Peroxisome proliferator-activated receptor-γ as a therapeutic target for hepatic fibrosis: from bench to bedside

Abstract

Hepatic fibrosis is a dynamic chronic liver disease occurring as a consequence of wound-healing responses to various hepatic injuries. This disorder is one of primary predictors for liver-associated morbidity and mortality worldwide. To date, no pharmacological agent has been approved for hepatic fibrosis or could be recommended for routine use in clinical context. Cellular and molecular understanding of hepatic fibrosis has revealed that peroxisome proliferator-activated receptor-γ (PPARγ), the functioning receptor for antidiabetic thiazolidinediones, plays a pivotal role in the pathobiology of hepatic stellate cells (HSCs), whose activation is the central event in the pathogenesis of hepatic fibrosis. Activation of PPARγ inhibits HSC collagen production and modulates HSC adipogenic phenotype at transcriptional and epigenetic levels. These molecular insights indicate PPARγ as a promising drug target for antifibrotic chemotherapy. Intensive animal studies have demonstrated that stimulation of PPARγ regulatory system through gene therapy approaches and PPARγ ligands has therapeutic promise for hepatic fibrosis induced by a variety of etiologies. At the same time, thiazolidinedione agents have been investigated for their clinical benefits primarily in patients with nonalcoholic steatohepatitis, a common metabolic liver disorder with high potential to progress to fibrosis and liver-related death. Although some studies have shown initial promise, none has established long-term efficacy in well-controlled randomized clinical trials. This comprehensive review covers the 10-year discoveries of the molecular basis for PPARγ regulation of HSC pathophysiology and then focuses on the animal investigations and clinical trials of various therapeutic modalities targeting PPARγ for hepatic fibrosis.

Introduction

Hepatic fibrosis is an integral clinicopathological condition of chronic liver disease predisposing to cirrhosis and hepatocellular carcinoma (HCC). It is characterized by overproduction of extracellular matrix (ECM), mainly type I and III collagens. The net deposition of ECM distorts hepatic architecture and leads to portal hypertension [1, 2]. The evolution of liver fibrosis is influenced by a broad spectrum of environmental and genetic factors that perform in concert. Globally, infection of hepatitis B virus (HBV) or C virus (HCV) and non-alcoholic fatty liver disease (NAFLD) are the most common etiologies for liver fibrosis. Although conventional therapies such as antiviral strategy can attenuate or prevent liver fibrosis, a large number of subjects are either resistant to causal treatment or diagnosed with end-stage fibrosis [3]. Currently, understanding of hepatic fibrosis has moved into some new frontiers. The pathogenesis of liver fibrosis due to NAFLD has been thought to be associated with the metabolic syndrome and insulin resistance (IR), suggesting that mediators of the metabolic syndrome are also effectors of liver fibrogenesis and correlate with a more severe fibrogenic progression [4]. Clinical studies have demonstrated that IR not only determines the responsiveness to interferon therapy for liver fibrosis in chronic HCV patients [5] but also is an independent predictor of advanced fibrosis in HCV infection and NAFLD [6, 7]. Furthermore, the overwhelming prevalence of obesity worldwide has also led to increasing incidence of nonalcoholic steatohepatitis (NASH), which implies a risk of HCC once fibrosis develops [8]. Therefore, hepatic fibrosis is a broader category subdivided by progressive status and complication with the metabolic syndrome under some circumstances.

The limitations in preventive measures and lack of established therapies for chronic liver diseases highlight the urgent need to develop novel antifibrotic agents. It has been growingly established that there is a vast complexity in the transcriptional control pathways in the diseased liver [9]. These molecular discoveries provide potential drug targets for intervention of hepatic fibrosis. Transcription factor peroxisome proliferator-activated receptor-γ (PPARγ) is one of the attractive targets. The dysfunction of PPARγ transcription plays a pivotal role in liver fibrosis progression [10]. Here, we will provide a brief but up-to-date summary on PPARγ and its regulatory system in the pathophysiology of liver fibrosis. Subsequently, we will comprehensively review preclinical and clinical studies of PPARγ agonists. The goal of this work is to highlight these findings that have potential to be translated into novel therapeutics for hepatic fibrosis.

An update on PPARγ molecular biology and physiology

PPARγ is predominantly present in liver and adipose tissue and belongs to the superfamily of nuclear receptors controlling the transcription of a subset of genes [11]. It has four major isoforms, namely γ1, γ2, γ3, and γ4, which differ only in their N-terminal sequence and have similar transcriptional activities [12]. Upon ligand binding, PPARγ activates gene expression by directly binding to specific DNA sequences called peroxisome proliferator response elements (PPRE) in target gene promoters as heterodimers with the retinoid X receptor (RXR). This process is termed ligand-dependent transactivation, which is linked to the recruitment of co-activators that promote assembly of general transcriptional machinery at the promoter [13]. In addition, PPARγ can exert biological functions via ligand-dependent transrepression and ligand-independent transrepression [14]. Ligand-activated PPARγ can interfere with other transcription factors such as nuclear factor-κB (NF-κB) and activator protein-1 (AP-1), leading to transrepression of gene expression [15, 16]. In contrast, ligand-independent transrepression of PPARγ is associated with the recruitment of co-repressors that function to antagonize the actions of co-activator complexes [17]. Moreover, ubiquitination [18] and phosphorylation [19] also contribute to the ligand-independent transrepression of PPARγ (Fig. 1).

Fig. 1.

Mechanisms of PPARγ regulation of gene transcription. a The primary structure of PPARγ mainly comprises a C-terminal ligand binding domain (LBD) that harbors the ligand-dependent activation function (AF2), a zinc-finger DNA binding domain (DBD), and an N-terminal AB domain that harbors a ligand-independent function (AF1). A hinge domain links the LBD to DBD. b Ligand-dependent transactivation. In the absence of ligands, PPARγ is bound to co-repressors and prevents gene transcription. Upon binding to ligands, PPARγ undergoes conformational changes. It dissociates the co-repressors and translocates to nucleus where it recruits co-activators that have histone acetylase activity, and forms a heterodimer with RXRα. The active transcription complex then binds to the DNA response elements termed PPRE located in the promoter regions of target genes, allowing transcription activation. A number of co-activator proteins have been identified and their combinatorial usage may provide the basis for cell type- or gene-specific transcriptional responses. c Ligand-dependent transrepression. Following activated by ligands, PPARγ can interfere with the activity and function of other transcription factors such as NF-κB and AP-1, leading to disruption of their binding to target gene promoters and thereby to transcription repression. In this mechanism, PPARγ does not bind to specific DNA response elements. This transcriptional model may underlie the ability of PPARγ to negatively regulate the inflammatory responses and collagen production in hepatic fibrosis. d Ligand-independent transrepression. In the absence of ligands, PPARγ can recruit co-repressors that antagonize the effects of co-activators, leading to transcription repression of direct target genes. The most defined co-repressors are NCoR (nuclear-receptor co-repressor) and SMRT (silencing mediator of retinoic-acid and thyroid hormone receptors). In addition, protein modifications such as ubiquitination and phosphorylation have also been shown to suppress the transcription activity of PPARγ and thus play a role in the ligand-independent transrepression mechanism

