The immune system is designed to protect the host from foreign invaders (1, 2). In doing so, cells of the immune system are educated to discriminate between endogenous ‘self’ tissues and exogenous ‘non-self’ components (1, 2). But errors do occur, with more than 80 clinically distinct autoimmune diseases being currently recognised affecting some 5% of North American and European populations (3, 4). In many of these disorders, the disease process begins months or even years before the appearance of clinical symptoms. This creates the opportunity to intervene early in the process to either treat the disease at a stage when it may be more responsive to treatment or to slow down the progress of autoimmune destruction.
Akin to the clearance of foreign antigens, immunity towards self-antigens is highly restricted to individual peptides within autoantigens (5–7). There is evidence to show that the highly diverse immune response seen in various autoimmune diseases such as systemic lupus erythematosus, multiple sclerosis and diabetes originate from reactivity to a single protein or even a single autoepitope which spreads in an ‘intramolecular’ or ‘intermolecular’ manner, selecting new antigenic targets within a single protein or in different proteins of a macromolecular complex (8, 9). Studies have naturally concentrated on the search for initial targets, with myelin basic protein, myelin oligodendroglial glycoprotein and proteolipid protein being prime candidates in multiple sclerosis and glutamic acid decarboxylase and insulin in insulin dependent diabetes.
In primary biliary cirrhosis (PBC), a chronic cholestatic liver disease characterised by progressive destruction of the small intrahepatic bile ducts, there is a consensus as to the primary and dominant target of the autoimmune response (10, 11). The disease is characterised by a highly specific antibody response directed against mitochondria (AMA) recognising the inner mitochondrial oxoacid dehydrogenase complex (OADC) and in particular the pyruvate dehydrogenase complex E2 subunit (PDC-E2) (10–12). The breakdown of immune-tolerance to self-PDC-E2 appears to be a fundamental step in the pathogenesis of PBC, there being an almost complete association between seropositivity for anti-PDC-E2 autoantibodies and present or future development of the bile duct lesions characteristic of the disease (13–15). AMA characteristic of PBC are only those directed against E2 subunits of the OADCs, including PDC, 2-oxoglutarate dehydrogenase complex (OGDC) and branched-chain 2-oxoacid (BCOADC) and also against subunits E1α, Elβ and E3 binding protein (E3BP), formerly known as protein X, of the PDC (12). More than 95% of the PBC patients have antibodies against PDC-E2 complex whose principal target is the inner lipoyl-binding domain (12).
The association of AMA with PBC is so striking that the diagnosis of PBC has to be questioned, as Dame Sheila Sherlock taught, in the absence of this biomarker.
There are a number of AMA features, which are puzzling (12, 16). Mitochondrial antigens are ubiquitous while the autoimmune response is confined to biliary epithelial cells of the small intrahepatic bile ducts. A protein complex of the inner mitochondrial membrane becomes a major B-cell and T-cell target in spite of being sheltered from the immune system by two membrane barriers. We do not know why OADC – and not other mitochondrial proteins – are targeted by the autoimmune assault. We are also unable to explain why the AMA pattern by Western blot differs from one patient to another (Fig. 1). While a minority of PBC sera react with all the subunits, most react with a combination thereof. Also, the intensity of the reaction varies (Fig. 1) (10, 12, 16). Finally, we cannot understand the relationship between antimitochondrial immunity and the pathogenesis of the disease (17–21).
Yet, there is a consensus that the study of pathogenetic mechanisms in PBC should start from AMA and its targets in view of the unique link between autoantibody and disease (13–15, 21). As shown in a key study by the Newcastle group, AMA can precede the clinical onset of the disease by two decades: once detected it predicts the inexorable progression to overt disease sometime in the future (13–15).
The contention as to whether AMA participate in the destruction of the bile ducts has received contradictory evidence from animal studies (12). AMA induction through the immunisation with recombinant PDC-E2 was not associated with a biochemical or histological picture of PBC (22). Conversely, another study shows that AMA induction is accompanied by lesions reminiscent of PBC, provided that heroic manoeuvres, such as neonatal thymectomy, were applied (23).
The question as to whether there is a hierarchy in the development of the individual AMA reactivities, as defined by Western blot, has been addressed in the past. Two early studies have shown that antibody responses to PDC-E2 and PDC-E3BP practically go always together and such simultaneous reactivity is believed to reflect cross-reactivity between significantly homologous inner lipoyl domains of the two subunits (24, 25). It is unclear whether the initial breakdown of self-tolerance is to PDC-E2 or to E3BP (Fig. 2). As recognition of the sequence partially shared in common between PDC-E2 and E3BP is the likely starting point of autoreactivity, it is important to define which of the two is indeed the first to be recognised as an autoantigen (12, 24, 25). Both studies have shown that that preabsorption of double-reactive PBC sera with PDC-E2 completely removes reactivity to PDC-E3BP, while incubation with E3BP does not completely abolish reactivity to PDC-E2 (24, 25). Thus, there are two distinct populations of antibodies reactive with PDC-E2, only one of which is fully cross-reactive with PDC-E3BP. These findings, according to the Newcastle's group are compatible with the original tolerance breakdown occurring against E3BP with consequent cross-reactivity with the homologous PDC-E2 sequence and further epitope spreading to other regions on E2 (25). According to Gershwin's group the presence of two or more epitopes on PDC-E2 and a single cross-reactive epitope on E3BP is consistent with the latter protein being cross-reactively recognised with E2, following an initial E2 tolerance breakdown (24). Epitope spreading on the E2 molecule would then follow (24). The proposed explanations are mirror image of each other, and to the practicing hepatologist may sound rather doctrinaire. Yet, it is important to identify the primum movens, the prime mover in the autoaggressive cascade. Interventions aimed at restraining specifically the autoimmune attack would need to start there.
As the antibodies reacting with the shared sequence belong to the immunoglobulin G class, and as antibodies of this class are only produced with the help of helper CD4 T-cells, the ‘precedence’ controversy may be shifted at T-cell level.
In the present issue of the Journal, McHugh et al. (26) of the Newcastle group have investigated CD4 T-cell responses to human PDC-E3BP, making use of the recently cloned autoantigen. They have found that only a minority of PBC patients (8/20, 40%) mount significant CD4 T-cell responses to PDC-E3BP, such responses being concomitant with CD4 T-cell responses against E2, while isolated responses to E3BP alone were practically absent (26). CD4 T-cell responses to E2 alone were present in the majority of patients (17/20, 85%) with nine patients showing reactivity to E2 but not to E3BP. On the basis of these data, McHugh et al. (26), conclude that PDC-E3BP is not a dominant T-cell autoantigen and that CD4 T-cell-dependent antibody responses are likely to result from cross-reactivity to PDC-E2, the latter being the original target of the autoimmune attack (26). Through a CD4 T-cell excursus now the views from Newcastle and Davis coincide, with E2 being the original focus of the autoimmune response, which in turn cross reactively recognises E3BP and spreads along the E2 molecules to involve additional E2 epitopes.