Potential conflict of interest: Nothing to report.
Infections of Reoviridae consisting of a double-stranded RNA (dsRNA) genome are a possible cause of biliary atresia (BA). The aim of the present study is to clarify the pathophysiological function of dsRNA viruses in the pathogenesis of BA. The expression of dsRNA pattern-recognizing receptors, Toll-like receptor 3 (TLR3), retinoic acid inducible gene I (RIG-I), melanoma differentiation-associated gene-5 (MDA-5), and dsRNA-activated protein kinase R (PKR) was constitutively detected in cultured human biliary epithelial cells (BECs). Stimulation with polyinosinic-polycytidylic acid [poly(I:C), a synthetic analog of viral dsRNA] induced the activation of transcription factors [nuclear factor (NF)-κB and interferon regulatory factor 3 (IRF3)] and the production of interferon-β1 (IFN-β1) and MxA as potent antiviral responses. Moreover, poly(I:C) up-regulated the expression of tumor necrosis factor–related apoptosis-inducing ligand (TRAIL), and both poly(I:C) and TRAIL reduced the viability of cultured human BECs by enhancing apoptosis. Experiments in vivo using tissue sections of extrahepatic bile ducts from patients with BA and controls (choledochal cysts and nonbiliary diseases) showed that the activation of NF-κB, interferon regulatory factor-3 (IRF-3), and PKR, and the enhancement of TRAIL and single-stranded DNA (ssDNA)–positive apoptosis were significant in BA, although extrahepatic bile ducts diffusely and constantly expressed TLR3 in all diseases. Conclusion: dsRNA viruses could directly induce the expression of TRAIL and apoptosis in human biliary epithelial cells as a result of the biliary innate immune response, supporting the notion that Reoviridae infections are directly associated with the pathogenesis of cholangiopathies in cases of BA. (HEPATOLOGY 2007.)
Biliary atresia (BA) is characterized by a progressive, inflammatory, and sclerosing cholangiopathy. Little is known about the etiology and pathogenesis of BA, but recent studies have demonstrated the presence of Reoviridae (type 3 reovirus and type C rotavirus) having a double-strand RNA (dsRNA) in liver tissue of patients with BA, although conflicting results have been reported.1–5 Moreover, the infection of newborn Balb/c-mice with Reoviridae including type A rhesus rotavirus and type 3 reovirus (Abney) leads to cholestasis and biliary obstruction resembling human BA.6, 7 However, the role of these viruses in the pathogenesis of cholangiopathies in patients with BA is still unknown.
Toll-like receptors (TLRs) are innate immune-recognition receptors that recognize pathogen-associated molecular patterns (PAMPs), and TLR3 recognizes dsRNA including dsRNA viruses.8 The stimulation of TLR3 by dsRNA transduces signals to activate the transcription factors nuclear factor-κB (NF-κB) and interferon regulatory factor 3 (IRF3) via Toll–interleukin-1 receptor domain-containing adaptor-inducing interferon beta (IFN-β1).8, 9 Moreover, the IFN-inducible helicase retinoic acid-induced protein I (RIG-I) and melanoma differentiation-associated gene-5 (MDA-5) also have been shown to bind dsRNA and regulate type I IFN–mediated responses to dsRNA.10 In addition, dsRNA-activated protein kinase R (PKR), a serine/threonine protein kinase, is also considered central to the interaction with dsRNA.11 The production of antiviral factors such as type I interferons (IFN-β1) and MxA by these virus-recognizing systems is characteristic of an immune response to viral infections.12
Although the apoptosis of biliary epithelial cells is speculated to play an important role in the obstructive cholangiopathy of BA, its mechanism is still unknown.13 Tumor necrosis factor receptor 1, CD95, and tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) receptors belonging to the tumor necrosis factor receptor superfamily contain a death domain and induce apoptosis by cross-linking their ligands. Human biliary epithelial cells (BECs) express CD95 and TRAIL receptors, but lack tumor necrosis factor receptor 1, suggesting that BECs are sensitive to CD95 ligand- and TRAIL-mediated apoptosis.14–16 TRAIL can interact with 4 different receptors, namely 2 agonistic receptors called DR4 (TRAIL-R1) and DR5 (TRAIL-R2), and 2 membrane-anchored decoy receptors designated DcR1 (TRAIL-R3) and DcR2 (TRAIL-R4). We speculated that clarifying the expression of these apoptosis-related molecules and susceptibility to apoptosis is important to understanding the mechanism of obstructive cholangiopathy in patients with BA.
