The Activator Protein 1 (AP-1) transcription factor subunit Fos-related antigen 1 (Fra-1) has been implicated in liver fibrosis. Here we used loss-of-function as well as switchable, cell type-specific, gain-of-function alleles for Fra-1 to investigate the relevance of Fra-1 expression in cholestatic liver injury and fibrosis. Our results indicate that Fra-1 is dispensable in three well-established, complementary models of liver fibrosis. However, broad Fra-1 expression in adult mice results in liver fibrosis, which is reversible, when ectopic Fra-1 is switched off. Interestingly, hepatocyte-specific Fra-1 expression is not sufficient to trigger the disease, although Fra-1 expression leads to dysregulation of fibrosis-associated genes. Both opn and cxcl9 are controlled by Fra-1 in gain-of-function and loss-of-function experiments. Importantly, Fra-1 attenuates liver damage in the 3,5-diethoxycarbonyl-1,4-dihydrocollidine-feeding cholestatic liver injury model. Strikingly, manipulating Fra-1 expression affects genes involved in hepatic transport and detoxification, in particular glutathione S-transferases. Molecular analyses indicate that Fra-1 binds to the promoters of cxcl9 and gstp1 in vivo. Furthermore, loss of Fra-1 sensitizes, while hepatic Fra-1 expression protects from acetaminophen-induced liver damage, a paradigm for glutathione-mediated acute liver failure. Conclusion: These data define a novel function of Fra-1/AP-1 in modulating the expression of detoxification genes and the adaptive response of the liver to bile acids/xenobiotic overload. (Hepatology 2014;58:261–273)
The liver performs a wide range of functions including nutrient synthesis, transformation and storage, as well as endogenous and exogenous substance detoxification. Studies using genetically modified mice revealed essential functions of the dimeric transcription factor Activator Protein 1 (AP-1) in controlling liver development, homeostasis, and disease. For example, c-Jun is critical for hepatocyte proliferation and survival during liver development, regeneration, inflammation, and cancer.[1-4] Although the close homologs junb and jund are dispensable for liver homeostasis,[4, 5] JunD-deficient mice are sensitive to tumor necrosis factor alpha (TNF-α)-mediated hepatitis and protected from carbon tetrachloride (CCl4)-induced liver fibrosis. Furthermore, junb and jund can substitute for c-jun during fetal liver development.[8, 9]
In contrast, the functions of Fos proteins in liver physiology are less well defined. c-Fos, FosB, Fra-1, and Fra-2 form AP-1 complexes by association with Jun proteins. Genetic inactivation of single fos genes has no obvious effect on liver homeostasis (reviewed) and the relevance of Fos proteins to liver disease in loss-of-function mouse models has not been reported. Interestingly, broad ectopic expression of Fra-1 or Fra-2 in transgenic mice resulted in increased bone mass, but also generalized fibrosis with predominant manifestation in the liver and lung.[11-14] In addition, Fra-1 transgenic mice (Fra-1Tg) developed an inflammation-associated ductular reaction preceding liver fibrosis, suggesting an involvement of Fra-1 in cholestatic liver disease.
Hepatic fibrosis is the final common endpoint of most chronic liver diseases. We set out to define the contribution of Fra-1/AP-1 to the pathogenesis of cholestasis and hepatic fibrosis using Fra-1 loss-of-function mice and novel mouse models with switchable ectopic Fra-1 alleles. Our results indicate that Fra-1 is dispensable for liver fibrosis in three independent experimental models. However, broad Fra-1 expression results in reversible periportal liver fibrosis affecting opn and cxcl9 expression in hepatocytes. Strikingly, these experiments revealed that Fra-1 modulates the expression of genes involved in xenobiotic detoxification and that Fra-1/AP-1 is relevant for acetaminophen (APAP)-induced liver failure. Thus, we identify Fra-1/AP-1 as a novel regulator of the detoxification function of the liver.
Mouse models provide alternative means to study human disease and evaluate therapeutic approaches. Fra-1Tg mice have been proposed as a model for chronic biliary disease, although this transgenic mouse presents a complex phenotype with osteosclerosis, generalized fibrosis, lipodystrophy, and lung alterations.[13, 31, 32] These facts limit its usefulness to dissect organ-specific functions of Fra-1. Here we generated novel mouse models harboring switchable, broad, or cell-specific Fra-1 alleles and investigated for the first time the physiological relevance of Fra-1 in liver disease using loss-of-function animals.
