Sorafenib prevents liver fibrosis in a non-alcoholic steatohepatitis (NASH) rodent model.

Liver fibrosis occurring as an outcome of non-alcoholic steatohepatitis (NASH) can precede the development of cirrhosis. We investigated the effects of sorafenib in preventing liver fibrosis in a rodent model of NASH. Adult Sprague-Dawley rats were fed a choline-deficient high-fat diet and exposed to diethylnitrosamine for 6 weeks. The NASH group (n=10) received vehicle and the sorafenib group (n=10) received 2.5 mg·kg(-1)·day(-1) by gavage. A control group (n=4) received only standard diet and vehicle. Following treatment, animals were sacrificed and liver tissue was collected for histologic examination, mRNA isolation, and analysis of mitochondrial function. Genes related to fibrosis (MMP9, TIMP1, TIMP2), oxidative stress (HSP60, HSP90, GST), and mitochondrial biogenesis (PGC1α) were evaluated by real-time quantitative polymerase chain reaction (RT-qPCR). Liver mitochondrial oxidation activity was measured by a polarographic method, and cytokines by enzyme-linked immunosorbent assay (ELISA). Sorafenib treatment restored mitochondrial function and reduced collagen deposition by nearly 63% compared to the NASH group. Sorafenib upregulated PGC1α and MMP9 and reduced TIMP1 and TIMP2 mRNA and IL-6 and IL-10 protein expression. There were no differences in HSP60, HSP90 and GST expression. Sorafenib modulated PGC1α expression, improved mitochondrial respiration and prevented collagen deposition. It may, therefore, be useful in the treatment of liver fibrosis in NASH.

Introduction

Liver cirrhosis represents a pathological response to chronic liver injury, regardless of etiology. Nonalcoholic fatty liver disease (NAFLD) encompasses a spectrum of liver diseases in which liver fibrosis can occur, usually preceding the development of cirrhosis and hepatocellular carcinoma (1).

The possibility of reversing hepatic fibrosis has prompted the investigation of promising antifibrotic therapies (2). Antifibrotic compounds are categorized by their mechanism of action, such as degradation of extracellular matrix, antioxidants, reduction of inflammation, and hepatic stellate cell (HSC) activation inhibitors, among others. Indeed, receptors including integrins, cytokines, growth factors such as transforming growth factor-β (TGF-β) and platelet-derived growth factor (PDGF), post-receptor signal regulators (Smads), and transcription factors such as peroxisome-proliferator-activated receptor gamma (PPAR-γ) have been suggested as potential targets for antifibrogenic therapeutic strategy (3,4).

Sorafenib is an antineoplastic drug that acts as a potent multikinase inhibitor of the vascular endothelial growth factor receptor (VEGFR), a PDGF receptor, and Raf, and has antitumor activity approved for the treatment of advanced renal cell carcinoma and hepatocellular carcinoma (HCC) (5,6). Sorafenib also has shown important antifibrotic effects on HSCs and liver endothelial cells. Experimental studies have found that sorafenib treatment reduced the number of activated HSCs (7), improved intrahepatic fibrosis, inflammation, and angiogenesis (8), and induced suppression of collagen accumulation and HSC growth (9). Recently, Yang et al. (10) demonstrated that besides antifibrotic, antiangiogenic and portal hypertensive effects, chronic antagonism of anti-VEGFR improves hepatic blood flow, hepatic venous dysregulation, inhibits leukocyte recruitment/activation, splanchnic blood pooling and ascites formation in NASH-cirrhotic rats. However, the processes underlying these effects have not yet been fully characterized.

In this study, we evaluated the effects of sorafenib on liver fibrosis in an experimental model that used diethylnitrosamine (DEN) combined with a choline-deficient high-fat diet to induce NASH and liver fibrosis in Sprague-Dawley rats (11).

