|Home | About | Journals | Submit | Contact Us | Français|
Several common autoimmune diseases, such as rheumatoid arthritis, systemic lupus erythematosus (SLE) and multiple sclerosis, are genetically linked to distinct human MHC class II molecules and other immune modulators. However, genetic predisposition is only one risk factor for the development of these diseases, and low concordance rates in monozygotic twins as well as geographical distribution of disease risk point towards environmental factors in the genesis of these diseases. Among these environmental factors, infections have been implicated in the onset and/or promotion of autoimmunity. In this review, we outline mechanisms by which pathogens can trigger autoimmune disease, and also pathways by which infection and immune control of infectious disease might be dysregulated during autoimmunity.
The immune system walks a fine line to distinguish self- from harmful non-self to preserve the integrity of the host. Deficits in this discrimination can result in susceptibility to infections or over-reactivity to harmless antigens, leading to immunopathology and autoimmunity. Therefore, it is not surprising that genetic factors that influence the sensitivity of the immune system are associated with autoimmune diseases, but might only manifest themselves as autoimmunity after stimulation by environmental factors, including viral infections. Unfortunately, this implies also that the over-reactive immune system of individuals susceptible to autoimmune disease might be triggered by more than one pathogen or even by more severe primary infections with common pathogens. Both factors make it difficult to associate distinct pathogens with particular autoimmune diseases. This review discusses the mechanisms by which pathogens could trigger autoimmune diseases and mechanisms by which autoimmune disease could alter the ability of the host to control infections and regulate the immune system. In discussing these aspects, we highlight recent studies showing the induction of autoimmune inflammation in the central nervous system (CNS) of mice following infection with Theiler’s murine encephalomyelitis virus (TMEV), and the dysregulation of immune responses against Epstein–Barr virus (EBV) in humans with the autoimmune disease multiple sclerosis. The discussed mechanisms and correlations between altered immune responses to these pathogens and autoimmune diseases could be developed as therapeutic targets or as diagnostic markers in the future.
As we discuss here, several mechanisms have been described to explain how viruses might trigger autoimmune diseases, including virus-induced general activation of the immune system and provision of viral antigens that specifically stimulate immune responses, which cross-react with self-antigens and therefore cause autoreactive immunopathologies.
The ability of the host to defend against invading pathogens is to a large extent mediated by a group of germline-encoded receptors known as pattern-recognition receptors (PRRs). These molecules include Toll-like receptors (TLRs), nucleotide-binding and oligomerization domain (NOD)-like receptors (NLRs), (RIG-I)-like helicases and a subset of C-type lectin receptors, which together recognize a large number of molecular patterns present in bacteria, viruses and fungi (reviewed in 1). The signalling pathways that are triggered by engagement of these molecules lead to cellular activation, which increases the antigen-presenting capacity of and the expression of co-stimulatory molecules by antigen-presenting cells (APCs), as well as their production of type I interferons, pro-inflammatory cytokines and chemokines, which initiate and direct the immune response against the invading pathogen. Microbial antigens as well as PRR-triggered inflammatory molecules drive the clonal expansion of pathogen-specific T and B cells. By triggering PRRs, stimulating early responses by the innate immune system and increasing the function of APCs, pathogens act as adjuvants for the immune response, while at the same time providing an antigen source to drive T-cell and B-cell activation and effector function (Figure 1). In this highly inflammatory environment, it is easy to envision how an aberrant destructive immune response can be triggered and/or escalated if autoreactive cells are present. There are several postulated mechanisms by which pathogenic infections can trigger autoimmune disease, but most evidence in animal models has been gathered to support the idea that cross-reactive immune responses cause autoimmunity due to similarities of viral and self antigens.
The well-documented degeneracy of antigen recognition by the T-cell receptor (TCR), which allow T-cell activation by different peptides bound to one or even several MHC molecules 2, implies that responses to foreign antigens could result in the activation of T cells that are also cross-reactive with self antigens. Similarly monoclonal antibodies, reflecting B-cell receptor specificity were found to recognize both microbial and self-antigens 3. This idea, known as molecular mimicry (Figure 2A), was first put forward by Fujinami and Oldstone 4,5. It is now generally accepted that a single T cell can respond to various distinct peptides, and that different peptide/MHC complexes can lead to cross-reactivity by the same TCR as long as the complexes have similar charge distribution and overall shape 6-8. This flexibility of TCR recognition is thought to be central to many immunological processes including thymic selection and the ability to recognize nearly all pathogen-derived peptides. A side effect of this is the induction of autoimmunity by microbial antigens. Vice versa the identification of such cross-reactivities can also help to uncover etiologic agents of autoimmune disease. In vitro studies have clearly shown that viral peptides with some degree of homology with self peptides can stimulate autoreactive T cells 6.
