2.1 Animals

Seven-week-old male C57BL/6 wild-type (WT) mice were used in this study. The mice were purchased from Taconic Farms, Inc (Samtako Bio Korea) and maintained in standard conditions (24 ± 2°C, 50 ± 5% humidity). They were fed a sterile standard chow diet and provided water ad libitum. Experimental procedures and animal management procedures were undertaken in accordance with the requirements of the Animal Care and Ethics Committees of Chonbuk National University. The animal facility of the Chonbuk National University is fully accredited by the National Association of Laboratory Animal Care.

2.2 Murine fibrosis induction and amlexanox treatment

Two different models of liver fibrosis were used in this study. To induce hepatotoxin-mediated liver fibrosis, CCl₄ (Sigma-Aldrich; CCl₄, 2:5 v/v in corn oil) or same volume of corn oil (Sigma-Aldrich) were administered intraperitoneally (i.p.) at a dose of 2 mL/kg bodyweight (BW) three times per week for up to 4 weeks. To induce biliary fibrosis, mice were fed with 0.1% diethoxycarbonyl-1, 4-dihydrocollidine (DDC, Saeronbio Inc) for 4 weeks. Based on a previous study showing that 100 mg/kg amlexanox was no side effect to experimental mouse,22 different doses of amlexanox (25, 50, and 100 mg/kg) were used in our in vivo experiments. Each group of mice was orally administered with amlexanox (25, 50, and 100 mg/kg) or same volume of vehicle once a day throughout the experimental period.

After 4 weeks of DDC diet feeding, mice were fed with normal diet for the next 1, 2 and 3 weeks to establish the model of fibrosis resolution as previously described with a slight modification (the resolution [RES] group).26 During normal diet feeding period, mice were orally administered with amlexanox (50 mg/kg) or same volume of vehicle once a day. Prior to necropsy, animals fasted for 8 hours. Standard necropsy techniques were used, and tissues were collected and prepared for future analysis.

2.3 Histopathologic examination

Livers were fixed in 10% phosphate-buffered formalin, routinely processed and then embedded in paraffin. Tissue sections (4 μm in thickness) were prepared using a microtome (HM-340E, Thermo Fisher Scientific Inc) and placed on glass slides. Haematoxylin and eosin (H&E) staining was performed according to standard techniques.

To evaluate the severity of liver fibrosis, liver sections were stained with Sirius-red staining (saturated aqueous solution of picric acid containing Direct Red 80; Sigma-Aldrich). After liver sections were stained, the percentage of red-stained collagen fibres was quantified by measuring Sirius-red-positive area per total liver section. Total liver section images were analysed for each animal using a light microscope (BX-51, Olympus Corp.) and digital imaging software (analySIS TS, Olympus Corp.).

2.4 Biochemical measurements

To evaluate the degree of liver injury, serum levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were determined using AM101-K spectrophotometric assay kits (ASAN Pharmaceutical). Absorbance values of serum biochemical were measured at wavelength of 490 nm using an EMax spectrophotometer (Molecular Devices).

2.5 Analysis of relative gene expression using quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was isolated from cells or tissues using Easy-Spin Total RNA extraction kit (GeneAll). Following degradation of remaining DNA using DNase I containing RNase inhibitor (Toyobo), samples were transcribed using ReverTra Ace^® qPCR RT Master Mix (Toyobo) according to the manufacturer's protocol. cDNA was used for real-time PCR on a CFX96™ Real-Time PCR Detection System (Bio-Rad Laboratories) using SYBR Green I as a double-strand DNA-specific binding. After the reaction was completed, specificity was verified by melting curve analysis. Quantification was performed by comparing Ct values of each sample normalized to Ct value of glyceraldehyde-3-phosphate dehydrogenase. Sequences of PCR primers are listed in Table 1.

