Leptin stimulates tissue inhibitor of metalloproteinase-1 in human hepatic stellate cells: respective roles of the JAK/STAT and JAK-mediated H2O2-dependant MAPK pathways.

Leptin is recognized as a profibrogenic hormone in the liver, but the mechanisms involved have not been clarified. The tissue inhibitor of metalloproteinase (TIMP)-1, which acts through inhibition of collagen degradation, is synthesized by activated hepatic stellate cells (HSC) in response to fibrogenic substances. The capacity of leptin to induce TIMP-1 and its signaling molecules were investigated in a human HSC cell line, LX-2. Leptin stimulated TIMP-1 protein, mRNA, and promoter activity. JAK1 and -2, as well as STAT3 and -5, were activated. After leptin, there was increased expression of tyrosine 1141-phosphorylated leptin receptor, which may contribute to STAT3 activation. AG 490, a JAK inhibitor, blocked JAK phosphorylation with concomitant inhibition of STAT activation, TIMP-1 mRNA expression, and promoter activity. Leptin also induced an oxidative stress, which was inhibited by AG 490, indicating a JAK mediation process. ERK1/2 MAPK and p38 were activated, which was prevented by catalase, indicating an H2O2-dependent mechanism. Catalase treatment resulted in total suppression of TIMP-1 mRNA expression and promoter activity. SB203580, a p38 inhibitor, prevented p38 activation and reduced TIMP-1 message half-life with down-regulation of TIMP-1 mRNA. These changes were reproduced by overexpression of the dominant negative p38alpha and p38beta mutants. PD098059, an ERK1/2 inhibitor, opposed ERK1/2 activation and TIMP-1 promoter activity, leading to TIMP-1 mRNA down-regulation. Thus, leptin has a direct action on liver fibrogenesis by stimulating TIMP-1 production in activated HSC. This process appears to be mediated by the JAK/STAT pathway via the leptin receptor long form and the H2O2-dependent p38 and ERK1/2 pathways via activated JAK.

various tissues, including the liver cells (5)(6)(7). Leptin's actions are mediated through the leptin receptor, which belongs to the class I cytokine receptor family, and shares common features with the interleukin-6 receptor (8). In humans and rodents, two major forms of the leptin receptor (OB-R) 1 are expressed. The long form OB-R (OB-R L ) is predominantly expressed in the hypothalamus and is present at low levels in peripheral tissues and specific cell types; it is the functional receptor isoform for leptin signaling (8 -10). In contrast, the short form (OB-R S ) is found in many organs and is considered to lack signaling capability.
In liver diseases, plasma leptin levels were reported to be increased in patients with alcoholic cirrhosis (11,12) regardless of body mass index (11), in nonalcoholic cirrhosis (13) and in nonalcoholic steatohepatitis (14), suggesting a possible involvement of leptin in the pathogenesis of liver fibrosis. Recent laboratory studies showed that hepatic fibrosis induced by chemical toxins and by Schistosoma mansoni was markedly decreased in leptin-deficient ob/ob mice (15)(16)(17), in ob/ob mice during the progression of experimental steatohepatitis (18) and in Zucker rats (fa/fa) (19), which lack a functional leptin receptor (compared with corresponding lean littermates). These findings implicated leptin as a mediator in the development of liver fibrosis but did not elucidate the mechanisms and cell types involved. One hypothesis suggests that leptin acts on the Kupffer cells and sinusoidal endothelium (in which OB-R L was detected) to release TGF-␤1 that, in turn, stimulates fibrogenesis in activated hepatic stellate cells (HSC) (15,19). Another view holds that leptin acts directly on HSC and triggers specific signal transduction systems, which alter collagen gene expression (16). In culture-activated rat HSC, rat HSC-T6, and human HSC-LX-1 cell lines, leptin was found to stimulate collagen I promoter activity (16), up-regulate mRNA (20,21), and increase protein production (21). Furthermore, Tang et al. (21) reported that leptin acts to enhance expression of the TGF-␤ type II receptor, which sensitizes HSC to the fibrogenic actions of TGF-␤1. Although leptin (15,22) and its receptors, either the long (16) or short forms (19,21), were detected in activated HSC and in immortalized HSC lines, the intracellular transduction molecules used by leptin signaling in HSC fibrogenesis have not been clarified.
Collagen I accumulation in liver fibrosis results, in part, from inhibition of its degradation by interstitial collagenase due to increased activity of the tissue inhibitor of metalloproteinase (TIMP)-1 (23). The latter is synthesized and secreted by activated HSC in response to fibrogenic cytokines in particular TGF-␤1 (24,25). Furthermore, serum TIMP-1 is increased in alcoholics with early fibrosis and can serve as a marker of precirrhotic and cirrhotic states (26). The regulation of TIMP-1 gene transcription has been extensively studied in HSC (27,28), but much less is known about ligand-induced intracellular signaling molecules that mediate its production. Leptin was found to increase TIMP-1 expression in human endothelial and vascular smooth muscle cells, although the pathways were not characterized (29). Thus far there is no report on the effects of leptin on TIMP-1 production and the associated signaling pathways in HSC.
