Temporal-spatial activation of apoptosis and epithelial injury in murine experimental biliary atresia


  • Nissa Erickson,

    1. Cincinnati Children's Hospital Medical Center and Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH
    Current affiliation:
    1. Department Pediatrics, University of Wisconsin, Madison, WI
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Sujit Kumar Mohanty,

    1. Cincinnati Children's Hospital Medical Center and Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Pranavkumar Shivakumar,

    1. Cincinnati Children's Hospital Medical Center and Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH
    Search for more papers by this author
  • Gregg Sabla,

    1. Cincinnati Children's Hospital Medical Center and Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH
    Search for more papers by this author
  • Ranajit Chakraborty,

    1. Department of Environmental Health, University of Cincinnati College of Medicine, Cincinnati, OH
    Search for more papers by this author
  • Jorge A. Bezerra

    Corresponding author
    1. Cincinnati Children's Hospital Medical Center and Department of Pediatrics, University of Cincinnati College of Medicine, Cincinnati, OH
    • Cincinnati Children's Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229-3039
    Search for more papers by this author
    • fax: 513-636-5581.

  • Potential conflict of interest: Nothing to report.


Biliary atresia is a fibro-inflammatory cholangiopathy that obstructs the extrahepatic bile ducts in young infants. Although the pathogenesis of the disease is undefined, studies in livers from affected children and neonatal mice with experimental biliary atresia have shown increased expression of proapoptosis molecules. Therefore, we hypothesized that apoptosis is a significant mechanism of injury to duct epithelium. To test this hypothesis, we quantified apoptosis using terminal transferase dUTP nick end labeling and active caspase-3 staining in livers and extrahepatic bile ducts from Balb/c mice infected with Rhesus rotavirus (RRV) within 24 hours of birth. RRV induced a significant increase in labeled cells in the portal tracts and in epithelial and subepithelial compartments of extrahepatic bile ducts, with onset within 3 days and peaks at 5–10 days. Exploring mechanisms of injury, we found increased biliary expression of caspases 1 and 4 and of interferon-gamma (IFNγ)–related and tumor necrosis factor-alpha (TNFα)–related genes. Using a cholangiocyte cell line, we found that neither IFNγ nor TNFα alone affected cell viability; however, simultaneous exposure to IFNγ and TNFα activated caspase-3 and decreased cell viability. Inhibition of caspase activity blocked apoptosis and restored viability to cultured cholangiocytes. In vivo, administration of the caspase inhibitor IDN-8050 decreased apoptosis in the duct epithelium and the extent of epithelial injury after RRV challenge. Conclusion: The biliary epithelium undergoes early activation of apoptosis in a mouse model of biliary atresia. The synergistic role of IFNγ and TNFα in activating caspase-3 in cholangiocytes and the decreased apoptosis following pharmacologic inhibition of caspases support a prominent role for apoptosis in the pathogenesis of experimental biliary atresia. (HEPATOLOGY 2008.)

Biliary atresia results from an inflammatory and fibrosing obstruction of extrahepatic bile ducts of unknown etiology. The onset of duct obstruction is restricted to the first few months of life and presents as neonatal cholestasis. Despite prompt diagnosis and surgical treatment, the disease progresses to end-stage cirrhosis in most patients and requires liver transplantation for long-term survival. Although the etiology of biliary injury is largely undefined, the presence of a coordinated activation of proinflammatory cytokines at the time of diagnosis suggests an important role of inflammatory circuits in the pathogenesis of disease.1, 2 Testing of this hypothesis in a murine model of biliary atresia revealed that interferon-gamma (IFNγ) provides critical biological cues for the inflammatory obstruction of bile ducts3 and that primed lymphocytes have a preferential tropism to bile ducts.4 A further functional relationship between lymphocytes and proinflammatory cytokines with pathogenesis of duct injury was implied by the finding that incubation of primed lymphocytes with homogenates from a murine cholangiocyte cell line triggered the release of IFNγ.5 However, the mechanisms by which cytokines may injure cholangiocytes and whether it occurs in early phases of biliary injury are not yet known.

In search of potential mechanisms of injury to bile ducts, we and other investigators found an up-regulation of apoptosis-related genes in extrahepatic bile ducts and livers of neonatal mice challenged with Rhesus rotavirus (RRV) type A.6, 7 These findings are in keeping with previous studies in humans reporting a high apoptosis index and increased expression of FasL in intrahepatic bile ducts of children with biliary atresia.8, 9 On the basis of these data, we hypothesized that apoptosis is a significant mechanism of injury to duct epithelium in biliary atresia. We tested this hypothesis using the experimental mouse model of RRV-induced biliary atresia to quantify apoptosis in livers and bile ducts and in vitro approaches to explore mechanisms of cytokine-induced apoptosis of cholangiocytes. We found a unique temporal and spatial activation of apoptosis along the biliary tract and a synergistic role for IFNγ and tumor necrosis factor-alpha (TNFα) in the promotion of cholangiocyte injury.


