Clin Liver Dis 12 (2008) 305321

Pathogenesis of Primary Biliary Cirrhosis

David E.J. Jones, MD, PhD
Liver Research Group, Institute of Cellular Medicine, The Medical School, Newcastle University, 4th Floor, William Leach Building, Framlington Place, Newcastle-upon-Tyne, NE2 4HH, United Kingdom

Primary biliary cirrhosis as a landmark disease Primary biliary cirrhosis (PBC) has acted, in many ways, as a landmark disease for establishing concepts regarding the pathophysiology of conditions in which autoimmune phenomena are prominent. PBC was one of the rst diseases in which the presence of serum autoantibodies was recognized [1]. Likewise, it was one of the rst diseases in which the identity of key autoantigens was established [2]. Understanding of the pathogenesis of PBC and, in particular, the precise role played by autoreactive immune responses remains, however, incomplete, with increasingly contradictory observations made in recent years. PBC is continuing, it seems, to act as a landmark disease by demonstrating that the pathogenesis of apparently autoimmune diseases is signicantly more complex than the initially simplistic models might have suggested [3]. PBC acts as a landmark disease in one other highly clinically relevant way by demonstrating that detailed understanding of autoimmune phenomena does not, as perhaps naively originally believed, lead inevitably to improvement in therapy. This lesson is starting to be learnt in other autoimmune diseases. The aim of this article is to present a contemporary perspective of the pathogenesis of PBC, to discuss the currently proposed models, and to draw realistic conclusions regarding where the current understanding of disease pathogenesis is likely to lead in terms of therapy in the near future. Key clinical observations that may guide understanding of primary biliary cirrhosis pathogenesis When considering the individual pathogenetic models for PBC, it is useful to remember the key clinical observations that characterize the disease (and which are, on occasion, overlooked in developing the concepts of
E-mail address: d.e.j.jones@ncl.ac.uk 1089-3261/08/$ - see front matter 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.cld.2008.02.004 liver.theclinics.com

306

JONES

pathogenesis). For any particular model to be valid, it must be able to explain, or at least be able to account for, each of these key clinical observations (Tables 13). It is sobering to realize that, in reality, none of the pathogenetic models proposed for PBC is able to truly account for all these key clinical observations. Autoantibody Autoantibodies frequently are seen in patients who have PBC and, typically, are present at very high titers. The universal nding of autoantibodies and associated autoreactive phenomena (discussed later) in patients who have PBC demonstrates the presence of self-directed immune responses in this disease. The key question is whether or not the presence of such responses represents autoreactivity (the presence of immune responses directed at self-antigens) or true autoimmunity (autoreactivity resulting directly in target cell damage and the induction of pathology). Any model based on autoreactive antibody responses playing a key role must be able to account for the development of disease in the small, but clearly delineated, subgroup of patients who have autoantibody-negative disease (described in detail in the article by Muratori and colleagues elsewhere in this issue). Elevated serum IgM Most patients who have PBC exhibit polyclonal elevation in serum IgM, which is typically not directed at mitochondrial or nuclear antigens [4]. At face value this phenomenon is suggestive of polyclonal activation of the B-cell compartment, with an associated failure of isotype switching. It is perhaps surprising that, despite being a robust and unusual clinical observation, this phenomenon, until recently, has been little studied. Recent data are supportive of this phenomenon representing aberrant B-cell activation [5]. Gender distribution of disease PBC is a disease predominantly, but not exclusively, aecting women. Pathogenetic models for PBC must explain this predominance while not invoking processes where female gender is obligatory (for example pregnancy). Heritability and disease associations PBC, and the disease-independent presence of antimitochondrial antibody (AMA), occurs more frequently in close relatives of patients who have PBC than in controls, and is associated with an increased incidence of autoimmune disease of other types (most particularly in non-organ specic autoimmune disease) in patients who have PBC and their rstdegree relatives [68]. These observations and the nding that PBC exhibits

PBC PATHOGENESIS

307

a higher concordance rate in monozygotic than dizygotic twin pairs point to a genetic component to disease susceptibility and suggest that this is expressed through genes that regulate the immune response [9]. Many of the familial risk/genetic studies of PBC have failed to distinguish, however, between true genetic risk and shared environmental exposure. Patient age Patients who have PBC typically present over age 40, an observation compatible with a signicant, additional, acquired component to disease pathogenesis that is encountered in adult life, is cumulative in nature with threshold levels not achieved until adult life, or is so slow in its eects that the phenotype of disease is achieved only in adult life. Disease recurrence post transplant Recurrence of PBC can occur after liver transplantation [10]. There seems to be no association between degree of HLA matching between donor and recipient (donors and recipients normally are not matched for HLA in liver transplantation, resulting in a random degree of coincidental matching in transplant populations) and risk for disease recurrence [10]. This argues that the disease, at least in its recurrent form, cannot be caused by a process that is directly HLA restricted. Tissue tropism of disease PBC principally is a disease of the small intrahepatic bile ducts, with loss of the biliary epithelial cells (BEC) lining these ducts underpinning much of the subsequent cholestatic damage. PBC, however, is not restricted to the liver as abnormality of salivary and lachrymal glands (with associated cell phenotypic change similar to that seen in BEC) also occurs [11].