Leptin signals from the OB-R L through the JAK/STAT pathway in the hypothalamic nuclei (30) and in various cell types as well. But leptin can also use other signaling cascades via JAK activation. Among these are p38 (31), extracellular signal-regulated kinase (ERK1/2) (2,(31)(32)(33)(34)(35)(36), and c-Jun terminal/stressactivated protein kinases (37) of the mitogen-activated protein kinase (MAPK) family members. Furthermore, leptin was found to generate increased amounts of H 2 O 2 in isolated vascular endothelial cells in association with atherogenic processes (37,38). H 2 O 2 was shown to stimulate p38 in cultureactivated rat HSC, resulting in ␣1(I) procollagen mRNA up-regulation (39), and to phosphorylate ERK1/2 in human fibrosarcoma cells, leading to increased matrix metalloproteinase-1 transcription (40). In addition, Sohara et al. (41) reported that ERK1/2 mediates TIMP-1 production in liver myofibroblasts in response to oncostatin M, which is, like leptin, a member of the interleukin-6 cytokine family that signals through class I cytokine receptors. These findings raise the possibility that leptin may stimulate TIMP-1 production, using the H 2 O 2 -dependent p38 and ERK1/2 signal transduction pathways via JAK activation.
Specifically, our aims were to assess whether the fibrogenic actions of leptin involve the induction of TIMP-1 in the LX-2 cell line-derived human HSC, the principal cells that mediate hepatic fibrogenesis (42,43). The respective roles of the JAK/ STAT and H 2 O 2 -dependent p38 and ERK1/2 MAPK signal transduction pathways in this process were evaluated. We also determined expression of the leptin receptor and its signaling capabilities in LX-2 cells in response to leptin.
Culture and Treatment of Hepatic Stellate Cells-LX-2, an immortalized human HSC line, was provided by Dr. S. L. Friedman, Mount Sinai School of Medicine, NY. Details for the generation of LX-2 have been described previously (44). Cells were seeded onto plastic tissue culture flasks at 0.9 ϫ 10 6 cells/ml in DMEM containing 5% fetal calf serum (FCS), 2 mM L-glutamine, 100 IU penicillin, and 100 mg/ml streptomycin and incubated at 37°C in a 5% CO 2 -air humidified atmosphere. The medium was changed 48 h after plating. At subconfluence, cells were washed in serum-free DMEM, and leptin with or without inhibitors was added. Leptin was used at concentrations of 25-100 ng/ml. The inhibitors and their concentrations were: 50 M JAK inhibitor AG 490 (45)(46)(47); 1000 units/ml catalase (49); 20 M p38 inhibitor SB203580 (39,48); 20 M SB202474 (an inactive analog of SB203580) (52); and 30 M ERK1/2 inhibitor PDO98059 (39,50). Me 2 SO was used as vehicle control for the inhibitors. In these experiments, LX-2 cells were used from passages 20 to 30.
Rat Hepatic Stellate Cells-HSCs were isolated from male Sprague-Dawley rats and cultured in DMEM supplemented with 10% FCS and antibiotics until subconfluence as previously described (39). Cells were trypsinized, subcultured, and used 3 days later (as passage 1). Animal experimental procedures followed the National Research Council's recommendations for animal care and were approved by the Institutional Animal Care and Use Committee.
TIMP-l mRNA Analysis by Northern Blot-Total cellular RNA was extracted from HSCs and used for Northern blot analysis as previously described (24). A cDNA probe for human TIMP-l or ␤-actin was labeled with [ 32 P]dCTP, using a random priming DNA labeling kit. Levels of mRNA were quantified by measuring the intensity of the bands on x-ray film, using the Evaluating Image Analysis Systems MCID TM (Imaging Research, Inc., St. Catherines, Ontario, Canada).
TIMP-l Protein Assay-TIMP-l in culture media of LX-2 cells and rat HSC was quantified by ELISA, using the Quantikine Human TIMP-1 ELISA kit according to the manufacturer's instruction. The absorbance was determined at 450 nm in an ELISA reader. Data were calculated against a standard curve and adjusted to nanograms/ml culture medium.
JAK and STAT Phosphorylation Assays-Whole cell lysates were extracted on ice using a lysis buffer containing 50 mM Tris-HCl, 1% Triton X-100, 10% glycerol, 150 mM NaC1 2 , 2 mM EDTA, 25 mM ␤-glycerophosphate, 1 mM sodium orthovanadate (Na 3 VO 4 ), and 1 mM phenylmethylsulfonyl fluoride. Proteinase inhibitor mixture was added to the buffer. The content was sonicated for 2 s to shear the DNA to reduce its viscosity, followed by centrifugation. The supernatants (200 l) were incubated with primary rabbit polyclonal antibodies against p-JAK1, p-JAK2, p-STAT1, p-STAT3, or pSTAT5 and then immobilized to agarose hydrazide beads by incubating overnight with gentle rocking at 4°C. The immune complexes were collected by centrifugation and washed in the lysis buffer. A 20-l sample was boiled for 5 min, cooled on ice, and subjected to 12% SDS-PAGE, followed by transblotting to a nitrocellulose membrane. After an overnight incubation with the respective primary antibodies (1:500 or 1:1000) at 4°C, the membrane was incubated with alkaline phosphatase-conjugated IgG (1:1000). Equal protein loading was controlled by immunoblotting of the corresponding nonphosphorylated JAK1, JAK2, STAT3, and STAT5, using rabbit polyclonal antibodies against the respective proteins. Immunoreactive proteins were revealed with Immun-Star Enhancer and Immun-Star Substrate (1:100) (Bio-Rad Laboratories, Hercules, CA) and then exposed to x-ray film. Signal intensities were analyzed using the Evaluating Image Analyzing System MCID.
