Primary biliary cirrhosis (PBC) is a progressive cholestatic disease of unknown etiology, which affects mostly middle aged women and is characterized by immune-mediated destruction of intrahepatic bile ducts, hepatic fibrosis, cirrhosis, and eventual liver failure.1 PBC exhibits several features characteristic of an autoimmune disease2: the uniform presence of the hallmark antimitochondrial antibody (AMA) directed against the pyruvate dehydrogenase complex (PDC)-E2, homogeneity of clinical manifestation and an association with other autoimmune diseases, the presence of autoreactive T cells in portal tracts, and the specific immune targeting of cholangiocytes. However, some features such as a lack of human leukocyte antigen bias in PBC, ineffectiveness of immunosuppressive therapy, and the presence of noncaseating granulomas are at odds with autoimmune mechanisms and point to a more complex multistep pathophysiology, possibly involving environmental and/or microbial triggers.3
No cure outside of liver transplantation exists for PBC. The only established treatment for PBC currently is ursodeoxycholic acid, which improves symptoms and serum biochemistry but only delays eventual liver failure and the need for transplantation.4 Development of a reliable, small, animal model with high fidelity to human disease is indispensable to study the pathophysiological mechanism of PBC and to test urgently needed novel therapeutic approaches. Multiple animal models of PBC have been proposed in the last decade, most notably: NOD.c3c4, dominant negative transforming growth factor β receptor type II mouse (dnTGFβRII), 2-octynoic acid-bovine serum albumin immunization (2OA-BSA), Novosphingobium aromaticivorans infection, and Ae2a,b−/− mouse.5 All of these models recapitulate some important features of PBC, including the hallmark AMA, but none demonstrate progressive fibrosis or cirrhosis, arguably the central clinical feature that determines morbidity, mortality, and the need for liver transplantation in PBC patients. This has led to an international quest to develop or optimize an animal PBC model that will have histological features of portal fibrosis.
In an elegant work published in this issue of HEPATOLOGY,6 Tsuda and colleagues from the Gershwin lab created a double-mutant mouse by crossing the dnTGFβRII mouse,7 a well- characterized mouse model of multiorgan autoimmune disease, with a knockout mouse lacking interleukin (IL)-12p35. This allowed investigators to study how deficiency of both IL-12 and IL-35, heterodimeric molecules that share the IL-12p35 subunit, affects the PBC-like phenotype in 12- and 24-week-old dnTGFβRII mice. The dnTGFβRII mice develop both colitis and cholangiopathy resembling PBC with circulating AMA.8 The IL-12p35−/−;dnTGFβRI double mutants were completely protected from colitis, whereas the cholangitis was only delayed, and once established was associated with the presence of periportal fibrosis in half of the 24-week-old mice. This group had previously crossed the same dnTGFβRII mouse with the p40−/− mouse, a functional knockout for both IL-12 and IL-23. These double mutants, which were used for phenotyping comparison in this study, were protected from both colitis and liver disease, suggesting liver-specific effects of p35 versus p40 deficiency. That the IL-12p40−/−;dnTGFβII double mutants do not develop cholangitis, whereas the IL-12p35−/−;dnTGFβII mice do, implies 2 possible explanations: either the lack of IL-12 in the presence of intact IL-23, or the loss of IL-35 (a recently discovered inhibitory cytokine that mediates regulatory T-cell function9) might be responsible for the fibrotic phenotype in IL-12p35−/−;dnTGFβRII mice.10
Understanding how particular cytokines in the IL-12 family modulate PBC-like disease in dnTGFβRII and other mouse models will immediately suggest a clinically promising drug target. In further studies characterizing this model, it will be important to dissect the relative contribution of IL-12, -23, and -35 in driving a PBC-like phenotype, and specifically the associated fibrotic response, using alternative approaches like exogenous cytokine administration or cytokine-neutralizing antibodies rather than genetic deletion. Although the interesting phenotype in IL-12p35−/−;dnTGFβRII provides an important advance in understanding the biology of the IL-12 family and warrants further studies in PBC, the utility of the IL-12p35−/−;dnTGFβRII mouse as a model to test novel therapies might be diminished until there is solid proof that PBC patients have deficient IL-12 signaling. In fact, based on the strong genetic association of IL-12 variants in PBC patients,11 and the well-established functional role of IL-12 in promoting autoimmunity in humans and mice via induction of Th1/Th17 responses,12 the opposite might be true. Further studies are necessary to explore the unclear role of IL-35 in human PBC.