Broad Fra-1 expression in adult Fra-1tetON mice largely recapitulated the phenotypes observed in Fra-1Tg mice with a randomly integrated H2Kb-fosl1-LTR transgene.[13, 14] Specifically, mutant mice developed periportal liver fibrosis, ductular reaction, cytokine dysregulation, and immune infiltrates. This finding has two implications. First, we unambiguously rule out a contribution from the transgene insertion site, fosl1 intronic sequences or the FBJ-sarcoma virus LTR sequence included in the H2Kb-fosl1-LTR transgene to the phenotype. Second, we establish that broad Fra-1 expression in adult mice is sufficient to trigger the disease.
An advantage of this new model is the possibility to switch off transgene expression. Dox withdrawal in Fra-1tetON mutant mice led to a striking improvement of the liver phenotype. Such “transgene addiction” demonstrates the requirement for Fra-1 for the maintenance of the disease and provides a rationale for experimentally addressing the functional relevance of Fra-1 in experimental liver fibrosis and dissecting its transcriptional targets.
Depending on the site of injury, fibrosis may develop in the hepatic parenchyma, as seen in chronic hepatitis and modeled in mice by CCl4 administration, or can be restricted to the portal areas in biliary diseases and in mice upon BDL or DDC feeding. Surprisingly, Fra-1 is largely dispensable in all three models, as indicated by the extent of liver injury, ductular reaction, and liver fibrosis in Fra-1Δembryo mice. We propose that Fra-1 can trigger fibrosis when ectopically expressed, but is dispensable in cholestasis and fibrosis models, because of compensation by other AP-1 members. Indeed, all Fos proteins are induced upon injury and can compensate for the absence of Fra-1. Such functional overlap has already been documented, e.g., Fra-1 can substitute for most of c-Fos functions.[33, 34] The best candidate for compensation is the closest homolog Fra-2, encoded by fosl2 as Fra-2 transgenic mice develop systemic fibrosis highly reminiscent of Fra-1 transgenics. In addition, Fra-2 is higher in untreated and in DDC- and CCL4-treated Fra-1Δembryo mutants. Analyzing mice with broad simultaneous fos11 and fosl2 inactivation would certainly be informative, but difficult to achieve, as fosl2 knockouts die within the first week after birth. A tissue-specific gene inactivation strategy is required and determining the cell type(s) where ectopic Fra-1 expression triggers the liver phenotype is essential to define the adequate gene inactivation strategy.
We investigated the contribution of hepatocytes using a hepatocyte-specific mouse line. Surprisingly, Fra-1hep-tetOFF mutants appeared healthy and displayed unaltered fibrosis upon DDC. This suggests that Fra-1 expression in other liver cell types independently or in addition to hepatocytes is required to trigger the fibrotic phenotype. Crossing the Fra-1Tg to a Rag2-deficient background attenuated the disease and the contribution of cholangiocytes was also suggested, on the basis of increased tgfb1 and pdgf expression. As Fra-1Tg fibroblasts express high amounts of tgfb1 and pdgfc, the contribution of hepatic stellate cells but also macrophages and endothelial cells deserves future investigation, using cell-specific tet-transactivator alleles. Nevertheless, some fibrosis-associated markers were found changed in Fra-1hep-tetOFF mutants and two genes: opn and cxcl9 were also deregulated in Fra-1 loss-of-function mutants, pointing to a specific and physiologically relevant regulation by Fra-1. Osteopontin, encoded by opn, is an extracellular matrix component and fibrotic marker. AP-1 regulates opn in vascular smooth muscle cells and macrophages[12, 36, 37] and opn was increased in Fra-1tetON and Fra-1hep-tetOFF mutants, while reduced in BDL-treated Fra-1Δembryo livers. Ectopic opn expression in hepatocytes triggers liver injury,[38, 39] but opn inactivation does not affect DDC-induced cholestasis. Thus, the Fra-1-dependent changes in opn expression are likely not sufficient to impact fibrosis development in Fra-1 mutant mice.
