To study liver fibrosis, male wild type and PPARβ/δ-null mice received intraperitoneal injections of CCl₄ twice per week for 6 weeks. The effect of activated PPARβ/δ on liver fibrosis was assessed by treating mice with the well-characterized selective ligand GW501516, in addition to CCl₄. CCl₄-treated wild type and PPARβ/δ-null mice developed moderate, centrolobular necrosis with inflammatory, periportal, neutrophil and Kupffer cell/macrophage infiltration. Calcium deposits were found in the necrotic areas (not shown). Liver pathology was slightly more developed in CCl₄-treated wild type compare to PPARβ/δ-null mice (Figure 1A). This indicated that, in the absence of exogenous activation, PPARβ/δ did only moderately impact liver fibrosis. However, wild type mice co-treated with CCl₄/GW501516 showed more severe centrolobular necrosis, marked neutrophil infiltration, and degenerated neutrophils and macrophages, including Kupffer cells. This result correlated with serum alanine aminotransferase (ALT) levels and liver weight (Figure 1B). These effects were not observed in similarly treated PPARβ/δ-null mice, which suggested that the GW501516 effect was dependent on PPARβ/δ expression. The expression profiles of Pparα and Pparγ under the different experimental conditions were similar in PPARβ/δ-null mice and wild type mice ( Additional file 1: Figure S1). This indicated that PPARβ/δ deletion probably did not trigger compensatory effects. Nonetheless, it is worth noting that CCl₄ treatment reduced the expression of Pparα and Pparγ by more than 50%.

Next, because CCl₄ is known to impact hepatic lipid homeostasis, we tested the distribution of neutral lipids 16 . The total triglycerides increased with CCl₄, and even more with CCl₄/GW501516 co-treatment in wild type mice, whereas the different treatments did not significantly affect triglyceride levels in PPARβ/δ-null mice (Figure 1C). The higher levels of triglycerides in the livers of CCl₄-treated wild type compared PPARβ/δ-null mice may indicate a moderate contribution of PPARβ/δ in the wild type animal, consistent with the liver pathology described above. On the contrary, neither treatment affected the levels of total free cholesterol or cholesterol esters in wild type or PPARβ/δ-null mice. The mechanisms underlying this PPARβ/δ-dependent accumulation of triglycerides, which triggered hepatic steatosis in co-treated wild type mice, is in accordance with marked fibrosis but remain to be explored.

The CCl₄/GW501516 treatment markedly increased Pparβ/δ expression in wild type mice. Importantly, the expression of pyruvate dehydrogenase kinase-4 (Pdk4) and Perilipin 2 (Plin2), two target genes of PPARβ/δ were also increased in wild type mice, which indicated transcriptional activation (Figure 1D). Note that GW501516 alone stimulated the expression of Pdk4 and Plin2 in wild type but not PPARβ/δ-null mice (not shown), while CCl₄ alone also stimulated the expression of Pdk4 in both wild type and PPARβ/δ-null mice and Plin2 in wild type mice, but the stimulation was highest in wild type co-treated mice.

Collectively, these results provided evidence that GW501516-dependent PPARβ/δ activity was enhanced in hepatic fibrotic tissues. This suggested that in this model, PPARβ/δ might exacerbate uncontrolled liver repair. This is consistent with the profibrotic effect of GW501516 reported by others, although their studies did not included null mice 15 .

In liver sections from untreated wild type and PPARβ/δ-null mice, F4/80 staining (macrophages, Kupffer cells) was weak (Figure 2A). However, in CCl₄-injured livers, we measured an important increase in staining, whereas the number and localization of recruited macrophages/Kupffer cells were similar in both genotypes. Most of the damage was located around blood vessels. Administration of GW501516 alone doubled the number of infiltrated macrophages/Kupffer cells in the wild type mice, but not in PPARβ/δ-null mice (not shown).