Material and Methods

Animals

Adult Sprague-Dawley rats weighing 250-300 g were housed in a temperature-, humidity-, and ventilation-controlled vivarium, with a 12-h light/dark cycle. All procedures for animal experimentation followed the ethical guidelines of the Helsinki Declaration of 1975 (NIH Publication No. 85-23, revised 1996) and the Guidelines of Animal Experimentation from the Faculdade de Medicina, Universidade de São Paulo.

Experimental procedures

NASH with fibrosis was induced in rats by a combination of a choline-deficient, high-fat diet (35% total fat, 54% trans fatty acid enriched; Rhoster Ltda., Brazil) and 13-15 mg/day diethylnitrosamine (DEN; Sigma Chemical, USA) administered in their drinking water (135 mg/L) as previously described (11) for 6 weeks. The animals were randomized to 3 treatment groups: NASH (n=10), sorafenib (n=10), and control (n=4). Animals in both the NASH and sorafenib groups received the study diet and DEN; sorafenib group animals received 2.5 mg·kg-1·day-1 sorafenib and NASH animals received vehicle (Ringer's solution) daily by gavage. Control animals were fed ad libitum with a standard diet (Nuvilab® Nutrientes Ltda., Brazil), and received Ringer's solution by gavage. Dietary intake was assessed by weighing the amount eaten every 3 days. After the 6-week period, the animals were anesthetized with 0.1 mL/kg intraperitoneal ketamine and sacrificed. Liver specimens were obtained for histopathological, gene expression, cytokine, and liver mitochondrial function analyses.

Histological analysis

After initial gross evaluation, liver tissue samples were fixed in 4% formaldehyde and processed for hematoxylin-eosin and picrosirius staining for histological analysis. Collagen content was determined by light microscopy (Nikon E-800, Japan) in picrosirius-stained histological sections. Since the distribution of injuries in the samples was heterogeneous and different damage patterns were observed, quantification was performed in areas with greater collagen deposition. The analysis was standardized for all animals and conducted in a blinded manner. Ten histological fields per sample were studied (total area of 3,031,614 μm2). The collagen fiber area was quantified with Image Pro-Plus 4.5 (Media Cybernetics Inc., USA), as described elsewhere (12). The results are reported as the percentage of the area occupied by collagen in relation to the total histological field.

Gene expression analysis

Tissue RNA extraction was performed by the Trizol® method and quantification of total extracted RNA was determined by spectrophotometry (Nanodrop ND-1000; Nanodrop Technologies, USA). The RNA preparation was considered free of proteins when the absorbance260/280 ratio was between 1.8 and 2.0. Real-time quantitative polymerase chain reaction (RT-qPCR) protocols were performed with a Rotor-Gene RG-3000 thermal cycler (Corbett Research, Australia) and SuperScript™ III Platinum® One-Step Quantitative RT-PCR System reagents (Invitrogen Life Technologies, USA) as recommended by the manufacturers. All RT-qPCR reactions were performed in duplicate for each sample of liver tissue for both the target gene and the β-actin control. The primers for the TIMP1, TIMP2, MMP9, HSP60, HSP90, GST, PGC1α, and β-actin genes were designed in the Primer3 Input program (http://frodo.wi.mit.edu/). The β-actin gene was used as the endogenous control. Relative quantification was calculated using the mathematical model described by Pfaffl (13). The primer sequences for each gene were as follows: HSP-60 forward, AGC AAA GGG GCT AAT CCA GT and reverse, TGA CAC CCT TTC TTC CAA CC; HSP-90 forward, GAT TGA CAT CAT CCC CAA CC and reverse, CTG CTC ATC ATC GTT GTG CT; GST forward, GGC GGA TCT GGA TGA AAT AGT TCT and reverse, CAA CGA GAT AAT CTT GTC CAT GGC; TIMP1, forward, TCC CCA GAA ATC ATC GAG AC and reverse, TCA GAT TAT GCC AGG GAA CC; TIMP2 forward, GCA TCA CCC AGA AGA AGA GC and reverse, GGG TCC TCG ATG TCA AGA AA; MMP-9 forward, CAA ACC CTG CGT ATT TCC AT and reverse, AGA GTA CTG CTT GCC CAG GA; PGC1α forward, CTA CAG ACA CCG CAC ACA TCG C and reverse, TCT CTC TGC TTG GCC CTT TCA G; β-actin forward, TGT CAC CAA CTG GGA CGA TA and reverse, GGG GTG TTG AAG GTC TCA AA; GAPDH forward, ATG ATT CTA CCC ACG GCA AG and reverse, CTG GAA GAT GGT GAT GGG TT.