Animal models in which molecular mimicry can trigger autoimmune disease are abundant. These include: TMEV-induced demyelinating disease (TMEV-IDD), a model of human multiple sclerosis in which intracerebral TMEV infection of mice leads to an autoimmune demyelinating disorder 30-40 days after infection 9; herpes simplex virus (HSV)-associated stromal keratitis, in which HSV infection leads to T-cell-mediated blindness in both humans and mice 10-12; diabetes model(s) 13; autoimmune demyelinating disease associated with Semliki Forest virus (SFV) 14; autoimmune myocarditis associated with coxsackie virus infection 15 and murine cytomegalovirus 16 (Table 1). Even so we primarily focus on a possible role of viruses in autoimmune diseases, other microbial pathogens have also been implicated in contributing to autoimmune disease by molecular mimicry, like for example streptococcus in rheumatoid myocarditis 17. There are also many less physiological scenarios that do not necessarily aim to closely model a particular disease but serve to reveal potential mechanisms through which immune responses to infections could lead to autoimmunity through molecular mimicry.
Along these lines, many investigations have used models of molecular identity, in which an exact microbial protein or epitope is transgenically expressed in a particular tissue. Under these conditions, animals are not generally susceptible to the development of spontaneous autoimmune disease. However, after infection with the microorganism containing the protein, autoimmunity in the transgenic protein-expressing organ ensues 18-21. Although these approaches are clearly artificial, they do indicate that T cells specific for a “self” antigen can become activated by infection with a microorganism containing an identical antigen, which provides appropriate innate immune signals, resulting in overt autoimmune disease. Even when the transgene-expressed antigen is also expressed in the thymus, so that normal mechanisms of negative selection significantly decrease the number of high-affinity T cells specific for the viral/self antigen, infection eventually results in autoimmunity 20. These experiments indicate that even T cells with low affinity for a self antigen and that have escaped negative selection, as would be the case for many self-antigen-specific responses, can be activated through molecular mimicry with a microbial antigen and cause disease. Similarly, in the TMEV-IDD model of multiple sclerosis, a severe rapid-onset demyelinating disease in the CNS is induced by intracerebral or peripheral infection with TMEV that has been engineered to express the immunodominant self epitope from myelin, proteolipid protein (PLP)139-151 22.
Several bacterial and viral peptides that mimic PLP139-151 23 have been identified, and these have been used in models that more directly address the possibility of autoimmune disease induced by molecular mimicry. TMEV engineered to express peptides derived from Haemophilus influenzae that mimic PLP139-151 (TMEV-HI, which shares 6 of 13 amino acids with PLP139-151) or murine hepatitis virus (TMEV-MHV, which shares only 3 of 13 amino acids with PLP139-151), induces a rapid onset, severe demyelinating disease that is similar to that induced by infection with TMEV expressing PLP139-151 22,24 itself. The H. influenzae mimic of PLP139-151 can also be processed and presented out of its own native flanking sequences when larger portions of the bacterial protein are expressed in TMEV, which further supports the potential role for molecular mimicry in a natural infection 25. Relevant to human disease, bacterial peptide mimics of the myelin basic protein (MBP) epitope 85-99, derived from such different pathogens as Mycobacterium tuberculosis, Bacillus subtilis and Staphylococcus aureus, induce demyelinating disease in mice that transgenically express a human MBP85-99-specific TCR as well as an HLA class II molecule that can present the peptide 26. Molecular mimicry was also shown in a model of diabetes in which lymphocytic choriomeningitis virus (LCMV) nucleoprotein was expressed under the control of the rat insulin promoter. Infection with Pichinde virus, which contains an epitope that mimics a subdominant epitope in the LCMV nucleoprotein, accelerated autoimmune disease that had already been established by previous infection with LCMV 13. HSV-induced stromal keratitis is mediated by corneal-antigen-specific T-cell responses that are induced following corneal HSV infection. This is a non-synthetic, natural model of autoimmune disease in which molecular mimicry can occur in the absence of genetic manipulation 10. The results from these experiments were called into question by a later study in which HSV-induced stromal keratitis could be induced in mice in the absence of T-cell responses against HSV 11. However, the former model allows for efficient HSV specific immune control, whereas in the absence of HSV specific immune control in the latter model, keratitis could be caused by pathogen induced immunopathology. Therefore, as different model systems with possibly different disease mechanisms were used to investigate HSV-induced stromal keratitis, the strong case for molecular mimicry put forth by the former study has yet to be ruled out 12.