BEC, biliary epithelial cells; CK7, cytokeratin 7; DAPI, 4′,6-diamidino-2-phenylindole; DMEM, Dulbecco's modified Eagle's medium; EHBD, extrahepatic bile duct; FITC, fluorescein isothiocyanate; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; IFNγ, interferon-gamma; IRF-1, interferon regulatory factor 1; mRNA, messenger RNA; NS, normal saline; PCR, polymerase chain reaction; RRV, Rhesus rotavirus; SD, standard deviation; SEC, subepithelial cells; STAT-1, signal transducer and activator of transcription 1; TNFα, tumor necrosis factor-alpha; TNFR-1, tumor necrosis factor receptor 1; TUNEL, terminal transferase dUTP nick end labeling.

Materials and Methods

Mouse Model of Biliary Atresia.

Biliary injury of neonatal onset was induced by the intraperitoneal administration of 1.5 × 106 focus forming units of RRV in 20 μL of saline into Balb/c mice within the first 24 hours of life as described previously3, 10, 11; the same volume of saline was used in control mice. All neonatal mice were raised by their mothers in a defined pathogen-free vivarium with a 12-hour dark-light cycle. RRV-injected and saline-injected mouse litters were housed in separate cages to prevent dissemination of the virus. Mice were monitored daily for development of jaundice of non–fur-covered skin, acholic stools, and weight changes, and they were sacrificed at different time points after RRV or saline challenge. The gross appearance of the liver and extrahepatic bile ducts was documented, and extrahepatic ducts were microdissected en bloc with the gallbladder. Tissues were snap-frozen in liquid nitrogen, embedded frozen in an optimal cutting temperature compound (Sakura Finetek USA, Inc., Torrance, CA), or formalin-fixed/paraffin-embedded for sectioning with an Olympus CUT4060 microtome (Olympus America, Inc., San Jose, CA).

Cell Line and Reagents.

The murine cholangiocyte cell line (mCL), established from Balb/c mice,3, 12 was grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum, 2 mM glutamine, and penicillin-streptomycin (Invitrogen, Grand Island, NY). Recombinant mouse IFNγ was purchased from Peprotech, Inc. (Rocky Hill, NJ), and recombinant mouse TNFα was purchased from R&D Systems (Minneapolis, MN). The pancaspase inhibitor Mx1013 was obtained from Cytovia, Inc. (San Diego, CA), and IDN-8050 was kindly provided by Dr. Richard Butt of Pfizer Global Research and Development (Sandwich, United Kingdom).

Detection of Apoptosis.

To quantify apoptosis, sections of paraffin-embedded livers and extrahepatic bile ducts were subjected to the terminal transferase dUTP nick end labeling (TUNEL) assay as described previously, with modifications.13 In brief, sections were deparaffinized, rehydrated, incubated with proteinase K for 20 minutes, washed in 0.5% hydrogen peroxide, incubated with a terminal transferase reaction mix that included terminal deoxynucleotidyl transferase enzyme and biotin-dUTP (Roche Diagnostics, Indianapolis, IN), stained with streptavidin–horseradish peroxidase (Dako, North America, Inc., Carpinteria, CA) and 3-amino-9-ethylcarbazole/dimethyl formamide (Sigma-Aldrich Chemical Co., St Louis, MO), and counterstained on Harris hematoxylin (Richard-Allan Scientific, Kalamazo, MI). To ensure analysis of a representative portion of the liver, TUNEL assays were performed on 10-12 sections from the left, median, and caudate lobes of the liver from each of 4–8 mice for each group at each time point; labeled cells were counted in at least 40 portal tracts of each section at 400× magnification (Zeiss microscope, Carl Zeiss, Inc., Thornwood, NY) and expressed as the average number per 10 portal tracts. Extrahepatic bile ducts (3–4 ducts per group and per time point) underwent serial transverse sections from the portal hepatis (proximal) toward the duodenum (distal) on its entirety. At least 10 serial sections from each of the proximal, middle, and distal thirds of the extrahepatic ducts were subjected to TUNEL staining, and labeled cells were counted and expressed as an average number per section.

