands in HSCs.

In this study, we have also provided evidence that SHP interacts with AP-1 by preventing its binding to the TIMP-1 promoter (Trim et al., 2000). The mechanism through which SHP inhibits TIMP-1 expression involves its direct interaction with JunD, the main AP-1 constituent in HSCs, and the formation of a SHP-JunD protein complex. Support for this concept comes from results of coimmunoprecipitation experiments carried out in SHP-overexpressing HSCs. In these experiments, we demonstrated that JunD is induced by thrombin and coimmunoprecipitates with SHP. Sequestration of JunD in a protein complex with SHP is likely responsible for prevention of JunD binding to the TIMP-1 promoter as demonstrated by the results of our EMSA experiments. Thus, we found that exposure of wild-type HSC-T6 to 6-ECDCA or SHP overexpression abrogates AP-1 binding to the TIMP-1 promoter. The results of the transactivation assay were also consistent with this concept, since SHP overexpression in HA-SHP was sufficient to abrogate the transactivation of the TIMP-1 promoter. Together, these results indicate that an FXR-SHP regulatory cascade mediates TIMP-1 inhibition caused by FXR ligands. We have now provided evidence that exposure of HSCs to FXR ligands prevents the physical interaction between JunD with SHP. The possibility of direct interaction of a nuclear receptor with JunD in HSCs has already been shown (Hazra et al., 2004) since peroxisome proliferator-activated receptor-γ overexpression leads to formation of a JunD-peroxisome proliferator-activated receptor-γ complex, which prevents JunD binding to the AP-1 binding site (Hazra et al., 2004).

Exposure to FXR ligands increases HSCs susceptibility to apoptogenic stimuli through a mechanism that involves SHP-mediated inhibition of TIMP-1 expression (Canbay et al., 2004). Thus, not only do FXR ligands decrease the expression of TIMP-1 but also addition of rTIMP-1 to HSCs attenuates apoptosis induced by 6-ECDCA and cycloheximide. The role of SHP in modulating HSCs sensitivity to apoptogenic stimuli is further shown by the fact that exposure of SHP-deficient cells, which express increased levels of TIMP-1, to cycloheximide greatly attenuated the apoptotic potential of 6-ECDCA. In contrast, treating SHP-overexpressing cells, in which TIMP-1 expression is reduced by ≈80%, increased the apoptotic potential of cycloheximide and 6-ECDCA.

Our in vivo data demonstrate that treatment of rats with 6-ECDCA protects against liver fibrosis development and accelerates liver fibrosis resolution induced by CCl₄. The results presented in Fig. 7 show that administration of 6-ECDCA results in a rapid (3–5-day) induction of SHP expression that associates with a dramatic down-regulation of TIMP-1 induction caused by CCl₄. Thus, whereas a single dose of CCl₄ causes an≈ 30-fold induction of TIMP-1 mRNA, simultaneous administration of 3 mg/kg 6-ECDCA, a dose of FXR ligand that protects against fibrosis development in bile duct ligated, and porcine serum administered rat reduced TIMP-1 induction by ≈60%, whereas 15 mg/kg UDCA had no protective effects (Paumgartner and Beuers, 2002). Furthermore, 6-ECDCA administration reduced the early, 3- and 5-day, up-regulation of α-SMA and α1(I)collagen mRNA induced by CCl₄, suggesting that the FXR ligand prevents the phenotype switch that associated with activation of HSCs in this model.

Administration of CCl₄ for 4 weeks is an established model of liver fibrosis. Using this model, we have shown that 3 mg/kg 6-ECDCA caused a 70 to 90% reduction of collagen deposition as assessed by liver morphometry and measurement of hepatic hydroxyproline content and α-SMA andα 1(I)collagen mRNA expression. The reduction in collagen deposition caused by 6-ECDCA treatment was associated with the reduction of the parenchymal area occupied by α-SMA-positive cells, suggesting a causal relationship between the decreased number of activated HSCs and the reduced accumulation of ECM components in CCL₄-intoxicated rats.

Here, we demonstrated that FXR activation in rats with established cirrhosis leads to accelerated resolution of liver fibrosis. The results shown in Fig. 7 demonstrate that administration of 3 mg/kg 6-ECDCA to rats leads to a rapid decrease (within 3–7 days) of α-SMA, α1(I)collagen, and TGFβ₁ mRNA. This effect correlates with reduction of TIMP-1 mRNA and a 100% increase in the activity of MMP2. Previous studies have shown that spontaneous resolution of liver fibrosis in the CCl₄ model is accompanied by an increased rate of apoptosis in HSCs, leading to a reduction of the number of activated HSCs. In accordance with our in vitro data, we found that administration of 6-ECDCA to rats with established cirrhosis increases the number of apoptotic figures in nonparenchymal cells, suggesting that an increased rate of HSCs apoptosis mediates the proresolution effect of 6-ECDCA in this model.

In conclusion, we have shown that FXR regulates the balance between profibrotic and antifibrotic mediators in HSCs. By demonstrating that a regulatory cascade involving FXR-SHP promotes the development of a quiescent phenotype and increases apoptosis of HSCs, this study establishes that FXR ligands may be beneficial in treatment of liver fibrosis.