Primary biliary cirrhosis as an autoimmune disease? PBC is associated with the almost universal development, and typically high level, of autoreactive immune responses [12]. The key outstanding issues regarding PBC as a putative autoimmune disease are the mechanism responsible for breakdown of tolerance to what are all highly conserved selfantigens and the nature of the role played by these responses (if any) in the development of target cell damage. The best characterized and most prevalent autoantibodies are directed against the 2-oxo-acid dehydrogenase family of multienzyme complexes (2-OADC) located on the inner mitochondrial membrane, all of which play key roles in energy homeostasis within the cell [13]. The predominant response is directed at the E2 and E3 binding protein (E3BP) components of pyruvate dehydrogenase complex (PDC). Antibodies

308

JONES

Table 1 Autoimmune model for primary biliary cirrhosis pathogenesis and the key clinical observations Clinical observation Autoantibody presence Reconcilable aspects (a) Clear link to autoimmunity (b) High titer suggests dominant eect (c) Emerging data regarding intracellular actions (d) Potential interaction with surface PDC resulting in ADCC Non-reconcilable/ unknown aspects (a) Development of AMA in non-PBC infectious disease settings in the apparent absence of disease development (b) Lack of disease development in animals undergoing AMA passive transfer and induced to express AMA (c) Autoantibody negative PBC is well described (a) High degree of isotype switching of the AMA response seen in most patients

Elevated IgM

Gender distribution Heritability and disease associations

(a) Clear link to abnormality in B-cell biology (b) Apparent role for TLR ligands (able to up-regulate antigen presentation in stimulating polyclonal IgM response) (c) High IgM component of the anti-LA immune response seen in very early disease (a) Female predominance common in autoimmunity (a) Familial risk a characteristic feature of autoimmune disease (b) Increased personal and familial risk of autoimmunity a feature of autoimmune disease (c) Genetic associations reported (a) Older age range than typical could represent exposure to disease trigger at later age or cumulative exposure to a rarer but retained factor until a trigger level of exposure is attained (b) Older age range could reect role for sex hormone triggering (eg, pregnancy or menopause) (a) Autoantibody remains post transplant suggesting ongoing process

(a) Eect more marked than for other autoimmune disease (b) Mechanism unknown (a) HLA association weaker than for other autoimmune diseases (b) Most studies fail to distinguish between genetic risk and shared environmental risk (ie, families sharing space) in terms of heritability of PBC (a) Age range atypical for autoimmune disease

Patient age

Post-transplant disease recurrence

(a) Recurrence across HLA boundaries argues against HLA-restricted T-cell responses causing recurrent disease (continued on next page)

PBC PATHOGENESIS

309

Table 1 (continued ) Clinical observation Tissue tropism of disease Reconcilable aspects (a) Secretory epithelial priming (antigen exposure after biliary excretion would result in addressin targeting of T-cell response to originating tissue type) (b) Intracellular eect of secretory IgA fully compatible Non-reconcilable/ unknown aspects (a) Autoantigens in PBC are universally expressed in all cell types

directed at PDC inhibit the function of the complex in vitro [14]. The mechanistic basis of this inhibitory eect and the apparent cross-reactivity of the antibody responses between PDC-E2 and PDC-E3BP [15] was determined with the observation that a lipoic acid (LA) cofactor, attached to PDC-E2, E3BP, and the other 2-OADC-E2 components within the inner mitochondrial space, represents a key shared component of the B-cell autoepitope within all these enzymes [1517]; antibodies reactive with free LA are postulated as representing the original immune motif of PBC, with reactivity against LA-containing proteins, such as PDC-E2 and E3BP, arising as a result of secondary epitope spreading [18]. More recently, CD4 T-cells directed against PDC-E2 were found to be present in the peripheral repertoire in patients who have PBC but not normal controls [1921] and enriched within the liver, suggesting localized recruitment [21]. Liver-enriched CD8 cells reactive with PDC-E2 also are described, again uniquely in patients who have PBC, and demonstrated to exhibit cytotoxic capability against PDC-E2-pulsed target cells [22]. The HLA class I and class IIrestricted epitopes within PDC-E2 are described in PBC, and both span the key LA-binding residue [19,23]. Although the nature of the autoreactive immune responses directed against PDC and related complexes is described in some detail, the role played by each component, if any, in disease pathogenesis remains unclear. With regard to the anti-PDC antibody response, there is no substantive evidence to suggest a mechanistic link to disease development, anti-PDC antibody development being seen in several infectious inammatory disease settings in the apparent absence of liver disease development. Furthermore, passive transfer of mouse anti-PDC into na ve recipient mice is not associated with the development of liver pathology [24]. These ndings do not, however, exclude a role for antibody in disease pathogenesis if an additional co-factor process is required in order for the eect to be observed. For example, antibody-directed cell cytotoxicity (ADCC) would not be observed in such human and animal disease settings if aberrant antigen expression on the target cells were necessary for the pathologic eect to be observed. This potentially is highly relevant given the observation that BEC from

310

JONES

Table 2 Infectious model for primary biliary cirrhosis pathogenesis and the key clinical observations Clinical observation Autoantibody presence Reconcilable aspects (a) Bacteria contain PDC and AMA are cross-reactive with bacterial PDC (b) Bacteria and viruses express TLR ligands to which increased sensitivity is seen in PBC and that stimulate AMA release in vitro (a) TLR ligands expressed by bacteria and viruses can stimulate IgM production by B cells in vitro (a) Postulated routes of bacterial infection (urinary tract) implicated more frequently in women (b) Female association of MMTV, the murine virus related to the PBC beta-retrovirus (a) Heritability of PBC may reect shared environmental exposure to infectious agents (b) Genetic factors determine the nature of the response to infectious disease (c) Disease clustering out with family units Nonreconcilable/unknown aspects (a) Bacterial exposure can induce PDC-reactive antibodies in the apparent absence of the induction of PBC (b) No clear antigenic link between beta-retrovirus and mitochondrial and nuclear autoantigens in PBC (a) Polyclonal IgM production is not a feature of other other bacterial and viral infections (a) No major dierences in infectious disease epidemiology or natural history between genders

Elevated IgM

Gender distribution

Heritability and disease associations

Patient age

Post-transplant disease recurrence Tissue tropism of disease

(a) Onset of infectious disease reects timing of exposure to pathogen, therefore fully compatible (a) Fully compatible

(a) No clear explanation for the increased personal and familial risk for other autoimmune disease diseases (b) Convergence data from monzygotic and dizygotic twins argue that the intra-familial risk is genetic rather than environmental in nature (a) Sporadic childhood cases reecting unusual exposure expected d

