We first investigated the UPR signature during the onset and progression of CBDL-induced liver fibrosis by analyzing mRNA or protein expression of different UPR markers. Mice were sacrificed on a weekly basis, starting from week 1 until week 6 after surgical induction. The mRNA levels of BiP were increased 6 weeks post CBDL. Activating transcription factor 4 (ATF4) and spliced X-box binding protein 1 (XBP1s) mRNA transcripts were not increased at any point following CBDL. Unspliced XBP1 (XBP1u) levels showed an increase at week 3 and growth arrest and DNA damage-inducible protein (GADD34) mRNA levels were significantly upregulated from week 3 until week 6 post-CBDL (Figure 1a). No increase in GADD34, ATF4 or phosphorylated eukaryotic initiation factor 2 α (peIF2α) protein levels could be detected, however, a gradual increase of CHOP expression in CBDL mice was seen on Western blot analysis starting at week 1 onwards to week 6 (Figure 1b). While c-Jun N-terminal kinase (JNK) activation is associated with apoptosis, we could not detect an increase in JNK phosphorylation in the time after CBDL (Figure 1b).

Since previous reports described the involvement of apoptosis in cholestatic liver damage and as CBDL was associated with induction of the pro-apoptotic ER stress marker CHOP in our study, we evaluated the expression of caspase 3, executor of both the intrinsic and extrinsic apoptotic pathway, ER stress-induced caspase 12, and tumor necrosis factor receptor superfamily member 1a (TNFRsf1a) and Fas-Associated protein with Death Domain (FADD), two markers involved in extrinsic apoptotic cell death. Caspase 3 mRNA expression was significantly upregulated starting at week 1 (with the exception of weeks 2 and 3), and caspase 12 was significantly increased at all timepoints after CBDL, compared to sham surgery (Figure 2a). TNFRsf1a and FADD were upregulated at all timepoints in CBDL mice compared to their matching control group (Figure 2a). Besides apoptosis, the possible involvement of inflammasome activation, which results in pyroptotic cell death, was assessed by measuring caspase 1 and nucleotide-binding, leucine-rich repeat and pyrin domains-containing protein 3 (NLRP3) inflammasome levels. Although caspase 1 gene expression was increased from week 3 until week 6 in CBDL mice (Figure 2b), no increase in NLRP3 protein expression was demonstrated (Figure 2c).

As we demonstrated the induction of ER stress and involvement of apoptotic cell death in this animal model of cholestatic liver disease, we next investigated the therapeutic potential of TUDCA, which is known for its ER stress-reducing and anti-apoptotic capacities, on the degree of CBDL-induced liver fibrosis. In a preventive and a therapeutic study, TUDCA was administered starting one week before and two weeks after surgical induction until the end of the experiments respectively. As expected, the CBDL procedure resulted in a significant increase in fibrotic area compared to sham surgery (Figure 3). Importantly, TUDCA treatment, both in the preventive and the therapeutic setting, led to a significant reduction of liver fibrosis (Figure 3). This was associated with reduced, however not significantly reduced, liver damage, as reflected by AST and ALT levels (Table 1).

In order to explore whether TUDCA exerts its beneficial effect on liver damage and fibrosis through a reduction in pro-apoptotic UPR signaling, UPR-induced CHOP expression and gene and protein expression of different pro-apoptotic markers after TUDCA treatment was studied. Western blot analysis confirmed the induction of the pro-apoptotic transcription factor CHOP at 6 weeks after CBDL and demonstrated reduced CHOP expression when TUDCA was given preventively (Figure 4a). Preventive supplementation with TUDCA also significantly reduced mRNA levels of caspase 3 and both preventive and therapeutic TUDCA treatment reduced cleaved caspase 3 and cleaved caspase 12 protein levels, compared to the CBDL controls (Figure 4b,c). TNFRsf1 expression was only decreased in the preventive setting whereas FADD was significantly lower following the preventive and therapeutic TUDCA treatments. Although an increased trend was observed in the CBDL and therapeutic TUDCA group, the anti-apoptotic Bcl2, was not significantly altered after the CBDL or TUDCA treatment (Figure 4b). No effect of TUDCA, neither preventively nor therapeutically, was seen on caspase 1 gene expression in CBDL mice nor on NLRP3 levels (Figure 4d,e).