Cells undergoing apoptosis were also detected in extrahepatic bile ducts by immunostaining with anti-active caspase-3 antibodies as described previously.14 This assay was performed in paraffin sections of individual extrahepatic bile ducts that were cut in their entirety in a longitudinal fashion to facilitate counting and expression of all labeled cells per bile duct. In brief, serial sections underwent antigen exposure by heating in sodium citrate (pH 6.0), inhibition of endogenous peroxidase by washes in 0.3% hydrogen peroxide/methanol, permeabilization with 0.2% Triton X in phosphate-buffered saline, incubation with a blocking solution (goat anti-rabbit kit from Vector Laboratories, Burlingame, CA), and then incubation with rabbit anti-active caspase-3 antibody (Cell Signalling Technologies, Danvers, MA) at 4°C overnight. Following washes, the sections underwent incubation with peroxidase-conjugated anti-rabbit immunoglobulin G and an avidin:biotinylated enzyme complex (ABC) solution and a final staining with 3,3′-diaminobenzidine reagent (Vectastain peroxidase and ABC development kits and 3,3′-diaminobenzidine from Vector Laboratories). Labeled cells were quantified at 400× magnification with a Zeiss microscope (Carl Zeiss), and the total number of cells was expressed per bile duct.

To examine the types of cells undergoing apoptosis in the liver and along extrahepatic bile ducts, we first fixed cryostat sections with cold acetone for 10 minutes and incubated them with M.O.M. mouse immunoglobulin G blocking reagent (Vector Laboratories). Then, we applied anti-active caspase-3 antibody (Cell Signaling Technologies) to the sections overnight at 4°C, which was followed by anti–cytokeratin 7 (anti-CK7) antibody for 1 hour at room temperature (BD Pharmigen, San Jose, CA); both were diluted to 1:100 vol/vol. After washes in phosphate-buffered saline, the sections were incubated with a 1:200 dilution of fluorescein isothiocyanate (FITC)–conjugated anti-mouse and Texas Red–conjugated anti-rabbit antibodies (both from Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) for 1 hour. The sections were then fixed and mounted in 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories), with specific signals detected with a Zeiss Axiophot 2 microscope (Carl Zeiss) using FITC, DAPI, and Rhodamine filters.

Caspase and Cytokine Expression by Real-Time Polymerase Chain Reaction (PCR).

The messenger RNA (mRNA) expression for caspases 1 and 4, signal transducer and activator of transcription 1 (STAT-1), interferon regulatory factor 1 (IRF-1), TNFα, and tumor necrosis factor receptor 1 (TNFR-1) was quantified by real-time PCR in extrahepatic bile ducts from RRV-injected and saline-injected mice. Total RNA was isolated using an RNeasy kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Purity of RNA was determined on the basis of the 260/280 absorbance ratios and integrity of ribosomal RNAs in agarose gels. Total RNA was reverse-transcribed, and real-time PCR was performed with Taq polymerase and primers for caspases 1 and 4, STAT-1, IRF-1, TNFα, TNFR-1, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; Table 1) as described previously.3, 6, 15

Table 1. Oligonucleotide Primers for Apoptosis-Related and Proinflammatory Molecules Employed in Real-Time PCR That Used Complementary DNA Pools of Extrahepatic Bile Ducts from Neonatal Mice Challenged with RRV or Normal Saline
GenePrimer SequenceAnnealing Temperature (°C)
Caspase-1Forward: 5′-CCTGGTCTTGTGACTTGGAGG-3′51
Caspase-4Forward: 5′-ATTCATTCCCAGATGCCCAC-3′53

Cholangiocyte Viability and Apoptosis In Vitro.

mCL cells (0.5 × 106/well) were seeded in 12-well microtiter plates and treated with 100 ng/mL IFNγ and/or 10 ng/mL TNFα. In some experiments, cells were pretreated with the caspase inhibitor Mx1013 for 1 hour before incubation with cytokines. At specific time points, the cells underwent trypsin–ethylene diamine tetraacetic acid treatment, and viability was measured by trypan blue exclusion of cells in suspension with a hemacytometer. To assess DNA fragmentation, mCL cells were lysed with 50 mM trishydroxymethylaminomethane–HCl (pH 8.0), 150 mM NaCl, 5 mM Na–ethylene diamine tetraacetic acid, 1.2% sodium dodecyl sulfate, and 10 mg/mL proteinase K overnight at 55°C, and this was followed by the addition of 400 μL of 5N NaCl, centrifugation at 2500g for 15 minutes, ethanol precipitation, and treatment with RNase A (200ug/mL). The solution underwent phenol/chloroform extraction, and DNA was electrophoresed on 1.5% agarose gel and stained with ethidium bromide to detect internucleosomal cleavage. To quantify apoptosis by immunohistochemical staining, mCL cells were fixed with acetone and cold methanol and incubated with normal goat serum followed by anti-active caspase-3 antibody (Cell Signalling Technologies) and FITC-conjugated secondary antibody. Immunofluorescence was detected in a Zeiss fluorescent microscope. Lastly, apoptosis was also quantified by enzyme-linked immunosorbent assay–based histone detection according to the instructions suggested by the manufacturer (Roche Diagnostics, Indianapolis, IN) and by detection of the immunocomplex at the optical density at 450 mm (reference: 490 nm).

