I am honored by the invitation to review recent advances in biliary atresia for The William F. Balistreri Festschrift. Dr. Balistreri (Bill)'s curiosity about the developmental rules that govern the adaptation of the neonatal liver to the extrauterine environment has served as inspiration to many of his trainees in Cincinnati, including those like myself who chose to study mechanisms of cholestasis and developmental biology of the liver. His scientific curiosity was heightened by the fundamental biological needs of the developing and mature liver, such as synthesis and export bile acids. Fueled by a passion for new discoveries and by a remarkable ability to stay up-to-date with scientific findings, Bill was able to forecast key areas of research that has driven the field forward. For example, he proposed nearly 20 years ago that "…familial forms of cholestasis, either idiopathic neonatal hepatitis or instances of intrahepatic cholestasis… would seem to be genetic diseases due to specific defects in the bile acid synthetic pathway or hepatic excretory process…" (1). Several lines of investigation now fully support this concept, as recently reviewed in a publication of a symposium on intrahepatic cholestasis (2). His eloquent writing on "saving livers while saving lives" highlights his enthusiasm for non-transplant options for the treatment of metabolic liver diseases (3), and represents editorial comments that will not age.
Obviously biliary atresia did not go unnoticed by Bill. Analyzing the key concepts of the disease in 1976, he proposed that "…extrahepatic biliary atresia is the result of the sporadic occurrence of a virus-induced, progressive obliteration of the extrahepatic bile ducts, with variable intrahepatic bile duct injury" (1). This has been a harder concept to tackle at the mechanistic level due to the multifactorial basis of disease (4). Today, there is growing enthusiasm for research on biliary atresia, which is fueled by the availability of an experimental model of disease and of powerful technologies that facilitate the broad screening for relevant functional signatures in the liver and biliary tract of affected patients.
PATHOGENESIS OF BILIARY ATRESIA: WHERE WE STOOD AT THE TURN OF THE CENTURY
A particularly challenging area of research is pathogenesis of biliary atresia owing to the multifactorial basis of disease. Until recently, theoretical considerations of the pathogenesis of biliary atresia were based largely on epidemiologic and clinical features, predisposing genetic factors, and pace of disease progression. Based on these observations, five mechanisms were proposed to be involved: 1) defect in morphogenesis of the biliary tract, 2) defect in fetal/prenatal circulation, 3) environmental toxin exposure, 4) viral infection, and 5) immunologic/inflammatory dysregulation (Table 1) (4).
Pathogenic mechanisms of disease are an important focus of ongoing research by several investigators, as supported by a focus on pathogenesis in 33 out of 184 recent publications on biliary atresia (http://www.ncbi.nlm.nih.gov/entrez/, July 2005). Instead of reviewing the key findings from these publications, I will summarize the recent data produced by the Cincinnati liver research group studying biliary atresia. I am a member of this research group, and have benefited greatly from Bill's teachings and vision. For the past three years, I have taken the liberty of borrowing elements of his vision and designed a research strategy to explore the biological mysteries of Billiary Atresia. The extra 'l" in "Billiary" is in keeping with the theme for this Festschrift, brings Bill to our daily research discussions, and simply shows our gratitude for his friendship and continued mentorship.
INVESTIGATING THE HEPATIC TRANSCRIPTOME IN CHILDREN WITH BILIARY ATRESIA
The pathogenesis of biliary atresia is multifactorial and probably results from the interplay between environmental and genetic factors that triggers an inflammatory and fibrosing obliteration of the extrahepatic bile ducts (4). In initial studies to search for biological processes driving disease pathogenesis, we undertook a large-scale gene expression analysis to identify physiologically relevant genomic signatures of biliary atresia in the livers of affected infants. In these studies, we compared the levels of gene expression in the livers of children with biliary atresia with children with other causes of neonatal cholestasis at different phases of disease using high-density microarrays (GeneChips) (5). Analysis of over 12,000 genes using highly stringent statistical parameters revealed a unique transcriptional footprint for patients with biliary atresia. This footprint was age-specific (<4 months of age), clustered at the time of diagnosis, and had a functional profile that implied a differential activation of hepatic lymphocytes. The profile included the downregulation of immunoglobulin genes and an overexpression of key proinflammatory genes, such as osteopontin and interferon-gamma (IFNγ).