Ligands for PPARγ include endogenous fatty acids and synthetic antidiabetic thiazolidinediones, as well as nonthiazolidinediones farglitazar and KR62776 [20]. Ligand activation of PPARγ commonly initiates transcription of adipogenesis-associated genes regulating adipocyte differentiation and lipid metabolism [21]. PPARγ is also genetically associated with type 2 diabetes and IR. PPARγ activation can restore insulin sensitivity via enhancing expression of glucose transporter type 4 and adiponectin that promotes fatty acid oxidation and insulin sensitivity [22]. Advances in PPARγ biology show that the pathophysiological functions are far beyond fatty acids and glucose. These new discoveries are briefly highlighted: (1) Activation of PPARγ attenuates matrix remolding [23] and oxidative stress [24], showing protection for tissue repair. (2) PPARγ regulates communication between metabolic and inflammatory reactions, exhibiting anti-inflammatory properties [25–27]. (3) PPARγ modulates cell division cycle, cell survival, and death [28, 29]. (4) PPARγ shows circadian expression in liver and adipose tissue [30]. (5) Activation of PPARγ is associated with bone loss and bone marrow adipogenesis [31]. Taken together, current knowledge suggests pleiotropic roles of PPARγ in a wide range of fundamental pathways with broad pathophysiological implications (Fig. 2). Given the importance of PPARγ in metabolic disorders and liver pathology [4], PPARγ has been suggested as a novel target for liver fibrosis therapy. Since persistent inflammatory infiltration contributes to the pathogenesis of liver fibrosis [32], the beneficial role of PPARγ in tissue repair and the anti-inflammatory property may further support the notion that targeting PPARγ is a promising strategy for treatment and prevention of liver fibrosis.

Fig. 2.

Representative natural and synthetic PPARγ ligands and ligand-specific PPARγ biological functions. PPARγ ligands include a wide range of chemical structures from endogenous and synthetic sources. The naturally occurring PPARγ agonist 15d-PGJ₂ is most commonly used in laboratory investigations. Thiazolidinediones including troglitazone, rosiglitazone, and pioglitazone represent an important class of therapeutic agents for type 2 diabetes. Meanwhile, nonthiazolidinedione PPARγ agonists have been increasingly developed with attractive pharmacological properties. Activation of PPARγ by ligands has been shown to produce varied transcriptional profiles and biological functions. PPARγ is originally thought to regulate adipogenesis and insulin sensitivity and thus serves as a master integrator of energy consumption and glucose homeostasis. Accumulating evidence shows that the biological outcomes of PPARγ activation are far beyond regulation of fat and glucose. PPARγ exhibits a pleiotropic role affecting a variety of pathophysiological pathways as illustrated in this schema, indicating that PPARγ may have broad implications in pharmacotherapy of diseases. The beneficial role in tissue repair and the anti-inflammatory property implicate PPARγ regulatory system in the management of hepatic fibrosis. Progress toward understanding PPARγ biology still continues

PPARγ regulation of stellate cell pathophysiology in hepatic fibrosis

Hepatic stellate cells (HSCs) are a type of mesenchymal cells anatomically located in the space of Disse between hepatocytes and liver sinusoidal endothelial cells. They contain a specialized organelle called lipid droplet responsible for hepatic retinoid and vitamin A storage [33]. Under physiological conditions, HSCs are quiescent and maintain liver lipid homoeostasis. Upon fibrogenic injury, HSCs proliferate and undergo myofibroblastic transdifferentiation, a dramatic phenotypic change from quiescent cells to myofibroblasts responsible for the overproduction of ECM. This process is termed HSC activation and has been defined as the central event in hepatic fibrogenesis. Activated HSCs characteristically express α-smooth muscle actin (α-SMA), and meanwhile lose their lipid droplets, making them no longer retain the adipogenic phenotype [34]. The molecular insights into PPARγ regulation of HSC pathophysiology in hepatic fibrosis are critically summarized and discussed below.

Inhibition of HSC activation and collagen production

Pathological proliferation and activation of HSCs correlate directly with depleted PPARγ expression. A pioneering study by Miyahara and colleagues demonstrated that the expression of PPARγ1 was reduced and its ability to bind to PPRE was dramatically impaired in activated HSCs, but treatment with PPARγ agonists was able to prevent HSC activation [35]. Treatment with PPARγ agonists also inhibited proliferation and induced apoptosis in HSCs, and markedly inhibited transforming growth factor (TGF)-β1-induced expression of connective tissue growth factor (CTGF), a key regulator of ECM production in hepatic fibrosis [36]. Because the effects of exogenous PPARγ ligands can be mediated by receptor-independent mechanisms [37], studies with ectopic PPARγ expression provide additional evidence supporting PPARγ as a key regulatory molecule during HSC activation. Infection with a recombinant adenoviral vector encoding PPARγ in activated HSCs resulted in restoration of morphologic characteristics of quiescent HSCs, such as reappearance of cytoplasmic lipid droplets and capacity to concentrate retinyl palmitate [38]. On the other hand, PPARγ inhibition of HSC activation leads to decreased collagen production due to the transrepression function of PPARγ. Ligand-induced PPARγ activation reduced collagen and fibronectin synthesis induced by TGF-β1 in human HSCs and directly inhibited procollagen type I promoter activity by interfering with AP-1 transactivation [39]. PPARγ was also able to directly interact with the transcription factor JunD but did not affect its expression, disrupting JunD binding to AP-1 promoter that promotes collagen I transcription in activated HSCs [38]. A site-directed mutagenesis study has identified a region proximal to -133 bp at α1(I) procollagen promoter as the action site, showing that PPARγ abrogated p300-facilitated NF-I transactivity for collagen expression in HSCs [40]. Furthermore, synergism between PPARγ and farnesoid X receptor (FXR) contributes to reducing profibrogenic activities of HSCs. For instance, ligands targeting FXR reversed the effect of depleted PPARγ expression and inhibited α1(I) procollagen production in rat HSCs [41]. PPARγ promoter could be activated by a FXR-SHP (small heterodimer partner) regulatory cascade that counter-regulated the pro-inflammatory phenotype in HSCs [42]. Studies have suggested that FXR was expressed in rat HSCs and modulated their functions [43, 44]. However, a recent report demonstrated the functional absence of FXR in mouse and human HSCs [45], which is in stark contrast to the above findings [43, 44]. An explanation involving an interspecies difference may be too simplistic and not reassuring. Therefore, further investigation is needed to address this discrepancy. The new findings would fundamentally influence novel therapies to be designed, given the role of FXR or FXR-PPARγ axis in hepatic fibrosis and the synergistic activity of FXR-PPARγ that could limit the side-effects due to single use of PPARγ agonists.