(a) Tropism of exposure to infectious agent would explain tropism of disease

(a) Infectious agents postulated to date have unclear tissue tropism

patients who have PBC show surface up-regulation of PDC-E2 [25]. Up-regulation of surface PDC is not, in isolation, however, sucient to cause disease [26]. Furthermore, the conclusion that antibody does not play a direct role in disease pathogenesis only relates to conventional paradigms for antibody-mediated cell damage. It is postulated, for example, that a specic pathogenetic role for AMA may be restricted to certain isotypes of the antibody, with particular focus on dimeric IgA, which has, in humans at least,

PBC PATHOGENESIS

311

Table 3 Cytopathic model for primary biliary cirrhosis pathogenesis and the key clinical observations Clinical observation Autoantibody presence Reconcilable aspects (a) Evidence of handling of self-antigens by cells in PBC, potentially rendering them immunogenic and giving rise to secondary autoimmunity Nonreconcilable/unknown aspects (a) Diculty in implicating autoimmunity as a secondary process when the phenomenon is universal in the disease (b) Apoptosis, oxidant stress, and cell senescence are universal processes but PBC is relatively rare (a) No clear potential mechanism (a) No clear potential mechanism (a) No clear potential mechanism

Elevated IgM Gender distribution Heritability and disease associations Patient age

d d d

Post-transplant disease recurrence Tissue tropism of disease

(a) Cell senescence/exhaustion component fully compatible with an age-related disease (a) Fully compatible

(a) No apparent intradisease association between degree of senescence and patient age d

(a) Potential role for bile-enriched factors in triggering cytopathic eects directly

(a) No such bile-enriched factor identied in practice

the property of undergoing transcytosis across BEC as part of the normal secretion pathway. Dimeric IgA anti-PDC can be identied in bile [27]. Antibody penetration into mitochondria, were it to occur, reasonably might be expected to block PDC enzymatic function and to exert signicant metabolic eects on the cell, including, potentially, the induction of apoptosis. In terms of underpinning clinical observations, an antibody-mediated pathogenetic eect has the conceptual advantage of explaining disease recurrence after transplant, given the near universal observation of ongoing AMA presence in post-transplant patients who have PBC. Direct antibody-mediated eects cannot explain the development of PBC in autoantibody-negative disease. There also are data supportive of a causal role for autoreactive T-cell responses directed in the pathogenesis of PBC. The kinetics of the HLA class Irestricted PDC-E2 specic response to PDC-E2 (precursor frequency is greatest in early disease when BEC damage is peaking) are compatible with an eecter response [22]. Moreover, in animal models of disease, the induction of autoreactive T-cell responses against PDC-E2 seemingly is associated with the development of PBC-like bile duct damage in spontaneous and induced models of the disease [2835]. Histologic study of liver tissue from patients who have PBC also is supportive given the presence of

312

JONES

a signicant T-cellrich portal tract inltrate around eected bile ducts and the evidence for, at some stages of the disease at least, apoptosis of BEC as a mechanism for cell loss (vide infra). Observations from human patients who have PBC (in whom regulatory T-cell deciency is reported) and a series of recently described genetically modied mice that spontaneously developed both histologic features suggestive of PBC and, in some cases, AMA (and that share the characteristic of disturbed T-cell regulation) point to impairment of T-cell regulation as a potentially key element in the pathogenesis of PBC and related disorders [3134,36]. Key clinical observations remain dicult to reconcile, however, with an autoreactive T-cell model. There is broad acceptance that response to corticosteroid therapy is limited in PBC. Signicantly greater response might be predicted if T-cell autoreactivity were a key component of disease pathogenesis. Furthermore, the observation that disease recurrence can occur across HLA boundaries post transplant (in patients who typically are receiving immunosuppressive therapies with a signicant T-cell directed component) also is dicult to reconcile with an HLA-restricted T-cell response a key component in the pathogenetic process. Also critical to the PDC-directed autoimmune response model for the pathogenesis of PBC is the question of how immune self-tolerance breaks down to PDC, an enzyme expressed in all nucleated cells in the body, and which, through its fundamental role in metabolism, plays a critical role in cell survival and function. Several themes are starting to emerge regarding the unique structural properties of PDC and the role that these properties may play in rendering it immunogenic. One theme is the importance of the complex, and highly conserved, quaternary structure of PDC and its fellow 2-OADC complexes (essential for enzymatic function) and the role that this may play in allowing induction of B-cell responses that are cross-reactive between PDCs from dierent species. A simple model for the pathogenesis of PBC, therefore, is induction of an immune response against microbial PDC, which subsequently is cross-reactive with self. The higher anity of the response to PDC of human compared with bacterial origin and the accompanying hierarchy of enzyme inhibitory capacity [37], however, argue against this simple model. Furthermore, there is the conceptual problem of why, given the universal presence of PDC in all commensal and pathogenic bacteria (and toll-like receptor [TLR] ligands able to activate antigen-presenting cells), the induction of PBC is not universal. Finally, and potentially critically, cross-reactivity between PDCs of dierent species is restricted to the B-cell compartment with even T-cell responses directed at other mammalian PDC forms (which show the highest sequence homology with human PDC) not directly associated with induction of T-cell autoreactivity [28]. A recent, and potentially exciting, possible explanation for the mechanism of breakdown of immune tolerance to PDC and its restricted expression has come with the observation that LA itself plays a fundamental role in the immunogenicity of PDC [18], and the nding that xenobiotics