Cholestatic liver disease is a major cause of liver fibrosis, which is triggered by hepatic cell death. It has been shown in different in vitro models that toxic bile acids, such as deoxycholic acid and glycochenodeoxycholic acid, cause ER stress and activation of the UPR [6,7,8,30]. In later studies, UPR activation has been investigated in different in vivo mouse models for cholestasis, which led to conflicting results [9,12,31,32]. However, none of these studies explored the UPR at different stages during early onset and progression of cholestatic liver disease. Therefore, we investigated the kinetics of hepatic UPR activation and the mode of associated hepatic cell death in a mouse model of cholestasis-induced liver fibrosis. In accordance with Mencin et al. and Tamaki et al. [12,31], we observed no increase in the UPR markers XBP1u and XBP1s (IRE1 pathway) and peIF2α and ATF4 (PERK pathway). GADD34 was increased starting from week 3 at the mRNA level, however, no significant differences in protein levels could be detected. Interestingly, we demonstrated gradually increasing levels of CHOP during the progression of liver fibrosis, which is one of the main UPR downstream effectors able to induce hepatic apoptotic and/or pyroptotic cell death when overexpressed [14]. Indeed, in case of excessive ER stress, CHOP promotes apoptosis through reduced expression of the anti-apoptotic protein Bcl-2, with subsequent Ca²⁺ influx into the cytoplasm [33]. Through m-Calpain, pro-caspase 12 is activated by cleavage, which further drives the caspase cascade [33]. Sequential activation of caspases plays a central role in the execution phase of apoptosis, in which caspase 3 is the final common executor of both the intrinsic and extrinsic apoptotic pathway. In the intrinsic pathway, mitochondrial permeabilization causes cytochrome c release, which activates caspase 9 and in turn caspase 3. In the extrinsic pathway, caspase 3 is activated by caspase 8 which is upstream bound and activated by FADD, which is bound to death receptors such as TNFR, forming the death-inducing signaling complex. In the present study, we demonstrate increased (cleaved) caspase 3 and 12, FADD and TNFRsf1a expression in fibrotic mice, indicative for the occurrence of apoptotic cell death during the initiation and progression of biliary liver damage and the development of fibrosis [34,35,36,37]. In addition, activation of CHOP is known to stimulate transcription of pro-caspase 1 and 11, which become activated upon cleavage and activated by NLRP3, the most important inflammasome [14]. Active caspase 1 and 11 in turn lead to apoptosis, through caspase 3, but also to pyroptosis, a programmed form of cell death, characterized by DNA damage, cell lysis and active IL-1β and IL-18 release [14]. Recently, the combination of CHOP-induced apoptosis and NLRP3 inflammasome-mediated pyroptosis has been demonstrated to be involved in the pathogenesis of experimental non-alcoholic steatohepatitis, which was successfully counteracted by treatment with TUDCA [14]. In our study, increased gene expression of caspase 1 in CBDL mice could be demonstrated, without, however, any change in NLRP3 protein levels, suggesting the absence of NLRP3 inflammasome-mediated pyroptosis. Summarized, these findings point towards a central role of CHOP-induced pro-apoptotic pathways in liver disease due to biliary etiology.

The chemical chaperone TUDCA, the hydrophilic taurine-conjugated form of UDCA, is known for its cytoprotective capacities of reducing ER stress and apoptotic cell death [15,20,29,30,38]. Since we demonstrated that CHOP-induced pro-apoptotic pathways are involved in cholestatic liver damage and fibrosis, and as TUDCA is known to be able to modulate these pathways in other diseases [25,27,39], we investigated its therapeutic potential to impede liver fibrosis in the CBDL mouse model. Importantly, both preventive and therapeutic treatment with TUDCA led to a significant reduction in the degree of liver fibrosis, which was accompanied by an improving trend for ALT and AST levels. These findings are in line with previous research which demonstrated that CHOP knockout mice develop less fibrosis [12]. A recent small double-blind randomized trial evaluated the therapeutic efficacy of TUDCA (750 mg daily) in liver cirrhosis, using UDCA as a parallel control, with a follow-up period of six months. They concluded that TUDCA therapy is safe and appears to be more effective than UDCA in improving biochemical parameters [40]. Our study evaluated the effect of TUDCA in a long term cholestasis mouse model, whereas other studies often study the effect during early cholestasis [41,42].

Lastly, we suggest that TUDCA indeed exerts its beneficial effect on liver damage and fibrosis through alleviating CHOP-induced apoptotic cell death, as treatment with TUDCA led to decreased CHOP expression and reduced levels of the pro-apoptotic markers caspase 3 and 12, TNFRsf1a and FADD.

In conclusion, this study indicates that CHOP-mediated apoptotic cell death contributes to the development of cholestasis-induced liver fibrosis and that treatment with TUDCA has a beneficial effect on liver fibrosis in this model of biliary fibrosis, which might, at least in part, be the result of its ER stress-reducing and anti-apoptotic properties. In this way, TUDCA might represent a therapeutic strategy for the treatment of biliary liver disorders.

Male 7-week-old Swiss and Sv129 mice, purchased from Harlan (Horst, The Netherlands) and Iffa Credo (Brussels, Belgium), were housed in the animal facility of the Faculty of Medicine and Health Sciences, Ghent University, Ghent, Belgium, and acclimatized under controlled conditions for one week prior to the experiments. They received care in accordance with Federation for Laboratory Animal Science Associations (FELASA) guidelines and the national guidelines for animal protection. The animal protocols used in this work were evaluated and approved by the Ethical Committee of Experimental Animals at the Faculty of Medicine and Health Sciences, Ghent University, Belgium, (ECD 13/12 approved on 29 April 2013 and ECD 14/67 approved on 3 November 2014).