Liver mitochondrial oxidation and phosphorylation activities

Liver mitochondria were prepared as previously described (14). The mitochondrial oxygen consumption was measured polarographically using a Gilson 5/6H Oxygraph (Gilson Medical Electronics, Inc., USA) at 28°C in a closed reaction vessel fitted with a Clark oxygen electrode (Yellow Springs Instruments Co., USA). The incubation medium consisted of 120 mM KCl, 2 mM sodium phosphate, 10 μM rotenone, and 1 mM EGTA (ethylene glycol-bis (2-aminoethylether)-N,N,N′,N′-tetraacetic acid) and was buffered at pH 7.3 with 5 mM Tris-HCl. Mitochondria were energized with potassium succinate as substrate at a final concentration of 10 mM. After a brief equilibration period, state 3 (activated state, S3) respiration was induced by the addition of 280 nmol adenosine diphosphate (ADP). The added ADP was phosphorylated to adenosine triphosphate (ATP) and the state 4 (basal state, S4) respiration was then measured. The ratio of oxygen consumption in the presence of ADP to that in the absence (respiratory control rate, RCR) and the ADP/O ratio was calculated as an index of mitochondrial oxidation and phosphorylation. RCR was calculated as oxygen consumption in S3/oxygen consumption in S4; ADP/O as moles of ATP formed from ADP per atom of oxygen consumed. S3 and S4 are reported as nanomoles of oxygen per milligram mitochondrial protein per minute (nmol.mg-1.min-1). Mitochondrial protein content was determined by the Lowry method (15).

Cytokine analysis

For the determination of cytokine levels, the liver tissue samples were ground in ice-cold radio-immunoprecipitation assay buffer. TNF-α, IL-6, and IL-10 were measured in each of the liver samples by enzyme-linked immunosorbent assay (ELISA) (R&D System Inc., USA), in accordance with the manufacturer's instructions (16). All measurements were made in duplicate, and the average values were used in the statistical analyses.

Statistical analysis

Data are reported as means with 95% confidence intervals (CIs). All analyses were performed with the statistical computing package R (version 2.15.2; The R Project for Statistical Computing, http://www.r-project.org). The Shapiro-Wilk normality test and Levene's variance homogeneity test were applied to verify normality and homogeneity of variance. Analysis of variance (ANOVA) followed by Tukey's post hoc test was used to identify statistical differences in homogeneous, normally distributed data. The Kruskal-Wallis non-parametric test was used to analyze data not normally distributed or homogeneous. Significance was set at P<0.05.