PBC PATHOGENESIS

313

exhibiting structural homology with LA can play a signicant role in triggering lipoate-directed immunoreactivity [35,38] Several strands of data support this emerging xenobiotic model. First, higher anity is reported to some synthetic molecules containing xenobiotic LA homologs than to LA itself; a nding compatible with the initial immune response directed at xenobiotics with subsequent cross-reactivity with LA [38]. Second, sensitization of experimental animals with constructs containing xenobiotics with structural homology to LA is associated with the development of autoreactive immune responses (at the T-cell and the B-cell levels) against native LA-containing autoantigens and, potentially crucially, the development of PBC-like liver lesions [35]. Finally, one of the key xenobiotics demonstrated as showing immune cross-reactivity with LA and able to induce autoreactivity to PDC in experimental animals with the associated development of PBC-like histologic lesions (6-bromohexanoic acid) recently has been demonstrated to compete eectively with (and under the appropriate energetic conditions out-compete) LA in terms of incorporation into PDC-E2 via the enzymes of the physiologic exogenous lipoylation pathway [39]. This gives rise to the possibility that xenobiotics present in the environment (6-bromohexanoic acid is an industrial substrate) that are immunologically cross-reactive with LA can undergo incorporation into PDC-E2, potentially rendering the whole complex immunogenic despite the protein backbone being fully self-origin. It is possible to conceive of an environmental contaminant accumulation model in which progressive incorporation of long-lasting halogenated xenobiotics over time results in eventual attainment of a threshold level that is immunologically signicant in terms of generating sucient altered-self PDC-E2 to trigger, under the appropriate costimulatory environment, autoreactive immune responses. This model ts with epidemiologic data suggesting that PBC has a specic geographic distribution, which is associated with potential environmental contamination (in terms of previous heavy industrial contamination or exposure to toxic waste) [40,41]. Studies addressing the issue of PDC tolerance breakdown in animal experimental models have pointed to an apparent hierarchy of tolerance maintenance as a key factor explaining the balance between PDC-directed autoreactivity and autoimmunity in PBC. The induction of breakdown of B-cell tolerance to PDC seems relatively easy to achieve, probably reecting the high degree of conservation of the quaternary structure of the complex [29]. The induction of breakdown of T-cell tolerance to PDC is signicantly more dicult to achieve, however; an observation compatible with this is the key threshold step for disease pathogenesis [29]. There are emerging data to suggest that one of the key mechanisms in the translation of breakdown of B-cell tolerance to self-PDC into T-cell tolerance breakdown is the capacity of activated B-cells to function as antigen-presenting cells, potentially able to present cross-reactive self-PDC concentrated through antigen-specic uptake by nonself-specic, but self-reactive, surface Ig. This possibility is strongly suggested by the observation that transfer

314

JONES

of activated B cells isolated from donor animals sensitized with non self-PDC (but not PDC-na ve donor animals) and pulsed ex vivo with self-PDC is sucient to induce T-cell tolerance breakdown to self-PDC in na ve recipient mice [24]. These observations raise the possibility that the B-cell response to PDC seen in PBC plays, in addition to any role related to release of pathogenetic antibody, a key upstream role in terms of mediating T-cell tolerance breakdown to self-PDC through augmented antigen presentation. Further complexity to the immune model for PBC pathogenesis is suggested by emerging datadthat dysregulation of the innate immune system also is a feature of the disease, potentially unmasking T-cell autoreactivity through altered APC function or down-regulation of regulatory T-cell function [42]. Peripheral blood mononuclear cells from patients who have PBC seem to have an increased sensitivity to TLR ligands suggesting chronic activation of innate immunity [5,43]. Particular attention has focused on the role played by TLR9 in B-cell activation (potentially a key activation factor for B-cell antigen presentation) as a potential explanation for polyclonal IgM production, one of the key clinical observations in PBC [5,44]. A potential in vivo mechanism for chronic TLR stimulation in PBC comes with the longstanding observation of the presence of bacterial cell wall components within the liver in patients who have PBC [45]. Although much of the work performed on the potential immune pathogenesis of PBC relates to the mitochondrial antigens, other nuclear-encoded antigens also are identied as providing reactivity in this disease and are important as markers and prognostic features in the disease. The nuclear antigens are described in detail by Muratori and colleagues elsewhere in this issue. Primary biliary cirrhosis as an infectious disease? The concept that PBC might have an infectious etiology dates back many years, starting with the observation of granulomatous change within the liver. More recently, the focus has been on a postulated retroviral etiology. One clinical observation above all is supportive of an infectious etiologyd that of rapid recurrence of the disease after liver transplantation. Further supporting evidence that argues for an infectious etiology comes from the data relating to BEC (and salivary epithelial cell) surface expression of PDC (or a cross-reacting antigen), which is a phenomenon unique to PBC [25] (in AMA-positive and AMA-negative PBC). This phenomenon, which also recurs early after liver transplantation, can be induced in cultured BEC from normal individuals when cocultured with tissue extracts from patients who have PBC [46]. Moreover, further supernatant passage from such cultures results in, once again, induction of surface PDC expression in na ve normal BEC cultures; observations interpreted as representing transmission of an infectious agent inducing up-regulation of PDC. This model was given