In Vivo Inhibition of Apoptosis.

To determine whether the inhibition of apoptosis decreases cholangiocyte injury in vivo, we administered the apoptosis inhibitor IDN-8050 (10 mg/kg) intraperitoneally to neonatal mice after RRV challenge. IDN-8050 is a pancaspase inhibitor with little effect on noncaspase proteases.16 The first dose was administered 8–12 hours after RRV inoculation, and new doses were administered every 12 hours for 5 days. For these experiments, control groups consisted of RRV inoculation with saline injections twice daily, whereas experimental groups were injected with IDN-8050 twice daily. All mice were monitored daily and sacrificed at several time points as described previously.

Statistical Analysis.

Values are expressed as mean ± standard deviation (SD), and differences between means were determined with the unpaired Student t test. However, because the sample sizes were insufficient for large sample approximation and the observations may not be normally distributed, their significance was tested by the permutation procedure following the algorithm suggested by Roff and Bentzen.17 For each test, we used 10,000 replications of permutations to compute the empirical P values, with significance set at P < 0.05.


RRV Induces Cellular Apoptosis in Portal Tracts and Extrahepatic Bile Ducts.

To determine whether apoptosis is a potential mechanism of duct injury in experimental biliary atresia, we first quantified TUNEL-labeled liver cells at different time points after RRV or saline challenge. We found that TUNEL-labeled cells were rare in livers following saline injection up to 18 days of age (Fig. 1A). After RRV challenge, labeled cells were inconspicuous within hepatic lobules but were present in most portal tracts at 6 days and increased to an average of ∼36 labeled cells per 10 portal tracts at 10 days (Fig. 1A). This single-staining technique did not allow for the precise identification of the cell types undergoing apoptosis in the portal tracts, but morphological features raised the possibility that labeled cells were both inflammatory cells and bile duct epithelial cells (Fig. 1B).

Figure 1.

Increased TUNEL labeling in livers after RRV challenge. (A) Number of TUNEL-positive cells in livers of neonatal Balb/c mice at different time points after intraperitoneal administration of RRV or normal saline (NS). The number of TUNEL-positive cells is expressed as the mean ± SD per 10 portal tracts for each group of mice; each group contains 4–8 mice per time point (*P = 0.043 for postinjection and control comparisons for each day examined). (B) Representative portal tract of a liver section stained by the TUNEL assay 10 days after RRV challenge; labeled cells display a brown color (measuring black bar on the right lower corner = 100 μm). The inset shows staining in ductular or nonparenchymal cells (magnification: 1000×).

Because the obstructive phenotype in experimental atresia is more localized to the extrahepatic bile ducts,3, 6, 10, 15, 18 we performed TUNEL staining on serial sections of ducts that were microdissected from groups of 3–4 mice sacrificed daily in the first week after RRV challenge and at days 10 and 14. We found no difference in the number of labeled cells between saline-challenged and RRV-challenged mice in the first 3 days. However, the number of TUNEL-labeled cells in bile ducts began to rise in RRV-challenged mice at day 4 and increased to ∼20-fold above saline controls at days 5 and 6 (Fig. 2A). In addition to this time-restricted surge in the number of labeled cells, we found that TUNEL labeling localized to the epithelial mucosa of extrahepatic bile ducts at 4 day (Fig. 2B) and was predominantly subepithelial at later time points when they populated the subepithelial compartment (Fig. 2C). Interestingly, microscopic examination of sections of bile ducts stained with hematoxylin/eosin did not show morphologic features typical of apoptosis in the first 3 days after RRV challenge. In contrast, the presence of nuclear condensation was inconspicuous at day 4 and easily seen at days 5–7 when the bile duct displayed extensive epithelial injury and luminal obstruction (data not shown). The temporal regulation and cell-specific onset of apoptosis suggested that apoptosis is a mechanism of injury of extrahepatic bile ducts.