To clarify the pathophysiological function of the Reoviridae in the cholangiopathy of BA, we examined the expression of dsRNA-recognizing receptors, the effects of dsRNA, and the apoptosis of biliary epithelial cells using cultured human BECs and liver tissues of BA patients.
A total of 14 patients with BA (average age, 2.2 months; male/female, 2/12), 18 patients with choledochal cysts (average age, 5 years; male/female, 4/14), and 6 patients with non-hepatobiliary disease (congenital heart anomaly; age-matched controls of BA; average age, 2.5 months; male/female, 3/3) were examined. These materials were surgically resected or autopsied common bile ducts, which were retrieved from the files of our laboratories. Formalin-fixed, paraffin-embedded sections were prepared for histological observation and immunohistochemistry.
In addition, 6 fresh liver specimens from each 2 patients with BA, primary biliary cirrhosis (PBC), and metastatic liver tumors were obtained from explant livers during liver transplantation (BA and PBC) and from surgically resected livers (metastatic liver tumors). Informed consent to conduct research was obtained from these patients. This study was approved by the Kanazawa University Ethics Committee.
Cultured Human BECs.
Six human intrahepatic BEC lines were established from the explanted livers with BA (BEC1 and BEC2), PBC (BEC3 and BEC4), and normal (BEC5 and BEC6, originated from nonneoplastic background livers of metastatic liver tumor showing normal histology) and were grown as monolayers.17 These cell lines had been confirmed to be BECs by the expression of a biliary-type cytokeratin, CK7. BECs were used between passages 4 and 8 for this study.
Pan-caspase inhibitor (Z-VAD-FMK, 40 μM, Calbiochem, Garmstadt, Germany), neutralizing TRAIL antibody (500 μg/ml, R&D, Minneapolis, MN), or the vehicle control [phosphate-buffered saline (PBS) or dimethyl sulfoxide] was preincubated with cultured cells for 1 hour at 37°C before stimulation with polyinosinic-polycytidylic acid [poly(I:C), a synthetic analog of viral dsRNA, 25 μg/ml (Invivogen, San Diego, CA)], recombinant human TRAIL (1 μg/ml, PeproTech, London, UK), or CD95 antibody (clone CH11, 0.5 μg/ml, MBL, Nagoya, Japan) for 1, 3, or 48 hours. These cells were used for the analyses of protein, mRNA, and cell viability according to the following methods.
NF-κB DNA-Binding Assay.
The activation NF-κB was evaluated by a sensitive multi-well colorimetric assay using a TransAM NF-κB Kit (Active Motif, Carlsbad, CA) following the manufacturer's instructions.
Reverse Transcription-Polymerase Chain Reaction and Real-Time Polymerase Chain Reaction.
For the evaluation of messenger ribonucleic acid (mRNAs) of TLR3, RIG-I, MDA-5, PKR, NF-κB, interferon regulatory factor-3 (IRF-3), IFN-β, MxA, TRAIL, TRAIL receptors (TRAIL-DR4, TRAIL-DR5, TRAIL-DcR1, and TRAIL-DcR2), CD95, and CD95 ligand in cultured BECs, total RNA was isolated from BECs, and 1 μg total RNA was reverse-transcribed with an oligo-(dT) primer and reverse transcriptase to synthesize complementary deoxyribonucleic acid. Human cultured cell lines (HuCCT1, HuH7, HepG2, and WI38) and normal tissues (pancreas and liver)18–27 were used as controls, and the complementary deoxyribonucleic acid was amplified by polymerase chain reaction (PCR) using the specific primers (Table 1). The PCR products were subjected to electrophoresis on 1.5% agarose gels containing ethidium bromide. Negative controls were obtained by replacing reverse transcriptase with ribonuclease-free and deoxyribonuclease-free water. In addition, to carry out relative quantification, real-time quantitative PCR was performed for measurements of IFN-β, MxA, TRAIL, and glyceraldehyde-3-phosphate dehydrogenase mRNAs according to a standard protocol using the SYBR Green PCR Master Mix and ABI-Prism 7700 Sequence Detection System (Applied Biosystems, Tokyo, Japan). Results are shown as relative mRNA expression compared with the level without any treatments (PBS).