PBC PATHOGENESIS

315

further stimulus by the identication of a beta-retroviral entity within culture supernatants in the passage system (described previously), and isolated from tissues from some (but not all) patients who have PBC [47]. This betaretrovirus has signicant sequence homology with mouse mammary tumor virus (MMTV), a widely used laboratory reagent [48]. These observations are provocative but remain unconrmed; the Selmi group were unable, in parallel experiments, to replicate several of the key ndings [49]. Although recent interest in terms of an infectious etiology for PBC has focused on the potential retroviral cause, there is an older literature that has linked bacterial infection with the induction of features of PBC. Potentially relevant observations include, in particular, the induction of AMA in patients who have mycobacterial and Escherichia coli infection (particularly in the context of chronic urinary tract infection) [50,51]. These observations make the point that infectious agents with their capacity to trigger crossreactive immune responses (and to provide the TLR ligands that apparently are critical for antigen-presenting cell activation in PBC) remain a potentially important source of possible disease triggers. The infectious etiology and the autoimmune etiology models for PBC pathogenesis, therefore, are inextricably linked. Primary biliary cirrhosis as a cytopathic disease? Ultimately, any model for PBC pathogenesis must be able to explain BEC loss and the development of progressive ductopenia. Initial observations identied the presence of apoptotic BEC in PBC liver [5254], an observation believed compatible, in particular, with an autoimmune etiology given the capacity for cytotoxic T cells to induce target cell apoptosis [22] and the presence of markers of T-cell cytotoxic functionality within the inammatory inltrates seen in PBC liver [55]. Apoptosis can occur, however, as a direct consequence of the eects of hydrophobic bile salts known to be retained within the liver in PBC [56,57]. BEC apoptosis seen after ductopenia resulting from another mechanism, therefore, could represent a consequence of early-stage PBC rather than a direct cause. This possibility is supported by the limited observations made in very early stage disease (when cytotoxic T-cell numbers are at their height within the portal tract inltrates) [22], which suggest that BEC apoptosis is limited in its extent [58,59]. These observations all suggest that, although BEC apoptosis undoubtedly plays an important role in the pathogenesis of PBC, the nature of that role seems more complex than previously believed. This has given rise to the concept that BEC damage, rather than representing the consequence of a single mechanism, may result from a complex series of processes potentially occurring at dierent stages in the disease process [3]. In addition to apoptosis there are emerging data to suggest that BEC loss can occur as a direct consequence of cellular senescence, potentially compounded by the eects of oxidative stress [3,60]. The BEC changes seen in

316

JONES

PBC liver are associated closely with myeloperoxidase-positive macrophages present within portal tract inltrates suggesting a potential role for chronic inammation in oxidative stressdriven or augmented BEC senescence. The evidence suggesting that senescence driven or augmented by oxidative stress contributes to the eventual phenotype of BEC loss gives rise to concepts of senescence acting as a mechanism for ab initio bile duct loss (where, for example, oxidative stress predominates). These data would also support a mechanism for consequential BEC loss as a result of epithelial exhaustion as proliferative homeostatic responses to BEC loss by other mechanisms fail, resulting in loss of the compensatory capacity, potentially regulated by the hedgehog pathway, which seems critical for the normal homeostatic response to BEC damage [3,61]. Observations in animal models of biliary obstruction and, more recently, in PBC itself, have suggested a further potential mechanism for BEC loss, that of epithelial to mesenchymal transition (EMT), a process in which epithelial cells undergo phenotypic reprogramming, losing epithelial markers and properties and gaining features more typical of broblasts [6268]. Phenotypic studies demonstrate an intermediate stage in which epithelial and mesenchymal markers are expressed, with eventual acquisition of a full mesenchymal functional and surface expression phenotype. This model is highly attractive as an explanation for loss of BEC and their replacement by broblasts. Studies using cultured BEC have demonstrated that EMT can be induced by transforming growth factor-beta (TGF-b). A role for TGF-b in the induction of EMT in vivo is suggested strongly by the presence in BEC of features of the TGF-b signaling pathway (nuclear pSMAD 2/3 expression) [68]. Potentially important in terms of the development of novel approaches to therapy, BEC EMT driven by TGF-b in culture is reversible by antagonism of the TGF-b eect through, for example, the use of hepatocyte growth factor (HGF). Furthermore, HGF reverses liver brosis secondary to bile duct ligation and ameliorates PBClike bile duct lesions in chronic graft-versus-host disease [64,69], suggesting potential in vivo therapeutic ecacy. HGF itself is expressed by BEC in aected bile ducts in PBC, suggesting that endogenous HGF production by BEC may represent a homeostatic response by BEC to EMT-inducing stimuli [63,70]. The identied mechanisms of BEC loss in PBC are not mutually exclusive. It is conceivable that all of these processes play key roles in dierent stages of the disease course. In one potential model, EMT could be occurring as an early process responsible for an initial ductopenic eect (an observation compatible with post-transplant recurrence observation) [67], with apoptosis occurring as an intermediate eect through the combined actions of hydrophobic bile salts and a cytotoxic T-cell response and senescence occurring as a late blow, as the homeostatic response of the BEC of regeneration through proliferation fails due to cellular exhaustion. The possibility that apoptosis may represent, in part at least, a secondary phenomenon in

PBC PATHOGENESIS

317

PBC, taken together with the altered antigen-expression patterns associated with apoptosis (mitochondrial autoantigens implicated in PBC pathogenesis are cleaved by key apoptotic cascade enzymes [71], potentially generating altered-self forms of PDC, which may be immunogenic) and the fact that BEC normally express specic phenotypic features suggestive of apoptosis resistance (which are lost in PBC), which alter the cleavage properties of PDC [72], gives rise to the challenging concept that autoreactivity in PBC may occur as a consequence of initial bile duct damage rather than representing its cause. In this model, autoreactive immune responses in PBC represent an epiphenomenon occurring in the disease or a booster secondary damage process. This model could go some way to explain the apparent lack of response to immunomodulatory therapy in PBC. An alternative potential explanation for the apparent presence of EMT in the earliest stages of the disease process is that this process and the counter-balancing capacity, demonstrated in vitro, of the cells to undergo reciprocal mesenchymal to epithelial transition represents an element of the homeostatic response to BEC damage, rather than the mechanism of damage per se, increasing the proliferative capacity of BEC in response to insult by cycling them through a mesenchymal phase, where such capacity is increased greatly compared with the epithelial phase, before reverting to the epithelial phenotype. This concept ts well with emerging ideas suggesting that the interactions of the cell types implicated in liver injury and repair are substantially more complex than previously believed, that they reect both sides of the balance of damage and repair (the ductular reaction, for example, representing an epithelial restorative and a pro-brogenic response) [61], and that they need to be viewed in a systematic way (tissue biology as opposed to cell biology) [73,74].