In keeping with the observation that specific T cells that have been primed by pathogens and cross-react with self antigens can cause autoimmunity in animal models, patients with autoimmune diseases such as SLE, rheumatoid arthritis and multiple sclerosis have been found to have higher frequencies and activation states and/or less co-stimulatory requirements of self-reactive lymphocytes 27-30. In multiple sclerosis, receptor analysis of T and B cells in CNS tissue and in the cerebrospinal fluid (CSF) showed clonal expansions in both populations, indicating that there is clonal reactivity to a restricted number of disease-relevant antigens 31-33. In addition, longitudinal studies provided evidence for long-term persistence of individual myelin-specific T-cell clones tracked over several years in the blood of patients with multiple sclerosis 34-36, indicating a strong, persisting memory response and/or ongoing autoantigen exposure at least for a subset of myelin-reactive T cells in multiple sclerosis.
We suggest that these memory responses reflect, at least in part, persisting clonal expansions of polyspecific T cells recognizing both self and virus antigens that have been found associated with human autoimmune diseases (Table 2). For example, high viral loads that occur during symptomatic primary EBV infection, resulting in infectious mononucleosis, are associated with an increased risk of developing multiple sclerosis 37-39, and could prime these polyspecific T-cell responses. Accordingly, patients with multiple sclerosis have predominant clonal expansions of T cells specific for the EBV-encoded nuclear antigen 1 (EBNA1), which is the most consistently recognized EBV-derived CD4+ T-cell antigen in healthy virus carriers, and EBNA1-specific T cells recognize myelin antigens more frequently than other autoantigens that are not associated with multiple sclerosis 40. Notably, myelin and EBNA1 cross-reactive T cells produce interferon-γ (IFNγ) but differ from EBNA1-monospecific cells in their capacity to produce additional cytokines, such as interleukin-2, which is indicative of polyfunctional T cells. Because T cells successively use their ability to produce more than one cytokine during differentiation, polyfunctional T cells are thought to be particularly important under conditions of antigen persistence and high antigen load because they are less susceptible to clonal exhaustion or activation-induced cell death 41. However, viral titres in circulating blood cells from patients with multiple sclerosis are similar to those detectable in healthy virus carriers 42 and patients with multiple sclerosis do not differ from healthy EBV carriers in the rate of EBV-induced B-cell transformation or in their ability to control the outgrowth of EBV-infected B cells in vitro 40, which argues against increased viral replication or impaired immune control of chronic EBV infection driving EBV-specific T-cell expansion in patients with multiple sclerosis. Instead, a more extensive priming of polyfunctional cross-reactive T cells during symptomatic primary EBV infection with high levels of viral load, and continuous restimulation caused by autoimmune tissue inflammation, could potentially establish and maintain a distinct repertoire of myelin-reactive virus-specific T cells, which could predispose for multiple sclerosis.
APCs that have become activated within the inflammatory milieu of a pathogenic infection can stimulate the activation and proliferation of autoreactive T or B cells in a process known as bystander activation. In this case, APCs present self antigen, obtained subsequent to tissue destruction and/or the uptake of local dying cells, to autoreactive cells 43,44 (Figure 2B). In addition to autoimmune responses that are initially primed by APCs and stimulated by bystander activation, additional autoantigen-specific T or B cells can be primed through epitope spreading 45 — a situation in which an immune response that is initiated by various stimuli, including microbial infection, trauma, transplanted tissue or autoimmunity, ‘spreads’ to include responses directed against a different portion of the same protein (intramolecular spreading) or a different protein (intermolecular spreading) (Figure 2B). Activating a broader set of T cells through epitope spreading is helpful in an antipathogen or antitumour immune response, because the pathogen or tumour cannot easily escape immune control with a single mutation in an immunogenic epitope. However, disease potentially arises when spread to and within self proteins occurs subsequent to the destruction of self tissue. Epitope spreading in animal models proceeds in an orderly, directed and hierarchical manner, such that more immunodominant epitopes elicit responses first, followed by less dominant responses. This type of spreading has been shown in experimental autoimmune encephalomyelitis (EAE), a non-infectious model of multiple sclerosis 46,47, as well as in TMEV-IDD 9,48-50 and in the non-obese diabetic (NOD) mouse model of type 1 diabetes 51 (S.D.M., unpublished observations). While these examples document epitope spreading within autoantigens and to additional autoantigens, the inflammatory environment of viral infections could support these immune response cascades by increasing the presentation of self antigens via provision of signals through PRRs.