Figure 2.

Increased TUNEL labeling in extrahepatic bile ducts (EHBDs) after RRV challenge. (A) Number (mean ± SD) of TUNEL-positive cells per cross section of EHBDs of neonatal Balb/c mice at different time points after intraperitoneal administration of RRV or normal saline (NS). Each group contains 3–4 mice per time point (*P < 0.003 for comparisons with controls at the depicted time points). TUNEL staining (brown) is seen (B) in cells lining the duct epithelium 3–4 days after RRV challenge or (C) in the subepithelial compartment 5 days after RRV challenge (magnification: 400×).

To directly test this possibility, we microdissected extrahepatic bile ducts from a new set of mice subjected to similar experimental challenges, embedded individual ducts, sectioned them in the longitudinal axis, and subjected all sections from each duct to immunostaining using anti-active caspase-3 antibodies. In order to evaluate the duct epithelium, we counted all labeled cells 3 days after RRV challenge (a time that might show cellular labeling without focal epithelial loss) and at 5 days (when there is a peak of TUNEL-labeled cells; Fig. 2B,C). We found higher numbers of active caspase-3–stained epithelial cells 3 days after RRV challenge than at 5 days (Fig. 3A). Notably, labeled cholangiocytes were present before the onset of morphologically evident injury to the epithelial lining (Fig. 3B). In contrast, apoptosis of cells within the subepithelial compartment was minimal at 3 days and rose substantially at 5 days; most of these cells displayed morphological features of inflammatory cells (not shown).

Figure 3.

Labeling of active caspase-3 in compartments of extrahepatic bile ducts (EHBDs) after RRV challenge. (A) Number (mean ± SD) of active caspase-3–stained cells per EHBD at 3 and 5 days after intraperitoneal administration of RRV soon after birth. Stained cells were counted in the entire extent of extrahepatic ducts and were expressed as stained biliary epithelial cells (BEC; staining in cells lining the duct epithelium) or subepithelial cells (SEC). n = 3 for EHBDs for each group and time point (*P < 0.05). (B) Representative longitudinal section of a segment of an EHBD 3 days after RRV challenge showing epithelial cells stained with active caspase-3 (arrows). The measuring black bar on the right lower corner equals 100 μm, and the inset is magnified at 1000×.

To more directly determine the cell types undergoing activation of caspase-3, we applied dual-immunostaining protocols to cryostat sections of livers and extrahepatic bile ducts from mice 3 and 6–7 days after RRV administration, respectively; the 3-day time point was chosen for extrahepatic bile ducts because later time points are associated with widespread loss of the duct epithelium. We found that antibodies to active caspase-3 produce specific signals within CK7-stained cells of intrahepatic (Fig. 4A–C) and extrahepatic (Fig. 4D–F) bile ducts. Active caspase-3–stained cells are also present in nonparenchymal cells of portal triads (Fig. 4A–C) and non–CK7-stained cells in the subepithelial compartment of extrahepatic bile ducts (Fig. 4G–I). Altogether, these data demonstrated a bimodal profile of apoptosis in extrahepatic bile ducts, in which caspase is activated in the epithelial lining at early phases after RRV challenge and is followed by a subsequent rise of apoptosis in inflammatory cells concurrent with the onset of jaundice.

Figure 4.

Cell-specific expression of active caspase-3 in the liver and extrahepatic bile duct. Incubation of cryostat sections of livers and extrahepatic bile ducts with anti-active caspase-3 and anti-CK7 antibodies shows the expression of active caspase-3 within CK7-stained cells in the epithelial lining of (A-C) intrahepatic (white arrows) and (A-F) extrahepatic (white arrows) bile ducts. Active caspase-3–stained cells in the liver are also detected (A-C) in nonparenchymal cells within expanded portal tracts (yellow arrows) and (G-I) in the subepithelial space of extrahepatic bile ducts (yellow arrows). Sections were harvested 3 days (extrahepatic bile ducts) and 7 days (liver) after RRV administration. (A,D,G) Active caspase-3 staining, (B,E,H) CK7 staining, and (C,F,I) overlay of active caspase-3 and CK7 images in a background of DAPI staining to depict individual cell nuclei (magnification: 400×).

Increased Expression of Apoptosis-Associated Genes in Extrahepatic Bile Ducts.