Consistent with these observations, GW501516/CCl₄-damaged wild type livers presented high levels of the pro-inflammatory markers such as macrophage inflammatory protein-1α (Mip-1α), monocyte chemoattractant protein-1 (Mcp-1), platelet-derived growth factor BB (Pdgfbb), tumor necrosis factor alpha (Tnf-α), transforming growth factor beta 1 (Tgfβ1) and the mouse homolog (F4/80) of the EGF-like module-containing mucin-like hormone receptor-like 1 in wild type mice (Figure 2B), which are known to be expressed in activated HSCs, infiltrated Kupffer cells, and other immune cells. Treatment with CCl₄ alone also induced the mRNA levels of these markers, but to a lower extent and in a PPARβ/δ-independent manner (Figure 2B). This was confirmed at the protein level for MIP-1α and MCP-1 (Figure 2C). Thus, GW501516-activated PPARβ/δ in fibrotic liver increased inflammation, most likely as a consequence of increased immune cell infiltration and HSC activation and proliferation.

Fibrosis is characterized by the deposition of ECM components. They are secreted by activated HSCs during liver repair. Histological sections stained with Sirus red showed normal distributions of collagen around liver blood vessels in untreated wild type and PPARβ/δ-null groups (Figure 3A). CCl₄ treatment caused a moderate increase in perilobular and centrolobular collagen distributions, widespread pericellular fibrosis, and centro-central fibrotic septa, which was more important in wild type compared to PPARβ/δ-null livers as determined by staining quantification ( Additional file 1: Figure S2). The damages were graded with an Ischak’s score of 2 in most liver sections. Co-administration of GW501516 and CCl₄ strongly enhanced collagen deposition at the centrolobular and periportal regions, and collagen fibers extended within the lobule and out to the surrounding hepatocytes in wild type mice, but not in similarly-treated PPARβ/δ-null mice. In wild type mice, CCl₄/GW501516 caused hepatic damage with an Ischak’s score of 3 in most liver sections. This result was supported by quantification of the Sirus Red staining ( Additional file 1: Figure S2) and by staining for fibrin with Martius/Scarlet/Blue (MSB) (Figure 3A right panels). These results showed that GW501516-activated PPARβ/δ in CCl₄-injured livers enhanced collagen deposition and thus, promoted fibrosis; however, this effect was not observed in PPARβ/δ-null mice.

In line with these observations, CCl₄ treatment increased pro-Col1α1 and pro-Col3α1 mRNA expression in both wild type and PPARβ/δ-null genotypes, though to a slightly lesser extent in the latter (Figure 3B), in agreement with the Sirus Red staining ( Additional file 1: Figure S2). Interestingly, the combined CCl₄/GW501516 treatment further induced the expression of these genes compared to CCl₄ alone only in wild type mice.

Taken together, our observations showed that activation of PPARβ/δ in CCl₄-injured livers strongly promoted collagen deposition, a hallmark of liver fibrosis.

During fibrogenesis, HSCs proliferate and transdifferentiate to myofibroblasts that express α-SMA 1 . Immunohistochemistry with α-SMA and Ki67 antibodies showed a quasi absence of staining in untreated wild type and PPARβ/δ-null liver sections (not shown). After chronic CCl₄ exposure, many α-SMA and Ki67 positive cells were observed within the lobule and in the fibrotic septa, at slightly higher levels in wild type compared to PPARβ/δ-null mice (Figure 4A). These results indicated that CCl₄ treatment induced the activation and proliferation of HSCs. Administration of GW501516 to CCl₄-injured livers further increased the lobular distribution and the number of activated and proliferative HSCs only in wild type mice. Since GW501516 alone had no effect on HSC proliferation in absence of CCl₄ treatment in wild type mice (not shown), this implied that CCl₄ activation of the HSCs was a prerequisite for the PPARβ/δ-dependent effect on cell proliferation.

Consistent with these results, CCl₄ administration increased the levels of α-Sma mRNA by 2-fold in both wild type and PPARβ/δ-null mice (Figure 4B). Combined administration of GW501516 and CCl₄ strongly increased the expression of both Desmin and α-Sma transcripts in wild type mice, but not in PPARβ/δ-null mice. This suggested that the agonist action was PPARβ/δ-dependent. These results demonstrated that ligand-activated PPARβ/δ increased the proliferation of activated HSCs in CCl₄-injured mouse liver, a cellular process that promotes and amplifies fibrosis.