Results

Sorafenib prevented NASH-related fibrosis

Histochemical analyses were performed to evaluate the degree of fibrosis in our NASH experimental model, as previously described. We observed a marked increase in mean collagen content in animals fed a choline-deficient high-fat diet plus DEN (7.35; 95%CI: 4.84-9.85) compared to control rats (0.52; 95%CI: 0.45-0.57) (P<0.001), indicating the effectiveness of this model in inducing early stages of fibrosis. Interestingly, sorafenib treatment reduced collagen content (3.41; 95%CI: 2.53-4.28) compared with the NASH group (P<0.001) (Figure 1). To understand the mechanisms underlying the latter effect, we investigated the processes associated with collagen production and degradation. No difference was observed in mean MMP9 mRNA transcription level in liver tissue from NASH (0.99; 95%CI: 0.90-1.07) compared to control rats (1.00; 95%CI: 0.95-1.07; P=0.834). However, mean MMP9 mRNA content was significantly increased in animals treated with sorafenib (1.61; 95%CI: 1.35-1.86) compared to both the control and NASH groups (P<0.001, Figure 2C). We observed that mean mRNA expression of TIMP1, an MMP9 inhibitor, was significantly lower in NASH animals (0.87; 95%CI: 0.80-0.94) than in controls (1.00; 95%CI: 0.96-1.04, P=0.036) and further impaired in sorafenib-treated rats (0.74; 95%CI: 0.68-0.81; P<0.001 vs controls and P=0.015 vs NASH; Figure 2A). The expression of TIMP2 was significantly lower in the sorafenib group (0.63; 95%CI: 0.55-0.71) than in controls (1.00; 95%CI: 0.97-1.03) and the NASH group (1.03; 95%CI: 0.94-1.12; both P<0.001; Figure 2B). Together these data suggest that collagen degradation was enhanced by treatment with sorafenib, resulting in less fibrosis.

Figure 1

Analysis of liver extracellular matrix by morphometry of collagen fibers stained in red with Sirius Red. A, Control animal (n=4); B, nonalcoholic steatohepatitis (NASH) animal that received only vehicle (n=10); C, NASH animal treated with 2.5 mg·kg-1·day-1 of sorafenib for 6 weeks (n=10); D, graph demonstrating that NASH animals treated with sorafenib presented a decrease in the area occupied by collagen fibers in liver parenchyma compared with NASH animals that received only vehicle (P<0.001, Tukey test).

Figure 2

A, B, C, Graphic representation of TIMP1, TIMP2, and MMP9 mRNA expression in control (n=4), nonalcoholic steatohepatitis (NASH; n=10), and sorafenib-treated animals (n=10). A, TIMP1 mRNA expression was significantly lower in NASH animals compared to control (P= 0.036, Tukey test) and further impaired in sorafenib-treated rats (P<0.001 vs control and P=0.015 vs NASH, Tukey test). B, TIMP2 was significantly lower in the sorafenib group compared to control and NASH groups (P<0.001, Tukey test). C, MMP9 mRNA content was increased in animals treated with sorafenib (P<0.001 vs control and NASH, non-parametric Tukey test). There was no statistical difference in MMP9 mRNA transcription levels in the liver from NASH animals compared to control rats (P= 0.834, non-parametric Tukey test). D, E, F, Graphic representation of IL-6, IL-10, and TNF-α in control, NASH, and sorafenib-treated animals. D, Protein levels of IL-6 were similar in the NASH and control groups (P=0.55, Tukey test). IL-6 levels were significantly lower in sorafenib animals compared to NASH animals (P=0.002, Tukey test). E, IL-10 protein levels were also lower in the sorafenib group compared to the NASH (P<0.001, non-parametric Tukey test) and control groups (P=0.039, non-parametric Tukey test). F, No differences in TNF-α protein levels were detected among the three groups (non-parametric Tukey test).

Anti-inflammatory effect of sorafenib on NASH-related fibrosis

Treatment with sorafenib was associated with a significant decrease in IL-6 and IL-10 protein expression, suggesting that it may indeed have a mild anti-inflammatory effect. Levels of IL-6 protein in the NASH group (20.30; 95%CI: 14.56-26.04) and control group (16.66; 95%CI: 14.89-18.44) did not differ (P=0.55). IL-6 levels were significantly lower in the sorafenib group (7.39; 95%CI: 4.68-9.75) than in the NASH group (P=0.002). IL-10 protein levels were also lower in the sorafenib group (3.86; 95%CI: 2.71-5.00) than in both the NASH (10.78; 95%CI: 7.49-13.59; P<0.001) and control (6.84; 95%CI: 5.46; 8.14; P=0.039) groups. No differences in TNF-α protein levels were detected among the 3 groups (Figure 2D,E,F).