To explore potential apoptosis-inducing pathways triggered by RRV in the biliary system of neonatal mice, we quantified the mRNA expression of caspases 1 and 4, TNFα, TNFR-1, and IFNγ-activated genes STAT-1 and IRF-1 using real-time PCR. We found higher mRNA expression of caspases 1 and 4 in extrahepatic bile ducts after RRV in comparison with saline controls, which reached statistical significance first on day 5 (Fig. 5A,B). A similar profile was noted for TNFα but not for its receptor (Fig. 5C,D) and for STAT-1 and IRF-1 (Fig. 5E,F). For all of these molecules, mRNA expression also tended to be higher at other time points after RRV challenge but did not reach statistical significance. On the basis of the finding that apoptosis is limited to subsets of epithelial cells (which may represent a small portion of the total cellular elements present in the bile ducts), we used an in vitro system to directly investigate the functional relationship between the increased expression of TNFα and IFNγ and the induction of apoptosis of the bile duct epithelium.

Figure 5.

Changes in biliary mRNA expression for apoptosis-related and proinflammatory molecules after neonatal RRV challenge. mRNA expression for (A) caspase 1, (B) caspase 4, (C) TNFα, (D) TNFR-1, (E) STAT-1, and (F) IRF-1 in extrahepatic bile ducts harvested from mice at different days after administration of RRV or normal saline (NS). Mean (±SD) levels of mRNA expression were determined by real-time PCR and are expressed as a ratio to GAPDH. *P < 0.05; n = 3–5 bile ducts per group and at each time point.

Synergy Between TNFα and IFNγ Induces Apoptosis in Cholangiocytes.

To determine whether TNFα and IFNγ induce apoptosis in bile duct epithelium, we cultured mCL cells with TNFα or IFNγ alone or with both cytokines at similar concentrations.19–21 Culturing cholangiocytes with either TNFα or IFNγ for 24 hours had no obvious effect on cell viability; in contrast, culturing with TNFα+IFNγ decreased viability to 45% (Fig. 6A). The lower viability was due to apoptosis as shown by extensive DNA laddering (Fig. 6B), increased histone exposure in cell lysates by enzyme-linked immunosorbent assay (Fig. 6C), and in situ detection of active caspase-3 in cultured cholangiocytes (Fig. 6D). Because cholangiocytes are targeted by RRV in neonatal mice,3, 22 we measured histone exposure in mCL cells following infection with RRV (but no exogenous TNFα or IFNγ). We found that RRV challenge did not induce apoptosis as measured by histone exposure (data not shown). To directly demonstrate that TNFα+IFNγ decreased cholangiocyte viability through activation of apoptosis, we incubated cholangiocytes with the pancaspase inhibitor Mx1013, and this was followed by culture with TNFα+IFNγ. Incubation with Mx1013 prevented apoptosis and maintained viability of cholangiocytes despite the presence of TNFα+IFNγ in the culture medium (Fig. 7A,B). These results clearly show that the induction of apoptosis in cholangiocytes by TNFα and IFNγ is caspase-dependent and requires the synergy of intracellular circuits that are operational only when TNFα-dependent and IFNγ-dependent pathways are activated simultaneously.

Figure 6.

Induction of apoptosis in cholangiocytes after culture with IFNγ and/or TNFα. Culture of the cholangiocyte cell line mCL with DMEM containing IFNγ (100 ng/mL) and TNFα (10 ng/mL) for 24 hours (A) decreased cell viability and (B) induced apoptosis as demonstrated by DNA fragmentation, (C) increased histone exposure, and (D) increased detection of active caspase-3 by immunostaining. The viability was measured by trypan blue exclusion and is expressed as a percentage of controls; the top panel in part D represents immunostaining of a control group in which mCL cells were cultured in DMEM without IFNγ or TNFα; results in parts A and C are expressed as mean ± SD. *P < 0.001 and **P = 0.008 for comparisons of each group against IFNγ+TNFα.

Figure 7.

Improved cholangiocyte viability by pancaspase inhibitor Mx1013. (A) Viability of the mCL cells cultured in DMEM with IFNγ (100 ng/mL) and TNFα (10 ng/mL) for 24 hours in the presence of 50–200 μM Mx1013. (B) Histone exposure (measured as absorbance units at 405–490) by mCL cells is shown after 12 hours of culture with IFNγ and/or TNFα in the presence of 50 μM Mx1013. Results are expressed as mean ± SD. *P < 0.01 for comparisons of each group against IFNγ+TNFα.

Decreased Injury to Bile Duct Epithelium by Administration of Pancaspase Inhibitor.

