Potential conflict of interest: Nothing to report.
JunD is implicated in the regulation of hepatic stellate cell (HSC) activation and liver fibrosis via its transcriptional regulation of the tissue inhibitor of metalloproteinases-1 (TIMP-1) gene. In the present study we found in vivo evidence of a role for JunD in fibrogenesis. Expression of JunD was demonstrated in alpha-SMA-positive activated HSCs of fibrotic rodents and human livers. The junD−/− mice were protected from carbon tetrachloride–induced fibrosis. The livers of injured junD−/− mice displayed significantly reduced formation of fibrotic crosslinked collagen and a smaller number of alpha-SMA-positive HSCs compared with those of wild-type (wt) mice. Hepatic TIMP-1 mRNA expression in injured junD−/− mice was 78% lower and in culture activated junD−/− HSCs was 50%-80% lower than that in wt mice. In examining the signal transduction mechanisms that regulate JunD-dependent TIMP-1 expression, we found a role for phosphorylation of the Ser100 residue of JunD but ruled out JNK as a mediator of this event, suggesting ERK1/2 is utilized. In conclusion, a signaling pathway for the development of fibrosis involves the regulation of TIMP-1 expression by phosphorylated JunD. (HEPATOLOGY 2006;44:1432–1440.)
Activated (or myofibroblastic) hepatic stellate cells (HSCs) are cellular mediators of liver fibrosis through their secretion of interstitial collagens and tissue inhibitor of metalloproteinase-1 (TIMP-1),1, 2 which is a broad specific inhibitor of the matrix metalloproteinases (MMPs) and prevents MMP-catalyzed degradation of interstitial collagens in order to promote collagen deposition. A second fibrogenic role identified for TIMP-1 is as a suppressor of HSC apoptosis.3
TIMP-1 gene transcription is induced during culture activation of isolated HSCs and is under the control of regulatory elements in the TIMP-1 promoter including activator protein-1 (AP-1) and RUNX binding sites.4–7 The AP-1 site is perfectly conserved in human and rodent TIMP-1 genes8 and binds homo- or heterodimers of Jun proteins and their Fos/ATF partners.9 Of the three Jun proteins (JunB, c-Jun, and JunD), JunD is the protein predominantly expressed in activated HSCs.4, 6 Homodimeric JunD is the optimal AP-1 regulator of TIMP-1 gene transcription in HSCs.6 JunD also regulates IL-6 and MMP-9 transcription in HSCs, indicating it may be a key profibrogenic regulator.6, 10 However, evidence is currently lacking about the in vivo role of JunD in liver fibrosis.
Mice lacking junD are viable and, apart from defects in the male reproductive system, are apparently normal.11 However, junD−/− fibroblasts are susceptible to stress-induced apoptosis and junD−/− mice are sensitive to TNF-α-mediated hepatitis.12 In addition, JunD is essential for halting the development of severe glomerular sclerosis, tubular dilation, and interstitial kidney fibrosis.13
Using the carbon tetrachloride (CCl4) model of chronic liver injury, we showed that junD−/− mice are protected against the development of liver fibrosis. We also demonstrated that junD−/− HSCs had lower expression of TIMP-1 than wild-type (wt) HSCs. To investigate tje signal transduction events regulating TIMP-1 gene transcription by JunD, we identified a requirement for phosphorylation of JunD at Ser100 and examined the role of Jun N-terminal kinase (JNK) and extracellular-signal-related kinase (ERK)1/2 pathways.
Isolation and Culture of Quiescent and Activated HSCs.
Primary rodent HSCs were isolated and maintained as previously described.3–7 The LX-2 line was described elsewhere.7 (L)-JNKI1, a cell-permeable peptide inhibitor of JNKs, SP600125 SAPK inhibitor II, and MAPKK inhibitor PD98059 were purchased from Calbiochem.
JunD−/− Mice and Experimental Model of Fibrosis.