Redox signaling in NASH was not the main target for sorafenib

No differences were observed in hepatic heat shock protein expression. Mean HSP60 expression was 1.00 (95%CI: 0.95-1.07) for controls, 0.98 (95%CI: 0.95-1.02) for the NASH group, and 0.95 (95%CI: 0.92-0.97) for the sorafenib group. The corresponding values for HSP90 were 1.00 (95%CI: 0.97-1.03) for controls, 1.03 (95%CI: 1.00-1.05) for the NASH group, and 1.06 (95%CI: 1.02-1.09) for the sorafinib group. No differences were observed in the expression of GST, an enzyme responsible for maintaining glutathione homeostasis and the main intracellular redox buffer. The values were 1.00 (95%CI: 0.99-1.02) for controls, 0.99 (95%CI: 0.97-1.02) for the NASH group, and 0.92 (95%CI: 0.86-0.98) for the sorafenib group (Figure 3A,B,C).

Figure 3

Graphic representation of HSP60, HSP90, and GST mRNA expression in control (n=4), nonalcoholic steatohepatitis (NASH; n=10), and sorafenib-treated animals (n=10). The mRNA expression of HSP60, HSP90, and GST did not differ among the three groups.

Sorafenib prevented mitochondrial dysfunction in NASH

As shown in Figure 4A, all stages of mitochondrial respiration in liver homogenates from NASH animals were markedly impaired compared to controls. Sorafenib treatment was associated with a significant improvement in mitochondrial function at all stages, restoring all parameters to almost the levels observed in control animals. Figure 4B shows that mean PGC1α mRNA levels were markedly lower in NASH subjects (0.40; 95%CI: 0.30-0.50) than in controls (P<0.001), but the control (1.00; 95%CI: 0.70-1.27) and the sorafenib (1.24; 95%CI: 0.71-1.69) groups were not significantly different (P=0.99). Mean PGC1α mRNA expression in the NASH group was significantly lower than in the sorafenib group; P<0.001). These results point strongly to the potent effect of sorafenib in correcting mitochondrial dysfunction, mainly by enhancing PGC1α expression.

Figure 4

A, Stages of mitochondrial respiration in liver homogenates of control (CTRL; n=4), nonalcoholic steatohepatitis (NASH; n=10), and sorafenib-treated animals (n=10). All stages of mitochondrial respiration in liver homogenates from NASH animals were markedly impaired compared to controls. Sorafenib treatment improved mitochondrial function, restoring the parameters to almost the levels observed in control animals. RCR: respiratory control rate; S3: activated state; S4: basal state respiration. *P<0.05 compared to CTRL group; #P<0.05 compared to NASH group (Tukey test). B, Graphic representation of PGC1α mRNA expression in control, NASH, and sorafenib-treated animals (P=0.99, control vs sorafenib group; P<0.001, control vs NASH; P<0.001, NASH vs sorafenib group; non-parametric Tukey test).

Discussion

This study showed that the antineoplastic drug sorafenib prevented the early stages of fibrosis in an experimental model of NASH. Sorafenib was shown not only to have mild anti-inflammatory properties, but also to prevent the mitochondrial dysfunction that characterizes this condition.

Treatment with sorafenib reduced collagen deposition by nearly 63% in this model. These data corroborate previous results by Hong et al. (17) that also demonstrated a greater than 60% reduction in collagen deposition in animals treated with sorafenib. Further investigation revealed that increased expression of MMP9 together with lower levels of TIMP1 and TIMP2 is crucial for matrix remodeling, despite possessing no activity against type I collagen (18). It may, however, exert an indirect effect on fibrosis resolution, since increased MMP9 activity markedly accelerates the onset of fibrinolysis (18-20). It has also been suggested that MMP9 accelerates apoptosis of HSCs, thereby contributing to fibrosis regression (21). In this context, our findings are the first to show an effect of sorafenib on MMP9 gene expression, thus uncovering a new mechanism whereby this drug may affect liver fibrosis.