The authors went on to characterize inflammatory infiltrates in IL-12p35−/−;dnTGFβRII mouse livers by fluorescence activated cell sorting analysis and cytokine secretion. They found that p35 deficiency led to an increase in hepatic CD8+ T cells, as well as a significant shift from Th1 to Th17 (IL-17) bias on stimulation of liver mononuclear cells compared with the p40−/− double mutants and parental dnTGFβRII. Of note, the Th17 (IL-17) pathway has been recently implicated as a potential driver of fibrogenesis in chronic liver diseases in several descriptive studies.13 In animal models of PBC, only a single prior study reported the significant presence of fibrosis (77% of mice), induced by activation of invariant natural killer T (iNKT) cells using a-galactosylceramide in the 2OA-BSA model.14 Although the fibrosis was not characterized in much detail, this observation is particularly intriguing because the immune mechanism by which N aromaticivorans infection leads to PBC-like disease in mice is iNKT dependent.15
Development of an animal model for a disease of unknown etiology is challenging, forcing investigators to rely solely on the presence of histological lesions and other disease-specific features (e.g., AMA), whereas the driving mechanism of disease is only assumed to be similar to the human correlate. Mouse is the preferred small animal model due to breeding and maintenance costs, and availability of genetically modified animals.16 However, the biggest obstacle in establishing a good mouse model of fibrosing liver disease, which by definition should be robust and progressive, is the rarely acknowledged fact that mice, as a species, are quite resistant to liver fibrosis (compared to rats). Thus, the fibrotic process in, for example, the popular hepatotoxin-induced models is very slow, and typically does not lead to cirrhosis even in extended experiments.17 In addition, fibrosis susceptibility is strongly influenced by genetic background and varies widely between inbred mouse strains.18 In some instances, such resistance can be successfully overcome by optimization of induction protocols19 and careful selection of a fibrosis-susceptible background. Most mouse models of PBC were created in C57Bl6 mice, which are relatively fibrosis resistant. Whether the popularity of this strain solely determined such choice is unclear, but this fact alone may be responsible for the failure to achieve a significant fibrotic phenotype in multiple models despite liver injury and inflammation. In the absence of systematic research into how far genetic background influences PBC-like phenotype, it seems reasonable to ask whether simply backcrossing the dnTGFβRII mutation on a fibrosis-susceptible genetic background, such as BALBc, may suffice to obtain a fibrotic phenotype.
Is fibrosis in the IL-12p35−/−;dnTGFβRII mouse model sufficiently robust and progressive to allow for the detection of antifibrotic efficacy of novel therapies? Perhaps even more importantly, is the driving force of fibrogenesis, especially given the nonphysiological absence of p35, sufficiently similar to that in PBC patients? Tsuda et al. employed the unbiased, quantitative biochemical hydroxyproline assay (the gold standard of fibrosis assessment) to measure the degree of fibrosis in this model (approximately 200 μg of hydroxyproline/gram of liver reported, or a 1.5× increase compared to controls at 24 weeks), and interpreted differences as highly significant. However, for preclinical studies that measure fibrosis as a drug efficacy readout, the scale of fibrotic changes indicates that fibrosis in this model might be too mild and progression too slow to detect any efficacy in real-life settings. Most commonly used models of sclerosing cholangiopathies that were validated for antifibrotic drug testing, for example Mdr2−/− or 3,5-diethoxycarbonyl-1,4-dihydrocollidine feeding,20, 21 develop much more robust fibrosis in a shorter time (on the order of 400-500 μg of hydroxyproline/gram of liver, or a 4- to 5-fold increase compared with nonfibrotic controls). Fibrosis in the IL-12p35−/−;dnTGFβRII mouse was also analyzed morphometrically in a single region of interest per animal, but this methodology makes it difficult to assess how frequent fibrotic lesions were present throughout the liver. The more commonly accepted analysis of randomly chosen multiple portal tracts (≥10 per mouse) would have been more instructive, especially in the typically heterogeneous fibrotic lesions in biliary-type fibrosis, which is associated with a much higher biopsy sampling error compared with, for example, chronic hepatitis.22 It is also impossible to determine from this study if fibrosis in IL-12p35−/−;dnTGFβRII mice progresses further after it is detected in the 24-week-old mouse; a more detailed longitudinal characterization of the fibrotic process is warranted. Further mechanistic studies should also determine the effector cell responsible for driving the fibrogenic response in this model (immune cells, activated cholangiocytes, or myofibroblasts), and whether the resultant fibrosis is due to increased fibrotic matrix production or impaired matrix removal.
The findings in this intriguing study shed new light on the role of cytokines of the IL-12 family in PBC and are an important step in deciphering the pathophysiology of this enigmatic disease. The interesting phenotype of IL-12p35−/−;dnTGFβRII mouse builds on prior work in PBC models, informs direction for further research, and poses many new questions. There is hope for these questions to be addressed in follow-up studies that will clarify the complex pathogenesis of PBC and suggest novel therapeutic targets.