Transcriptional regulation of HSC adipogenic phenotype

HSCs indeed have an adipocyte phenotype, despite the absence of adipocyte-specific PPARγ2 isoform in HSCs [38]. Activation of HSCs shares similar cellular and molecular mechanisms with differentiation of preadipocytes [46]. However, the adipogenic phenotype of HSCs is lost concomitantly with profound changes in expression profile of adipocyte-specific genes during activation [47, 48]. There has been growing evidence that PPARγ is a pivotal molecule for adipogenic transcriptional regulation and for maintenance of quiescence in HSCs. Ectopic expression of PPARγ restored the coordinated expression of a panel of transcription factors associated with adipogenesis in cultured rat HSCs, including CCAAT/enhancer binding protein α (C/EBPα), C/EBPβ, C/EBPδ, liver X receptor α, and sterol regulatory element-binding protein 1c, whose expression was rapidly depleted during HSC transdifferentiation to myofibroblastic cells. These genetic events induced adipogenic differentiation in HSCs and led to HSC phenotypic reversal to be quiescent [49]. Moreover, a cascade of transcription factors involving PEX16 and catalase, two downstream genes of PPARγ, and adipsin, a typical adipogenesis marker, controlled the HSC switch between the lipid storing and the myofibroblast phenotypes in liver fibrotic processes [50]. PPARγ was also able to induce HSC adipogenic phenotype and quiescence via albumin that plays a direct role in the formation of cytoplasmic lipid droplets, suggesting that albumin may be a downstream effector of PPARγ in HSCs [51].

HSC transdifferentiation is manifested by a shift of the phenotype from adipogenic nature to myogenic lineages due to repressed adipogenic transcription. Core to this alteration is the depleted expression of PPARγ, the master adipogenic regulator. Restoration of PPARγ expression can achieve the fat-storing phenotype that favors suppression of HSC activation marker genes. However, there are some paradoxes. Hepatic PPARγ overexpression has been linked to exacerbated steatosis by mechanisms involving activation of lipogenic genes and de novo lipogenesis as well as increased hepatic triglyceride concentrations [52]. PPARγ could induce adipogenic conversion of hepatocytes in PPARα^−/− mice via upregulating adipocyte-specific genes and lipid accumulation [53]. These suggest that PPARγ may play a role in stimulating hepatic lipogenesis. Therefore, hepatic PPARγ overexpression, if without cellular selectivity, is likely to promote a lipogenic microenvironment in the liver, which renders the liver more sensitive to various profibrogenic stimuli. Moreover, enhanced adipogenic transcriptional regulation participates in the pathogenesis of NAFLD. Interestingly, under these adipogenic conditions, HSCs are still activated continuously and lose their adipogenic phenotype. Thus, activated HSCs seem to be resistant to the adipogenic signals that make hepatocytes steatotic [54]. More recently, a study has demonstrated that loss of retinoid-containing lipid droplets during HSC transdifferentiation did not affect liver fibrogenesis but reduced hepatocarcinogenesis [55], indicating that reversal of HSC adipogenic phenotype may have no potential for antifibrogenesis. All these findings constitute a lipid myth concerning HSC transdifferentiation. It is postulated that there should be some regulatory commonalities between HSC adipogenesis and hepatocyte lipogenesis under pathological conditions, which are poorly understood. A detailed understanding of these lipid paradoxes in hepatic fibrogenesis and the critical involvement of PPARγ will help provide new insights into HSC pathophysiology.

Epigenetic regulation of HSC transdifferentiation

PPARγ-mediated molecular events that orchestrate the coordinated expression of adipogenic genes in HSCs can also be modulated at the epigenetic level. It was found that diminished PPARγ expression was associated with DNA methylation mediated by methyl-CpG binding protein 2 (MeCP2) in activated HSCs, suggesting PPARγ modulation of chromatin packaging. DNA methylation inhibitor enhanced PPARγ expression in cultured HSCs and blocked HSC myofibroblastic transdifferentiation [56]. Some details of how PPARγ expression was suppressed by MeCP2 were subsequently elucidated, pointing to the role of microRNAs (miRNAs). Downregulation of miRNA-132 released a translational block on MeCP2, which bound to the 5′ end of PPARγ gene, where it facilitated H3K9-mediated methylation and recruited the transcription repressor HP1α. MeCP2 also promoted EZH2 (an epigenetic regulatory protein) expression and H3K27 methylation to form a repressive chromatin structure in the 3′ exons of PPARγ [57]. These molecular recognitions indicate that MeCP2 and DNA methylation exert epigenetic control over HSC adipogenic phenotype and that the repressed PPARγ expression is induced at least partially by a combination of miRNA-132, MeCP2, and EZH2 in a relay pathway. Moreover, the epigenetic PPARγ repression has been shown to be involved in necdin-Wnt pathway that mediated the anti-adipogenic HSC transdifferentiation. It was found that necdin, a melanoma antigen family protein that inhibits adipogenesis, was selectively expressed in HSCs and stimulated HSC myofibroblastic transdifferentiation via transactivation of canonical Wnt and Wnt-mediated suppression of PPARγ. Silencing of necdin abrogated the recruitment of MeCP2 and HP1α to PPARγ promoter and decreased H3K27 dimethylation at the exon 5 locus, leading to restoration of PPARγ expression and reversal of culture-activated HSCs to be quiescent cells [58]. More recently, delta-like 1 homolog (DLK1), which was selectively expressed in HSCs in the adult rodent liver, was shown to be critically involved in epigenetic repression of PPARγ. DLK1 knockdown reversed activated HSCs to be fat-storing quiescent cells via restoring PPARγ expression dependent on suppression of canonical Wnts [59]. These findings reinforce the concept that loss of adipogenic regulation underlies the shift of HSC cell fate to myofibroblastic cells, given the established role for DLK1 as an anti-adipogenic mediator [60]. Collectively, these discoveries preliminarily address the epigenetic mechanisms for PPARγ-mediated HSC transdifferentiation. Other epigenetic events such as mRNA stabilization [61] and miRNA interactions [62] have also been characterized in HSC pathophysiology in hepatic fibrosis, increasingly touching on the fundamental question that how HSC transdifferentiation is achieved by global mechanisms. Given the notion that the increased collagen expression in liver fibrosis is primarily post-transcriptional [63], it would be attractive to further explore the regulatory role of PPARγ in these epigenetic mechanisms, because modulators that selectively interfere with these pathways could be developed to pharmacologically reduce hepatic fibrosis. For example, rosmarinic acid and baicalin, two active phytochemicals in the herbal prescription Yang-Gan-Wan, have been recently shown to epigenetically derepress PPARγ expression via suppressing canonical Wnt signaling in HSCs, leading to attenuated HSC activation and reduced fibrogenesis [64].

