Cytokine-induced acceleration of cholangitis, polyglandular pathology and autoantibodies: an important mechanism in development of PBC?


Primary biliary cirrhosis (PBC) is defined by the presence of highly specific circulating autoantibodies that recognise mitochondrial and/or nuclear antigens (AMA, antimitochondrial antibodies; ANA, antinuclear antibodies), progressive chronic non-suppurative granulomatous cholangitis and frequently extrahepatic polyglandular disorder. Because of the presence of autoantibodies, apparent immune-mediated destruction of glandular tissue by auto-specific T lymphocytes and associations with other autoimmune conditions, PBC has conventionally been considered autoimmune in origin although precise aetiopathogenetic mechanisms remain unknown (1). A key objective during the past two decades has been development of an acceptable animal model with which to conduct investigations into both pathogenetic mechanisms and possible therapies for PBC; this has proven a difficult task.

Liver cirrhosis can be induced, both in animals and in man, by a range of infectious and toxic agents but to date none of the numerous animal models postulated as models for PBC has proven ideal. Simultaneous coexistence of major hallmark features of PBC, (AMA, ANA, biliary tract damage, polyglandular disorder) have not been concurrently represented in one individual model. Some models based on inoculation with autoantigen exhibit autoantibody responses but not hepatic or extrahepatic pathology (2); others show hepatic and extrahepatic glandular manifestations similar to those observed in PBC but not autoantibodies (3). In one recent model both AMA and hepatic lesions were observed (4) but the model has been criticised because Freunds adjuvant, which was used in the inoculum, can alone induce hepatic lesions (5). Despite problems in establishing an acceptable animal model of PBC, such studies have nonetheless conclusively demonstrated that AMA are not pathogenetic and that additional mechanisms are involved in generation of glandular lesions.

Most models of PBC to date have employed a single inductive agent (e.g. inoculation with autoantigen) but PBC is likely multi-factorial in origin. Solitary inducers are thus unlikely to produce the simultaneous expression of characteristic features that distinguish PBC from cirrhosis of other causes. Furthermore the known lengthy latent period between exposure to the inductive agent(s) and appearance of signs of PBC may impose restrictions on development of a suitable model in species whose life spans are relatively short. In the current issue of Liver International (p. 595) Okada et al. describe an adaptation to the female C57BL/6 murine strain in which some 60% of females have an inherent predisposition to spontaneously develop liver cirrhosis and polyglandular disorders at 18–24 months of age (normal life span being 24 months); anti-bile duct antibodies are present in 14% of affected animals (6, 7). Here Okada et al. have succeeded in inducing early onset of PBC-like cholangitis in female C57BL/6 mice at around 4 months of age by administering multiple injections of polyinosinic acid (poly I:C), a synthetic double stranded RNA known to induce interferon (IFN)-α. All sera of treated mice showed elevation of IFN-α 3 h after injection of poly I:C (for up to 16 weeks). A smaller sample of sera tested for a wider range of cytokines showed poly I:C induced increases in interleukin (IL)-10, IL-12, monocyte chemoattractant protein 1 (MCP-1), IFN-γ and IL-6. Significant mononuclear cell infiltration (CD4+, CD8α+, CD11b+, CD11c+) of the liver including portal tracts and bile ducts of all treated mice was observed after 8 weeks of treatment with the number of portal tracts infiltrated by mononuclear cells increasing subsequently up until 16 weeks. Increased serum alanine amino-transferase (ALT) and alkaline phosphatase (ALP) were observed as were circulating autoantibodies (confirmed by ELISA and immunocytochemistry). AMA with specificity for pyruvate dehydrogenase complex (PDC), E1 and E2 components (appropriate for PBC) were confirmed by immunoblotting and ANA (confirmed by immunocytochemistry) were observed from week 4 of administration of poly I:C. Extrahepatic manifestations including pancreatitis and salivitis were also observed. This model therefore apparently simultaneously fulfils multiple criteria required for a model of PBC, (e.g female dominance, presence of specific autoantibody, biliary tract lesions, extrahepatic glandular pathology). The authors suggest that accelerated induction of PBC-like cholangitis in this model is related to elevation of specific soluble cytokines by poly I:C. The study calls for reexamination of the extent of involvement of cytokine-mediated mechanisms in the generation of PBC.