Summary There is a wealth of clinical and experimental data now linked to understanding the pathogenesis of PBC. The balance of the available data strongly supports an autoimmune model. Inconsistencies within this model, however, must be acknowledged. Paradoxic observations (in particular the unusual response pattern to immunomodulatory therapies and the recurrence of disease after transplantation across HLA boundaries) need to be explained to reconcile this model. The concept of an infectious etiology is controversial but the possibility that infection acts as a trigger to disease renders the infection hypothesis fully compatible with a principally autoimmune mechanism for disease pathogenesis. Although there is little evidence that PBC represents a disease in which cytopathic processes occur as the primary eect, BEC, rather than acting as passive victims in the process, are active participants and, in all probability, ductopenia represents as much a failure of BEC to be able to respond normally to insult as a direct

318

JONES

consequence of that insult. It may be, therefore, that all of the autoimmune, infectious, and cytopathic pathogenetic models are correct. Where does this leave us, therefore, with regard to therapy? Understanding the antigen specicity of B-cells and auto reactive T-cell responses in PBC has not inuenced therapy directly (and is unlikely to so in the future). Approaches aimed at tolerizing autoreactive immune responses to PDC have proved, perhaps unsurprisingly, ineective in terms of liver injury. PBC joins, therefore, the long list of autoimmune diseases where the na ve model of increased understanding of antigen specicity followed directly by the development of novel approaches to therapy has led to disappointment. It may be, however, that the goal of developing novel approaches to treatment of this disease will be realized by studying not the intermediate process in the disease (probably autoreactive immune responses) but the earliest stages of the disease when initial insult occurs and, in particular, the stages of BEC damage. PBC is a disease in which BEC are damaged and lost and perhaps the focus in terms of therapy should be directed precisely at this most critical process.

References
[1] Mackay IR. Primary biliary cirrhosis showing a high titre of autoantibody: report of a case study. N Engl J Med 1958;258:1858. [2] Yeaman SJ, Fussey SP, Danner DJ, et al. Primary biliary cirrhosis: identication of two major M2 mitochondrial autoantigens. Lancet 1988;1(8594):106770. [3] Jones DEJ. Pathogensis of primary biliary cirrhosis. Gut 2007;56:161524. [4] Mutimer DJ, Fussey SP, Yeaman SJ, et al. Frequency of IgG and IgM autoantibodies to four specic M2 mitochondrial autoantigens in primary biliary cirrhosis. Hepatology 1989;10: 4037. [5] Kikucki K, Lian ZX, Yang GX, et al. Bacterial CpG induces hyper-IgM production in CD27() memory B cells in primary biliary cirrhosis. Gastroenterology 2005;128:30412. [6] Jones DE, Watt FE, Metcalf JV, et al. Familial primary biliary cirrhosis reassessed: a geographically-based population study. J Hepatol 1999;30:4027. [7] Watt FE, James OFW, Jones DEJ. Patterns of autoimmunity in PBC patients and their families. QJM 2004;97:397406. [8] Lazaridis KN, Juran BD, Boe GM, et al. Increased prevalence of anti-mitochondrial antibodies in rst-degree relatives of patients with primary biliary cirrhosis. Hepatology 2007; 46:78592. [9] Invernizzi P, Selmi C, Mackay IR, et al. From bases to basis: linking genetics to causation in primary biliary cirrhosis. Clin Gastroenterol Hepatol 2005;3:40110. [10] Liermann Garcia RF, Evangalista Garcia C, McMaster P, et al. Transplantation for primary biliary cirrhosis:retrospective analysis of 400 patients in a single centre. Hepatology 2001;33: 227. [11] Joplin R, Johnson GD, Matthews JB, et al. Distribution of pyruvate dehydrogenase dihydrolipoamide acetyltransferase (PDC-E2) and another mitochondrial marker in salivary gland and biliary epithelium from patients with primary biliary cirrhosis. Hepatology 1994;19:137580. [12] Palmer JM, Kirby JA, Jones DEJ. The immunology of primary biliary cirrhosis: the end of the beginning. Clin Exp Immunol 2002;129:1917.