Oxidative Stress Assessment-Oxidative stress was assessed as previously described using fluorescent probes (42). The amount of fluorescence was determined in a spectrofluorometer.

TIMP-1 Induction by Leptin
Hyrothidine is oxidized by O 2 . produced by the cells. Hence, the loss of fluorescence is proportional to the amount of superoxide generated. Hydroethidine fluorescence was measured at 352 nm for excitation and 434 nm for emission. Lipid Peroxidation-This was assessed by the addition of cis-parinaric acid to LX-2 culture at a final concentration of 5 M. Subsequent to peroxidative stress, cis-parinaric acid is degraded, resulting in decreased intensity. The loss of fluorescence is proportional to lipid peroxidation. The cis-parinaric acid fluorescence was measured at 325 nm for excitation and 413 nm for emission.
Reduced Glutathione Measurement-GSH levels in LX-2 cells were determined, using the Cayman's GSH assay kit according to the manufacturer's instruction.
p38 and ERKI/2 MAPK Phosphorylation Assays-These were performed by Western blots, using the components provided in the Phos-phoPlus p38 MAPK and ERKl/2 MAPK antibody kits as previously described (39,51). p38 activation was assayed using rabbit polyclonal phospho-p38 (Thr-180/Tyr-182) antibody. ERK1/2 phosphorylation was detected using the rabbit polyclonal phospho-ERK1/2 (Thr-202/Tyr-204) antibody. The antibodies were used at 1:1000 dilution. Horseradish peroxidase-conjugated anti-rabbit IgG (1:2000) was used as the secondary antibody. Equal protein loading was controlled by immunoblotting of corresponding nonphosphorylated p38 or ERK1/2 proteins. Immunoreactive proteins on the blots were visualized using the LumiGLO chemiluminescent reagents and then exposed to x-ray film. Signal intensities were quantified with the Evaluating Image Analysis System MCID.
p38 Kinase Activity Assay-This was performed by the detection of phosphorylation of activating transcription factor (ATF)-2, a substrate of p38, using the p38 MAP Kinase Assay kit as described previously (39). Whole cell lysates (200 l) were immunoprecipitated with a 20-l aliquot of monoclonal phospho-p38 (Thr-180/Tyr-182) antibody immobilized to agarose beads. The immunoprecipitated pellet was incubated with 2 g of ATF-2 fusion protein in the presence of 100 M ATP and 50 l of kinase buffer for 30 min at 30°C. A 20-l sample was resolved on a 12% SDS-PAGE gel, transblotted to nitrocellulose membrane, probed with phospho-ATF-2 (Thr-71) antibody (l:1000) as the primary antibody and horseradish peroxidase-conjugated anti-rabbit IgG (1:2000) as the secondary antibody. Immunoreactive proteins were visualized and quantified as described above. Equal protein loading was controlled by immunoblotting of the corresponding nonphosphorylated ATF-2.
Transfection of TIMP-l Promoter and Chloramphenicol Acetyltransferase Assay-The TIMP-1 promoter function was studied using the CAT reporter plasmid (pBLCA T3) containing nucleotides Ϫ102 to 96 (minimal promoter) of the human TIMP-1 gene (27). For transient transfection, LX-2 cells were grown on 24-well plates (3 ϫ l0 5 cells/well) to 90% confluence in DMEM supplemented with 5% FCS. The cell medium was changed, and transfection was performed using the Lipo-fectAMINE TM kit per the manufacturer's protocol. 1 l of LF2000 reagent was diluted to 50 l with DMEM and incubated for 5 min at room temperature. To this was added 320 ng of the reporter plasmid DNA or the promoterless pBLCAT3 suspended in 50 l of DMEM. The mixture was incubated at room temperature for 20 min to allow the formation of DNA⅐LF2000 complex. Finally, the mixture (100 l) was added to the culture, mixed gently, and incubated with the cells at 37°C for 24 h. Thereafter, the cells were treated with leptin in the absence or presence of AG 490, catalase, SB203580, or PD098059 for 24 h and then harvested for CAT activity assay, using the CAT ELISA kit and following the manufacturer's protocol. Cell lysates or CAT standard (200 l) were added to the wells of an ELISA plate precoated with a polyclonal antibody to CAT and incubated at 37°C for 1 h. Next, the bound CAT was sequentially reacted with an anti-CAT antibody labeled with digoxigenin and anti-digoxigenin antibody conjugated to peroxidase. Finally, the peroxidase substrate 2,2-azino-di-ethylbenzthiazoline sulfonate was added to yield a colored reaction product, and its absorbance was measured at 405 nm in an ELISA plate reader. The sensitivity of the assay was Ն50 pg/ml. Data were calculated against a standard curve and expressed as nanograms/mg of protein.