Secondary biliary fibrosis was induced by common bile duct ligation (CBDL) at the age of eight weeks, as described previously [29]. In brief, under isoflurane (Isoflo^®, Abbott, Louvain-la-Neuve, Belgium) inhalation anaesthesia, a midline abdominal incision was made and the common bile duct was isolated and ligated with two knots of a non-resorbable suture (Silkan^® 5/0, Braun Aesculap, Tuttlingen, Germany). The first ligature was made below the junction of the hepatic ducts and the second was made above the entrance of the pancreatic duct. The common bile duct was resected between the two ligatures after which the abdomen was closed by suturing the abdominal muscle and skin in two separate layers. Control mice were sham-operated; the common bile duct was isolated but not ligated.

In order to study the hepatic UPR and cell death during the time course of biliary fibrosis following CBDL, male Swiss mice were euthanized on a weekly basis, for six consecutive weeks, starting at week 1 after surgical induction (6 sham and 6 CBDL groups, 7–14 animals in each group). At sacrifice, liver samples were immediately harvested into liquid nitrogen and kept at −80 °C for protein and mRNA extraction. A portion was fixed in 4% formaldehyde for 24 h and imbedded in paraffin for histological analysis.

The therapeutic potential of TUDCA in CBDL mice was studied in both a preventive and therapeutic study. For the preventive study, male Sv129 CBDL and sham mice received 10 mg/kg/day TUDCA (Calbiochem, Overijse, Belgium) in their drinking water, starting one week before surgical induction till the moment of euthanasia. In the therapeutic setting, TUDCA (10 mg/kg/day) was added to the drinking water, starting 14 days after surgical induction till euthanasia. Control CBDL and sham mice received normal drinking water. This makes a total of five groups, a sham group with TUDCA, a sham group without TUDCA, a CBDL control group without TUDCA treatment, a preventive CBDL group and a therapeutic CBDL group (6–10 animals/group, Table 1). Six weeks after sham or CBDL surgery, retro-orbital blood was taken under ketamine (100 mg/kg, Ceva santé, Brussels, Belgium) and xylazin (10 mg/kg, Kela, Sint-Niklaas, Belgium) anesthesia after which the animals were euthanized by cervical dislocation. Liver samples were collected as described above.

The degree of liver fibrosis was scored quantitatively with Cell^D software (version 3.4, Olympus, Berchem, Belgium) on Sirius red-stained liver sections. A minimum of ten fields per liver section per mouse were blindly evaluated and scored by two independent researchers (AP and SR).

RNA was extracted from 20 mg of frozen liver tissue preserved in RNA-later (Ambion, Thermo Fisher, Ghent, Belgium), according to the manufacturer’s guidelines (Rneasy Mini Kit, Qiagen, Venlo, The Netherlands) and measured for purity and quantity with spectrophotometry (Nanodrop, Thermo Scientific, Waltham, MA, USA). cDNA was made out of one microgram of mRNA by reverse transcription using the iScript cDNA synthesis kit (BioRad, Temse, Belgium) according to the manufacturer’s instructions. Diluted cDNA was subjected to 45 cycles of quantitative PCR amplification using SYBR Green mix (Sensimix, Bioline Reagents Ltd., London, UK) and 2 µM of each primer. A two-step program was run on a LightCyclerR 480 (Roche, Vilvoorde, Belgium). Melting curve analysis confirmed primer specificities. All reactions were run in duplicate and normalized to reference genes that showed stable expression in all samples (GAPDH, SDHA, HMBS and HPRT). The double ΔCq method was used for processing of qPCR data in which data were compared to the PBS control group. The PCR-efficiency of each primer pair was calculated using a standard curve of reference cDNA. Amplification efficiency was determined using the formula 10^−1/slope − 1. The primer set sequences are listed in Table S1.

Total protein extract was obtained by homogenizing tissue in RIPA buffer (PBS, 0.5% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 5.5% β-glycerophosphate, 1 mM dithiothreitol and complete protease and phosphatase inhibitors (Roche Diagnostics, Vilvoorde, Belgium)). The total protein yield was determined using Bradford reagent (Biorad, Temse, Belgium) and 30 µg protein was fractionated by SDS-PAGE. Protein lysates from multiple mice per group were pooled and transferred to a nitrocellulose membrane which was subsequently blocked for 1 h with 5% BSA, incubated overnight at 4 °C with primary antibodies in blocking buffer (Table S2), followed by 1 h incubation at room temperature with horse radish peroxidase-conjugated secondary antibodies (goat-anti rabbit sc2004, Santa Cruz, Heidelberg, Germany or goat anti-mouse sc2005, Santa Cruz). Clarity western ECL substrate (Biorad, Temse, Belgium) was used to visualize the proteins. GAPDH and β-tubulin were used as loading control proteins or total protein load was used for normalization.

Data analysis was performed using GraphPad (GraphPad Software Inc., San Diego, CA, USA) and SPSS 21 (IMB Corp, Armonk, NY, USA). Differences between groups were calculated using t-test (for two groups) or one-way ANOVA analysis with Sidak’s correction (for multiple groups) for normally distributed data. For non-parametric data, multiple groups were compared using the Kruskal-Wallis test with Dunn’s post hoc test. Reported p-values were two-sided and considered significant when lower than 0.05.