To investigate the extent to which caspase-dependent apoptosis of biliary epithelium contributes to the pathogenesis of experimental biliary atresia, we inoculated RRV into newborn Balb/c mice, and this was followed by twice daily intraperitoneal administration of the pancaspase inhibitor IDN-8050 for 5 days; control mice were age-matched mice that received saline followed by twice daily injection of IDN-8050 and a separate group of mice challenged with RRV and not subjected to treatment with the caspase inhibitor. We used IDN-8050 because the administration of even low doses of Mx1013 resulted in high mortality in saline-injected mice. The administration of IDN-8050 alone had no impact on the growth, appearance, or survival of neonatal mice during the 5 days of injection or thereafter. The administration of IDN-8050 improved short-term survival in RRV-challenged mice, with survival of 67% of mice (4 of 6 mice) 12 days after RRV challenge, a time point at which the survival of RRV-challenged controls (not treated with IDN-8050) was limited to 38% (5 of 13 mice), with the bile ducts of all nonsurviving mice showing the typical obstruction of extrahepatic bile ducts.

To determine whether the administration of IDN-8050 modified the temporal-spatial induction of apoptosis by RRV, we examined extrahepatic bile ducts from groups of experimental and control mice at 3, 5, and 7 days. Using immunostaining to detect active caspase-3 as a marker of apoptosis, we found that IDN-8050 administration decreased the number of bile duct epithelial cells undergoing apoptosis during the entire duration of twice daily doses (3–5 days) and resulted in a mild decrease in labeled cells in the subepithelial compartment at 5 days and a significant decrease at 7 days (Fig. 8A,B). Interestingly, bile ducts displayed a remarkable surge in the number of labeled cells 2 days after cessation of IDN-8050 treatment, which implied a rebound in apoptosis to potentially lead to onset or extension of biliary injury. To further investigate this possibility, we used hematoxylin/eosin-stained sections to examine the entire segment of extrahepatic bile ducts for the presence of epithelial injury. As reported previously in RRV-challenged mice,3, 15 duct injury began early in RRV-challenged mice and progressed to complete obstruction within 7 days; however, this course was modified by a delayed onset of injury and decreased incidence of luminal obstruction at 7 days when mice received IDN-8050 (Fig. 9).

Figure 8.

Decrease in bile duct apoptosis following administration of IDN-8050. Mean (±SD) number of active caspase-3–stained cells per extrahepatic bile duct (EHBD) at 3, 5, and 7 days after intraperitoneal administration of RRV soon after birth, with an experimental group also receiving twice daily administration of IDN-8050 intraperitoneally. Panel A depicts a significant decrease in the number of stained biliary epithelial cells (BEC; staining in cells lining the duct epithelium) during IDN-8050 treatment, whereas panel B depicts the number of stained subepithelial cells (SEC). The last day for IDN-8050 treatment was day 5 after RRV challenge (arrow). n = 3 for EHBDs for each group and time point; *P < 0.01 and **P < 0.01 for comparisons of IDN-treated and untreated groups or different time points of IDN-treated groups as depicted.

Figure 9.

Decrease in bile duct injury following administration of IDN-8050. Extrahepatic bile ducts were microdissected from mice challenged with RRV and an experimental group also receiving twice daily administration of IDN-8050 intraperitoneally. The entire ducts underwent serial sections, staining with hematoxylin/eosin, and scoring based on the extent of epithelial injury along the length of the duct or the presence of luminal obstruction. Filled or open circles represent individual mice.


Our findings provide evidence that induction of apoptosis by RRV infection soon after birth is a prominent mechanism of bile duct injury in experimental biliary atresia. The mechanism is under temporal and spatial regulation, with onset of epithelial apoptosis along the length of the biliary tree in early phases of injury. Interestingly, the earlier onset and peak of apoptosis within the extrahepatic components of this tree imply that the injury begins in extrahepatic bile ducts and extends to intrahepatic ducts as the disease progresses in neonatal mice. Exploring the potential molecular mechanisms driving apoptosis, we found that IFNγ and TNFα, produced in the biliary microenvironment, emerged as potential triggers of synergistic intracellular circuits that result in the activation of caspases and decreased viability of cholangiocytes. Notably, inhibition of caspases in vivo significantly decreased epithelial apoptosis and duct injury. Collectively, these data suggest that epithelial apoptosis is an important initiator of duct injury following viral challenge to the neonate and that it is mediated, at least in part, by the simultaneous expression of IFNγ and TNFα.