Transfection of Dominant Negative p38 Mutants (p38␣dn and p38␤dn)-To determine the specificity of p38 activation, and its role in TIMP-1 gene expression induced by leptin, transfections of p38␣dn and p38␤dn were performed. LX-2 cells were grown in 24-well plates as described above in the TIMP-1 promoter transfection protocol. 800 ng of pCDNA3 plasmids containing p38␣dn or p38␤dn (52) were transiently transfected into the cells, using the LipofectAMINE TM kit per the manufacturer's instruction. Additionally, cells were cotransfected with the p38-negative mutants and TIMP-l promoter. Transfected cells were treated with leptin for 24 h for analysis.
TIMP-l mRNA Stability Determination-To assess whether p38 and ERK1/2 regulate TIMP-l gene expression at the post-transcriptional level, LX-2 cells were treated with leptin (75 ng/ml) for 24 h to induce TIMP-l mRNA. This was followed by actinomycin D (10 g/ml) treatment for 20 min to block the transcription (53). The culture medium was changed, and fresh medium containing SB203580 (20 M) or PD098059 (30 M) was added. After 2, 8, and 12 h of incubation, total RNA from LX-2 cells was isolated for Northern blot analysis of TIMP-l mRNA levels, and the decay time course in the absence or presence of the inhibitors was analyzed. Similar experiments were performed with LX-2 cells transiently transfected with p38␣dn and p38␤dn.
Western Blot Analysis of Leptin Receptor (OB-R)-Protein lysates of LX-2 cells treated or not with leptin (75 ng/ml) for 24 h were analyzed for OB-R expression by Western blot using 12% SDS-PAGE. The primary antibody was rabbit anti-human OB-R (H300) antibody. For the detection of Tyr-1141-phosphorylated OB-R, a goat anti-human phospho-OB-R (Tyr-1141) antibody was used. Mouse monoclonal antibody against GADPH was used as the control for equal protein loading. Immunoreactive proteins were visualized with Immun-Star Enhancer and Immun-Star Substrate (1:100) and then exposed to x-ray film. Signal intensities were analyzed using the Evaluating Image Analyzing System MCID.
Immunocytochemical Staining-LX-2 cells grown on 4-chambered glass slides were incubated with monoclonal anti-vimentin, anti-␣-SMA, or anti-desmin antibodies, followed by detection with the Super Sensitive MultiLink kit. The immunoreaction was visualized by using the chromogen diaminobenzidine.
Protein Determination-Cell lysate protein content was determined using the BCA protein assay kit from Pierce, Rockford, IL. Statistics-Data are reported as means Ϯ S.E. Statistical analysis was performed by analysis of variance followed by Student-Newman Keuls tests for multiple comparisons between treatment groups using Instat (3.01) and Sigma Stat (2.0) software (Jandel Scientific Software, San Rafael, CA). p Ͻ 0.05 was considered to be significant.

RESULTS
Leptin Increases TIMP-1 mRNA and Protein: Dose-dependent Effect-Human LX-2 HSC expressed TIMP-1 mRNA of 0.9 kb, as analyzed by Northern blot (Fig. 1). Leptin treatment for 24 h elicited a dose-dependent increase in TIMP-1 mRNA, reaching a maximal level of 3.6-fold at 75 ng/ml leptin. This effect was accompanied by a 3.4-fold increase in TIMP-1 protein in the culture media. The induction of TIMP-1 by leptin in LX-2 cells was reproduced in culture-activated rat HSC (data not shown).
Phosphorylation of JAK and STAT after Leptin: Time-dependent Effect-A significant increased phosphorylation was first detected at 10 min for both JAK1 and JAK2 in LX-2 cells and reached a maximum (about 4.5-fold, respectively) between 30 and 60 min after 75 ng/ml leptin (Fig. 2A). The values returned to control levels at 240 min. There was an associated increase in phosphorylation of STAT3 (3.4-fold) and STAT5 (2.5-fold) at 30 min which remained maximal until 60 min (Fig.  2B). A significant increase in p-STAT3, but not p-STAT5, was still evident at 360 min. The total JAK and STAT protein contents did not change after leptin. No phosphorylation of STAT1 by leptin was detected (data not shown).
JAK Inhibitor AG-490 Inhibits JAK Phosphorylation and Its Effects on STAT Phosphorylation and TIMP-1 mRNA-Treatment of LX-2 cells with AG 490 for 30 min completely blocked the phosphorylation of JAK1 and JAK2 induced by leptin (75 ng/ml) (Fig. 3A). This treatment also resulted in total inhibition of STAT3 and STAT5 phosphorylation (Fig. 3B), as well as TIMP-1 mRNA expression (Fig. 3C), Me 2 SO, in a concentration (2.1 mM) equivalent to that present in AG 490, had no effect on TIMP-1 mRNA expression, whether induced by leptin or not. 1 h after leptin (6.5-fold, 3.3-fold, and 2.6-fold versus controls at 0 time, respectively). At 24 h, the H 2 O 2 level was still 2.9 times higher than in control, whereas superoxide and lipid peroxidation values returned to control levels. Cellular GSH fell in a time-dependent manner after leptin treatment, with a 39% decrease after 1 h and a further 59% decrease at 24 h. Fig. 4B shows that AG 490 treatment of LX-2 cells for 1 h prevented H 2 O 2 and superoxide formation, lipid peroxidation, and restored the decreased cellular GSH level seen after leptin, suggesting a mediation by activated JAK1 and JAK2 of the leptininduced oxidative stress.