An even broader form of bystander activation is achieved by cross-linking MHC class II molecules on APCs with TCRs comprising a certain Vβ domain by superantigens. T-cell populations that are stimulated in this manner could potentially contain a subset of T cells specific for a self antigen 52. There are multiple examples in which superantigens are involved in diseases such as EAE, arthritis and inflammatory bowel disease, making superantigens another mechanism by which bystander activation can initiate, or at the least exacerbate autoimmunity in mouse models 53-55. In these studies, staphylococcal, mycoplasma- and enteric microbiota-derived superantigens were shown to amplify, but not initiate, autoimmune T cell responses (Table 1). Furthermore, the association of certain genotypes of the superantigen-encoding endogenous retrovirus HERV-K18, which is transactivated by EBV 56, with multiple sclerosis has been reported 57. However, Vβ7+ and Vβ13+ T-cell populations, which are stimulated by the retroviral superantigen, do not seem to be selectively expanded in patients with multiple sclerosis. Nevertheless, viral-antigen-specific and/or superantigen-expanded T cells might participate in the development or maintenance of autoimmune disease. So although molecular mimicry might initially prime autoreactive T cells, these responses could be amplified by superantigen-mediated expansion and by activation of autoantigen-specific T cells that express a given Vβ chain that is targeted by microbial superantigens.
Infections can affect the immune response in many ways, and mechanisms such as molecular mimicry and bystander activation are certainly not the only ways in which pathogens might trigger or accelerate autoimmune disease. A recent study showed that in a spontaneous animal model of SLE, lipid raft aggregation on T cells, which was induced by Cholera toxin B from Vibrio cholerae in this particular study, but could be induced by several microorganisms or toxins, enhanced T-cell signalling and exacerbated SLE 58. Furthermore, viral infections could also directly maintain autoreactive effector cells or autoantigen-presenting cells 59. As one example, persistent infection of microglia with TMEV has been shown to cause upregulation of MHC and costimulatory molecules and enhance the ability of these cells to function as effective APCs 60. In another example, EBV immortalizes B cells and assists in their differentiation to long-lived memory B cells 61. In addition, even in infected memory B cells, that usually do not express the latent EBV proteins that are associated with immortalization, non-translated viral RNAs contribute to resistance to cell death 62. These mechanisms could preserve autoreactive B cells or a reservoir of APCs capable of presenting autoantigens to promote autoimmunity. For example, a reservoir of EBV-infected B cells was recently found in submeningeal aggregates of brains from patients with multiple sclerosis63. Although several causal relationships between pathogen infection and autoimmunity have been identified in animal models and correlations have been drawn in human autoimmune diseases, pathogen-derived triggers of autoimmunity have been difficult to identify, because evidence of autoimmunity is likely to become clinically apparent only after a considerable period of subclinical autoimmune responses, at which time the pathogen might have already been cleared and/or the antiviral immune responses might have subsided, the so-called ‘hit-and-run’ hypothesis.
All of these mechanisms are interrelated, non mutually exclusive, and dynamic, so the idea of microbial infection eliciting autoimmunity must be viewed not as a defined event that occurs through a particular mechanism, but as a process that can occur through many pathways simultaneously and/or sequentially (Figure 2). For example, epitope spreading can be initiated through molecular mimicry as illustrated by the activation of PLP178-191-specific T cells in SJL mice in which autoimmunity was induced by infection with TMEV expressing either PLP139-151 or a PLP139-151 mimic peptide 22, or following bystander damage. Molecular mimicry can initially activate autoreactive T cells that then expand and become pathogenic through bystander activation, or vice versa. As a result, it can be difficult to distinguish between the postulated mechanisms, even in seemingly simple animal models 5,11,12,64.
Animal studies have made it clear that in principle, infections can trigger autoimmune responses. This must be distinguished, however, from the elicitation of overt autoimmune disease as a direct result of microbial infection, which might be more difficult to establish.
Autoreactive adaptive immune cells are unavoidably present in the periphery in humans and animals. These cells can exist because their cognate self-antigen was not expressed in the thymus and the antigen will therefore only become apparent to the immune system after tissue destruction as a result of infection or trauma. Alternatively, whereas many autoreactive T cells are deleted in the thymus during development, some T cells that make their way to the periphery might be highly specific for a microbial antigen, but also have lesser affinity for a self antigen. The presence of autoreactive cells in the periphery, however, does not necessarily predispose for clinical autoimmune disease.