To assess whether these findings resulted from decreased infiltration of B lymphocytes in the livers of infants with biliary atresia, we examined liver sections of subjects with biliary atresia and other causes of neonatal cholestasis (disease controls). Samples from both groups of patients had similar degrees of portal expansion and inflammation, with foci of extramedullary hematopoiesis observed in all samples in a similar fashion (5). Using immunohistochemical labeling with antibodies against CD20 and CD79a (B lymphocyte markers), we found no difference in the number of labeled cells. These findings suggested that the differential expression of several immunity/inflammation genes in biliary atresia was due to selective functional states of lymphocytes within the hepatic microenvironment in early stages of disease. Equally notable was the simultaneous increase in the expression of proinflammatory genes at the time of diagnosis, with a persistent increase in the expression of IFNγ at later phases of disease. This implied that the proinflamatory, Th1-like commitment of lymphocytes begins during the early stage of disease and continues through the stages of advanced fibrosis. Although the findings were intriguing, we could not discern whether the proinflammatory commitment of lymphocytes was directly related to pathogenesis of biliary atresia, or simply represented a non-specific inflammatory response to an injury to the biliary tract. Experiments in humans to address this type of question at a mechanistic level faced remarkable challenges, such as the difficulty in obtaining samples at the onset of biliary atresia, or from healthy individuals to serve as controls. Therefore, we took the challenge to the laboratory and designed mechanistic studies in a murine model of rotavirus-induced biliary atresia.
USING AN EXPERIMENTAL MODEL OF BILIARY ATRESIA TO IDENTIFY MECHANISMS OF DISEASE
Pioneering studies by other investigators in 1993 and 1997 demonstrated that infection of Balb/c mice with rhesus rotavirus (RRV) in the first 24 hours of life results in biliary obstruction and recapitulates the dramatic phenotype of biliary atresia. We established this model in our laboratory, and observed, at first hand, the generalized jaundice, acholic stools, and bilirubinuria that develop by the end of the first week of life following RRV challenge (6,7). We found that the onset of jaundice coincides with the expansion of portal triads by lymphocytic infiltrates and bile duct proliferation, with a typical obstruction of extrahepatic bile ducts by inflammatory cells (Fig. 1), and duct stenosis of variable length. We also found that RRV has a unique tropism to cholangiocytes of extrahepatic bile ducts and livers 3-7 days after viral inoculation. Interestingly, mice are able to clear the virus from the hepatobiliary system during the course of cholestasis and later phases of duct obstruction, as demonstrated by the lack of detectable expression of mRNA encoding viral proteins or live virus at 14 days (8).
Similar to the livers of children with biliary atresia, livers of mice challenged with RRV displayed infiltration of lymphocytes at the time of duct obstruction. Applying dual-stain flow-cytometric analysis, we found significant increases in the number of CD3/CD4+ and CD3/CD8+ lymphocytes in the liver, as well as the hepatic overexpression of murine Ifnγ and IL-12 genes when compared to saline-injected control mice. Based on the previous findings of increased expression of IFNγ in livers of children with biliary atresia and on the role of IFNγ as an effector of the Th1 phenotype, we hypothesized that Ifnγ has a regulatory role in the obstruction of bile ducts in the murine model of experimental biliary atresia. To directly test this hypothesis, we inoculated RRV into mice carrying the genetic inactivation of the Ifnγ gene (Ifnγ−/− mice). Injection of RRV into Ifnγ−/− mice resulted in the development of symptoms of cholestasis 3-5 days later, but the symptoms were short-lived, with resolution and long-term survival in ~80% of the mice. Analysis of extrahepatic bile ducts showed that the loss of Ifnγ did not prevent the onset of inflammation in the liver of extrahepatic bile ducts, but the degree of inflammation was significantly reduced as demonstrated by a decrease in the population of CD3/CD4+ and CD3/CD8+ lymphocytes in the liver. More notably, the lumen of extrahepatic bile ducts of Ifnγ−/− mice was not obstructed by inflammatory cells, despite the presence of lymphocytes in the ductal wall (Fig. 1). Interestingly, the daily administration of recombinant Ifnγ beginning 24 hours after RRV challenge restored the atresia phenotype, with complete obstruction of extrahepatic bile ducts by inflammatory cells (8).