Cell lysates of poly(I:C)-stimulated or unstimulated cultured cells (10 μg protein/lane) were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis. Separated proteins were transferred to a nitrocellulose membrane; the membrane was blocked in 5% bovine serum albumin and then probed for 1 hour with primary antibodies against human TRAIL [goat immunoglobulin G (IgG), 0.2 μg/ml, R&D), phospho-IRF-3 (rabbit IgG, 2 μg/ml, Upstate, Lake Placid, NY), IRF-3 (rabbit IgG, 0.4 μg/ml, Santa Cruz, Santa Cruz, CA), PKR (rabbit IgG, 1 μg/ml, Epitomics, Burlingame, CA), phospho-PKR (rabbit IgG, 1 μg/ml, Epitomics), and cleaved caspase 3 (rabbit IgG, 0.26 μg/ml, Cell Signaling, Danvers, MA). After a wash, the membrane was incubated for 1 hour with Envision+ system (Dako, Tokyo, Japan) or Simple Staining Kit (Nichirei, Tokyo, Japan), and visualized with the benzidine reaction.
Cell Viability Assessment.
Approximately 1 × 104 cells/well placed into 96-well plates were treated with poly(I:C), recombinant TRAIL, or CD95 antibody (positive control). Cell viability was measured 48 hours later with a microplate reader using the tetrazolium salt (Roche, Penzberg, Germany) and presented as a percentage of the value for the control culture.
The deparaffinized and rehydrated sections for studying TLR3, NF-κB, and TRAIL were microwaved in 10 mM citrate buffer for 20 minutes in a microwave oven; those for IRF-3 and phospho-PKR were treated in an ethylenediamine tetraacetic acid pressure cooker. After the blocking of endogenous peroxidase, these sections were incubated at 4°C overnight with antibodies against TLR3 (rabbit IgG, 1 μg/ml, Santa Cruz), IRF-3, NF-κB (mouse IgG1, 2 μg/ml, Santa Cruz), single-stranded DNA (ssDNA, rabbit IgG, 0.25 μg/ml, Dako), phospho-PKR, or TRAIL, and then at room temperature for 1 hour with Envision+ system or Simple Staining Kit. After a benzidine reaction, sections were lightly counterstained with hematoxylin. As a negative control, isotype-matched immunoglobulin was used as the primary antibody. We primarily examined BECs lining extrahepatic bile ducts in each case and, in BA, evaluated the inflamed biliary epithelium in the proximal area of obstructive lesion.
As for in vitro study, results were obtained from 2 independent experiments using BEC1-BEC6. Data were analyzed using the paired t test; P < 0.05 was considered statistically significant.
Baseline Expression of dsRNA-Recognizing Receptors and Transcription Factors in BECs.
We first examined the expression of dsRNA-recognizing receptors (TLR3, RIG-I, MDA-5, and PKR) and transcription factors (NF-κB and IRF-3) in BECs. These mRNAs were basically detected in all 6 cultured BEC lines by reverse transcription-polymerase chain reaction analysis (Fig. 1A), although a slight difference was found among BEC lines. In addition, immunohistochemistry showed that TLR3 protein was constantly expressed in the cytoplasm of cultured BECs (Fig. 1B) and of epithelial cells lining extrahepatic bile ducts (Fig. 1C, D), and its distribution and intensity did not differ between BA and controls.