The junD−/− mice supplied had a mixed C57Bl6/129Sv background with exchange of the junD sequence for lacZ. The cohorts of wt and junD−/− mice used for the study were generated from litters produced at similar times from a large group of breeding pairs of heterozygote mice in order to minimize the risk of drift of their genetic background. Genotypes were determined by PCR using primers for junD and lacZ. Fibrosis was induced by twice-weekly I.P. injection of 1 μL/g of body weight of CCl4and olive oil (1:4). Animals were injured for 8 weeks prior to culling 1, 4, and 7 days after the final injection. Acute CCl4 injury was carried out by administration of a single dose of CCl4.
Histological and Morphometric Assessment of Fibrosis.
Sirius Red staining was performed on paraffin sections by treating with 0.2%-0.5% phosphomolybdic acid for 15 minutes, before immersing in 0.1% Picric Sirius Red for 3 hours. Scoring and ranking for fibrosis was performed blinded, and the Mann-Whitney rank-sum test was used for analysis. For morphometric analysis, Sirius Red sections were captured at 200× magnification using a Zeiss AxioCam HR CCD camera. Images were processed using an automation macro in order to assign a value for percentage collagen (Sirius Red staining) coverage per field.
Human livers were obtained with local ethical permission (LREC ref CA/5192). Immunohistochemical staining in formalin-fixed tissue was performed according to standard protocols. JunD staining was carried out by incubation overnight with polyclonal anti-JunD (sc-74, SantaCruz Biotechnology) at a 1:60 dilution at 4°C. JunD expression was visualized by 3,3′-diaminobenzidine tetrahydrochloride (DAB) staining in the absence (JunD single stain) or the presence (JunD/α-SMA dual stain) of nickel. Anti-rat smooth muscle alpha-actin (α-SMA) antibody (Serotec) was used at a 1:40 dilution. Immunostaining of macrophages and neutrophils was carried out using anti-CD68 (Serotec—rat antimouse, MCA 1957, 1:150) or antibody clone 7/4 (Serotech, 1:100), respectively.
Quantitative RT-PCR was performed in triplicate using an ABI Prism 7700 Thermal Cycler, and transcript expression was determined using the ΔΔCT calculation. The primers and probes used for TIMP-1 were: forward, 5′-GCA TGG ACA TTT ATT CTC CAC TGT; reverse, 5′-TCT CTA GGA GCC CCG ATC TG; and probe, 5′-CAG CCC CTG CCG CCA TCA.
Phosphorylated JunD was detected with anti-phosphoJun antibody (1:500) that recognizes a phosphorylated epitope common to the Ser73 of c-Jun and the Ser100 of JunD (Upstate). Monoclonal antibodies (Sigma, Poole, UK) were employed for detection of α-SMA (1:1000), desmin (1:1000), and β-actin (1:1000). The p75 was detected using a rabbit polyclonal antiserum (Promega, UK). Sheep antirabbit or goat antimouse horseradish peroxidase (HRP) antibody conjugates were used at 1:2000.
Plasmid DNAs and Reporter Assays.
TIMP-1 promoter activity was determined in transfected cells using human TIMP-1 promoter-CAT or -luciferase (TIMP-1-Luc) vectors.4–7 The pCMV2JunD was previously described.6 The sRSP-HA-SAPKβp54KK-RR contained a mutated JNK/SAPKβ (SAPKβ K55K56-RR) provided by Dr. I. Clark (UEA, Norwich, UK). The pSG5JunDSer100-Ala,14 was provided by Dr. L.F. Lau (Chicago, IL).The JIP1 Jun binding domain (JIP1 aa 127-282) construct was donated by Dr. A. Whitmarsh (Manchester, UK).
JunD Is Expressed in Myofibroblasts in Rodent Fibrosis and Human Liver Disease.
Staining of normal rat liver with anti-JunD revealed strong nuclear staining of the hepatocytes (Fig. 1), confirming JunD is expressed in rodent hepatocytes.15, 16 CCl4-injured rat livers displayed nuclear JunD staining in cells with a myofibroblast-like morphology in fibrotic tissue. Dual α-SMA (brown) and JunD (black) staining confirmed that approximately 60% of myofibroblastic HSCs expressed JunD (Fig. 1B). JunD was also present in myofibroblast-like cells associated with fibrotic tissue in HCV-induced cirrhosis and primary biliary cirrhosis (Fig. 1C).