The role of apoptosis as a potential pathogenic mechanism of biliary atresia was first suggested by the report of increased TUNEL labeling in intrahepatic bile ducts of affected infants.8 Similar findings were reproduced in different studies, which reported an overexpression of the apoptosis promoter FasL in biliary cells that were also stained by the TUNEL assay9 and an inverse correlation between the expression of E-cadherin and TUNEL-labeled cells.23 Recognizing the relevance of these findings, we made use of the rotavirus-induced mouse model of experimental atresia to more directly explore the relationship between apoptosis and the onset of biliary injury, as well as the potential mechanisms triggering apoptosis.10, 11 Previous reports using this mouse model did not identify highly coordinated expression for FasL or E-cadherin in large-scale gene expression analyses in the liver and extrahepatic bile ducts but displayed a unique molecular signature containing soluble mediators (for example, granzymes A and B), intracellular circuits (for example, nuclear factor kappa B), and executioners (for example, caspases) of apoptosis.6, 7 In the current study, TUNEL-labeled and caspase-3–labeled cells were easily identified in portal tracts and extrahepatic bile ducts in early phases of experimental atresia (soon after rotavirus challenge) and during progression to luminal obstruction. Although the distribution of labeled cells resided along the epithelium of extrahepatic bile ducts within 3 days of viral challenge, the highest number of labeled cells occurred around the time of duct obstruction, and they were primarily populating the subepithelial compartment. The expression of active caspase-3 in this prominent cluster of inflammatory cells probably represents the native response to ultimately remove most of the expanded T cells via programmed cell death.24, 25 In contrast, the expression of active caspase-3 in duct epithelial cells at early phases following viral challenge (3 days) and immediately before the detection of extensive epithelial damage (4–5 days) supports a chief role for apoptosis in the pathogenesis of duct injury. Although cholangiocytes are primary targets of neonatal rotavirus infection,3, 22 we did not find evidence of apoptosis in cultured cholangiocytes following rotavirus challenge alone. Instead, the exposure of cholangiocytes to IFNγ and TNFα induced apoptosis, and this suggests that the early expression of extracellular ligands may constitute key triggers of epithelial injury.

The simultaneous requirement of IFNγ and TNFα to trigger apoptosis in cholangiocytes provides insight into how cytokines may work in concert to initiate epithelial injury. IFNγ alone has been shown to play a critical role in the tropism of lymphocytes to extrahepatic bile ducts following neonatal rotavirus challenge and in the control of the duct lumen obstruction by inflammatory cells.3 In contrast, the inhibition of TNFα-mediated signaling in vivo did not prevent the atresia phenotype in neonatal mice.26 TNFα is a pleiotropic cytokine that modulates cell hyperplasia or apoptosis in different cells lineages but may require other paracrine or autocrine elements to influence cell survival.27 Consistent with this principle, we found that the apoptosis-inducing properties of TNFα required costimulation of cholangiocytes with IFNγ. Because epithelial injury appears to occur in mice with a deficiency in TNFα-mediated signaling after neonatal rotavirus challenge,26 it is likely that other circuits may exist in parallel (in synergy with IFNγ or as an accessory pathway) to ultimately injure cholangiocytes. Despite the seemingly limited contribution of TNFα as a single molecule, the collective data underscore a dual role for IFNγ in the pathogenesis of bile duct injury: (1) as a cofactor that initiates epithelial injury by the coordinated induction of apoptosis in cholangiocytes with TNFα and (2) as a late regulator of duct injury by modulating the inflammatory obstruction of the duct lumen.

In summary, we have found an activation of apoptosis in intrahepatic and extrahepatic bile ducts that is restricted in time and anatomical compartment in a well-established mouse model of experimental biliary atresia. The activation of apoptosis in the duct epithelium precedes the onset of obvious cholangitis and can be recapitulated in vitro by the synergistic effects of IFNγ and TNFα. In this context, future studies defining the cellular sources of individual cytokines and the point(s) of crosstalk between their intracellular circuits will provide a more comprehensive understanding of the molecular mechanisms of injury to the neonatal bile duct. Whether activated by these two cytokines or by other paracrine or autocrine circuits, apoptosis is executed by caspases, as suggested by the restoration of cell viability in vitro and the decreased injury of extrahepatic bile ducts when the experiments were reproduced in the presence of a caspase inhibitor. Altogether, these data add experimental support to the proposed involvement of apoptosis in the pathogenesis of biliary atresia.


IDN-8050 was a gift from Dr. Richard Butt of Pfizer Global Research and Development, Sandwich, United Kingdom.