PBC PATHOGENESIS

319

[13] Yeaman SJ, Kirby JA, Jones DEJ. Autoreactive responses to pyruvate dehydrogenase complex in the pathogenesis of primary biliary cirrhosis. Immunol Rev 2000;174:23849. [14] Van de Water J, Fregeau D, Davis P, et al. Autoantibodies of primary biliary cirrhosis recognise dihydrolipoamide acetyltransferase and inhibit enzyme function. J Immunol 1988;141:23214. [15] Palmer JM, Jones DEJ, Quinn J, et al. Characterisation of the autoantibody responses to recombinant E3 binding protein (protein X) of pyruvate dehydrogenase in primary biliary cirrhosis. Hepatology 1999;30:216. [16] Quinn J, Diamond AG, Palmer JM, et al. Lipoylated and unlipoylated domains of human PDC-E2 as autoantigens in primary biliary cirrhosis: signicance of lipoate attachment. Hepatology 1993;18:138491. [17] Koike K, Ishbashi H, Koike M. Immunoreactivity of porcine heart dihydrolipoamide acetyl- and succinyl-transferases (PDC-E2, OGDC-E2) with primary biliary cirrhosis sera: characterisation of the autoantigenic region and eects of enzymatic delipoylation and relipoylation. Hepatology 1998;27:146774. [18] Bruggraber SF, Leung PS, Amano K, et al. Autoreactivity to lipoate and a conjugated form of lipoate in primary biliary cirrhosis. Gastroenterology 2003;125:170513. [19] Shimoda S, Nakamura M, Ishibashi H, et al. HLA DRB4 0101-restricted immunodominant T-cell autoepitope of pyruvate dehydrogenase complex in primary biliary cirrhosis: evidence of molecular mimicry in human autoimmune disease. J Exp Med 1995;181:183545. [20] Jones DEJ, Palmer JM, Yeaman SJ, et al. T-cell responses to the components of pyruvate dehydrogenase complex in primary biliary cirrhosis. Hepatology 1995;21:9951002. [21] Shimoda S, Van de Water J, Ansari A, et al. Identication and precursor frequency analysis of a common T-cell epitope motif in mitochondrial autoantigens in primary biliary cirrhosis. J Clin Invest 1998;102:183140. [22] Kita H, Matsumura S, He X, et al. Quantitative and functional analysis of PDC-E2-specic autoreactive cytotoxic T lymphocytes in primary biliary cirrhosis. J Clin Invest 2002;109: 123140. [23] Kita H, Lian Z, Van de Water J, et al. Identication of HLA-A2-restricted CD8 cytotoxic T-cell responses in primary biliary cirrhosis: T-cell activation is augmented by immune complexes cross-presented by dendritic cells. J Exp Med 2002;195:11323. [24] Robe AJ, Kirby J, Jones DEJ, et al. A Key role for autoreactive B-cells in the breakdown of T-cell tolerance to pyruvate dehydrogenase complex. Hepatology 2005;41:110612. [25] Joplin R, Gordon L, Lindsay J, et al. Membrane dihydrolipoamide acetyltransferase (E2) on human biliary epithelial cells in primary biliary cirrhosis. Lancet 1992;339:934. [26] Inamura K, Tsuji H, Nakamoto Y, et al. Transgenic mice aberrantly expressing pyruvate dehydrogenase complex E2 component on biliary epithelail cells do not show primary biliary cirrhosis. Clin Exp Immunol 2006;145:93100. [27] Kisand K, Kisand K, Salupere V, et al. Enzyme -linked immunosorbent assays for the determination of IgG, IgA, and IgM autoantibodies to pyruvate dehydrogenase in primary biliary cirrhosis. Int J Clin Lab Res 1994;24:98101. [28] Jones DEJ, Palmer JM, Yeaman SJ, et al. Breakdown of tolerance to pyruvate dehydrogenase complex in experimental autoimmune cholangitis a murine model of primary biliary cirrhosis. Hepatology 1999;30:6570. [29] Jones DEJ, Palmer JM, Bennett K, et al. Investigation of a mechanism for accelerated breakdown of immune-tolerance to the primary biliary cirrhosis associated autoantigen,pyruvate dehydrogenase complex. Lab Invest 2002;82:2119. [30] Jones DEJ, Robe AJ, Kirby JA, et al. Adopotive transfer of self-PDC-reactive T-cells into naive mice induces portal tract and bile duct changes characteristic of primary biliary cirrhosis (PBC). J Hepatol 2006;44:S10. [31] Irie J, Wu Y, Wicker LS, et al. NOD.c3c4 congenic mice develop autoimmune liver disease that serologically and pathogenetically models human primary biliary cirrhosis. J Exp Med 2006;203:120919.

320

JONES

[32] Oertelt S, Lian ZX, Cheng CM, et al. Anti-mitochondrial antibodies and primary biliary cirrhosis. J Immunol 2006;177:165560. [33] Wakabayashi K, Lian ZX, Moritoki Y, et al. IL-2 receptor alpha (-/-) mice and the development of primary biliary cirrhosis. Hepatology 44:12409. [34] Gershwin ME, Chuang Y-H, Yang G-X, et al. NOD.2445 mice, a spontaneous model of human primary biliary cirrhosis, develop specic autoreactive T cells. Hepatology 2007; 46:543A. [35] Leung PSC, Park O, Tsuneyama K, et al. Induction of primary biliary cirrhosis in guinea pigs following chemical xenobiotic immunization. J Immunol 2007;179:26517. [36] Lan RY, Cheng C, Lian ZX, et al. Liver-targeted and peripheral blood alterations of regulatroy T-cells in primary biliary cirrhosis. Hepatology 2006;43:72937. [37] Teoh KL, Mackay IR, Rowley MJ, et al. Enzyme inhibitory autoantibodies to pyruvate dehydrogenase complex in primary biliary cirrhosis dier for mammalian, yeast and bacterial enzymes: implications for molecular mimicry. Hepatology 1994;19:102933. [38] Long SA, Quan C, Van de Water J, et al. Immunoreactivity of organic mimeotopes of the E2 component of pyruvate dehydrogenase: connecting xenobiotics with primary biliary cirrhosis. J Immunol 2001;167:295663. [39] Walden HR, Palmer JM, Kirby JA, et al. Substrate promiscuity in the pyruvate dehydrogenase complex (PDC) lipoylation pathway:xenobiotic incorporation as a potential mechanism for the pathogenesis of primary biliary cirrhosis. Hepatology 2007;46:545A. [40] Prince MI, Chetwynd A, Diggle P, et al. The geographical distribution of primary biliary cirrhosis in a well-dened cohort. Hepatology 2001;34:10838. [41] Ala A, Stanca CM, Bu-Ghanim M, et al. Increased prevalence of primary biliary cirrhosis near Superfund waste sites. Hepatology 2006;43:52531. [42] Pasare C, Medzhitov R. Toll pathway-dependent blockade of CD4CD25 T cell-mediated suppression by dendritic cells. Science 2003;299:10336 [comment]. [43] Mao TK, Lian Z-X, Selmi C, et al. Altered monocyte responses to dened TLR ligands in patient with primary biliary cirrhosis. Hepatology 2005;42:8028. [44] Moritoki Y, Lian Z, Wul H, et al. AMA propduction in primary biliary cirrhosis is promoted by the TLR9 ligand CpG and suppressed by potassium channel blockers. Hepatology 2007;45:31422. [45] Haruta I, Hashimoto E, Kato Y, et al. Lipoteichoic acid may aect the pathogenesis of bile duct damage in primary biliary cirrhosis. Autoimmunity 2006;39:12935. [46] Sadamoto T, Joplin R, Keogh A, et al. Expression of pyruvate-dehydrogenase complex PDC-E2 on biliary epithelial cells induced by lymph nodes from primary biliary cirrhosis. Lancet 1998;352:15956. [47] Xu L, Shen S, Guo L, et al. Does a betaretrovirus infection trigger primary biliary cirrhosis? Proc Natl Acad Sci U S A 2003;100:84549. [48] Xu L, Sakalian M, Shen Z, et al. Cloning the human betaretrovirus proviral genome form patients with primary biliary cirrhosis. Hepatology 2004;39:1516. [49] Selmi C, Ross SR, Ansari AA, et al. Lack of immunological or molecular evidence for a role of mouse mammary tumor retrovirus in primary biliary cirrhosis. Gastroenterology 2004; 127:493501. [50] Klein R, Wiebel M, Engelhart S, et al. Sera from patients with tuberculosis recognise the M2a-epitope (E2-subunit of pyruvate dehydrogenase complex) specic for primary biliary cirrhosis. Clin Exp Immunol 1993;92:30816. [51] Butler P, Valle F, Hamilton-Miller JM, et al. M2 mitochondrial antbodies and urinary rough mutant bacteris in patients with primary biliary cirrhosis and in patients with recurrent bacteriuria. J Hepatol 1993;17:40814. [52] Kuroki T, Seki S, Kawakita N, et al. Expression of antigens related to apoptosis and cell proliferation in chronic nonsuppurative destructive cholangitis in primary biliary cirrhosis. Virchows Arch 1996;429:11929.