Poly(I:C)-Induced Activation of Transcription Factors and PKR in Cultured BECs.
To investigate the dsRNA-induced activation of transcription factors (NF-κB and IRF-3) and PKR, we treated cultured BECs with poly(I:C) and evaluated the activation of these molecules. As shown in Fig. 2A, the NF-κB–DNA-binding assay showed that the treatment with poly(I:C) resulted in the activation of NF-κB in BECs. Western blotting showed that phospho-PKR and phospho-IRF-3 were detected in poly(I:C)-treated BECs, but not untreated samples (PBS, vehicle controls), although PKR and IRF-3 were found irrespective of treatment, demonstrating that poly(I:C) activates PKR and IRF-3 (Fig. 2B). A significant difference among the cell lines was not found. In addition, the immunohistochemistry for NF-κB, IRF-3, and phospho-PKR demonstrated the nuclear translocation of NF-κB and IRF-3, suggesting activated forms, and the enhanced expression of phospho-PKR in poly(I:C)-treated BECs (Fig. 2C).
Poly(I:C)-Induced Expression of IFN-β and MxA in Cultured BECs.
To assess the biological relevance of the poly(I:C)-induced activation of NF-κB and IRF-3, we examined the effect of poly(I:C) on cultured BECs by measuring the production of antiviral factors IFN-β and MxA. Reverse transcription-polymerase chain reaction showed that the amplicon of IFN-β mRNA could not be detected in any of the cell lines without stimulant, but de novo expression appeared in all the poly(I:C)-treated cell lines (Fig. 3A). This de novo expression was partially, not completely, inhibited by pretreatment with an NF-κB–specific inhibitor, isohelenin, suggesting that the poly(I:C)-induced production of IFN-β involves other transcription factors, probably IRF3, as well as NF-κB. In contrast, MxA mRNA was detected under basal conditions and up-regulated by the treatment with poly(I:C). This up-regulation was completely inhibited by isohelenin, suggesting that the production is dependent on NF-κB. Moreover, we confirmed the statistical significance of the finding in a quantitative analysis using real-time PCR (Fig. 3B).
Detection of Transcription Factors and Phospho-PKR in Extrahepatic Bile Ducts.
NF-κB and IRF-3 were expressed constantly in the cytoplasm in extrahepatic bile ducts in BA and controls (choledochal cysts and nonhepatobiliary diseases); however, in BA patients, nuclear expression indicating the active forms was also detected (Fig. 4). Moreover, the expression of phospho-PKR (activated PKR) was also found in the cytoplasm of inflamed bile ducts in BA (Fig. 4).
Expression of Apoptosis-Related Molecules and Effects of dsRNA in Cultured BECs.
Amplification of TRAIL, TRAIL-related receptors (TRAIL-DR4, TRAIL-DR5, TRAIL-DcR1, TRAIL-DcR2), and CD95 mRNAs was detected in all 6 BEC lines (Fig. 5A). Moreover, the expression of TRAIL, but not the others, was up-regulated at 3 hours after the treatment with poly(I:C) (Fig. 5B), and this up-regulation was completely inhibited with a return to the basal level on pretreatment with isohelenin (Fig. 5B). CD95 ligand was not detected in any cell lines irrespective of treatments (Fig. 5A). As shown in Fig. 5C, real-time PCR analysis showed that treatment with poly(I:C) significantly up-regulated the expression of TRAIL mRNA 42.3-fold.
Effect on Viability of dsRNA and TRAIL in Cultured BECs.
To investigate the effects of poly(I:C) and TRAIL on cultured BECs, we analyzed cell viability using a water-soluble tetrazolium salt assay. As shown in Fig. 6A, the treatments with poly(I:C) and TRAIL reduced cell viability after 48 hours. The reduction in viability caused by poly(I:C) as well as TRAIL was inhibited by pretreatment with the pancaspase inhibitor and neutralizing anti-TRAIL antibody. Moreover, immunohistochemistry for ssDNA and a Western blot analysis of the cleaved caspase 3 revealed increased ssDNA-positive apoptosis and the presence of cleaved caspase 3 in poly(I:C)-treated BECs (Fig. 6B and 6C, respectively). These suggest that poly(I:C)-induced apoptosis is dependent on the TRAIL-mediated caspase pathway.