Targeting PPARγ for hepatic fibrosis: data from animal studies

Gene therapy

The aforementioned molecular discoveries strongly highlight the critical role for PPARγ as a potential therapeutic target for hepatic fibrosis. Accumulating in vivo evidence supports this notion. PPARγ-deficient rodents subjected to bile duct ligation (BDL) [65] or fed the methionine- and choline-deficient (MCD) diet [66] developed more severe fibrosis or steatohepatitis than the wild-type ones. These observations raise a concept that gene therapy for PPARγ activation can be a therapeutic approach for hepatic fibrosis. More recently, this hypothesis has been tested in several studies with animal models. Adenovirus-mediated PPARγ overexpression downregulated a set of fibrogenic genes leading to inhibited HSC activation and reduced fibrogenesis in C57BL6 mice with MCD diet-induced hepatic fibrosis [67]. These effects were recaptured in activated human HSCs, where ectopic PPARγ overexpression arrested cell cycle at G0/G1 checkpoint and downregulated mRNA expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) and TIMP-2, two proteins preventing matrix metalloproteinase (MMP)-directed degradation of collagen, and resultantly decreased collage production [67]. These findings support the notion that normalized PPARγ function favors matrix degradation conditions in fibrogenic HSCs. In addition, adenovirus-mediated PPARγ overexpression ameliorated hepatic inflammation and fibrosis in nutritional steatohepatitis in mice fed the MCD diet via upregulating adiponectin and hemeoxygenase-1, and downregulating tumor necrosis factor-α (TNF-α), interleukin (IL)-6, MMP-2, and MMP-9 [68]. Similar treatments also decreased HSC accumulation and collagen production in BDL-induced fibrosis in rats [65]. Furthermore, lentiviral-mediated PPARγ overexpression decreased HSC proliferation and type I collagen expression and effectively ameliorated the sustained fibrosis in CCl₄-intoxiciated rats. However, this manipulation showed no significant benefits in the early phase of the fibrosis [69]. Although current data on gene therapy targeting PPARγ have indeed shown therapeutic benefits for liver fibrosis, how to achieve HSC-selective delivery needs to be further addressed, especially given the aforementioned lipid paradoxes. Encouragingly, several kinds of delivery system targeting activated HSCs have been explored with enhanced targeted potential in vivo [70]. It is tempting to develop HSC-targeted PPARγ delivery to improve the antifibrotic effects but with less off-target adverse effects.

Thiazolidinedione agonists

Antidiabetic thiazolidinediones, which were identified to be potent PPARγ-selective ligands for the first time in 1995 [71], have shown therapeutic promise for hepatic fibrosis in intensive animal investigations. Troglitazone is the first marketed thiazolidinedione agent, but it has been withdrawn due to its severe hepatotoxicity [72]. Despite this, troglitazone is still useful in the proof-of-concept studies with cell culture systems or animal models. Marra and coworkers [73] first reported that troglitazone inhibited HSC proliferation and chemotaxis in cultured human HSCs. Subsequently, they investigated the in vivo effects of troglitazone in rats with BDL-induced fibrosis [74]. Troglitazone treatment retarded liver fibrosis as indicated by significant decrease in type I collagen expression and hydroxyproline level, and also attenuated HSC activation as indicated by decreased α-SMA expression. Histochemical evaluation and reduced activity of γ-glutamyltransferase (γ-GT) further indicated troglitazone inhibition of ductular reaction. Additionally, troglitazone was found to affect epithelial-mesenchymal interaction, a significant ongoing controversy concerning the origin of fibrogenic cells in hepatic fibrosis [75–79], in this chronic cholestasis system.

Several studies have described the antifibrotic properties of rosiglitazone. Tahan and coworkers [80] first assessed the effects of rosiglitazone on a MCD diet model of NASH in rats, showing that rosiglitazone ameliorated inflammation and improved alanine transaminase (ALT), alkaline phosphatase (ALP), and IL-6 levels in the induction study and IL-1β, IL-6 and TNF-α levels in the treatment study. The anti-inflammatory property of rosiglitazone implicated in antifibrotic therapy was recaptured in mice with liver fibrosis caused by Schistosoma japonicum infection, where rosiglitazone substantially inhibited the expression of TGF-β1, α-SMA, and type I collagen concomitant with reduction in inflammatory mediators including IL-6, TNF-α, and NF-κB [81]. In nutrition-related fibrosis in mice, rosiglitazone significantly attenuated MCD diet-induced fibrosis by repressing profibrogenic factors TGF-β1 and CTGF, and also ameliorated nutritional fibrosing steatohepatitis evidenced clearly by morphological and biochemical observations [82]. Recently, a study aiming at characterizing a mice model that may faithfully recapitulate the pathophysiology of human NASH showed that rosiglitazone substantially attenuated liver steatosis, inflammation, and fibrosis via stimulation of antioxidant gene expression and mitochondrial β-oxidation, and suppression of oxidative stress [83]. Furthermore, rosiglitazone could be synergistic with agonists targeting retinoic acid receptor (RAR) in reducing thioacetamide-induced hepatic fibrosis in rats via inhibiting HSC proliferation and reducing TGF-β1 and TNF-α [84], and also protected against alcoholic fatty liver correlated closely with adiponectin signaling and the stimulated sirtuin 1-AMP-activated kinase system in mice [85]. However, Garcia-Ruiz and coworkers revealed that rosiglitazone was ineffective in ob/ob mice with NASH. Rosiglitazone treatment for 12 weeks improved neither the NAFLD inflammatory activity nor the mitochondrial injury in the model. On the contrary, rosiglitazone increased liver steatosis, particularly microvesicular steatosis and oxidative stress. The anti-inflammatory property of rosiglitazone was absent in this controlled study [86]. These conflicting results may be due to the different pathogenetic pathways of the NAFLD models. Young ob/ob mice are known to have fatty livers without overt histological evidence of steatohepatitis, whereas MCD diet-fed rats, a model used in previous studies [80], exhibit characteristics of NASH with histological evidence of both significant hepatic steatosis and inflammation.