2.6 Western blot analysis

Liver tissues or cells were directly homogenized on ice with an extraction buffer (T-PER, Thermo Fisher Scientific Inc,) for 5 minutes. After centrifugation at 13 000 g for 15 minutes at 4°C, protein concentration in the supernatant was measured using Pierce BCA Protein Assay kit (Thermo Fisher Scientific Inc) according to the manufacturer's protocol. Equal amounts of protein were then subjected to sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE). After transferring to polyvinylidene difluoride (PVDF) membrane, blocking was carried out using 5% bovine serum albumin in Tris-buffered saline (20 mmol L⁻¹ Tris, 150 mmol L⁻¹ NaCl, pH 7.4) containing 0.05% Tween-20 at room temperature for 1 hour. The membrane was then incubated with primary antibodies diluted 1:1000 in blocking buffer at 4°C overnight. The following antibodies were used: rabbit antimouse α-smooth muscle actin (αSMA) antibody (Abcam), rabbit antimouse TGFβ antibody (Cell Signaling), rabbit antimouse NF-κB or phospho-NF-κB (pNF-κB, Cell Signaling), rabbit antimouse STAT3 or phospho-STAT3 (pSTAT3, Cell Signaling), rabbit antimouse pTBK1 (Cell Signaling), rabbit antimouse pIKKε (Cell Signaling), rabbit antimouse B-cell lymphoma 2 (Bcl2, Cell Signaling), rabbit antimouse Bax (Cell Signaling) and rabbit antimouse β-actin (Santa Cruz Biotechnology Inc). To detect antigen-antibody complexes, peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology Inc) were diluted 1:2000 in blocking buffer and incubated at room temperature for 1 hour. Protein bands were visualized with enhanced chemiluminescence detection system using ImageQuant™ LAS 500 (GE Healthcare Life Science). Expression levels of protein were quantified with ImageQuant™ TL software.

2.7 Isolation of liver cell fractions

Liver cells were fractionated into various cell populations as described by our previous studies.27, 28 Briefly, mouse livers were digested by type I collagenase (1 mL/min) perfusion. Liver cells isolated from WT mice were suspended and centrifuged at 50 g for 3 minutes. Following centrifugation, the pellet representing hepatocytes was resuspended, filtered and washed several times using DMEM (Lonza) supplemented with 10% foetal bovine serum (FBS, Thermo Fisher Scientific Inc), 100 IU/mL penicillin and 100 μg/mL streptomycin. Viability of hepatocytes was assessed using trypan blue (Sigma-Aldrich). It was over 85%. To further isolate hepatic cells, 3-layer discontinuous density gradient centrifugation was performed with 20%, 11.5% OptiPrep (Sigma-Aldrich) and buffer to obtain non-parenchymal cell fraction, and HSC fraction, respectively. Kupffer cell (KC) fractions were positively selected from the MNC fraction by magnetic cell sorting using anti-F4/80 antibody (Miltenyi Biotec). HSC layer between 11.5% OptiPrep and buffer was carefully collected. HSC fraction was purified by negative selection of contaminating KCs by magnetic cell sorting with appropriate antibody. Cell fractions were immediately homogenized for RNA extraction or washed twice with PBS, and resuspended in RPMI-1640 media for cell culture.

2.8 Lactate dehydrogenase (LDH) assay

Cell death was evaluated using a cytotoxicity detection kit (Sigma-Aldrich) based on the measurement of activity of LDH released from the cytosol into the culture medium following the manufacturer's instruction. Absorbance of sample was measured at wavelength of 490 nm using an EMax spectrophotometer (Molecular Devices).

2.9 Cell culture and treatment

Hepatocytes (5.0 × 10⁵ cells/well) or KCs (1.0 × 10⁶ cells/well) isolated from WT mice were plated into 12-well plates and then cultured at 37°C in a 5% CO₂ incubator with DMEM or RPMI-1640 media containing 10% FBS, 100 IU/mL penicillin and 100 μg/mL streptomycin. To mimic in vivo condition, primary hepatocytes and KCs were co-cultured in 12-well trans-well plate (Sigma-Aldrich) at a ratio 1:4 of KCs/hepatocytes. And co-cultured cells were treated with 0.3% CCl₄ with or without 50 μmol L⁻¹ amlexanox for 24 hours.

To investigate the effect of amlexanox to inflammation, isolated KCs (5.0 × 10⁵ cells/well) were seeded in 24 wells and treated with indicated concentration of amlexanox or vehicle for 24 hours. Lipopolysaccharide (LPS, 1 μg/mL) was used to induce inflammation in KCs.