The molecular mechanisms underlying PPARβ/δ regulation of activated HSC proliferation after liver injury are not known. To address this question, we first explored whether treatment with GW501516 also impacted gene expression in human activated HSC LX-2 cells, which express key genes for hepatic fibrosis and are phenotypically similar to primary activated human HSCs in vivo 17 . These cells are in a pre-activated state and they progressively express activation markers after cultivation 18 . Therefore, they present some similarity with CCl₄ activated HSCs. The results presented in Figure 5 showed that the expression of genes stimulated in the mouse liver after CCl₄/GW501516 treatment (see Figure 1 4) was also enhanced by GW501516 in human HSC LX-2 cells. This observation prompted us to use these cells to identify the signaling pathways involved in HSC proliferation. For this purpose, we stably knocked down (KD) PPARβ/δ in human LX-2 stellate cells with lentiviral constructs that contained short interfering RNAs (siRNAs) against PPARβ/δ mRNA. This resulted in a 90% reduction in PPARβ/δ mRNA expression (Figure 6A).

The proliferation of LX-2 cells was measured in a [³H-methyl]-thymidine incorporation assay. Treatment with GW501516 increased proliferation by 2.5 fold compared to control DMSO-treated cells. This effect was blunted in PPARβ/δ KD LX-2 cells (Figure 6B). To elucidate the cascade of events between activated PPARβ/δ and increased LX-2 cell proliferation, we specifically inhibited several signaling pathways that might be involved in activated HSC proliferation. The MAPK extracellular signal-regulated kinase 1/2 (Erk1/2) pathway inhibitor, PD98049 (MEK1 inhibitor), had no effect on GW501516-inducible LX-2 cell proliferation (Figure 6C). In contrast, pre-incubation with the PI3K pathway inhibitor, LY294002, followed by exposure to GW501516 for 48 h inhibited PPARβ/δ-dependent LX-2 cell proliferation at the dose of 100 nM (Figure 6D). This implicated a PI3K-dependent pathway in GW501516-induced HSC proliferation. Next, we applied inhibitors of two PI3K downstream targets, the stress-activated protein kinase/c-Jun NH2-terminal kinase (SAPK/JNK) and p38 MAPK. These inhibitors (JNK inhibitor II and SB202190, respectively) also abolished activated PPARβ/δ-dependent stimulation of LX-2 cell proliferation (Figure 6E and 6F). This result identified two main signaling pathways, SAPK/JNK and p38 MAPK, which are involved in PPARβ/δ-induced HSC proliferation.

Because the PI3K pathway appeared to be required for PPARβ/δ-induced LX-2 cell proliferation, we analyzed the phosphorylation level of the PI3K downstream target Akt, a well-validated marker for PI3K activity. We observed a PPARβ/δ-dependent increase in Akt phosphorylation on serine 473 in control LX-2 cells (Figure 7A). This effect was inhibited by the PI3K inhibitor LY294002 and was blunted in PPARβ/δ KD LX-2 cells. Moreover, the Akt protein expression level was not modified in control or PPARβ/δ KD LX-2 cells. This indicated that Akt phosphorylation was both PPARβ/δ- and PI3K-dependent.

It is known that PI3K phosphorylation and stimulation of PKC are among the earliest events in the activation of MLK3, a MAPK kinase kinase (MAPKKK). MLK3 stimulates the MAPKKs MKK3/6 and MKK4, which finally activate p38 and JNK MAPKs in the last steps of initiating HSC proliferation 19 20 21 22 . Among the different PKC isoforms tested in control LX-2 cells, GW501516 induced only the phosphorylation of PKCα/βII on Thr638/641. This phosphorylation was not observed in similarly treated PPARβ/δKD LX-2 cells (Figure 7B). In addition, GW501516-activated PPARβ/δ had no effect on PKCα/βII protein expression levels. Interestingly, GW501516 increased both MLK3 protein expression and phosphorylation specifically in control LX-2 cells (Figure 7C). This effect was blunted by inhibitors of PI3K (LY294002) and PKC (Gö6983). In line with these results, GW501516 treatment in control LX-2 cells induced a PPARβ/δ-dependent phosphorylation of p38 at Thr180/Tyr182 and JNK at Tyr183/Thr185 (Figure 7D). This effect was also dependent on PI3K and PKC activation, as shown by LY294002 and Gö6983 treatments, which abolished the GW501516-induced phosphorylation of p38 and JNK (Figure 7D).