Much work has focused on pioglitazone in experimental liver fibrosis treatment. Kon and coworkers [87] first studied the therapeutic effects of pioglitazone on early phase liver fibrosis caused by CCl₄ intoxication in rats, showing that pioglitazone inhibited the promoter activity of α1(I) procollagen gene and HSC proliferation and prevented hepatic inflammation and necrosis. Similar results were later obtained by others using the same rodent model [88]. Moreover, pioglitazone prevented the activation of Kupffer cells that may produce toxic mediators resulting in hepatocyte impairment and profibrogenic responses in the liver [87]. Consistent with this, pioglitazone suppressed the sensitization of Kupffer cell to lipopolysaccharide (LPS) and attenuated ethanol-induced liver injury in rats [89]. It was elucidated that pioglitazone prevention of LPS-induced liver injury was mediated by suppression of TNF-α secretion from Kupffer cells [90]. Another report also confirmed that the therapeutic effects of pioglitazone on hepatic inflammation and necrosis induced by ethanol and LPS were dependent on suppression of TNF-α [91]. Furthermore, Tomita and coworkers investigated the mechanisms underlying pioglitazone protection against alcoholic liver injury. They demonstrated that pioglitazone significantly attenuated steatosis and lipid peroxidation in rats through activating hepatocyte growth factor (HGF)/c-Met signaling, enhancing very low-density lipoprotein-dependent lipid retrieval, and suppressing triglyceride synthesis [92]. In choline-deficient l-amino acid-defined (CDAA) diet-induced NASH model, pioglitazone not only improved hepatic steatosis and prevented fibrogenesis in the rat liver, but also reduced the mRNA expression of TIMP-1 and TIMP-2, which were upregulated during fibrogenesis [93]. In the same model, pioglitazone at a clinical dose for type 2 diabetes ameliorated hepatic fibrosis via downregulation of procollagen, α-SMA and TGF-β1, as well as serum hyaluronic acid [94]. A recent mechanistic investigation showed that pioglitazone inhibited the suppressor of cytokine signaling-3 (SOCS-3), a key intermediator between impaired inflammatory and insulin pathways, in rats fed a high-cholesterol and fructose diet [95]. This may at least partially explain pioglitazone’s benefits in dietary hepatic injury.

Although the aforementioned data strongly favor the therapeutic potential of pioglitazone for hepatic fibrosis, not all studies have reached consistent conclusions. Leclercq and coworkers assessed pioglitazone efficacy for established liver fibrosis in three different rat models caused by CCl₄ injection, CDAA diet, and BDL. Early administration of pioglitazone reduced hepatic fibrosis, type I collagen content, and profibrotic genes, as well as the number of activated HSCs, but did not significantly affect necroinflammation. When pioglitazone treatment was initiated in the later course of pathogenesis, no antifibrotic effects were observed. Similarly, pioglitazone reduced the severity of fibrosis induced by CDAA diet when introduced early, while delayed treatment remained ineffective. In contrast, pioglitazone failed to interrupt progression of fibrosis due to BDL, irrespective of the timing of its administration [96]. These results suggest the limited efficacy of pioglitazone for established fibrosis and imply that the therapeutic outcomes can be affected by the nature of hepatic injury, disease duration, and the time when medication is introduced. It is also presumed that PPARγ ligands cannot be sufficient to reverse the pathogenesis at the advanced stage when a larger number of HSCs have been activated. Later, this group further used BALB/c and C57BL6/J mice injected with CCl₄ to perform in vivo and in vitro evaluation of pioglitazone’s antifibrotic potential. The data showed that pioglitazone neither prevented hepatic fibrogenesis in two different strains of mice, nor impacted collagen I gene expression and HSC activation, although serum adiponectin concentration was elevated [97]. Furthermore, pioglitazone did not inhibit the myofibroblastic transdifferentiation in cultured HSCs isolated form the liver tissues [97]. These are in sharp contrast to other reports on the antifibrotic properties of pioglitazone in various models of fibrosis in rats [87–95]. The interspecies variations and polymorphisms in the expression and tissue distribution of PPARγ may offer explanations for the therapeutic failure of pioglitazone in mice. Noteworthy, these paradoxical results somewhat complicate the interpretation of the antifibrotic profiles of thiazolidinediones.

Nonthiazolidinedione agonists

Several nonthiazolidinedione PPARγ agonists such as KR62776 and GW570 also exhibited therapeutic benefits in experimental fibrosis. Two independent studies with CCl₄ toxication rat model demonstrated that these two agents prevented hepatic fibrogenic responses to injury via inhibiting HSC activation and abrogating collagen production [98, 99]. Of interest is that KR62776 selectively stimulated rat HSC apoptosis with little effects on hepatoma cells, suggesting that KR62776 is likely to possess specific inhibitory actions on HSCs without affecting the hepatic origin cells [98]. This selectivity could be explained by the finding that KR62776 adopted a mode distinct from that of thiazolidinediones for binding to the receptor domain [100]. Furthermore, PPARγ ligands from natural products also deserve attention. We have reported that the nontoxic natural compound curcumin was a potent PPARγ activator and was able to disrupt TGF-β and NF-κB signalings, leading to inhibited expression of CTGF and ECM in activated HSCs, and that curcumin also protected against liver injury and fibrogenesis by attenuating oxidative stress and suppressing inflammation in rats [101–103]. These discoveries expand the structural type of PPARγ agonists with antifibrotic potential. This is truly illuminating given the urgent need to develop novel antifibrotics in the present-day where chronic liver diseases are overwhelmingly widespread. The preclinical animal studies of targeting PPARγ regulatory system for hepatic fibrosis are summarized in Table 1.

Table 1.