It is clear that in many cases, an infection is necessary for the development of overt disease, even when abundant autoreactive T cells are present. A cogent example is that although demyelinating disease is readily induced in mice either by priming with PLP139-151 peptide in complete Freund’s adjuvant (CFA), or by infecting with TMEV expressing a PLP139-151 mimic, priming with PLP139-151 mimics in CFA does not induce overt disease, despite the fact that T cells from mice primed with mimic peptides robustly respond to PLP139-151 23,24,44. Presumably, TLR engagement and other innate immune stimuli that are present upon infection with TMEV expressing the mimic peptide allow APCs to provide the necessary signals for full-blown activation and optimal migration of autoreactive T cells 60. So, the nature of a pathogen, which directs the type of immune response that is elicited, can profoundly influence the potential for development of autoimmune disease, and could in fact increase or decrease the likelihood of autoimmunity in the presence of autoreactive cells. In this regard, T helper 1 (TH1)- and TH17-polarized T-cell responses have been proposed to accelerate autoimmunity, whereas TH2-polarized responses might confer protection 65. Also, in the case of molecular mimicry, the virus-encoded mimic itself has an important role, as a peptide that partially mimics a self antigen (known as an altered peptide ligand), depending on the context of the infection, could have a tolerizing rather than an immunizing effect 60.
Even the presence of autoreactive T cells together with the presence of an appropriate infection might not lead to autoimmune disease. For example, in Pichinde virus infection of RIP-LCMV-NP mice, although the mimic-encoding Pichinde virus was not sufficient to initiate overt autoimmunity, it was able to accelerate autoimmune disease that had already been established by infection with LCMV 13. So, viral “adjuvant” and self mimics might in some cases be able to trigger autoimmune disease only if autoreactive cells are already ‘primed’ to some degree, such that the autoreactive T cells have been previously activated and exist at a higher precursor frequency 66. The affinity of TCRs for various self-peptide–MHC complexes might also have a key role in the development of autoimmune disease. Indeed, a threshold level of TCR affinity has been shown to be important for the establishment of autoimmunity 67. In RIP-LCMV mice, whether or not the self antigen was expressed in the thymus during development (which affects T-cell affinity) has a significant impact on the speed with which autoimmune disease develops 20. TLR engagement alone can be sufficient to induce the correct environment for the development of autoimmune disease if autoreactive T cells are of high enough affinity for self antigen 20,68. However, as most T cells will have low affinity for self under physiological situations, studies in which TCR affinity for self-antigen is low may have greater relevance to human autoimmune disease.
The potential for the development of overt disease is clearly dependent on the presence of autoreactive T cells. However, whether overt disease actually occurs can depend on various other coincident events, including the number of autoreactive T cells present, the avidity and affinity of these cells (determined by co-receptor expression and binding to MHC/peptide complexes, respectively), and the presence of innate inflammatory signals required for activation and differentiation of those T cells to a pathogenic phenotype. Despite the requirement for all of these elements, it is clear that they do not need to happen at the same time or in the same place to elicit autoimmune disease.
In many animal models, autoimmune responses are triggered during the initial or acute response to an infection, and autoimmune disease occurs exclusively in the infected organ, such as during corneal HSV infection leading to stromal keratitis 10 11,12. Along these lines, submeningeal reservoirs of EBV infected B cells have been reported in the brains of patients with multiple sclerosis 63, but it remains unclear if these focus pathogenic immune responses to the diseased tissue. Models in which infection directs autoreactive responses to distinct tissues provide a simple system in which to study the pathological mechanisms of infection-induced autoimmunity. However, in most cases a robust immune response to a pathogenic infection in the target organ is usually not associated with the development of autoimmunity in humans. None of the proposed mechanisms for the development of infection-induced autoimmunity excludes the possibility that disease can occur temporally and/or spatially distal from the site of the initiating infection (Figure 3). Animal models that allow investigators to study this aspect of infection-induced autoimmune disease are few, but they might provide important insights relevant to human disease.
Autoimmune CNS demyelinating disease can be triggered by molecular mimicry when the pathogen containing the mimic epitope does not infect the CNS itself. When mice that express an LCMV protein in the CNS were peripherally infected with LCMV, autoimmune responses occurred in the CNS despite the fact that LCMV was not detectable in that organ 21. In non-transgenically manipulated mice, pancreatropic Coxsackie virus engineered to express PLP139-151, induces CNS demyelinating disease associated with PLP139-151-specific T-cell responses, also in the absence of any apparent infection in the CNS itself (S.D.M.,unpublished observations).
The fact that the various mechanisms for infection-induced autoimmunity discussed here are non-mutually exclusive make them both more complicated and more plausible as potential causes for human autoimmune disease. For example, molecular mimicry and adjuvant effects of pathogens might work early on during the development of autoimmune responses, whereas bystander activation through the inflammatory environment of infections and/or superantigens might exacerbate autoimmune responses later on. However, as we consider the potentially multi-mechanistic and multi-step nature of autoimmunity, it is important to remember that an established autoimmune response can also have effects on pathogen-directed immune responses occurring in the same organ or elsewhere in the body.
The flip side of autoimmunity being driven by viral infections is that autoreactive immune responses or even only predisposition to these might affect the development of antiviral immune responses. This might alter their composition, the viral set-point during chronic infections and the anatomical distribution of virus-specific lymphocytes. These alterations could be used as diagnostic markers for autoimmune disease activity, but might not regulate the autoimmune disease itself.