p38 MAPK Activation by Leptin and Its Inhibition by JAK Inhibitor AG 490, Catalase, p38 Inhibitor SB203580, and p38 Dominant Negative Mutants-Because leptin induced oxidative stress and because the latter stimulated p38 MAPK in activated rat HSC (39), we evaluated whether leptin induces the activation of p38 through H 2 O 2 in LX-2 cells. Fig. 5A shows phosphorylation of p38 by leptin in a dose-and time-dependent manner. A maximal stimulation was elicited by leptin at 75 ng/ml, and, at this concentration, a significant increase was first observed at 1 h (2.5-fold), and it peaked (about 3.6-fold) from 2 to 24 h. Treatment of LX-2 cells with AG 490 or catalase for 2 h prevented leptin-induced p38 phosphorylation (Fig. 5B), demonstrating involvement of p-JAK1 and -2 and H 2 O 2 in the process; p38 phosphorylation was also inhibited by the p38 inhibitor SB203580 but not by its inactive analog SB202474 or

TIMP-1 Induction by Leptin
the ERK1/2 inhibitor PD098559. Transfection of p38␣dn and p38␤dn into LX-2 cells resulted in a total suppression of leptininduced p38 phosphorylation. Leptin also increased the kinase activity of p38 (Fig. 5C), and the rise was abolished by AG-490, catalase, and SB203580 as well as by overexpressed p38␣dn and p38␤dn in LX-2 cells, but not by SB202474 or PD098059, further substantiating the participation of JAK and H 2 O 2 in the activation of p38 induced by leptin. ERK1/2 MAPK Phosphorylation by Leptin and Its Inhibition by JAK Inhibitor AG 490, Catalase, and ERK1/2 Inhibitor PD098059 -Because leptin has been shown to activate ERK1/2 in a variety of cell types (32-36), we assessed phosphorylation of ERK1/2 by leptin in LX-2 cells and the involvement of JAKs and H 2 O 2 in this process. Fig. 6A shows a dose-dependent stimulation of ERK1/2 phosphorylation by leptin, with a maximal effect (5-fold) at 75 ng/ml. At this leptin concentration, increased phosphorylation of ERK1/2 was observed at 1 h (3fold) and was maximal (4.3-fold) at 2 h (Fig. 6A). A significant effect was still evident at 24 h. Treatment with AG 490 or catalase for 2 h abolished the leptin stimulation of ERK1/2 phosphorylation (Fig. 6B), demonstrating an involvement of p-JAK1 and -2 and H 2 O 2 in the process. PD098059 prevented the rise in ERK1/2 activation induced by leptin. SB203580, SB202474, or the p38 negative mutants had no effect on ERK1/2 phosphorylation.
Inhibition of Leptin-induced TIMP-1 mRNA by Catalase, p38 Inhibitor SB203580, Dominant Negative p38 Mutants, and ERK1/2 Inhibitor PD09859 -To determine whether H 2 O 2 -dependent MAPK signaling participates in the up-regulation of TIMP-1 mRNA by leptin, LX-2 cells were treated with leptin (75 ng/ml) in the presence or absence of the respective inhibitors, and TIMP mRNA was analyzed after a 24-h incubation. As shown in Fig. 7A, catalase prevented the 4-fold rise in TIMP-1 mRNA induced by leptin. SB203580, but not its inactive analog SB202474, and overexpressed p38␣dn and p38␤dn halved the level of TIMP-1 mRNA after leptin (Fig. 7B), implicating a role for p38 in the induction of TIMP-1 mRNA by leptin. PD098059, like the p38 inhibitor, reduced the rise in TIMP-1 mRNA level to one-half (Fig. 7C). These results suggest that the up-regulation of TIMP-1 mRNA by leptin is mediated, at least in part, by H 2 O 2 through the p38 and ERK1/2 MAPK signaling pathways.
Leptin Stimulates TIMP-1 Promoter Activity and Its Inhibition by JAK Inhibitor AG-490, Catalase, and ERK1/2 Inhibitor PD098059 -To determine whether leptin-induced TIMP-1 analyzed by Western blot, using a phospho-p38 (Thr-180/Tyr-182) antibody as the primary antibody. The intensity of bands on the blots was normalized to that of total p38, detected by a nonphosphorylated p38 antibody, and values are presented as -fold change relative to controls (without leptin or at 0 h), assigned a value of 1. The time-dependent effect of leptin on p38 phosphorylation was studied using 75 ng/ml leptin, because this concentration elicited a maximal level of p38 phosphorylation. Upper panels are typical Western blots, and lower panels are histograms of data of three separate analyses. *, p Ͻ 0.05 and **,

TIMP-1 Induction by Leptin
gene expression is regulated at the transcriptional level in LX-2 cells, the activity of TIMP-1 promoter was assayed in cells with transfections of the human TIMP-1 minimal promoter (nucleotides Ϫ102 to 96). Leptin (75 ng/ml) treatment raised the promoter activity 3-fold (Fig. 8). The rise was prevented equally by AG 490, catalase, and PD098059, but not SB203580 or overexpressed p38␣dn and p38␤dn. Transfection with the minimal promoter in the absence of leptin treatment caused a slight increase in the promoter activity compared with transfection with the promoterless empty pBLCAT vector, but the change was not affected by any of the inhibitors tested. These results implicate that leptin-induced TIMP-1 gene expression in LX-2 cells is mediated by activated JAK1 and JAK2, which, in turn, activate the ERK1/2 signaling pathway via H 2 O 2 formation and suggest that p38 is not involved in the transcriptional regulation of the TIMP-1 gene.