Primary HSCs (1.0 × 10⁶ cells/well) were plated onto 12-wells plate and cultured for up to 7 days post-isolation for cell activation. Culture media were changed every other day. Human HSC line (LX-2, 1.0 × 10⁶ cells/well) was routinely cultured in 12-well plate. To evaluate the roles of IKKε/TBK1 on HSCs, quiescent (culture day 1) or activated primary HSCs (culture day 7) were treated with indicated concentration of amlexanox or vehicle for 24 hours. And LX-2 cells were treated with 10 ng/mL human recombinant TGFβ with or without amlexanox treatment for 24 hours. To activate or inhibit STAT3, cells were treated with colivelin (1 μmol L⁻¹, Cayman Chemical) or SPI (10 μmol L⁻¹, BioVision, Inc) with 50 μmol L⁻¹ amlexanox for 24 hours.

2.10 Statistical analysis

All data were expressed as the mean ± standard error. Differences between multiple groups were compared using one-way analysis of variance (ANOVA) using SAS version 9.1 (SAS Institute Inc). Duncan's multiple range test (DMRT) was used for individual comparisons. Experimental groups marked by different letters represent significant differences among groups. Differences between two groups were compared using a two-tailed Student's t test. A P < .05 was considered statistically significant.

3.1 Increased activities of TBK1 and IKKε were found in fibrotic livers in mice

Consistent with previous finding showing increased activity of TBK1 and IKKε in the livers of mice with obesity,22 we also found significantly increased phosphorylation levels of TBK1 and IKKε in the livers of mice fed with DDC diet compared with those of mice fed with chow diet (Figure 1A). We observed similar patterns of phosphorylation levels of these proteins in hepatotoxin-induced fibrotic livers (Figure 1B). To more clearly demonstrate these effects, phosphorylation levels of TBK1 and IKKε were evaluated in HSCs isolated from chow diet- or DDC diet-fed mice for 4 weeks. The results showed that increased protein levels of pTBK1 and pIKKε were observed in HSCs isolated from fibrotic livers (Figure 1C). Consistently, mRNA expression levels of TBK1 and IKKε were significantly increased in HSCs isolated from fibrotic livers compared with HSCs isolated from non-fibrotic livers (Figure 1D). Although there was no significant difference of expression levels of these genes in hepatocytes, IKKε but not TBK1 was markedly up-regulated in KCs isolated from fibrotic livers (Figure 1D). These findings suggest that pTBK1 and pIKKε might affect the pathogenesis of liver fibrosis. These findings suggest that pTBK1 and pIKKε might affect the pathogenesis of liver fibrosis.

3.2 Treatment of amlexanox alleviates the severity of biliary fibrosis in mice

To further determine the roles of TBK1 and IKKε in the pathogenesis of biliary fibrosis, chow diet- or DDC diet-fed mice were orally administered with vehicle or indicated dose of amlexanox daily. The changes in bodyweight during the experimental period did not differ among the chow diet-fed groups (Figure S1). However, reduced body weight in mice with DDC diet was recovery by treatment of amlexanox in a dose-dependent manner. Also, DDC diet-fed mice had decreased protein levels of pTBK1 and pIKKε in livers (Figure 2A). Based on histopathologic analysis using H&E and Sirius-red staining, inhibition of TBK1 and IKKε by amlexanox markedly reduced biliary hyperplasia and collagen deposition in the livers of mice fed with DDC diet (Figure 2B,C). Consistent with these findings, we observed significantly lower serum levels of ALT and AST in amlexanox-treated mice with fibrosis than those in vehicle-treated mice with fibrosis (Figure 2D). Additionally, amlexanox-administered mice showed dose-dependently reduced IL-1β and TNFα levels together with reduced NF-κB activation in fibrotic livers. (Figure 2E,F). In line with histopathologic observation (Figure 2B,C), we found lower expression levels of pro-fibrogenic genes such as TGFβ, alpha-1 type I collagen (Col1α1) and tissue inhibitors of metalloproteinase-1 (TIMP1) in fibrotic livers of mice treated with amlexanox than those of vehicle-treated mice (Figure 2G). These results were further supported by Western blot analysis, showing that reduced protein levels of α-smooth muscle actin (αSMA) were observed in fibrotic livers of mice treated with amlexanox (Figure 2H). These findings suggest that suppression of TBK1 and IKKε by using amlexanox can attenuate cholestasis-induced chronic liver injury and its associated fibro-inflammatory responses in mice through modulating NF-κB activation.