Collectively, these results were consistent with our data on HSC proliferation (Figure 6), and suggested that GW501516 stimulated HSC proliferation by activating p38 and JNK MAPKs, via an upstream signaling pathway involving PI3K, PKCα/βII and MLK3 (Figure 7E).

To test whether the activity of PPARβ/δ may also be relevant to the development of human liver fibrosis, its levels were measured in healthy subjects and patients with alcoholic fibrosis/cirrhosis. In diseased livers, there was a clear trend towards higher PPARβ/δ expression, and more heterogeneous expression was observed among fibrotic livers compared to healthy livers (Figure 8A). This heterogeneity was observed for all the mRNAs tested and may reflect differences in the severity of fibrosis between diseased individuals. Importantly, two well-established PPARβ/δ target genes, phosphoinositide dependent kinase 1 (PDPK1) and transforming growth factor beta-1 (TGFβ1), showed increased expression in diseased livers, which may reflect higher PPARβ/δ transcriptional activity (Figure 8A). However, the expression of PLIN2 and PDK4 was not increased (Figure 8A). Furthermore, the expression of inflammatory (MCP-1; Figure 8B) and fibrosis (pro-COL1α1 Figure 8C) marker genes was significantly increased in the biopsy samples, in agreement with results obtained in mouse liver. The expression of pro-COL3α1 and α-SMA showed a similar trend, albeit without reaching statistical significance (Figure 8C). Together, these results obtained from human subjects suggest that a similar mechanism of fibrosis development also exists in man, but a direct mechanistic implication of PPARβ/δ in this species remains to be substantiated.

CCl₄ was obtained from VWR International and olive oil was from Sigma Aldrich. GW501516 was synthesized by Kissei Pharmaceutical Co. Ltd. (Matshumoto, Japan).

Wild type and PPARβ/δ-null 23 6–8 week-old male mice of a mixed genetic background Sv129/C57BL/6, were maintained at 23 °C on a 12-h light–dark cycle with free access to water and a standard diet. To induce liver fibrosis, 6 wild type and 6 PPARβ/δ-null mice received repeated intraperitoneal injections (1 μl/g body weight) of CCl₄:olive oil (1:1) twice per week for 6 weeks. Injection of olive oil alone served as a control. In addition to the CCl₄-treatment, 6 wild type and 6 PPARβ/δ-null male mice received 10 μg/kg/day of GW501516 in 0.5% carboxymethyl cellulose (CMC), or GW501516 and CMC alone by gavage once per day for 6 weeks. At the end of the experimental period, blood samples were collected by retro-orbital puncture for measurement of the liver damage specific enzyme alanine transaminase (ALT) and neutral lipid analysis and the mice were then killed by cervical dislocation. After weighing, livers were either rapidly frozen in liquid nitrogen for later analyses or immediately prepared for immunocytochemistry studies and pathological examinations. All treatments were repeated in 3 independent experiments (n=6/genotype). All experiments involving animals were approved by the Veterinary Office of the Canton Vaud (Switzerland) in accordance with the Federal Swiss Veterinary Office Guidelines and conformed to the European Commission Directive 86/609/EEC and the “Guide for the Care and Use of Laboratory Animals” (NIH publication 86–23 revised 1985).

Hepatic lipids were determined by gas chromatography 40 .

Liver biopsies were collected by transparietal puncture from 8 healthy individuals and 12 patients (10 males, 2 females; aged 48–69 years) with alcohol-caused liver fibrosis or cirrhosis, diagnosed on clinical, biological, and histological grounds 41 . Total RNA was isolated from the liver biopsies with TRIzol reagent (Invitrogen, Carlsbad, CA) and gene transcription was analyzed by quantitative reverse-transcription PCR. All clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki.