The activation of innate immune cells can be initiated by both pathogen-associated “stranger” signals 69 and damage-associated, altered-self “danger” signals 70. These apparently disparate mechanisms trigger inflammation through common means, as stranger and danger signals both ligate PRRs. TLRs have a particularly instructive role in innate immune responses against microbial pathogens, as well as the subsequent induction of adaptive immune responses. Both experimental infections in mice lacking individual TLRs and key molecules of the TLR signaling pathways 71 as well as natural infections in humans with primary immunodeficiencies that selectively impair TLR responses 72 clearly show the crucial role of TLRs in shaping protective antiviral immunity.
A role for TLR signaling in the induction and maintenance of autoimmune diseases was first highlighted by Leadbetter et al. 73 who showed that immunoglobulin in the blood provokes autoimmune responses when immune cells recognize it as a complex with self-DNA. In B-cell receptor (BCR)-transgenic mice in which most B cells express surface antibodies with low affinity for self-IgG2a, which is the rheumatoid-factor specificity, immunoglobulin neither activates the B cells nor makes them tolerant unless these mice are crossed onto an autoimmune-prone lpr background; self-IgG2a in the offspring of this cross is immunogenic, resulting in high titres of circulating rheumatoid-factor autoantibodies, a diagnostic marker of autoimmune disease 74. This study found that immune complexes consisting of self-IgG2a and self-DNA, triggering both surface BCRs and endosomal TLRs, were necessary and sufficient for the loss of self-tolerance in this model. Similar models have been reported for RNA-containing immune complexes and TLR7/TLR8 activation 75,76. So, endogenous TLR ligands such as self-DNA or RNA or nucleic acid-associated proteins could act as adjuvants in autoimmune diseases that are characterized by prominent tissue damage or impaired removal of apoptotic cell or necrotic debris 77, and could assist in the priming of anti-viral immune responses.
Furthermore, the induction and maintenance of autoimmune tissue inflammation crucially depends on the cytokine polarization profile of pathogenic TH cells in animal models of T-cell-mediated autoimmune diseases 65,78,79. Both TH1 and TH17 cells are thought to coordinate autoimmune inflammation in these diseases, presumably through distinct pathways 80. Although the TH1-cell cytokine IFNγ can inhibit the generation of TH17 cells, it also reinforces TH1-cell differentiation 78, which is instrumental in establishing protective antiviral immune responses. Therefore, the TH1-polarizing milieu of autoimmune diseases might support superior anti-viral immune responses 81,82.
Although it has not yet been shown in experimental models, we suggest that the increased availability of intrinsic “danger-signals” released through autoimmune tissue damage probably both affects host responses to microbial pathogens at sites of autoimmune inflammation and enhances pathogen-specific innate and adaptive immune responses.
In addition to the adjuvant activity of autoimmunity, which might enhance pathogen-specific immune responses, autoimmunity can also affect pathogens that persist in lymphocytes. HTLV-1 and EBV are classical examples, with HTLV-1 persisting in memory T cells and EBV persisting in memory B cells. Both viruses seem to establish latency without detectable antigenic protein expression in these cellular reservoirs 83,84. Reactivation of HTLV-1 occurs upon TCR and co-stimulatory molecule engagement 85. Similarly, lytic replication of EBV can only be observed in plasma cells 86, and it can be induced by cross-linking of surface immunoglobulin on infected B cells 87. Therefore, autoimmunity could trigger reactivation of these pathogens (Figure 4), as has been documented in the case of EBV reactivation by malaria antigens 88. Indeed, patients with SLE have abnormally high frequencies of EBV-infected cells that have aberrant expression of the immediate early lytic antigen BZLF1 in peripheral blood 89. Interestingly, increased cell-associated viral loads correlated with autoimmune disease activity. Moreover, increased EBV loads correlated with EBV-specific CD8+ T-cell responses in patients with SLE 90, which had decreased cytotoxicity as a sign of exhaustion 91, possibly due to persistent restimulation by the high antigenic load. By contrast, CD4+ T-cell responses to EBV were also upregulated in patients with SLE, but negatively correlated with viral loads, suggesting enhanced immunoprotection by these cells 90. Whereas EBV viral loads are up to 40-fold increased in patients with SLE, they are 10-fold increased in patients with rheumatoid arthritis 92. Again, CD8+ T-cell responses to EBV antigens positively correlate with these increased viral loads 93. Also in rheumatoid arthritis the increased antigen load seems to cause further differentiation of EBV-specific CD8+ T cells, resulting in the presence of a subpopulation of terminally differentiated, and presumably costimulation-insensitive, CD27−CD28− cells that are rarely observed in healthy EBV carriers 93,94. In addition to EBV-specific T-cell responses, subdominant antibody responses and broadened responses to dominant EBV antigens are also observed in patients with rheumatoid arthritis93, again suggesting that the increased EBV antigen load in these patients hyperstimulates EBV-specific humoral and cell-mediated immune responses. Although these increased immune responses maintain EBV-specific immune control in most cases, the increased autoantigen-mediated stimulation of the B-cell compartment can even result in lymphoma development by driving B cells to hypermutation and germinal centre reactions, which increases the risk of acquiring transforming mutations. In the case of rheumatoid arthritis, Hodgkin lymphomas, including EBV+ tumours, are more strongly associated with the autoimmune disease than non-Hodgkin lymphomas 95. These studies suggest that lymphotropic pathogens such as EBV can be affected by autoimmune stimulation of host immune cells, leading to increased viral titres, increased immune responses against the pathogen and even pathogen-associated malignancies. Collectively, the evidence indicates that dysregulation of EBV-specific immune responses is a feature of rheumatoid arthritis and SLE, and is probably driven by autoantigen-mediated activation of EBV-infected B cells.