JunD−/− Mice Are Protected Against Fibrosis Induced by Chronic Injury of the Liver From Carbon Tetrachloride.
PCR detection of lacZ and the absence of junD PCR product identified junD−/− mice, whereas the reverse result identified wt mice (Fig. 2A). Wt and junD−/− mice were injured for 8 weeks and their livers harvested 1, 4, and 7 days following the final administration of CCl4. Figure 2B shows representative Sirius Red–stained liver sections from peak injured (day 1) and recovering (days 4 and 7) mice. Wt mice responded as expected, with evidence of bridging fibrosis on both day 1 and day 4 postinjury. Prolonged recovery (day 7) was associated with partial resolution of fibrosis. The junD−/− livers displayed significant attenuation of the fibrotic response, with only thin incomplete tracts of crosslinking collagen observed at peak injury. On day 4 very slight evidence of fibrosis was apparent, and on day 7 the junD−/− livers appeared normal. Sections from all 43 mice in the study were randomized and graded according to severity of fibrosis (from 43, indicating the most fibrosis, to 1, indicating the least fibrosis) prior to calculation of mean rank at each time point. The junD−/− livers consistently scored a lower mean ranking than the wt livers (Fig. 3A). Fibrosis was also histopathologically graded (from 4 = cirrhosis to 0 = normal). Mean peak fibrosis scores (day 1) were 1.9 (± 0.20 SEr) for wt livers compared with 0.8 (± 0.15) for junD−/− livers. Morphometric quantification of Sirius Red staining also demonstrated reduced accumulation of crosslinked collagen in the junD−/− livers relative to the wt livers (Fig. 3B). Analysis of hepatic procollagen I mRNA expression revealed high levels at peak injury (day 1) that diminished with recovery; however, no differences in expression were observed between the wt and junD−/− livers (Fig. 3C).
The low level of fibrosis observed in junD−/− mice may have been a result of attenuation of the injury process. However, serum ALT activity was similar in chronically injured wt and junD−/− mice (Supplementary Fig. 1A: available at: http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). ALT and AST levels were also similarly raised in wt and junD−/− mice following acute CCl4 injury (Table 1). H&E staining revealed similar histological features of hepatocyte damage and inflammation in the 2 genotypes (Supplementary Fig. 1B), and no statistically significant differences in neutrophil or macrophage recruitment were observed (Supplementary Fig. 1C).
Table 1. Serum Aminotransferases for wt and JunD−/− Mice in Response to Acute CCl4 Injury
Reduced α-SMA-Positive Cells and Expression of TIMP-1 in Injured JunD−/− Livers.
The number of activated HSCs populating the injured liver was assessed by counting α-SMA-positive cells. α-SMA-positive staining was associated with bands of bridging fibrosis in the wt and junD−/− livers; however, a smaller number of α-SMA-positive cells was observed in the latter (Fig. 4A–B). This observation was supported biochemically by demonstration of lower levels of hepatic α-SMA protein (Fig. 4C). Expression of TIMP-1 was elevated in HSCs of fibrotic liver.2 At peak injury (day 1), hepatic TIMP-1 mRNA expression was 78% lower in junD−/− mice than in wt mice (Fig. 5A). As previously reported,17 expression of hepatic TIMP-1 transcripts diminished to close to background levels with recovery.
junD−/− HSCs Express Reduced Levels of TIMP-1 mRNA.
We observed no obvious attenuation of the morphological activation of cultured junD−/− HSCs or differences in their proliferation or apoptotic index. However, junD−/− HSCs expressed 50%-80% fewer TIMP-1 transcripts than wt HSCs (Fig. 5B and Fig. 8A). The absence of JunD did not have a general phenotypic influence on cultured HSCs because other phenotypic markers (p75 neurotrophic receptor, desmin, and α-SMA) were expressed at similar levels in the wt and junD−/− HSCs (Fig. 5C). Similar expression of α-SMA in the wt and junD−/− HSCs indicated low expression of α-SMA in injured junD−/− livers (Fig. 4C) is a consequence of a reduced number of α-SMA-positive myofibroblasts (Fig. 4B). Although we are unable to explain this phenotype, we cannot rule out the possibility of in vivo differences in HSC activation, proliferation, and apoptosis not reproduced in the culture model.