MMP9 gene expression and protein activity are tightly regulated, being induced by reactive oxygen species and inflammatory mediators. Therefore, we sought further to determine whether the observed increase in MMP9 expression resulted from changes in those signaling pathways.

Since inflammatory cytokines play a key role in fibrogenesis, and persistent inflammation always precedes liver fibrosis and its progression to more severe forms (22,23), our data suggest that lower IL-6 and IL-10 levels were present in sorafenib-treated than in NASH rats, suggesting that the beneficial effect of sorafenib may be at least partially due to its anti-inflammatory activity.

Mitochondria play a central role in fat oxidation and cellular energy metabolism, being responsible for oxidative phosphorylation and fatty acid β-oxidation (24). Indeed, various changes in mitochondrial metabolism have been implicated in NASH pathogenesis and carcinogenesis, secondary to their role in pathways associated with cellular energy metabolism, free radical generation, and apoptosis (25-32). In NASH, mitochondrial dysfunction has been associated with the development, perpetuation, and worsening of the disease (26). The loss of mitochondrial function leads to the accumulation of lipid deposits in the cytosol and participates at various levels in NASH pathogenesis. It not only impairs fat metabolism but also increases oxidative stress and cytokine production, triggering cell death, inflammation, and fibrosis (33,34). We report here that mitochondrial respiration is severely impaired in NASH. All steps of the respiratory chain are affected, suggesting a global mitochondrial dysfunction. We demonstrated here, as in a previous study from our group (35), a significant reduction in RCR in the NASH group, suggesting mitochondrial dysfunction. Moreover, we observed a recovery of RCR, similar to the control group, in the animals treated with sorafenib. However, we cannot exclude interference by sorafenib in mitochondrial biogenesis, which differs in tissues depending on the numbers of mitochondria, and the participation of other cells, such as fibroblasts, to explain this finding.

The transcriptional coactivator PPAR-γ coactivator-1 (PGC-1) is a main regulator of mitochondrial function and biogenesis. PGC-1 interacts with peroxisome proliferator-activated receptor alpha (PPAR-α) to increase mitochondrial β-oxidation of fatty acids, increasing the expression of nuclear respiratory factor-1 and mitochondrial transcription factor A, which in turn increase the number of mitochondria and their oxidative phosphorylation capacity (36). Based on these findings, we examined whether PGC1α was involved in this process. PGC1α is the main regulator of mitochondrial biogenesis and oxidative phosphorylation. It is highly expressed in the liver, and its expression may be modulated by physiological and pathological stimuli. We observed that PGC1α expression decreased during NASH development, reaching a nadir when fibrosis was fully present. Interestingly, both PGC1α expression and mitochondrial respiration were restored by treatment with sorafenib. However, although these results suggest that sorafenib modulates PGC1α expression, improves mitochondrial respiration, and prevents collagen deposition, our data do not support these as the primary actions of sorafenib. We do not know how sorafenib affects PGC1α expression, which leads to improvement of mitochondrial respiration. Further studies are needed to clarify the exact role of sorafenib in the enhanced PGC1α expression and in the global improvement in mitochondrial respiration.

In summary, we have shown that liver fibrosis after NASH development is characterized by enhanced collagen deposition and MMP-9 activity. These findings were not clearly related to imbalances in redox signaling or inflammatory mediators. Metabolic abnormalities, characterized by mitochondrial dysfunction and impaired PGC1α expression, were seen in this experimental model and were corrected by sorafenib treatment. Sorafenib, by preventing mitochondrial dysfunction, may contribute to the treatment of not only liver diseases but also of other metabolic and degenerative diseases.