Preclinical animal studies of PPARγ-targeted therapy for hepatic fibrosis

Agent	Animal model	Mode of action	References
Ectopic PPARγ	BDL (rat)	Reduce activated HSCs, inhibit collagen production	[65]
	MCD diet (mouse)	Inhibit HSC activation, anti-inflammation, upregulate adiponectin, reduce ECM	[67, 68]
	CCl4 (rat)	Inhibit HSC activation, reduce collagen production	[69]
Troglitazone	BDL (rat)	Attenuate HSC activation, reduce γ-GT level	[74]
Rosiglitazone	MCD diet (rat, mouse)	Improve transaminases, anti-inflammation, inhibit TGF-β and CTGF	[80, 82]
	Schistosoma japonicum (mouse)	Reduce TGF-β and collagen production, anti-inflammation	[81]
	High-fat diet (mouse)	Anti-inflammation, enhance mitochondrial β-oxidation, inhibit oxidative stress	[83]
	Thioacetamide (rat)	Synergistic with RAR ligands, inhibit HSC proliferation, reduce TGF-β1, anti-inflammation	[84]
	Ethanol (mice)	Stimulate adiponectin signaling, activate sirtuin 1-AMP-activated kinase system	[85]
	ob/ob mice with NASH	Ineffective, increase oxidative stress and steatosis	[86]
Pioglitazone	CCl4 (rat)	Reduce HSC proliferation, inhibit α1(I) procollagen gene promoter, inhibit Kupffer cell activation, anti-inflammation	[87, 88]
	CCl4 (mouse)	Ineffective	[97]
	Ethanol (rat)	Inhibit Kupffer cell activation, suppress TNF-α, activate HGF/c-met signaling, anti-inflammation	[89–92]
	CDAA diet (rat)	Reduce matrix, inhibit TGF-β and collagen production, downregulate SOCS-3	[93–95]
	BDL (rat)	Ineffective	[96]
KR62776	CCl4 (rat)	Inhibit HSC activation, stimulate HSC apoptosis	[98]
GW570	BDL (rat), CCl4 (rat)	Inhibit collagen production	[99]
Curcumin	CCl4 (rat)	Inhibit oxidative stress, anti-inflammation, reduce HSC activation	[103]

Targeting PPARγ for NASH and hepatic fibrosis: data from clinical trials

NASH is characterized by excessive fat accumulation in the liver coexisting with cell injury and inflammation. It is suggested that between 10 and 29 % of individuals with NASH develop fibrosis and even cirrhosis within 10 years [104] and 4 to 27 % of individuals with NASH-induced cirrhosis develop HCC [105]. The pathophysiology of NASH is generally defined by a “two-hit” theory. The first hit is considered as the increase of free fatty acids in hepatocytes, which inhibits β-oxidation and further amplifies the accumulation of fatty acids. The second hit involves all of the mechanisms that contribute to the development of inflammation and fibrosis [106]. Thus, NASH is closely linked to the metabolic syndrome such as abnormal lipid metabolism and IR. In general, drugs that reverse IR may be beneficial in NASH. Indeed, the past decade has witnessed intensive clinical evaluations of PPARγ agonists, mainly thiazolidinediones, for treating NASH and hepatic fibrosis.

Troglitazone therapy

In 2001, Caldwell and coworkers published the first open-label pilot study of troglitazone in NASH patients, which was completed before this drug was withdrawn. In this trial, 10 women with biopsy-proven NASH, some of whom complicated with obesity, type 2 diabetes or cirrhosis, were received troglitazone therapy for up to 6 months. Although 70 % of subjects had normalized ALT levels, only mild histological improvements were observed in follow-up biopsies [107]. This short-term trial indicates an absent correlation between liver enzyme levels and histology, and the limited efficacy of troglitazone for persistent steatohepatitis at the histological level. Later, this group surveyed the extended follow-up data of NASH patients, who were previously treated with troglitazone, to assess how lifestyle modifications affect clinical and histological parameters related to NASH recurrence. They reported that short-term troglitazone therapy led to histological improvements and discontinued progression of fibrosis in two subjects, who had sustained exercise programs and weight reduction [108]. These benefits cannot reflect long-term effects of troglitazone, but suggest that thiazolidinedione therapy combined with sustained lifestyle changes could be preventive for NASH relapse that is not inevitable after medication discontinuation.

Rosiglitazone therapy

Neuschwander-Tetri and coworkers performed the first open-label uncontrolled trial of rosiglitazone in 30 overweight patients with well-characterized NASH. Their interim results showed that rosiglitazone therapy for 24 weeks restored insulin sensitivity, reduced hepatic fat content, and improved biochemical evidence of hepatocyte injury, indicating that IR is indeed critical to NASH pathogenesis [109]. Their final results revealed that treatment for 48 weeks significantly improved insulin sensitivity and histological markers of NASH [110]. Importantly, although there was no significant improvement in overall fibrosis, some patients had improved zone 3 fibrosis, suggesting that hepatic fibrosis may occur in the absence of characteristic features of NASH. Whether longer treatment durations are associated with resolution of the zone 3 fibrosis or improvement in the overall fibrosis remains uncertain; and the nature of zone 3 fibrosis may be worthy of detailed examination in future trials. Of equal importance regarding this trial is the 6-month post-treatment follow-up evaluation, showing that liver enzyme levels had increased to near pretreatment levels [110]. This implies that permanent use of thiazolidinediones for insulin sensitization is probably essential in preventing progressive liver injury in patients with response to the initial treatment. Moreover, in a prospective, longitudinal controlled study enrolling 74 individuals with NASH, 48 weeks of rosiglitazone treatment reduced histological features of steatohepatitis including steatosis and ballooning in contrast to the diet and exercise group, but fibrosis was not significantly improved [111]. Examination of liver biopsies from NASH patients demonstrated that rosiglitazone effects were associated with increased crystalline inclusions in hepatocyte mitochondria [112]. However, whether these are adaptive or pathological remains unknown and further studies are warranted to assess hepatic mitochondrial function during thiazolidinedione therapy for NASH. With regard to drug safety, an open-label, inadequately controlled trial enrolling 68 type 2 diabetes patients with NASH showed that rosiglitazone treatment for 24 weeks led to normalized ALT levels in one-third of patients without serious adverse events [113]. More recently, a small-scale uncontrolled study recaptured the similar results in 27 NASH subjects for 6-month treatment, showing that rosiglitazone was well tolerated and significantly improved liver function without adversely affecting the lipid profiles [114].

Ratziu and coworkers [115] conducted a randomized placebo-controlled trial of rosiglitazone in 63 patients with NASH. Rosiglitazone improved steatosis and transaminase levels, but did not affect any other liver histological lesions including fibrosis and necroinflammation. Rosiglitazone was also proven safe in general, despite weight gain and an asymptomatic low-grade reduction in hemoglobin level occurred in some patients. This study had a limitation of low baseline NAS (NAFLD activity score, a histological feature scoring system with reasonable inter-rater reproducibility addressing the full spectrum of lesions of NAFLD [116]), leaving little room for improvements. Thus, this group enlarged the investigation, in which 53 patients who had been subjected to a 1-year randomized trial were enrolled in an open-label extension trial for 2 additional years [117]. Rosiglitazone showed substantial antisteatogenic effects only in the first year of treatment without additional benefits in longer therapy, despite the maintained effects on insulin sensitivity and transaminase levels. In addition, no significant improvements in fibrosis during 2 or 3 years of treatment were seen, which was consistent with their data in 2008 [115], rendering it improbable that the inefficacy observed after only 1 year of treatment could be due to the short treatment duration. An explanation could be that ameliorated steatosis and decreased aminotransferases are directly associated with improved insulin sensitivity, whereas fibrosis may not be reduced significantly by simply correcting IR.