Although the aetiopathogenesis of PBC remains uncertain, the majority of evidence supports the hypothesis that autoimmune mechanisms play a significant role. An important principle of autoimmune disease is the ability to induce lesions in an experimental animal by immunization with specific autoantigen and many attempts have been made to induce PBC in laboratory animals (including mice, rats, rabbits, guineapigs and primates) by inoculation with the major mitochondrial antigen in PBC, PDC-E2. In such studies an appropriate antibody response is induced but not other manifestations of PBC. The observation of circulating autoantibody with specificity for mitochondrial antigens associated with PBC in the current study is therefore surprising. Previous reports of spontaneous development of liver pathology in the C57BL/6 mouse have described anti-bile duct antibodies but not AMA (6, 7). Furthermore, previous models that were not based on inoculation with PDC-E2, for example graft versus host disease models, failed to elicit an appropriate autoantibody response. A conclusion drawn by Okada et al. is that increase in IFN (or other unidentified upregulated cytokine(s) is alone sufficient to induce a range of PBC-like changes including an autoantibody response. This surprising finding questions the mechanism by which auatoantibody arises in PBC. Conventional wisdom dictates that auto-sensitivity to an antigen must arise through breakage of tolerance following exposure to excess of that antigen or similar cross-reactive antigen. Jones et al. in their SJL/J model have conducted detailed studies of interplay between autologous and prokaryotic antigens in the generation of PBC, demonstrating a PBC-like condition in female SJL/J mice in which tolerance to PDC-E2 was broken by coinoculation with autologous and foreign (prokaryotic) PDC-E2 (4). Here Okada et al. propose a model in which an autoantibody response is apparently initiated independently of autoantigen and possibly based on cytokine disruption.

A number of proinflammatory cytokines are known to be elevated in the local portal tract microenvironment in PBC contributing to development and perpetuation of chronic inflammatory reaction around bile ducts and it is now thought that biliary epithelial cells (BEC) may actively participate in the inflammatory process (8). BEC respond to proinflammatory cytokines by expressing a range of immunologically important molecules including lymphocyte adhesion molecules, major histocompatibility complex (MHC) antigens and chemo-attractants (9). IFN-γ and IL-2 are predominant cytokines in the PBC liver (10) and BEC respond to IFN-γ by induction of MHC antigens and lymphocyte adhesion molecules (11) while Sharma et al. (12) demonstrated alteration in Il-1 production in PBC. Cytokines have also been shown to influence levels of specific autoantigens in cells. The size and number of nuclear dots (SP100 nuclear autoantigen) was increased in HEp2 cells exposed to IFN-α, -β and -γ (13) and in a preliminary report IL1 (but not IFN-γ or -α) was found to influence the amount of PDC-E2 in BEC (14). These data suggest that cytokines are capable of upregulating autoantigen on targets and that this may then lead to breakage of immunological tolerance.

Positive AMA precedes development of liver pathology in PBC by many years. Extended asymptomatic periods of up to 20 years between initial detection of AMA in serum and emergence of bile duct lesions have demonstrated (15). In the current study, Okada et al. have detected AMA in a only a minority of mice after 4 weeks of poly I:C treatment and even after eight weeks of treatment only 27.8% of mice exhibited positive AMA whereas all animals showed evidence of mononuclear cell infiltration of the liver at this stage. Thus although in common with PBC in humans, not all mice became AMA positive (87.5% AMA positivity after 24 weeks), in contrast to observations in humans, positive AMA appeared not precede PBC-like lesions in the livers of C57BL/6 mice. Furthermore, anomalous distribution and amount of PDC-E2 autoantigens observed in BEC targets in liver biopsies of patients with PBC (16, 17) was not reported to be present on bile ducts in the C57BL/6 model. Thus the sequence of events appears to differ between human PBC and C57BL/6, poly I:C-induced cholangitis. Nonetheless, the possibility of cytokine induction of PBC-like features remains intriguing. Th1 cytokines including IFNγ predominate over Th2 in the PBC liver but the mechanism for such upregulation is unknown. IFNs are synthesised by a number of immune and non-immune cells as a response to viral infection, can inhibit virus replication in infected cells and contribute to inflammatory responses by stimulating expression of MHC and leucocyte adhesion molecules. In view of recent reports of possible viral involvement in the multi-factorial process that leads to the development of PBC (18) further studies of the possible role of IFN, in induction of PBC characteristics would appear worthwhile.


I am grateful to Prof. James Neuberger for proof reading and helpful suggestions.