3.3 Treatment of amlexanox attenuates CCl₄-induced liver fibrosis

To further explore the role of TBK1 and IKKε in fibrotic responses, we next performed additional experiments using hepatotoxin-induced liver fibrosis model. As shown in Figure 3A, amlexanox dose-dependently decreased the protein levels of pTBK1 and pIKKε in fibrotic livers induced by CCl₄. Similar to the results from Figure 2, inhibition of TBK1 and IKKε significantly reduced hepatic fibrogenesis and bridging fibrosis in CCl₄-treated mice as confirmed by histopathologic observation (Figure 3B,C). Also, amlexanox-treated mice showed significantly lower serum levels of ALT and AST in fibrotic milieu (Figure 3D). Consistently, markedly reduced inflammation and its related NF-κB activation were observed by amlexanox administration in the livers of mice treated with CCl₄. (Figure 3E,F). These findings together with results from Figure 2 strongly suggest that amlexanox can exert beneficial effects in mice with liver fibrosis induced by CCl₄.

3.4 Inhibition of TBK1 and IKKε promotes fibrosis resolution

As treatment of amlexanox attenuated the progression of liver fibrosis based on above results, we then investigated the roles of TBK1 and IKKε in fibrosis regression by feeding a DDC diet for 4 weeks, followed by feeding a chow diet plus amlexanox (50 mg/kg) for 1, 2 and 3 weeks (RES group; total 5, 6 and 7 weeks, respectively). Compared with 4 weeks DDC diet feeding, RES group showed markedly reduced liver fibrosis, as revealed by Sirius-red staining. And such effects were further accelerated by treatment of amlexanox (Figure S2A). Also, we observed significantly lower expression levels of pro-fibrogenic genes in amlexanox-treated mice in RES group (Figure S2B). Consistently, amlexanox-treated mice showed reduced serum biochemical and expression levels pro-inflammatory genes during fibrosis resolution (Figure S2C,D). For more detailed analysis, we chose RES group after being shifted to a normal diet for 2 weeks. As shown in Figure 4A, significantly lower protein levels of pTBK1 and pIKKε were found in the RES treated with amlexanox. Amlexanox-treated group showed lower liver damage and fibrogenesis in mice after 2 weeks of fibrosis resolution, as confirmed by histopathologic analysis and Sirius-red staining (Figure 4B,C). Additionally, inhibition of TBK1 and IKKε significantly reduced serum levels of ALT and AST in mice during fibrosis resolution (Figure 4D). Significant lower inflammation was also observed in fibrotic livers of mice treated with amlexanox compared to those of mice treated with vehicle, as confirmed by reduced protein levels of pro-inflammatory genes (Figure 4E). Consistently, administration of amlexanox resulted in significant lower protein levels of pNF-κB by amlexanox in the livers of mice after 2 weeks of fibrosis resolution (Figure 4F). In line with histopathologic examination, RES group showed significantly lower liver fibrosis in mice treated with amlexanox, as revealed by decreased expression levels of pro-fibrogenic genes (Figure 4G). Furthermore, we found significantly lower protein levels of αSMA in fibrotic livers of mice treated with amlexanox during fibrosis resolution (Figure 4H). These findings suggest that amlexanox can promote the resolution of liver fibrosis in mice.

3.5 Treatment of amlexanox significantly reduces inflammatory responses in KCs

As TBK1 and IKKε were highly expressed in KCs and HSCs (Figure 1D), we focused on KCs and HSCs. In this part, we first investigated the roles of TBK1 and IKKε on KCs in inflammatory responses. As the KCs promote the progression of liver fibrosis by producing inflammatory cytokines,29 the inflammatory responses were evaluated by performing in vitro study. As shown in Figure S3A,B, treatment of amlexanox significantly decreased protein levels of pTBK1 and pIKKε in KCs and its related inflammatory gene expression. To further demonstrate these effects, LPS were co-treated with amlexanox to evaluate the role of TBK1 and IKKε on inflammation in KCs. As shown in Figure 5A, increased pTBK1 and pIKKε by LPS were markedly decreased by amlexanox in KCs. Consistently, dose-dependently decreased protein levels of pNF-κB were observed in LPS-treated KCs by treatment of amlexanox (Figure 5B). Also, expression levels of TNFα and IL-1β were decreased in LPS-treated KCs by treatment of amlexanox (Figure 5C,D).