Given the modest effects of rosiglitazone on NASH resolution and fibrosis regression, studies aimed at combination therapies that synergistically improve IR seem to be laudable. Recently, two open-label, randomized trials have assessed the efficacy of rosiglitazone and metformin in combination versus rosiglitazone or metformin alone. Rosiglitazone therapy seemed to be more effective in metabolic control and histological improvement in 64 NAFLD patients with impaired glucose metabolism and NAFLD activity score of at least 5 in liver biopsy [118]. Although combination therapy effectively decreased the NAFLD score in follow-up biopsy, fibrosis did not change significantly after the treatment [118]. Another trial enrolling 137 subjects (108 completed the trial) with biopsy-proven NASH showed that 48 weeks of combination of rosiglitazone and metformin conferred no greater benefit than rosiglitazone alone in all histologic parameters including steatosis, hepatocellular inflammation, and fibrosis, leading to the conclusion that adjuvant therapies were ineffective [119].

Pioglitazone therapy

Shadid and colleagues [120] first examined pioglitazone’s effects on liver fatty acid metabolism in a trial enrolling 20 upper-body obese subjects, five of whom with NASH, showing that pioglitazone improved liver function in NASH. Subsequently, a pilot study involving 18 non-diabetic patients with NASH demonstrated that pioglitazone therapy for 48 weeks led to improvement in biochemical and histological features of NASH, as well as resolution of fibrosis [121]. However, this study was not well controlled and had a narrow margin of statistical significance for fibrosis improvement. Belfort and coworkers performed the first placebo-controlled trial using diet plus pioglitazone in 55 patients with NASH for 6 months. Pioglitazone led to clear metabolic and histological improvements, but no significant reduction in fibrosis was observed [122]. A similar placebo-controlled trial enrolling 74 non-diabetic patients with histological NASH for 1 year demonstrated that pioglitazone resulted in improvements in metabolic and histological parameters, most notably in liver injury and fibrosis indicated by biopsy [123]. This trial offered clear evidence of pioglitazone benefits for fibrosis, which was contrary to the Belfort’s data [122]. Furthermore, a pilot study, in which 21 NASH patients received 48-week pioglitazone therapy with significant improvements and 13 of them were followed for repeated liver biopsy for an additional 48 weeks showed that pioglitazone discontinuation resulted in elevated serum ALT levels and marked worsening of parenchymal inflammation and steatosis, but no change in fibrosis, suggesting that long-term therapy with pioglitazone may be necessary to maintain improvements in NASH patients [124]. Furthermore, several lines of clinical evidence indicated that improvements in liver histology during pioglitazone treatment for NASH might correlate with increased secretion of adiponectin by peripheral adipocytes [125], and with restoration of adipose tissue insulin sensitivity, as well as increased oxidation of free fatty acids in the liver [126].

There are also several clinical reports about pioglitazone versus vitamin E for NASH therapy. It is proposed that the antioxidant vitamin E may have benefits for NASH, because oxidative stress contributes critically to the pathogenesis of NASH and hepatic fibrosis [127]. A small-sized pilot study showed that pioglitazone combined with vitamin E led to greater improvements in NASH histology, but vitamin E alone did not significantly reduce histological parameters other than steatosis [128]. This group later planned a randomized, double-blind, placebo-controlled trial for 96 weeks to evaluate the therapeutic efficacy of pioglitazone and vitamin E in 247 non-diabetic patients with NASH [129]. The results demonstrated that pioglitazone had no benefits over placebo for the primary outcome, but led to significant improvements in some histological features of NASH, such as steatosis and lobular inflammation. However, neither pioglitazone nor vitamin E ameliorated fibrosis and portal inflammation [130]. A conclusion still cannot be drawn about the relative efficacy of pioglitazone and vitamin E in NASH treatment. Larger extended trials are warranted to assess the long-term and sustained effects of pioglitazone on the biochemical and histological features of NASH and hepatic fibrosis.

Farglitazar therapy

More recently, McHutchison and coworkers [131] evaluated the antifibrotic potential of farglitazar, a nonthiazolidinedione insulin-sensitizing agent approximately 100-fold more potent than rosiglitazone in its ability to both bind to and activate PPARγ in a large-scale randomized placebo-controlled trial in 265 patients with HCV infection and biopsy-proven fibrosis of Ishak stages 2–4. Treatment with farglitazar for 52 weeks failed to improve immunohistochemical markers of fibrogenesis, including well-accepted biological markers of HSC activation and collagen synthesis. There was also no significant improvement in other more standard histological assessments such as ranked assessment of the paired biopsy specimens, and Ishak fibrosis and necroinflammation scores, although minor decreases in steatosis and serum ALT values of uncertain clinical significance were observed. Of note, farglitazar failure in clinically reducing fibrosis is in contrast to the preclinical evidence that farglitazar inhibited activation of isolated HSCs and decreased collagen content in a rodent model of fibrosis [131]. It is possible that animal models of hepatic fibrosis that mimic injury over weeks to months are unable to reflect the duration of injury and fibrogenesis in HCV infection that results in maturation of ECM and scar tissue over many years [132]. The cross-linked ECM components in established fibrosis may be resistant to resorption with the antifibrotic agents, which have no effects on the underlying cause of injury (e.g., HCV infection) and are given over a relatively short period of time.