Detection of TRAIL and Apoptosis in Extrahepatic Bile Ducts.
TRAIL was expressed in the cytoplasm of epithelial cells lining extrahepatic bile ducts at various levels of intensity in the patients with BA and controls (choledochal cysts and nonhepatobiliary diseases). However, in BA patients, the biliary epithelium showed an enhanced level of expression (Fig. 7). Moreover, many ssDNA-positive apoptoses were found in the biliary epithelium of BA patients, though they were scanty in controls (Fig. 7).
Viral infections involving the Reoviridae, cytomegalovirus, and Epstein-Barr virus are suspected of being part of the cause of BA.1–3 Reoviridae having a dsRNA genome, in particular, are characterized by epithelial tropism, and rotavirus type A is the most frequent causative agent in cases of acute infantile diarrhea in young children. Therefore, BECs likely are also a target of the viruses directly causing biliary diseases.
BECs have been reported to express a functionally active multiple TLRs and respond to the corresponding bacterial PAMPs such as lipopolysaccharide.14, 17, 18, 29, 30 In the TLR family, TLR3 recognizes dsRNA and is located in the endosomal membranes with the ligand-binding domain facing the lumen of the endosomes and the signaling domain positioned on the cytoplasmic side. Therefore, TLR3 allows cells to detect dsRNA that is phagocytosed from the extracellular space, where it is released by infected cells that are undergoing lysis or necrosis.31 In this study, we demonstrated the TLR3 expression in cultured human BECs and the diffuse expression of TLR3 in extrahepatic bile ducts of patients with BA. In addition, BECs produced antiviral products (IFN-β and MxA) on exposure to a synthetic analog of viral dsRNA, poly(I:C). Al-Masri et al.32 reported the increased expression of MxA in bile ducts of patients with BA, suggesting that biliary innate immunity against any virus in vivo is associated with the pathogenesis of BA and supporting the current results obtained using cultured human BECs. In contrast, cytosolic receptors for intracellular dsRNA such as RIG-I and MDA-5 are essential for innate immune responses to intracellular viral replication.10 The current study showed that RIG-I and MDA-5 are also expressed in cultured BECs, suggesting that RIG-I and MDA-5 also function as a sensor of viral infections in biliary cells. It is necessary to examine the distribution of RIG-I and MDA-5 in the biliary tree to clarify the role of these receptors in the pathogenesis of BA. However, immunohistochemical detection could not be performed using human tissue, because no antibodies for RIG-I and MDA-5 are commercially available.
In general, epithelial cells including BECs initially have no direct contact with the dsRNA of Reoviridae and cannot directly recognize it, because genomic dsRNA is surrounded by 3 concentric layers of protein in the mature virus. The direct exposure of biliary epithelial cells to the dsRNA would occur if (1) the virus successfully infected biliary epithelial cells and its genome replicated within the cytosol; and (2) BECs were exposed to genomic dsRNA derived from broken-up infected cells. During an infection of host cells, single-stranded RNA viruses as well as dsRNA viruses generate RNA-RNA strand pairs in the process of RNA-dependent RNA synthesis. Some DNA viruses also produce dsRNA during their life cycle. Therefore, in the former situation, biliary epithelial cells could respond to most viruses by recognizing via RIG-I and MDA-5 in the cytosol of infected cells. In contrast, the latter situation is restricted to dsRNA viruses, mostly the Reoviridae: dsRNA of this genus enters cells through the endocytic machinery and is directly recognized by TLR3. Which situation and which receptor is important to the pathogenesis of BA is unknown, because the exact distribution of Reoviridae in the liver and common bile ducts has not been clarified. Even though these dsRNA viruses do not directly infect BECs in cases of BA, biliary cells could react to extracellular components of dsRNA released from infected dying cells.1–3