Family-based genetic epidemiological studies provide unequivocal evidence that the susceptibility for autoimmune diseases is inherited, and genome-wide microsatellite screens and large-scale single nucleotide polymorphism (SNP) association studies have identified chromosomal loci that are associated with specific disorders such as SLE, rheumatoid arthritis, type 1 diabetes and multiple sclerosis. HLA-DR and -DQ alleles within the HLA class II region on chromosome 6p21 are by far the strongest risk-conferring genes for all of the aforementioned conditions 96,97,98,99. Although the MHC region has proven difficult to dissect because of its intense and variable patterns of linkage disequilibrium, there is evidence that additional loci in the HLA class III and HLA class I genomic regions and loci telomeric of genes encoding the classical MHC might have independent associations with autoimmune disease 96. Furthermore, less robust susceptibility effects have been identified in non-MHC regions. Among these, the ITGAM-ITGAX region on chromosome 6p11, which encodes the α-chain of the αMβ2-integrin (also known as Mac-1, CR3 and CD11b), which is important for neutrophil and monocyte adherence to stimulated endothelium as well as the clearance of immune complexes, was associated with SLE in multiple studies 97,100,101, and the IL7RA region on chromosome 5p13 and IL2RA on 10p15 were identified as loci associated with multiple sclerosis 102-104. Epistatic interactions between these risk-conferring and protective allelic variants are thought to define the overall genetic threshold for disease susceptibility 8.
We propose that immune functions of autoimmune susceptibility genes and their products probably affect host-pathogen interactions in patients with autoimmune diseases 72. Along these lines, it was shown that CD8+ T cells recognizing an immunodominant EBV epitope, restricted by HLA-B8, cross-reacted on HLA-B*4402, presumably presenting a self-peptide 105. This cross-reactivity was strong enough to mediate alloreactivity against HLA-B*4402+ cells, and result in deletion of this EBV specificity in HLA-B*4402+/HLA-B8+ individuals by negative selection, which ablates T cells of this specificity from the repertoire of EBV-specific immune responses in these individuals. Incomplete deletion of these alloreactive T cells in genetically susceptible individuals could however result in autoreactive and EBV-specific T cells. Similar to the genetic variation of the host predisposing for particular viral peptide presentation and recognition, as documented by the just discussed example, genetic variation of viruses might also favour peptide presentation that can stimulate cross-reactive T cell specificities. For example, the EBV strain B95-8 carries a point mutation in the HLA-B8 restricted CD8+ T-cell epitope discussed above, which affects T-cell recognition of the virus through inefficient binding to HLA-B8 106. Vice versa, autoreactive T cell specificities could be preferentially triggered by certain virus strains encoding peptide epitopes that stimulate cross-reactive T cell specificities. With increased sequence variation also mapping to EBNA1 107, especially an association between multiple sclerosis and a particular EBV strain remains possible, and sequence variation in the viral strain might enhance particular EBV-specific T- and B-cell responses that could participate in the autoimmunity. Therefore, the particular HLA background of an individual and the distinct viral strains carried by the individual could select for T-cell specificities with autoreactive capacity during anti-viral immune responses.
Considering this potential mechanism, it is apparent that disease promoting or protective effects of gene-environment interactions should ideally be investigated in patients with autoimmune diseases in comparison to syngeneic controls such as non-affected monozygotic twins 108. As this is not feasible, we suggest that patients and controls should at least be matched for expression of large-size risk alleles such as HLA-DR and HLA-DQ allelic variants as this strategy minimizes the risk that detected differences in pathogen-specific immune responses are consequence rather than cause of disease susceptibility.