Outcome interpretation and the safety issues

To date, the only treatment for NASH is moderate weight loss and lifestyle modification. However, the majority of patients fail to execute these lifestyle measures persistently, and thus pharmacological treatment should be considered [106]. Current clinical trials testing PPARγ agonists in NASH are mainly based on the logical assumption that improving IR will lead to an improvement in liver disease. The outcomes generally demonstrate that reduction in aminotransferases and loss of steatosis are the two most reproducible hepatic effects of PPARγ agonists, but there are still a large proportion of subjects who are non-responders or only partial responders. Noteworthy, no studies have convincingly shown significant histological improvement for NASH fibrosis, although two uncontrolled trials [110, 121] and one controlled trial [123] reported improved fibrosis. Several meta-analyses also suggest the modest efficacy of thiazolidinediones for fibrosis [133–135], except one indicating that pioglitazone significantly improves fibrosis in NASH [136]. There is awareness that small sample size, short treatment duration, and heterogeneity of trials may influence the evaluation of drug efficacy for fibrosis. In addition to these, some other issues should also be taken into consideration. First, thiazolidinediones may act directly on adipose tissue, with secondary effects on the liver, because they improved glucose homeostasis and triglyceride levels in the absence of liver PPARγ [137], suggesting that hepatic PPARγ expression contributes relatively little to the beneficial effects of thiazolidinediones. It can thus be postulated that thiazolidinediones improve NASH as a result of its primary insulin-sensitizing effects on adipose tissue, which reduces the flux of free fatty acids from adipose tissue to the liver; whereas PPARγ activation may induce steatogenic responses in the liver [52, 53, 86] and compromise thiazolidinediones’ benefits for liver histology. Therefore, correcting IR is not enough to reverse liver injury in NASH, and different pathways, such as inflammatory and fibrotic cascades, should be targeted in order to achieve higher antifibrotic potency. Second, although PPARγ activation inhibits HSC activation and ECM production, and has therapeutic potential for liver fibrosis, it should be noted that PPARγ expression is dramatically depleted during excessive HSC activation [34–36, 38]. Under this condition, ligands probably cannot fully activate PPARγ signaling pathways, due to too less amount of receptor (PPARγ) available for binding, which is supported by the finding that rosiglitazone did not affect NASH in PPARγ-deficient mice fed the MCD diet [66]. This may also account for some cases where thiazolidinediones are ineffective in rodents with established fibrosis [96, 97]. It will be tempting to test whether PPARγ ligands combined with PPARγ-targeted gene therapy can coordinately yield more potent antifibrotic efficacy in animals and humans. The third challenging issue is that the lack of solid evidence-based justification of therapeutic endpoints may produce negative histological results, because the interplay between IR and hepatic effects remains to be defined. Some limitations in current fibrosis biomarkers, such as lack of sensitivity during the initial stages of fibrosis and lack of etiology-specificity, and the sampling variability of liver biopsy may affect examination of the outcomes. More sensitive biomarkers for fibrosis assessment are critically needed in evaluation of drug efficacy, particularly in situations where control of the liver injury is of principal importance, and the optimal biomarkers should be able to predict both liver histology and clinically significant endpoints [138]. It is anticipated that proteomic and metabonomic approaches may afford the possibility to characterize sensitive and specific biomarkers or biomarker panels for assessment of hepatic fibrosis [139, 140].

Weight loss can upregulate PPARγ expression and improve insulin sensitivity [141, 142], which may reinforce the idea that thiazolidinediones combined with lifestyle modification have synergistic benefits in NASH and fibrosis [108, 111]. However, these agents are approved for use only in diabetes patients; any extension of their indications should carefully assess the adverse effects and safety issues in non-diabetics, especially given that many studies [108–110, 124] argue for prolonged therapy in NASH. Apart from weight gain and fluid retention, cardiovascular effects, such as myocardial infarction, are emerging in thiazolidinedione therapy [143]. Recently, rosiglitazone has been withdrawn from the European market [144] and limited by the U.S. Food and Drug Administration [145], because of cardiovascular risk. It is still unclear whether thiazolidinedione-associated weight gain and cardiovascular events would ultimately limit the long-term benefits in patients with NASH and related liver diseases [133, 134]. Furthermore, a large proportion of patients with progressing liver diseases have high liver enzyme levels, whereas thiazolidinedione treatments are usually recommended to be avoided in these patients due to potential hepatotoxicity, especially given that there have been clinical reports on the rosiglitazone- and pioglitazone-induced hepatotoxicity [146]. Recently, thiazolidinediones have also been linked to cancer, since pioglitazone long-term use was associated with an increased risk of bladder cancer [147, 148]. Moreover, ligand-activated PPARγ mutant was also likely to enhance Angptl4-mediated angiogenesis and promote ErbB2 oncogene-induced tumorigenesis, indicating the necessity of epidemiologic analysis in patients treated with PPARγ agonists for possible effects on tumor development [149]. These findings imply a potential risk for patients with HCC or complicated with other cancers when receiving thiazolidinedione therapy. In summary, it is clear that available data represent clues for subsequent trials and highlight the methodological challenges of trial design in this disease. Whether thiazolidinedione efficacy and safety can be successfully demonstrated by better selection of NASH or fibrosis patients in non-diabetic population needs to be further determined before clinical recommendations.

Prospects for novel PPARγ modulators for hepatic fibrosis

The 10-year discoveries on PPARγ in hepatic fibrosis from bench to bedside not only portray a novel paradigm of hepatic fibrosis but also illuminate a framework for developing new antifibrotic PPARγ modulators. Some specific issues that remain unanswered should be prioritized in future studies.

Novel agents targeting PPARγ for antifibrogenesis should have selectivity and remove some of the side-effects associated with thiazolidinediones. It was found that activation of PPARγ2 not only induced lipogenesis in hepatocytes [150], which further led to HSC proliferation and apoptosis resistance [151], but also inhibited hepatocyte growth by disrupting cyclin D1 transcription [152]. These findings indicate that future PPARγ ligands for hepatic fibrosis should be HSC-specific with high target selectivity toward PPARγ1 isoform so as to maximize the therapeutic benefits of PPARγ activation. More intriguingly, several non-agonist PPARγ ligands have recently been developed exhibiting a separate biochemical activity, i.e., blocking cyclin-dependent kinase 5-mediated phosphorylation of PPARγ. Their unique mode of binding to PPARγ made them have no serious side-effects of thiazolidinediones such as fluid retention and weight gain [153, 154]. These new compounds clearly suggest that the development of safer PPARγ ligands for hepatic fibrosis is feasible.

Regulatory mechanisms of HSC by PPARγ are becoming clear, but their roles in wound response of the liver as a whole are still elusive and complex. For example, it is unclear whether reversal of HSC activation by PPARγ impairs liver regenerative responses. It has been increasingly understood that HSC activation also has pro-regenerative purposes, because the soluble factors (e.g., HGF) secreted by activated HSCs may be essential for liver regeneration [155]. Hence the best therapeutic strategy for hepatic fibrosis should focus on suppressing the former while promoting the latter. It is conceivable that inhibition of HSC activation combined with protection of hepatocytes may represent an optimal approach. Pharmacological potential for crosstalk of ligands between PPARα and PPARγ may exactly afford this possibility. Activation of PPARα, which is mostly expressed in hepatocytes in the liver, stimulates fatty acid transport and β-oxidation to prevent triglyceride accumulation, and controls inflammatory responses [156]. A number of data showed that PPARα agonists prevented liver steatosis and inflammation and reversed fibrosis in animals [157–160] and had histological benefits in NASH patients [161, 162]. Based on these promising effects of PPAR ligands, it can be proposed that dual agonists for PPARα/γ could be more perspective for therapeutic development against NASH and fibrosis owing to their optimal effects in the liver. Currently, such dual agonists have been successfully developed for pharmacological evaluation [163–165]. However, in vivo investigations testing their efficacy for the treatment of hepatic fibrosis have not been reported so far. Such studies may be highly illuminating and should be planned.