Autoantibodies are an important serological feature of autoimmune diseases and may be of significant value in their diagnosis or prognosis.1 Some autoantibody types are of particular diagnostic value, in that they are highly disease-specific, notably those that are a component of the disease process, such as autoantibodies to the acetylcholine receptor in myasthenia gravis.2 Other antibodies are pathognomonic for a particular disease without having obvious pathogenic effects, such as antimitochondrial antibodies in primary biliary cirrhosis.3 A particular clue to disease pathogenesis is provided by antibodies to transglutaminase in celiac disease: the disease-causing antigen is gliadin, in fact primarily deaminated gliadin, which is recognized by specific T cells. Transglutaminase is the enzyme responsible for deamination of gliadin, and therefore directly related to the key antigen, without itself being the key target antigen of the inflammatory condition.4 Many of the important autoantibodies associated with autoimmune diseases are directed towards posttranslational protein modifications or enzymes, and it may well be, that the pathogenic key event may not be the autoimmune recognition of the modified proteins, but aberrant ligand binding or modification.5 Indeed, the example of celiac disease demonstrates the importance of identifying the target antigens of autoantibodies, because these can be related to possible pathogenic processes.
In autoimmune hepatitis (AIH), most patients present with autoantibodies. Common autoantibodies in AIH include those binding to perinuclear antigen of neutrophils (atypical p-ANCA), nuclear antigens (ANA), filamentous actin (SMA), the SEPSECS molecule (SLA/LP), or cytochrome P450 2D6 (LKM-1).6 However, most of the known autoantibodies are not disease-specific (p-ANCA, ANA, SMA, LKM-1). Only SEPSECS (SLA/LP) autoantibodies are disease-specific autoantibodies, but are present in less than a third of the patients (SLA/LP).7 Therefore, the identification of additional autoantibody reactivities is desirable. This may not only help the diagnosis of AIH, but might also provide clues to the pathogenesis of this condition. In an effort to identify disease-specific antibodies, Tahiri et al.8 compared the reactivities of 65 AIH sera and 50 control sera to liver plasma membrane proteins by applying proteomics techniques. By restricting the analysis to plasma membrane antigens, the authors were hoping to increase the likelihood to detect pathogenic and hence disease-specific autoantibodies. The idea is that target antigens for pathogenetic autoantibodies should be accessible to these antibodies at the cell surface. However, as in other liver-specific autoimmune diseases,3 the majority of the detected autoantigens are not exclusively expressed in the liver. The authors argue that a possible clue to the organ-specific manifestation of autoimmune disease, despite ubiquitous expression of the relevant autoantigens, may be an immunogenic translocation of autoantigens from other cellular compartments to the plasma membrane. It remains to be seen whether this translocation is a possible trigger of pathogenic immunity, or whether inflammatory stress had caused the translocation of the autoantigens to the plasma membrane; indeed, the majority of the detected autoantigens were stress proteins such as heat shock proteins. Be that as it may, the most remarkable finding of the study is that, despite the large number of sera studied, the detected autoantibodies were all non–disease specific and thus unlikely pathogenic. Indeed, it is remarkable how most of the antibodies are present in normal sera, and the only difference detectable is one of titer. This titer difference mostly is marginal considering that AIH patient sera have much higher immunoglobulin levels than normal control sera.
There are several possible explanations for the inability to detect disease-specific antigens in this study: (1) the analysis did not include proteins with higher molecular weights (that is, >100,000); thus, the authors may have missed large potentially disease-specific antigens9; (2) they may have missed relevant antigens by restricting the analysis to plasma membrane antigens, which were, moreover, derived from rat and not human hepatocytes; or (3) it may as well be, contrary to our expectations, that disease-specific autoantibodies are not present in the majority of patients with AIH.
Indeed, there are fewer disease-specific autoantibodies known than one might expect, and several autoimmune diseases could thus far not be correlated to a specific immune reactivity. In fact, “natural” autoantibodies, which are antibodies detected in the absence of known immunizations, can be demonstrated in healthy subjects.10 It often comes down to a question of definition and methodology: The presence of autoantibodies in healthy subjects is frequently obscured by defining reactivity thresholds for antibody detection assays, which eliminate the “negligible background” of normal sera. However, these autoantibodies fluctuate in healthy subjects in response to various triggers, and may even rise above detection thresholds, which has long been interpreted as “false-positive” test result. Natural autoantibodies (as well as T cells) appear to recognize a relatively conserved set of autoantigens.11 Many autoantigens that are recognized by natural autoantibodies are also the target antigens of autoantibodies associated with autoimmune disease. Because T and B cell receptors are inherently polyspecific,12 some degree of autoreactivity is probably inevitable, and it is possible that humans acquire these autoreactivities accidentally during immune responses to foreign antigens, by ways of molecular mimicry between self and nonself.13 However, natural autoantibodies do not merely seem to originate by accident; even newborns manifest natural antibodies that recognize relatively uniform sets of antigens prior to individual immune experience.14, 15 Thus, it appears as if natural antibodies, though frequently detected, are not related to autoimmune disease. Even though their titer may be increased in autoimmune disease conditions, they usually do not serve specific diagnostic purposes, and they are unlikely to explain the pathogenesis of autoimmune diseases.
Nevertheless, there are good reasons to believe that natural autoantibodies, although non–disease specific, can still be helpful in the diagnosis of some autoimmune diseases. The analysis of the clonotypes of autoreactive lymphocytes has revealed that, within the repertoire of lymphocytes that recognize a given autoantigen, some clonotypes are more relevant than others, in that they drive pathogenesis, while others do not.16, 17 In other words, it is not necessarily the reactivity to a particular autoantigen per se that causes disease, but the appearance of aggressive “driver clones” of limited epitope-specificity that propagate pathogenic autoimmunity.17 Thus, the presence of autoantibodies to distinct epitopes within an autoantigen may signify disease-specific pathogenic immune activity, while the recognition of other epitopes of the same autoantigen by natural autoantibodies may be non–disease specific. This is illustrated, for example, by the finding that the recognition of a specific epitope within the MOG protein distinguishes the multiple sclerosis–specific autoantibody response to MOG from that in nondemyelinating disease.18
The study of autoantibodies has thus far usually been based on a rather simple one-to-one paradigm of a direct antigen-autoantibody interaction as the key pathogenetic process. However, the immune system, like the neural system is based on a complex multitude of molecular interactions, and pathogenetic processes are influenced by a multitude of factors. This is likely to also be the case in autoimmune liver disease. Modern diagnostic methodology, in combination with computational analysis may be able to grasp this multifactorial process and define disease-specific reactivity patterns,19 which can be used for diagnostic purposes and provide new clues to the pathogenesis. Therefore, the diagnostic value of natural autoantibodies may be improved by zooming in from analysis of whole autoantigens to the analysis of dominant autoepitopes. Alternatively, zooming out to the analysis of global patterns of reactivity to sets of antigens could also improve the diagnostic relevance of natural autoantibodies.15 Indeed, such pattern analysis of natural antibodies reacting to various self and non-self antigens served the reliable prediction of future disease in prediabetic mice.19 Because Tahiri et al. have now demonstrated that the plasma membrane antigens targeted by autoantibodies in AIH are the same as those targeted by natural autoantibodies,8 it will be possible to analyze whether there are disease-specific epitopes within the recognized autoantigens or disease-specific reactivity patterns.