In epithelial cells, the pleiotropic transcription factor NF-κB plays a central role in regulating genes that govern the onset of mucosal inflammatory responses. A variety of stimuli lead to the rapid degradation of inhibitory protein I-κ and the subsequent translocation of NF-κB to the nucleus, where it regulates gene transcription. IRF-3 is also a transcription factor associated with anti-viral responses via TLR3, and the production of IFN-β depends on IRF-3/ISRE3.8, 9 In addition, PKR is directly activated on binding dsRNA to undergo dimerization and autophosphorylation leading to the degradation of I–κ.33 In this study, we showed that treatment with poly(I:C) activates NF-κB, PKR, and IRF-3 in cultured BECs. Moreover, poly(I:C) induced the production of IFN-β in BECs, and this production was partially inhibited by an NF-κB–specific inhibitor, isohelenin. This is consistent with the finding that TLR3 signaling leads to the activation of NF-κB and IRF-3 and consequent production of IFN-β. In contrast, the production of MxA triggered by treatment with poly(I:C) was completely inhibited by isohelenin, suggesting that the induction of MxA production depends solely on NF-κB. The current in vivo study showed increased expression of nuclear (activated) NF-κB and IRF-3, and of phosphorylated PKR in epithelial cells lining the extrahepatic bile ducts of patients with BA. BECs, through the actions of dsRNA-recognizing receptors and downstream signaling, may play a role in the innate immunity of the biliary tree after a viral infection by producing anti-viral factors that limit viral replication in the cells.
Here, we conducted a study of apoptosis to clarify the pathophysiological functions of dsRNA viruses in BECs, because biliary apoptosis may play an important role in the obstructive cholangiopathy of BA.13 The current study showed that BECs express apoptosis-inducing receptors (CD95, TRAIL-DR4, and TRAIL-DR5), apoptosis-inhibiting decoy receptors (TRAIL-DcR1 and TRAIL-DcR2), and an apoptosis-inducing ligand (TRAIL). Among them, the expression of TRAIL, but not the others, was up-regulated by treatment with poly(I:C) in cultured BECs. This up-regulation was inhibited by isohelenin, suggesting that the induction of TRAIL depends on the NF-κB pathway. Moreover, treatments with poly(I:C) and TRAIL significantly reduced the viability of cultured BECs accompanying ssDNA-positive biliary apoptosis and cleaved caspase 3, and this reduction in cell viability was significantly inhibited by a caspase inhibitor, suggesting that the poly(I:C)- and TRAIL-induced reduction of cell viability is caused by the mechanism of biliary cell apoptosis. In addition, the poly(I:C)-induced reduction in cell viability was inhibited by neutralizing TRAIL antibody, suggesting that poly(I:C)-induced biliary apoptosis depends on the production of TRAIL. The in vivo study also indicated that enhanced TRAIL expression and ssDNA-positive apoptosis were found in the biliary epithelium of patients with BA, in contrast with controls. These findings suggest that the induction of TRAIL expression and biliary apoptosis is caused by an innate immune response to dsRNA viruses and is closely associated with the obstruction of extrahepatic bile ducts in cases of BA.
In this study, we explored the role of innate immune responses to dsRNA in BECs to better understand the pathogenesis of BA, which is associated with infections of dsRNA viruses. Consequently, the constitutive expression of dsRNA-recognizing receptors and the production of antiviral products in BECs as a response to a challenge with the viral dsRNA analog poly(I:C) were demonstrated. In addition, poly(I:C) markedly up-regulated the expression of TRAIL and the induction of biliary apoptosis. The in vivo study showed an enhancement of TRAIL expression, apoptotic biliary cells, and the activation of transcription factors and PKR in the biliary epithelium lining extrahepatic bile ducts in patients with BA. Taken together, these findings suggest that BECs not only directly participate in the antiviral innate immune response through the production of antiviral effectors to prevent viral replication by secreting antibiotics in response to dsRNA, but also play a role in the generation of apoptotic responses, supporting that infection of Reoviridae are directly associated with the initial pathogenesis of obstructive cholangiopathy in cases of BA.