Besides genetic factors on both the pathogen and the host side that might boost distinct virus-specific immune responses, the preferential homing of primed antiviral T cells might also falsely implicate pathogens in the immunopathology of autoimmune diseases (Figure 5). For example, lytic EBV infection was suspected to contribute to rheumatoid arthritis after it was found that lytic EBV antigen-specific T cells were enriched in inflamed joints 109. Indeed CD8+ T cells specific for immediate early and early lytic EBV antigens were first cloned from the synovial fluid of patients with rheumatoid arthritis 109. These specificities, which are now recognized to be among the most frequent T-cell responses, develop during persistent infection with EBV 110. Similarly the first lytic EBV antigen-specific CD4+ T-cell responses were also cloned from this source 111. Interestingly, it was later found that these lytic EBV antigen-specific T cells home to a variety of autoimmune inflamed tissues 112, including knees affected by rheumatoid arthritis, eyes of patients with uveitis, and the brains of patients with multiple sclerosis. These data were interpreted as primarily reflecting the migration of EBV-specific T cells in response to inflammatory chemokines, such as the CXCR3 ligand CXCL10, rather than a direct involvement of EBV-directed immunity in the immunopathology of the autoimmune diseases. Similarly pathogen infected lymphocytes could preferentially migrate to inflamed tissues and this localization could be erroneously interpreted as a contribution of the pathogen in the autoimmune pathology, whereas it reflects in reality only the normal migratory behavior of infected host cells. The enrichment of EBV infected B cells in tertiary lymphoid tissue in post mortem CNS tissue form patients with multiple sclerosis might result from changes in the migratory behaviour of infected B cells 63, and needs to be further evaluated. In particular, CXCL13-mediated recruitment of CXCR5+ B cells to multiple sclerosis lesions should be evaluated along these lines 113. Therefore, the enrichment of both pathogen-specific and pathogen-infected lymphocytes at sites of autoimmunity might be informative with respect to the migration behaviour of pathogen-infected host cells and lymphocytes targeting them, rather than indicating the involvement of the respective pathogens in the immunopathology of autoimmune diseases. Nevertheless, a better understanding of these mechanisms could suggest how certain pathogens focus the autoimmune reactivity of a sensitized immune system to certain organs and reveal the diagnostic usefulness of monitoring antiviral immune responses in patients with autoimmune diseases.
Besides genetic predisposition to autoimmunity, environmental factors are known to be involved in the initiation and promotion of autoimmune diseases. Among these, viral infections are main candidate environmental factors due to their capacity to elicit strong immune activation and to induce autoimmune diseases in animal models, as well as the correlation of viruses with autoimmune diseases in humans, as described in the recent studies highlighted in this review. These studies suggest that viruses can trigger autoimmunity through molecular mimicry and their adjuvant effects during initiation of disease, and can promote autoimmune responses through bystander activation or epitope spreading via inflammation and/or superantigens. However, an association of dysregulated antiviral immune responses with a given autoimmune disease has to be interpreted with caution, because these can be differently primed in individuals with ongoing autoimmune disease or with a genetic predisposition to autoimmune disease. Furthermore, the autoimmune disease can alter virus infection by affecting its host cells, and might lead to redistribution of antiviral lymphocytes to sites of autoreactive tissue inflammation. These changes might prove to be useful as diagnostic markers for autoimmune reactivity, and might be harnessed therapeutically. However, any therapeutic approach that targets these responses should be done with caution so that immune control against the pathogen is not compromised.
Work in the authors’ laboratories is supported by the Dana Foundation’s Neuroimmunology program, the Arnold and Mabel Beckman Foundation, the Alexandrine and Alexander Sinsheimer Foundation, the Burroughs Wellcome Fund, the Starr Foundation, the National Cancer Institute (R01CA108609 and R01CA101741), the National Institute of Allergy and Infectious Diseases (RFP-NIH-NIAID-DAIDS-BAA-06-19), the Foundation for the National Institutes of Health (Grand Challenges in Global Health), and an Institutional Clinical and Translational Science Award (to the Rockefeller University Hospital) to C.M.; by a Dana Foundation and Irvington Institute’s Human Immunology Fellowship, a Pilot Grant from the National Multiple Sclerosis Society (PP1145) and by an Institutional Clinical and Translational Science Pilot and Collaborative Project Grant (to the Rockefeller University Hospital) to J.D.L.; by the National Institute for Neurological Diseases and Stroke (R01 NS-023349, R01 NS-040460 and R01 NS-030871), the National Multiple Sclerosis Society (RG 3793-A-7), and the Myelin Repair Foundation to S.D.M. and M.T.G.