Serine Phosphorylation of JunD Is Required For Transactivation of TIMP-1 Promoter in HSCs.
Serine phosphorylation is a mechanism for regulating AP-1-dependent gene transcription.18, 19 The N-terminus of JunD contains 3 MAPK-regulated phospho-acceptor sites (Ser90, Ser100, and Thr117) that are phosphorylated in response to ERK1/2 and JNK activation.14, 20 To determine if HSCs support phosphorylation of JunD, human LX-2 HSCs were transfected with the JunD expression vector. JunD-transfected LX-2 HSCs expressed 2 major protein species (39 and 43 kDa) reactive with both anti-phosphoJun and anti-JunD (Fig. 6A). To determine if HSCs in the injured liver express JunD in a phosphorylated form, in vivo activated HSCs were isolated from mice following acute injury with CCl4 or sham injury with olive oil (2 animals per group). Immunoblot analysis with anti-phosphoJun detected 2 protein species similar in size to the isoforms of JunD detected in JunD-transfected LX-2 HSCs (Fig. 6B). We next investigated the functional relevance of Ser100 phosphorylation by determining if mutant JunDSer100-Ala is attenuated for transactivation of the TIMP-1 promoter. Overexpression of wt JunD resulted in a 2-fold increase in the induction of TIMP-1 promoter activity in rat HSCs; in contrast, JunDSer100-Ala did not stimulate the promoter (Fig. 6C).
JNKs Do Not Stimulate TIMP-1 Gene Transcription in Activated HSCs.
Ser100 of JunD is a target of JNK-dependent phosphorylation.19 Overexpression of dominant-negative JNK2 or the JIP1 JNK-binding domain (JIP1aa127-282 functions as a repressor of JNK1, -2, and -3)21 did not inhibit TIMP-1 promoter activity (Fig. 7A–B). Indeed, we reproducibly observed a stimulatory effect of JNK inhibition. These results contrast with the anticipated inhibitory effects of dominant-negative JNK2 and JIP1aa127-282 on serum-induced TIMP-1 promoter activity in 3T3 cells (Fig. 7A–B). The Ser100 of JunD is also an efficient target for phosphorylation by ERK1/2.14, 20 Treatment of rat HSCs with the ERK1/2 inhibitor PD98059 resulted in 40% repression of TIMP-1 promoter activity but had no effect on promoter activity in 3T3 cells (Fig. 7C). As further confirmation that ERK1/2 has a role as a regulator of JunD-dependent TIMP-1 expression, we determined the effects of the pharmacological JNK inhibitors SP600125 and PD98059 on expression of endogenous TIMP-1 mRNA in wt and junD−/− HSCs. SP600125 did not attenuate TIMP-1 expression in either cell type (Fig. 8A). PD98059 treatment of wt HSCs inhibited TIMP-1 expression by greater than 50% but had no effect on expression in junD−/− HSCs, confirming JunD to be the downstream target of ERK1/2. Finally, we confirmed these findings in human HSCs. TIMP-1 promoter activity in LX-2 was unaffected by a cell-permeable peptide inhibitor of JNK activity but was inhibited 60% by PD98059 (Fig. 8B).