Collectively, these data demonstrate that Ifnγ plays a critical role in the inflammatory obstruction of extrahepatic bile ducts in an experimental model of biliary atresia, and may constitute a therapeutic target to stop disease progression in children. These data also suggest that the pathogenic mechanisms of biliary atresia may obey a biological continuum previously not recognized. The initiating events of this continuum, namely the immediate jaundice and inflammation, were not dependent on Ifnγ. In contrast, the progression to duct obliteration by lymphocytes and fibrosis was prevented by the loss of Ifnγ (Fig. 2).
MINING THE BILIARY TRANSCRIPTOME IN EXPERIMENTAL BILIARY ATRESIA
In addition to the opportunities to perform mechanistic studies of disease, the murine model of experimental biliary atresia also provides a unique access to the extrahepatic biliary tract at the onset and different phases of progression to bile duct obstruction. In order to be accepted as an adequate model to study pathogenic mechanisms of disease, this model must reproduce important clinical features that are exclusive to biliary atresia: 1) the onset of disease restricted to the neonatal period, and 2) an injury that targets primarily the biliary system. Both features are met by the murine model of RRV-induced experimental biliary atresia, as supported by unpublished experiments in which we inoculated RRV into Balb/c mice at different time points beginning at 7 days of age, and were unable to induce injury or obstruction in extrahepatic bile ducts. Therefore, we continued to use the experimental model in studies of pathogenesis of bile duct injury and obstruction.
Departing from the traditional experimental approach of liver-based studies to investigate potential pathogenic mechanisms of biliary atresia, we microdissected the extrahepatic bile duct and gall bladder en bloc from neonatal mice at 3, 7, and 14 days after RRV or saline injection within 24 hours of birth. We then used these samples to isolate RNA and hybridize against the Affymetrix MOE430 GeneChip, an oligonucleotide-based high-density microarray that displays the expression of ~41,000 gene products. After creating a data expression platform, we applied highly stringent statistical criteria to identify the genes in the biliary tract that are differentially expressed at early and later phases of bile duct injury and obstruction. We found a total of 310 genes that increased in the RRV group above saline-injected controls in at least one time point (9). Functional analysis of these genes showed a predominant activation of genes encoding immunity regulators, enzymes, and structural proteins throughout the study period, with a time-restricted expression of genes related to nucleic acid binding, transport, apoptosis, and signal transduction. These genes formed a highly select transcriptional footprint that will guide future studies exploring how individual gene or gene groups work in concert to regulate the onset of duct injury and progression to obstruction in experimental biliary atresia.
The Ifnγ signature in early stages of experimental atresia. Temporal and functional analyses of individual genes from the transcriptional footprint identified a predominant immunity network in early phases of biliary injury and obstruction. At the onset of injury (i.e., 3 days after RRV challenge), the bile ducts displayed high expression levels of genes encoding antigen-processing/presenting molecules, major histocompatibility complex class II-associated antigens, and genes that regulate the expression of interferons, such as interferon response factors (Irf)-7 and −9 and interferon-receptor 1-bound protein. This molecular signature of immunity genes was further amplified at the time of lumen obstruction by inflammatory cells (i.e., 7 days after RRV challenge), with the overexpression of genes involved in antigen presentation and the Th1 polarization of inflammatory cells. Specifically, these genes were IL2, tumor necrosis factor-alpha (TNFα)-1b, TNFα-inducible protein 2, and a gene group forming an Ifnγ network: Ifnγ-induced guanosine triphosphatases, Stat1, Ifnγ-induced chemokines (such as Cxcl9 and Cxcl10), and major histocompatibility complex II genes. These findings revealed a temporal activation of inducers of interferons in early phases of biliary injury, followed by the expression of an Ifnγ-driven molecular network that is known to polarize lymphocytes to a proinflammatory profile (Fig. 3). The molecular signature of immunity genes is in keeping with the findings of Ifnγ as a key pathogenic mechanism of biliary atresia, and gives a much broader view of the signals that work in cooperation to target the biliary system in response to a neonatal insult by RRV.