KC-derived TGFβ is one of the most important cytokines involved in the pathogenesis of liver fibrosis.29, 30 Similar to these findings, TGFβ is abundantly expressed in KCs isolated from non-fibrotic or fibrotic livers (Figure 5E). Thus, we next verified whether treatment of amlexanox affects TGFβ expression in KCs. Treatment of amlexanox significantly reduced the expression levels of TGFβ in KCs with or without LPS stimulation (Figure 5F). In addition, co-culture experiments of hepatocytes-KCs showed that the expression levels of TGFβ (Figure 5G) and pro-inflammatory genes (Figure S3C) were remarkably decreased by treatment of amlexanox in CCl₄-mediated inflammatory milieu, and even in normal condition. Consistent with these findings, mice administered with amlexanox had significantly lower protein levels of TGFβ in fibrotic livers, as confirmed by our in vivo experiments (Figure S4). Taken together, treatment of amlexanox can suppress the production of TGFβ and pro-inflammatory cytokines in KCs by inhibiting the activity of NF-κB.

3.6 Inhibition of TBK1 and IKKε suppresses HSC activation and its related fibrogenic responses

It has been well-known that HSCs play a critical role in the pathogenesis of liver fibrosis.31 Also, HSCs are one of the main cellular sources of TBK1 and IKKε (Figure 1D). Therefore, we determined the roles of TBK1 and IKKε on LX-2 cells or primary murine HSCs by using amlexanox. As shown in Figure S5A,B, dose-dependently decreased protein levels of pTBK1 and pIKKε also were found in non-activated LX2 cells and HSCs upon amlexanox treatment. Also, the expression levels of pro-fibrogenic genes were significantly decreased by treatment of amlexanox (Figure S5C,D). Next, TGFβ-treated LX-2 cells and activated HSCs (7 days post-isolation) were used to determine the role of amlexanox on activated HSCs. Significantly reduced protein levels of pTBK1 and pIKKε were observed in activated LX-2 cell and HSCs (Figure 6A,B). Additionally, inhibition of TBK1 and IKKε markedly decreased protein levels of αSMA in activated LX-2 cells and HSCs (Figure 6C,D). Consistently, these cells showed lower expression levels of pro-fibrogenic genes by treatment of amlexanox (Figure 6E,F). As NF-κB is closely associated with the activation of HSCs,32 we next evaluated whether treatment of amlexanox affects activation of NF-κB in HSCs and found dose-dependently decreased protein levels of pNF-κB in activated HSCs or LX-2 cells upon amlexanox treatment (Figure 6G,H). These findings suggest that inhibition of TBK1 and IKKε can suppress the activation of HSCs and its related fibrogenic responses.

3.7 Anti-fibrotic effects by amlexanox treatment on HSCs are partially dependent on inhibition of STAT3

Considering that activation of STAT3 is involved in HSC activation and proliferation,33, 34 we further verified activation of STAT3 in activated LX-2 cells and primary HSCs and found that pSTAT3 was markedly reduced when TBK1 and IKKε were inhibited (Figure 7A,B). To confirm the inhibition of STAT3 by amlexanox resulting in anti-fibrotic effects, activated primary HSCs or TGFβ-treated LX-2 cells were cultured with amlexanox plus colivelin, STAT3 activator, to increase the activity of STAT3. As shown in Figure 7C,D, treatment of colivelin significantly restored reduced protein levels of pSTAT3 induced by amlexanox in activated HSCs and LX-2 cells. Consistently, significantly decreased protein levels of αSMA by treatment of amlexanox were abolished by treatment of colivelin in activated HSCs and LX-2 cells. Similarly, colivelin treatment reversed the decreased mRNA expression levels of pro-fibrogenic genes by treatment of amlexanox in activated HSCs and LX2 cells (Figure 7E,F). Next, we further verified the STAT3 dependency of amlexanox-mediated inhibition on activation of HSCs by using a STAT3 activation inhibitor (SPI). The results showed that the anti-fibrotic effects of amlexanox were amplified by treatment of SPI in activated HSCs (Figure 7G,H). These findings indicate that amlexanox-provided anti-fibrotic effects on activated HSC, at least partially, are dependent on STAT3 inhibition.