Total RNA was extracted from frozen mouse liver samples, from human liver biopsies, or from LX-2 cells with TRIzol reagent (Invitrogen), according to the manufacturer’s instructions. Single-stranded cDNA templates were generated by reverse transcription with Superscript II reverse transcriptase (Invitrogen). For qRT-PCR (TAQ MAN), the cDNA equivalent of 10 ng of total RNA was amplified. All primers, including those specific for amplifying mRNA of mouse or human PPARβ/δ, PPARα , PPARγ TNF-α, MCP-1, MIP-1α, TGF-β1, PDGFBB, pro-Col1α1, pro-Col3α1, α-SMA, Desmin, PDPK1, and PDK4, were purchased from Applied Biosystems. Fluorescence was quantified with an ABI Prism 7900HT SDS system (Applied Biosystems). The following housekeeping genes were used to normalize mouse liver samples: Eef1A1, mRps9, mGapdh; human biopsies: RPS18, hTBP; and LX-2 cells: hGUSB. Relative mRNA expression levels were calculated with the comparative Ct method (User Bulletin # 2, Applied Biosystems) and qBase software. All values represent means from treated samples compared to means from control samples (vehicle- or olive oil-treated wild type mice or DMSO-treated LX-2 cells), which were set to 1.

Liver specimens were fixed in 4% paraformaldehyde (PAF) for 24 h, and then embedded in paraffin. Tissues sections (4 μm) were stained with hematoxylin/eosin (H/E), for routine examination, or with Sirus Red for Col1α1 and Col3α1 visualization.

To determine the degree of necroinflammatory liver injury, mouse liver sections were submitted to blind histopathologic examinations (grading according to Ischak’s score) by an independent pathologist (see Acknowledgements).

When indicated, control and LX-2 cells treated with siRNA against PPARβ/δ were first pre-incubated for 30 min in serum-free media with 20μM LY294002 (PI3K inhibitor) or 7 nM Gö6983 (PKC inhibitor, Calbiochem) before the addition of GW501516 for 24 h. Total cell proteins were extracted in ice-cold lysis buffer (10 mM Tris–HCl pH 7.5, 150 mM NaCl, 5 mM NaF, 1% Triton X-100, 0.1% SDS) supplemented with the following protease inhibitors: 2 μg/ml Aprotinin, 1 μg/ml Leupeptin, 2 μg/ml Pepstatin A, 1 mM phenylmethylsulfonyl fluoride, 1% deoxycholic acid, and 1 mM Na₃VO_4. After quantification, 30 μg of proteins were separated by SDS-PAGE and subjected to immunoblotting. All primary antibodies were diluted at 1/1000 and incubated overnight in 1× TBS 0.1% Tween-20, 5% nonfat milk. Anti-phospho Akt (Ser473), anti-phospho PKCα/βII (Thr638/641), anti-phospho MLK3 (Thr277/Ser281), anti-phospho p38 MAPK (Thr180/Tyr182), anti-phospho JNK/SAPK (Tyr183/Tyr185), anti-Akt, anti-PKCα, anti-MLK3, anti-p38 MAPK, and anti-SAPK/JNK were from Cell Signaling; anti-β-tubulin (loading control) was from BD Pharmingen^TM. The signals were detected with an ECL detection kit (Amersham Pharmacia Biotech), according to the manufacturer's instructions. The ScanImage densitometry program was used for quantification, and signals were normalized to the β-tubulin signal.

Control LX-2 and PPARβ/δ KD LX-2 cells were plated in 24-well culture plates at a seeding density of 3×10⁴ and incubated in DMEM with 2% FCS. One day later, they received serum-free medium for 24 h. Thereafter, they were treated with 0.01% DMSO (control), or 100 nM GW501516 for 48 h in serum-free medium. Alternatively, cells were pre-incubated for 30 min with 10 nM or 100 nM of SB202190 (p38 MAPK inhibitor), JNK inhibitor II, LY294002 (PI3K inhibitor), or PD98059 (MEK1 inhibitor) before the addition of 100 nM GW501516 for 48 h in serum-free medium. During the last 13 h of these treatments, 1 μCi/well of [³H-methyl]-thymidine was added. Then, the culture media were discarded, plates were placed on ice, cells were washed with ice-cold PBS, and then fixed in 500 μl ice-cold 10% trichloroacetic acid (TCA) for 20 min. TCA was removed, and 100 μl cell dissociation solution (0.2M NaOH, 1% SDS) was added to each well. Cells were incubated with gentle shaking for 10 min at room temperature. The samples were neutralized with 100 μl of 0.2 M HCl and transferred to vials with 3 ml scintillation cocktail. The scintillation counts (cpm) for treated control LX-2 cells and PPARβ/δ KD LX-2 cells were expressed as a percentage of the counts measured for the DMSO treated LX-2 cells, which were set to 100%.