Cxcl9 is a chemokine implicated in T, endothelial, and stellate cell chemotaxis, associated with human hepatic fibrosis and antifibrosis in mice.[23, 41] Cxcl9 was consistently down-regulated in Fra-1tetON and Fra-1hep-tetOFF mutant livers, while increased in BDL-treated Fra-1-deficient animals. Furthermore, Fra-1 binds to the mouse cxcl9 promoter in liver samples and primary hepatocytes, likely with c-Jun as a dimerizing partner. Cxcl9 is induced by interferon-gamma (IFNγ)/Stat1 and can be potentiated by numerous transcription factors. Further studies will define whether Fra-1 actively represses cxcl9 transcription or interacts with cis-acting regulators at the cxcl9 promoter.
Bile acid levels are mainly regulated by transporters in the liver and intestine, while detoxification and excretion of harmful substances is an important hepatic function. Our analysis revealed an intriguing interaction between Fra-1 and genes involved in the liver's adaptive response to bile acids and xenobiotic overload. We observed an overall more pronounced adaptive response in Fra-1hep-tetOFF and Fra-1Δembryo mutant livers, with lower expression of bile acid uptake transporters. Consistently, circulating bile acids were increased in DDC-fed Fra-1hep-tetOFF mutants. The reason for unchanged bile acids in Fra-1Δembryo mutants upon BDL or DDC is not clear, but as these mice lack Fra-1 in all organs, a contribution from the adaptive response in the intestine or kidneys cannot be excluded. In addition, Fra-1 mutants exhibited altered basal or BDL/DDC-inducible expression of enzymes metabolizing xenochemicals and toxic endogenous compounds, such as cytochrome P450 members as well as gsta1, gsta2, and gstp1. Furthermore, Fra-1 and c-Jun bound the gstp1 promoter and c-Jun∼Fra-1 forced dimers activated transcription from a gstp1 reporter, indicating that Fra-1/AP-1 is a relevant transcriptional modulator of gstp1. GSTA1 induction has been associated with increased AP-1 activity in human hepatocyte cultures and the mouse gsta1 and gsta2 promoters harbor potential AP-1 binding sites. Thus, these genes might also be direct transcriptional targets of Fra-1/AP-1.
GSTs catalyze the conjugation of toxic compounds with GSH, thus facilitating their excretion. Besides gsta1, gsta2, and gstp1, several genes deregulated in Fra-1 mutants are connected to glutathione metabolism. For instance, slco1a1, slco1a4 can mediate GSH export from hepatocytes, while abcc2 mediate biliary excretion of GSH-conjugates. GSH attenuates hepatic oxidative stress, liver injury, and necrosis. Interestingly, Fra-1hep-tetOFF mutants had decreased liver injury upon DDC and were protected from APAP-induced liver damage, a paradigm for GSH-dependent acute liver failure, while Fra-1Δembryo mutants were more sensitive. Interestingly, cxcl9, opn, xenobiotic transporter, and detoxification enzymes were largely unaffected in Fra-2-overexpressing mutants and these mice were not protected from APAP-mediated liver injury. Our findings therefore imply a specific and novel role for Fra-1/AP-1 in the metabolic response of the liver to toxic compounds, likely through modulating GSH and xenobiotic handling. Preliminary data suggest that Fra-1 is increased in patients with APAP-induced liver injury (not shown), indicating that our findings are relevant for human disease.
ERK are the main kinases responsible for Fra-1 induction and stabilization.[28, 45] Consistent with increased Fra-1, increased ERK1/2 phosphorylation was observed in APAP-treated livers. While Jun N-terminal kinases (JNKs) have been extensively studied in APAP-induced liver injury, the function of ERK and their targets has not been reported. JNK inhibition is hepatoprotective in APAP-treated mice. However, the function of JNK has mostly been ascribed to its role in maintaining mitochondrial integrity, rather than affecting Jun proteins or APAP/GSH metabolism.[47, 48] Given that Fra-1 expression is not affected by JNK, including upon APAP, and because our genetic data indicate that Fra-1 inhibition produces the opposite outcome as JNK inhibition, we propose that the ERK/Fra-1 pathway is hepatoprotective in APAP-mediated liver injury by regulating GSH metabolism, thus counteracting the deleterious function of JNK in the mitochondria.
We thank Drs. A. Bozec, P. Hasselblatt, K. Matsuo, C. Oesterreicher, and M. Petruzzelli for valuable suggestions; the CNIO Transgenic Core Unit for ES cell injection; and G. Luque and G. Medrano for technical help with mouse procedures.