We and others have previously provided circumstantial evidence that JunD operates as a profibrogenic transcription factor.4–6, 10, 23–25 JunD is also the major Jun family protein expressed in activated pancreatic stellate cells, indicating a generic role for JunD in fibrogenesis.26 Studies with junD−/− mice and cells have revealed new physiological functions of JunD, including regulation of spermatogenesis, control of cell senescence and apoptosis, prevention of renal cell proliferation and hyperplasia, regulation of lymphocyte proliferation and differentiation, inhibition of tumor angiogenesis, and cardiac hypertrophy.11–13, 27–29 Our studies of junD−/− mice have provided direct evidence that JunD functions as a profibrogenic factor in diseased liver. Interestingly, this is in contrast with the apparently antifibrogenic function that JunD has in the kidney following partial nephrectomy.13 These conflicting findings most likely reflect the mechanisms responsible for fibrosis in the 2 experimental models and the different cellular origins of fibrogenic myofibroblasts in the liver and kidney. Increased renal fibrosis in junD−/− mice correlates with a late second wave of tubular epithelial cell proliferation normally repressed by JunD.13 A major contributor to myofibroblasts in diseased kidneys is the epithelial-to-mesenchymal transition (EMT).30 Hence, the late second wave of tubular epithelial cell proliferation in partial nephrectomized junD−/− mice would increase the number of cells available for EMT, whereas the accompanying perturbation of normal kidney tubule development in these mice would act as a stimulus for EMT. To date it has not been reported that EMT is a major fibrogenic mechanism in the liver; instead, CCl4-induced liver fibrosis is mediated by transdifferentiation of quiescent HSCs and portal fibroblasts to proliferative myofibroblasts. It is therefore plausible that JunD may play either a pro- or antifibrogenic role depending on the tissue and nature of the injury.
Absence of JunD resulted in lowered expression of TIMP-1 mRNA in injured livers and in culture activated junD−/− HSCs relative to wt HSCs. The molecular ratio of TIMP-1 to MMPs is a determinant of the rate of progression and regression of liver fibrosis. Expression of interstitial collagenase is relatively constant during the development and resolution phases of fibrosis. By contrast, hepatic TIMP-1 expression increases during fibrogenesis and subsequently diminishes to basal level during resolution.17 Mice with liver-targeted transgenic overexpression of TIMP-1 displayed enhanced experimentally induced fibrosis and suppressed spontaneous resolution.31, 32 By contrast, hepatic overexpression of MMPs ameliorates experimental fibrosis.33 Augmentation of spontaneous resolution of liver fibrosis via stimulation of HSC apoptosis is associated with a 40% diminution of hepatic TIMP-1 and a 30% elevation of hepatic MMP activity.34 Hence, the 78% difference between junD−/− and wt mice in level of hepatic TIMP-1 would significantly attenuate collagenolytic activity and result in reduced matrix deposition.
The N-terminus of JunD contains 3 MAPK target residues (Ser90, Ser100, and Thr117) phosphorylated by both the ERK1/2 and the JNK pathways.14, 20 JunD is phosphorylated at Ser100 in activated HSCs and is necessary for stimulation of TIMP-1 gene transcription. We propose that Ser100 phosphorylated JunD homodimers operate in concert with either RUNX1A or RUNX2 to generate high-level TIMP-1 expression by activated HSCs.4–7 Although JunD is a less efficient substrate for JNK than c-Jun, it is a more efficient substrate than c-Jun for phosphorylation by ERK1/2.20 This latter property of JunD is a result of the presence of both N-terminal D and C-terminal DEF MAPK docking signals, which in combination increase affinity for ERK1/2; in contrast, c-Jun lacks the DEF motif. We have shown that suppression of JNK leads if anything to stimulation of TIMP-1 promoter activity, whereas inhibition of ERK1/2 represses promoter activity. Activated HSCs express weak JNK activity because of the repressive influence of NF-κB. This appears necessary for preventing JNK-mediated apoptosis.34 Because JunD has a relatively weak affinity for JNK, and as there are a wide number of JNK substrates,35 active JNK in HSCs may be insufficient to drive JunD phosphorylation. Inhibition of ERK1/2 exerted a significant inhibitory effect on TIMP-1 expression, although only when JunD is expressed. We therefore propose that maximal TIMP-1 gene transcription in HSCs requires ERK1/2 phosphorylation of JunD.
In summary, we have described a new profibrogenic pathway in which ERK1/2 activation stimulates JunD-directed elevation of TIMP-1 expression in activated HSCs. Strategies that target JunD or its phosphorylation at Ser100 by ERK1/2 may be of benefit in the treatment of hepatic fibrosis.