Simultaneous quantifications of MCP-1 and MIP-1a levels in the livers of treated wild type and PPARβ/δ-null mice were performed with the mouse cytokine/chemokine LINCOplex KIT, 96-well plate assay (LINKO Research, MCYTO-70K) on a Luminex^R100. Similar sized pieces of frozen liver samples were homogenized with a power homogenizer (Polytron) in 1 ml of ice-cold PBS with the complete protease inhibitor cocktail. Lysates were incubated on ice for 10 min, followed by centrifugation at 13,000 rpm for 20 min at 4 °C. Supernatants were collected, and centrifugation was repeated several times, until the sample was clear of debris. The isolated proteins were quantified with the Bradford assay (BioRad). An immunoassay was run with 25 μl protein lysate to determine the cytokine/chemokine levels (pg/ml) in the liver according to the manufacturer’s instructions. All data were normalized to the protein concentration, and values were expressed relative to the value measured in vehicle-treated (olive oil) wild type mice, which was set to 1 (n=6 for each treated group).

To detect proliferating and activated HSCs, paraffin-embedded liver sections (4 μm) were double immunostained with the anti-αSMA antibody, as a marker for activated hepatic stellate cells, and anti-Ki67, as a marker for cell proliferation. Briefly, after deparaffinization, the antigen unmasking step was performed in 0.01 M citrate buffer, pH 6.0, by heating sections to 100 °C in a microwave oven for 20 min. After washing, sections were blocked in 1% BSA/1×PBS for 30 min at room temperature, and then incubated overnight at 4 °C with mouse anti-α-SMA (1:50) and rabbit anti-Ki67 (1:100) or with anti-F4/80 (1:10) antibody in blocking buffer. After washing, slides were incubated with the appropriate secondary antibodies for 40 min in blocking buffer. Secondary antibodies were anti-mouse IgG Cy3 (1:100) and anti-rabbit IgG FITC (1:400) for anti-α SMA/anti-Ki67, or anti-rat A568 (1:200) for anti-F4/80. After washing 3 times, slides were incubated for 5 min in DAPI for nuclear staining. Then, the sections were rinsed in water and mounted with DABCO for confocal-microscopy.

The following sequence was chosen to target the mouse PPARβ/δ sequence: 5'-GCACATCTACAACGCCTAC-3'. This sequence was 100% identical to the human PPARβ/δ sequence. A BLAST search ensured that the sequences would not target other RNAs in a nonspecific manner. The short interfering RNA (siRNA) was cloned into a pLV-TH lentivirus vector under the control of the polymerase III-dependent H1 promoter 42 . In addition, an internal cassette allowed expression of the green fluorescent protein (GFP) marker gene under the control of the elongation factor (EF-1) α promoter 43 . In our study, the empty pLV-TH vector, which contained all the features, but not the siRNA, was designated the control, and the pLV-TH vector containing PPARβ/δ siRNA was designated psiPPARβ/δ. All recombinant lentiviruses were produced by transient transfection of 293T cells according to standard protocols. Briefly, subconfluent 293T cells were co-transfected with 20 μg of the control vector, pLV-TH, or the PPARβ/δ-targeted vector, psiPPARβ/δ, 15 μg of pCMV-∆R8.91, and 5 μg of pMD2G-VSVG by calcium phosphate precipitation. The medium was changed after 16 h, and the recombinant lentiviruses were harvested 24 h later. The lentiviral infection efficiency in LX2 cells was monitored by the percentage of GFP-expressing cells detected by FACS analysis. At an infection multiplicity of 60, 90% of the LX-2 cells expressed GFP 48 h after transduction. The infected cells were then harvested, and total RNA was extracted. A qRT-PCR analysis demonstrated that PPARβ/δ was knocked down (KD) by 90% in the PPARβ/δ KD LX-2 cells compared to control-infected LX-2 cells (Figure 7A).

Data are expressed as means ± SEM or SD for treated wild type and PPARβ/δ-null mice (n=6), or as the means ± SEM or SD of several independent experiments performed in triplicate for LX-2 cells. Statistical significance was determined with the Student's t-test.