Beyond the Ifnγ network. In addition to the molecular network related to Ifnγ, the transcriptional footprint also identified the simultaneous activation of novel biologic processes at the onset of duct obstruction: complement cascade and apoptosis. Among the complement components, the overexpression of C3a receptor-1, C1qα and C1qβ is suggestive of the activation of the complement cascade during duct injury and obstruction. There was also a time-restricted activation of apoptosis genes, with increased expression of genes that either trigger (e.g.: granzymes A and B) or drive (e.g.: caspases) apoptosis. Although apoptosis has been found in bile duct cells of infants with biliary atresia (10), we found no previous reports of an association between abnormalities of the complement system and pathogenesis of biliary atresia in humans. Therefore, these findings provide the basis for future studies exploring how these processes may work in parallel with the Ifnγ-related circuits to regulate duct injury and obstruction.
A diverse profile at later phases of duct obstruction. Consistent with the importance of the careful staging of the pathogenesis of experimental biliary atresia, the number of immunity genes with differential levels of expression 14 days after the RRV challenge decreased when compared to other time points (day 3 = 28, day 7 = 65, day 14 = 9). At 14 days, the inflammatory infiltration of the duct lumen is replaced by extracellular matrix, which results in the formation of atretic segments. Despite these morphological findings, the analysis of the transcriptome did not show a significant increase in the number of genes regulating extracellular matrix production or clearance. Although the low number of genes may be due to a minimal representation of viable cells at this late phase of biliary obstruction, it is possible that the microdissection of extrahepatic bile ducts harvested from mice at 14 days after RRV may not have included a greater proportion of the narrowed atretic segments. Therefore, future studies exploring the molecular regulators of the late phases of experimental biliary atresia will require the use of high-resolution laser-capture microscopy to focus exclusively on atretic segments of the extrahepatic bile ducts.
SEARCHING FOR MOLECULAR BASIS OF CLINICAL PHENOTYPES
The combination of patient- and animal-based experiments offers a powerful strategy to discover molecular processes that may play key regulatory roles in disease pathogenesis. A greater understanding of the biological basis for disease phenotypes, however, will be possible only if one continues to rigorously interrogate the relationship(s) between patients' samples and the carefully defined (and matched) set of clinical variables. Therefore, we began this line of experiments by searching for molecular signatures that can discriminate the two clinical forms of biliary atresia: embryonic and perinatal. The embryonic form comprises the group of infants with biliary atresia who also have congenital non-hepatic malformations. This group of patients comprise ~20% of the cases and may present earlier with more severe liver disease. The remaining ~80% of patients have no non-hepatic anomalies and are classified as the perinatal form of biliary atresia (11). To determine if the clinical forms can be differentiated at the transcriptional level and to search for molecular mechanisms underlying the phenotypic differences, we performed a genome-wide comparative analysis using livers of patients with embryonic and perinatal forms of biliary atresia obtained at the time of diagnosis. For the analysis, infants were matched by age at diagnosis, degree of liver injury (as evidenced by serum levels of aminotransferases) and cholestasis (as measured by serum levels of conjugated bilirubin), and by the status of synthetic function (as indicated by levels of albumin and prothrombin time/international normalization ratio) (12).
Data filtering and two-way cluster analysis of the gene expression platform identified 230 genes with an expression profile that is highly distinctive of the embryonic and perinatal clinical phenotypes. Functionally, the profile did not reveal a single higher order function for a specific cell type; instead, it uncovered a coordinated expression of regulatory genes (DNA and RNA processing, imprinting, laterality, signal transduction, transcription regulation, and cell cycle control). These regulatory genes were predominantly represented in the embryonic form (45% of genes), with a unique pattern of expression of genes involved in chromatin integrity/function (SMARCA-1, RYBP, and HDAC3). Notably, we also found a uniform overexpression of five imprinted genes (IGF2, Peg3, PEG10, MEG3, and IPW) 1.8-2.9-fold above levels of infants with the perinatal form. In addition, a similar fold increase (1.8-3.5) in expression was also noted in comparison to three infants with other types of neonatal cholestasis. The expression of the two genes with highest fold increases, IGF2 and PEG10, may be closely regulated by HDAC3, a gene involved in regulation of chromatin structure through deacetylation of histones (13). Collectively, the coordinate expression of genes regulating transcription through modifications of chromatin structure/function and of imprinted genes points to a potential role of epigenetic factors in modulating the phenotypes of biliary atresia. Epigenetic factors have been shown to be important clinical determinants in complex diseases, such as colorectal cancer (14). Validation of this claim in biliary atresia will require formal studies of loss of imprinting for index genes, such as IGF2, as well as patterns of DNA methylation in affected infants.