p38 Inhibitor SB203580, but Not ERK1/2 Inhibitor PD098059, Inhibits Leptin-induced TIMP-1 Message Stability-Because p38 activation does not appear to regulate TIMP-1 mRNA expression at the transcriptional level in LX-2 cells, we tested whether p38 acts to stabilize TIMP-1 message induced by leptin. Treatment of LX-2 cells with SB203580 resulted in an increased decay of TIMP-1 mRNA with an approximately half-life (t1 ⁄2 ) of 4 h compared with a t1 ⁄2 of 10 h without SB203580 treatment, suggesting a stabilization of TIMP-1 message by p38 (data not shown). This effect of p38 was reproduced by overexpression of p38␣dn and p38␤dn in LX-2 cells. By contrast, in LX-2 cells treated with PD098059, no change in TIMP-1 mRNA decay was observed. These results demonstrate that the leptin-induced TIMP-1 mRNA expression via the p38 pathway involves a stabilization of the TIMP-1 message.
Expression of Leptin Receptors (OB-R) in LX-2 Cells-With the OB-R H300 antibody, which reacts with an amino acid sequence 541-840 of the extracellular and transmembrane domains that are common to both the long and short forms of OB-R (3,8), multiple immunoreactive bands were revealed by Western blot near 102 kDa, representing the short isoforms of OB-R, and a single band between 122 and 204 kDa, representing the long form of OB-R (7, 16), but their intensities were not increased after leptin treatment (Fig. 9A). In contrast, when probed with the anti-human p-OB-R (Tyr-1141) antibody directed at the Tyr-1141-phosphorylated cytoplasmic domain at the C terminus of the long form OB-R, an immunoreactive band between 122 and 204 kDa was detected, and there was a striking FIG. 6. Increased ERK1/2 MAPK phosphorylation by leptin and its inhibition by AG 490, catalase, and ERK1/2 inhibitor PD098059. A, dosedependent and time course of ERK1/2 activation by leptin. LX-2 cells were incubated with leptin at 25-100 ng/ml for 2 h. Phosphorylated ERK1/2 protein content was analyzed by Western blot, using rabbit polyclonal anti-phospho-ERK1/2 (Thr-202/Tyr-204) antibody as the primary antibody. The intensity of the bands on the blots was normalized to that of total ERK1/2, detected by anti-nonphosphorylated ERK1/2 antibody, and values are presented as -fold change relative to controls (without leptin or at 0 h), assigned a value of 1. Because 75 ng/ml leptin elicited a maximal level ERK1/2 phosphorylation, this concentration was used to assess the time-dependent effect of leptin on ERK1/2 phosphorylation. Upper panels are typical Western blots, and lower panels are histograms of data of three separate analyses. *, p Ͻ 0.05 and **, p Ͻ 0.01 versus control. B, inhibition of leptin-induced ERK1/2 phosphorylation. LX-2 cells were treated for 2 h with leptin (75 ng/ml) alone or with AG 490 (50 M), catalase (1000 units/ml), PD098059 (30 M), or SB203580 (20 M). LX-2 cells transfected with p38␣dn and p38␤dn were likewise incubated with 75 ng/ml leptin. Phosphorylated and total ERK1/2 were analyzed as in A. The numbers above the immunoblots refer to the mean values of three separate analyses.
increase in its immunoreactivity after leptin treatment (Fig. 9B), suggesting up-regulation of Tyr-1141 phosphorylation of OB-R L .
Cytochemical Staining and Ultrastructure of LX-2 Cells-LX-2 cells stained positively for vimentin and ␣-SMA, but negatively for desmin. Staining for glial fibrillary acidic protein, synaptophysin, calponin, and myosin was also observed. Viewed under the phase contrast microscope and by comparison to activated rat or human HSC in culture, the majority of LX-2 cells displayed a lesser stellate shape with only a modest amount of visible stress fibrils. By electron microscopy, LX-2 cells contained modest amounts of rough and smooth endoplasmic reticulum, free ribosomes, Golgi apparatus, mitochondria, and a fair number of lipid droplets. Filamentous structures were not prominent, in agreement with the images by phase contrast microscopy. DISCUSSION This study demonstrates that leptin has the capacity to stimulate TIMP-1 gene expression and to increase TIMP-1

TIMP-1 Induction by Leptin
protein production in activated human LX-2 HSC. This effect is associated with increased phosphorylation of JAK1 and JAK2 as well as of STAT3 and STAT5, accompanied by increased expression Tyr-1141-phosphorylated OB-R L . Leptin also activates the H 2 O 2 -dependent ERK1/2 and p38 MAPK pathways via activated JAK. ERK1/2 appears to act at the transcriptional level through stimulation of the TIMP-1 promoter activity, whereas p38 acts at the post-transcriptional level through stabilization of the TIMP-1 mRNA. These signaling pathways are schematically illustrated in Fig. 10.