PBC PATHOGENESIS

321

[53] Harada K, Ozaki S, Gershwin ME, et al. Enhanced apoptosis relates to bile duct loss in primary biliary cirrhosis. Hepatology 1997;26:1399405. [54] Tinmouth J, Lee M, Wanless IR, et al. Apoptosis of biliary epithelial cells in primary biliary cirrhosis and primary sclerosing cholangitis. Liver 2002;22:22834. [55] Martinez OM, Villanueva JC, Gershwin ME, et al. Cytokine patterns and cytotoxic mediators in primary biliary cirrhosis. Hepatology 1995;21:1139. [56] Rodrigues CMP, Fan GS, Ma XM, et al. A novel role for ursodeoxycholic acid in inhibiting apoptosis by modulating mitochondrial membrane perturbation. J Clin Invest 1998;101: 27909. [57] Paumgartner G, Beuers U. Ursodeoxycholic acid in cholestatic liver disease: mechanisms of action and therapeutic use revisited. Hepatology 2002;36:52531. [58] Ballardini G, Guidi M, Susca M, et al. Bile duct apoptosis is a rare event in primary biliary cirrhosis. Dig Liver Dis 2001;33:1516. [59] Onori P, Alvaro A, Floreani A, et al. Activation of the IGF1 system characterizes cholangiocyte survival during progression of primary biliary cirrhosis. J Histochem Cytochem 2007; 55:32734. [60] Aboutwerat A, Pemberton PW, Smith A, et al. Oxidant stress is a signicant feature of primary biliary cirrhosis. Biochim Biophys Acta 2003;1637:14250. [61] Jung Y, McCall SJ, Li YX, et al. Bile ductules and stromal cells express hedgehog ligands and/or hedgehog target genes in primary biliary cirrhosis. Hepatology 2007;45:10916. [62] Lee JM, Dedhar S, Kalluri R, et al. The epithelial-mesenchymal transition: new insights in signaling, development, and disease. J Cell Biol 2006;172:97381. [63] Wasilenko S, Mason AL. New insights from recurrent primary biliary cirrhosis in liver transplantation: the paradox of BEComing a broblast. Hepatology 2007;45:83740. [64] Xia J-L, Dai C, Michalopoulos GK, et al. Hepatocyte growth factor attenuates liver brosis induced by bile duct ligation. Am J Pathol 2006;168:150012. [65] Sicklick JK, Choi SS, Bustamante M, et al. Evidence for epithelial to mesenchymal transition in adult liver cells. Am J Physiol Gastrointest Liver Physiol 2006;291:G57583. [66] Rygiel KA, Robertson H, Burt AD, et al. Demonstration of the transition of intrahepatic biliary epithelial cells to broblasts during chronic inammatory liver diseases. J Hepatol 2006;44:S241. [67] Robertson H, Kirby JA, Yip W, et al. Biliary epithelial-mesenchymal transtion in posttransplantation recurrence of primary biliary cirrhosis. Hepatology 2007;45:97781. [68] Rygiel KA, Robertson H, Pekalski M, et al. Epithelial-mesenchymal transition contributes to portal tract brogenesis during human chronic liver disease. Lab Invest 2008;88:11223. [69] Kuroiwa T, Iwasaki T, Imado T, et al. Hepatocyte growth factor prevents lupus nephritis in a murine lupus model of chronic graft-versus-host disease. Arthritis Res Ther 2006;8:R123. [70] Sasaki M, Ikeda H, Kataoka H, et al. Augmented expression of hepatocyte growth factor activator inhibitor type 1 (HAI-1) in intrahepatic small bile ducts in primary biliary cirrhosis. Virchows Archiv 2006;449:46271. [71] Matsumura S, Van de Water J, Kita H, et al. Contribution to antimitochondrial antibody production: cleavage of pyruvate dehydroganse complex-E2 by apoptosis-related proteases. Hepatology 2002;35:1422. [72] Odin JA, Huebert RC, Casciola-Rosen L, et al. Bcl-2-dependent oxidation of pyruvate dehydrogenase-E2, a primary biliary cirrhosis autoantigen, during apoptosis. J Clin Invest 2001;108:22332. [73] Richardson MM, Jonsson JR, Powell EE, et al. Progressive brosis in nonalcoholic steatohepatitis: association with altered regeneration and a ductual reaction. Gastroenterology 2007;133:8090. [74] Theise ND, Kuwahara R. The tissue biology of ductular reactions in human chronic liver disease. Gastroenterology 2007;133:3502.