SUMMATION OF KEY CONCEPTS
- Biliary atresia is the most common cause of pathologic jaundice in young infants and results from the obliteration of the extrahepatic bile ducts by an inflammatory and fibro-obliterative process. Despite the multifactorial basis of disease, recent studies have begun dissecting the biological processes underlying the pathogenesis of bile duct injury and obstruction.
- Livers of infants with biliary atresia have molecular signatures of proinflammatory genes, with a predominant overexpression of the INFγ gene.
- Inactivation of the Ifnγ gene in mice completely prevents the inflammatory obstruction of extrahepatic bile ducts in an experimental model of RRV-induced biliary atresia.
- Investigation of the biliary transcriptome in mice uncovered a rich, Ifnγ-dependent molecular circuit, and identified apoptosis and activation of the complement cascade as processes that may work in synergism to promote duct obstruction.
- Comparative analysis of the hepatic transcriptome in children with the embryonic and perinatal forms of biliary atresia produced a preliminary transcriptional profile that differentiates the clinical forms, and identified a functional signature that is consistent with a potential role of epigenetic forces in determining the embryonic phenotype of biliary atresia.
This work was supported by the NIH grant DK-64008 (to J.A.B.) and R24 DK 064403, the Cincinnati DDRDC. The author is grateful to the members of the Laboratory of Liver Research for their inquisitive mind, hard work, and enthusiasm for studies of biliary atresia, and to Dr. William Balistreri for his past, present, and future mentorship.
1. Balistreri WF. Neonatal cholestasis: lessons from the past, issues for the future. Semin Liver Dis
2. Balistreri WF, Bezerra JA, Jansen P, et al. Intrahepatic cholestasis: summary of an American Association for the Study of Liver Diseases single-topic conference. Hepatology
3. Balistreri WF. Nontransplant options for the treatment of metabolic liver disease: saving livers while saving lives. Hepatology
4. Balistreri WF, Grand R, Hoofnagle JH,et al. Biliary atresia
: current concepts and research directions. Summary of a symposium. Hepatology
5. Bezerra JA, Tiao G, Ryckman FC, et al. Genetic induction of proinflammatory immunity
in children with biliary atresia
6. Riepenhoff-Talty M, Schaekel K, Clark HF, et al. Group A rotaviruses produce extrahepatic biliary obstruction in orally inoculated newborn mice. Pediatr Res
7. Petersen C, Biermanns D, Kuske M, et al. New aspects in a murine model for extrahepatic biliary atresia
. J Pediatr Surg
8. Shivakumar P, Campbell KM, Sabla GE, et al. Obstruction of extrahepatic bile ducts by lymphocytes is regulated by IFN-gamma in experimental biliary atresia
. J Clin Invest
9. Carvalho E, Liu C, Shivakumar P, et al. Analysis of the Biliary Transcriptome in Experimental Biliary Atresia
10. Liu C, Chiu J-H, Chin T, et al. Expression of fas ligand on bile ductule epithelium in biliary atresia
- a poor prognostic factor. J Pediatr Surg
11. Carmi R, Magee CA, Neill CA, et al. Extrahepatic biliary atresia
and associated anomalies: etiologic heterogeneity suggested by distinctive patterns of associations. Am J Med Genet
12. Zhang DZ, Sabla G, Shivakumar P, et al. Coordinate expression of regulatory genes differentiates embryonic and perinatal forms of biliary atresia
13. Lynch CA, Tycko B, Bestor TH, et al. Reactivation of a silenced H19 gene in human rhabdomyosarcoma by demethylation of DNA but not by histone hyperacetylation. Mol Cancer
14. Cui H, Cruz-Correa M, Giardiello FM, et al. Loss of IGF2 imprinting: a potential marker of colorectal cancer risk. Science