LX-2 is an immortalized cell line derived from human HSC and has the features of activated HSC (44). Our cytochemical staining showed that LX-2 cells express ␣-SMA, a marker for activated HSC. Vimentin staining of LX-2 cells reveals their mesenchymal origin and the lack of desmin reflects their origin from human HSC (54). Glial fibrillary acidic protein and synaptophysin were observed in human HSC (55). The presence of calponin and myosin, which are smooth muscle-associated contractile proteins (56), suggests the acquisition by LX-2 of a contractile property, a characteristic of HSC.
All isoforms of the leptin receptor contain an identical extracellular domain of about 816 amino acids and a transmembrane domain of 23 amino acids but differ in their intracellular domains (3,8). OB-R S , the major short form of the receptor, has FIG. 8. Leptin stimulates TIMP-1 promoter activity and its inhibition by AG 490, catalase, and PD098059 but not by SB203580 or overexpression of dominant negative p38 mutants. LX-2 cells were transiently transfected with the minimal TIMP-1 promoter (nucleotides Ϫ109 to 96) contained in the pBLCAT reporter plasmid or with the empty vector plasmid. Transfected cells were treated with leptin (75 ng/ml) alone or with the inhibitors as indicated. LX-2 cells cotransfected with the TIMP-1 promoter and p38␣dn and p38␤dn were likewise incubated with leptin. After 24 h, LX-2 cell extracts were prepared for quantification of TIMP-1 promoter activity, using a CAT ELISA kit as described under "Materials and Methods." CAT activity is expressed as mean nanograms/mg of protein Ϯ S.E. of three separate experiments. ***, p Ͻ 0.001 versus minimal promoter without leptin; ### , p Ͻ 0.001 versus minimal promoter plus leptin.

FIG. 9. Leptin receptor (OB-R) expression in LX-2 cells and effect of leptin.
Detection of OB-R was performed by Western blot, using rabbit polyclonal anti-human OB-R (H300) antibody (A) and anti-human pOB-R (Tyr-1141) antibody (B). Multiple immunoreactive bands near 102 kDa and a band between 122 and 204 kDa were detected with by the OB-R(H300) antibody. In contrast, a band between 122 and 204 kDa was detected with the p-OB-R (Tyr-1141) antibody, and its intensity was increased after leptin.
a cytoplasmic region of about 34 amino acids, whereas OB-R L has a long 303-amino acid cytoplasmic domain, which contains box 1 and box 2 motifs for interaction with the receptor-associated JAK and a box 3 motif for STAT3 activation. Leptin binding to OB-R L triggers activation of JAK2 (and possibly JAK1), which then phosphorylates Tyr-1141 (homologous to murine Tyr-1138) at the extreme C terminus of the cytoplasmic domain of the receptor (see Fig. 10). The phosphorylated Tyr-1141 binds STAT3 from the cytosol, which becomes activated by JAK2 (32,33,57,58). The activated STAT protein translocates to the nucleus to stimulate transcription of early response genes and late target genes, including, among others, TIMP-1 gene (67, 68). A previous study by Saxena et al. (16) reported expression of OB-R L protein and a corresponding 361-bp cDNA product in primary culture-activated rat HSC and in the immortalized rat HSC-T6. Others found only the short isoforms of the receptor in activated rat HSC (19) and in the human LX-1 HSC (21). Our immunoblot data showed that unstimulated LX-2 cells express both the long and short OB-R forms. More importantly, we found an enhanced expression of Tyr-1141phosphorylated OB-R L after leptin, which could explain the increased STAT3 activation observed here, and this mechanism is consistent with the mediation of STAT3 activation by Tyr-1141 phosphorylation of the OB-R L that has been described in other cell systems (9,32,33,57,58). The cytoplasmic domain of OB-R S does not contain tyrosine residues, and its signaling capability remains controversial (5-7, 9, 10, 57, 61). Because its expression in LX-2 cells was not increased by  (67, 70). Activated JAK1 and -2 also induce H 2 O 2 formation that, in turn, activates ERK1/2 and p38. ERK1/2 stimulates TIMP-1 promoter activity with enhanced TIMP-1 mRNA expression. p38 acts through a mechanism involving a stabilization of the TIMP-1 message with up-regulation of TIMP-1 expression. leptin, as shown by our immunoblot analysis (Fig. 9), OB-R S activity is most likely not involved in the events elicited by leptin observed in this study. Nevertheless, its coexpression with OB-R L in the unstimulated LX-2 cells raises the possibility that OB-R S may have a role in modulating the activities of OB-R L , such as that which occurs in some organ systems (30,58).
Our finding of increased STAT3 phosphorylation after leptin in LX-2 cells is consistent with the observation by Saxena et al. (16) in primary cultured rat HSC. We found, in addition, that the leptin-induced STAT3 phosphorylation was preceded by JAK1 and JAK2 activation and was inhibited by AG 490, indicating mediation of STAT3 activation by JAK. AG 490 treatment also resulted in a total suppression of TIMP-1 promoter activity (Fig. 8) as well as TIMP-1 mRNA expression (Fig. 3). It is known that the TIMP-1 promoter contains a binding site for STAT3 (59,62). It is expected that STAT-DNA binding activity at this site could contribute to the enhanced transcription of TIMP-1 gene observed after leptin. Such a mechanism has been described in human renal mesangial cells in response to thrombin (60) and in hepatoma cells after oncostatin M stimulation (59). Because this study did not examine STAT3 nuclear translocation and its binding to the TIMP-1 promoter in leptintreated LX-2 cells, a direct role for the STAT pathway in TIMP-1 gene activation has yet to be established. STAT5 was also activated by leptin, although at a lower level than STAT3. The role of STAT5 in leptin signaling is not known, but it was found to be activated together with STAT3 in COS and hepatoma cell lines cotransfected with OB-R L and STAT cDNAs (6,9,10), and in porcine medullary cells after leptin treatment (35). Functionally, STAT5 regulates expression of milk proteins in the response of mammary tissue to prolactin (63).
Leptin could act through signaling cascades other than the JAK-STAT pathway (8,63). Reactive oxygen species, in particular H 2 O 2 , act as second messengers that mediate diverse signal transduction pathways and are potent mediators of fibrogenesis in HSC (39). The present data revealed that leptin induces oxidative stress in LX-2 cells. This effect was mediated by activated JAK1 and -2, because it was prevented by AG 490. It was reported that leptin induces H 2 O 2 formation through increased fatty oxidation in mitochondria of endothelial cells (38). It is not presently known whether this occurs in the LX-2 cell line; this could be the subject of a future study. Our data also revealed that leptin-induced H 2 O 2 contributes to TIMP-1 gene expression in LX-2 cells, because treatment with catalase, an antioxidative enzyme that degrades H 2 O 2 , resulted in total suppression of TIMP-1 promoter activity (Fig. 8) and of TIMP-1 mRNA expression (Fig. 7). This action of H 2 O 2 on TIMP-1 gene expression is mediated through activation of p38 and ERK1/2 MAPK signaling pathways, as discussed below.
The time course data in Figs. 2A, 4A, and 5A revealed that activation of p38 after leptin is preceded by JAK activation and H 2 O 2 formation. Furthermore, p38 activation was dependent on H 2 O 2 , because it was prevented by catalase. The stimulatory effect of p38 on TIMP-1 mRNA was specific, because it was abolished by the p38 inhibitor SB203580, but not by its inactive analog SB202474. p38 MAPK appears to act by stabilizing leptin-induced TIMP-1 mRNA, because SB203580 treatment destabilized the message with a reduction in the message halflife. In addition, this effect of SB203580 was reproduced by overexpression of the dominant negative p38 mutants in LX-2 cells. In contrast to ERK1/2, p38 signaling had no effect on the activity of TIMP-1 promoter (see below). ERK1/2 MAPK was activated by leptin in LX-2 cells with a similar kinetics as p38. Like p38, the leptin-induced ERK1/2 activation was mediated by JAK1 and -2 through H 2 O 2 . ERK1/2 acts to regulate TIMP-1 gene expression at the transcription level through stimulation of TIMP-1 promoter activity. The effect of ERK1/2 signaling on TIMP-1 promoter activity was specific, because it was blocked by the ERK1/2 inhibitor PD098059, but not by SB203580 or overexpressing dominant negative p38 mutants. ERK1/2 activation by leptin or its receptors has been reported in a number of cell systems (32)(33)(34)(35)64). In COS and Chinese hamster ovary cells transfected with the leptin receptors, ERK1/2 can be activated via two pathways. One pathway requires phosphorylation of Tyr-985 in the cytoplasmic domain of OB-R L , whereas the second one occurs independently of tyrosine phosphorylation of OB-R L (32,33,64). In addition, although overexpressed OB-R S in COS cells is unable to phosphorylate STAT3, it can, nevertheless, activate ERK1/2, although much less than that initiated by OB-R L . However, the significance and relevance of this effect for in vivo leptin signaling is unclear (64). Which of these receptor signaling mechanisms that mediate the activation of ERK1/2 in response to leptin in LX-2 cells is not known; additional studies are needed to answer this question.
In conclusion, the capacity of leptin to stimulate TIMP-1 gene activation and to increase its protein production in human HSC implicates that the hormone has a direct action on the stellate cells and could explain its fibrogenic role in the liver. Key future studies are needed to determine the binding of STAT3 to the TIMP-1 promoter and to evaluate possible crosstalk between the STAT pathway and the MAPK pathways (45,64). It is important to note that the signaling molecules and pathways observed in cell lines could differ considerably from those actually activated in vivo (8,30). Nevertheless, the present study illustrates that the human LX-2 HSC line is a useful tool for the elucidation of molecular mechanisms of hepatic fibrogenesis that could be potential targets for therapeutic intervention with antifibrotic agents.