|Home | About | Journals | Submit | Contact Us | Français|
Despite having long been postulated, compelling evidence for the theory that microbial triggers drive autoimmunity has only recently been reported. A specific association between Novosphingobium aromaticivorans, an ubiquitous alphaproteobacterium, and primary biliary cirrhosis (PBC) has been uncovered in patients with PBC. Notably, the association between Novosphingobium infection and PBC has been confirmed in a mouse model in which infection leads to the development of liver lesions resembling PBC concomitant with the production of anti-PDC-E2 antibodies that cross-react with conserved PDC-E2 epitopes shared by Novosphingobium. The discovery of infectious triggers of autoimmunity is likely to change our current concepts about the etiology of various autoimmune syndromes and may suggest new and simpler ways to diagnose and treat these debilitating diseases.
The underlying etiologies of autoimmune diseases have been difficult to elucidate, as they are driven by complex interactions between environmental factors and genetic traits [1–3]. Although autoimmunity often clusters together in individuals and families, indicating the potential for a broad-spectrum genetic defect in immunological tolerance mechanisms, the genetic factors leading to the development of immune responses against specific antigens in a tissue and/or organ-specific manner remain largely unknown. This compartmentalization of autoimmune responses is particularly surprising, as many autoantigens are ubiquitously expressed.
Although many environmental factors have been associated with the development of autoimmune diseases, infectious agents have long been considered as potential culprits for the induction of these disorders [4–13]. For example, rheumatic fever represents one of the prototypes of autoimmune disease of infectious origin in clinical practice . Other well-documented examples include autoimmune gastritis and Guillain–Barré syndrome [15–17] (summarized in Table 1). Despite these examples, the establishment of a strong association between a specific microbial infection and the development of autoimmune disease in patients remains a challenge.
Infection-induced models of autoimmune disorders, involving either the induction of nonspecific inflammation or the ectopic expression of self-antigens on microbes and/or host tissue in animals [18–20], have shed light on the potential role of infections in the development of autoimmune diseases and have led to the evolution of the following general concepts for the induction of autoimmunity by microbial infection:
The structural relationship between mammalian and bacterial structures may elicit cross-reactive immune responses. While this may contribute to the eradication of bacteria, ‘mimic’ specific T or B cells can cross-react with self-epitopes, thus leading to tissue pathology and autoimmunity .
This may apply to Guillain–Barré syndrome mediated by antilipopolysaccharide antibodies to Campylobacter jejuni cross-reacting with human gangliosides , or to rheumatic fever caused by an antistreptococcal immune response that cross-reacts with cardiac myosin , and possibly to late-phase Lyme disease . Similar mechanisms may be associated with Helicobacter pylori infection and chronic atrophic/autoimmune gastritis [16,17]. Molecular mimicry might also be playing a role in the pathogenesis of primary biliary cirrhosis (PBC), a chronic cholestatic liver disease. In support of the potential role of molecular mimicry in the pathogenesis of PBC, it has recently been shown that undetected infections with an ubiquitous alphaproteobacterium, Novosphingobium aromaticivorans, are strongly associated with PBC in humans [31–35]. Based on these clinical reports, we have begun studying the mechanisms by which infection with Novosphingobium triggers autoimmunity and propagates autoreactive T and B cells in a mouse model resembling PBC without the need for ectopic antigen presentation. We provide compelling evidence that activation of natural killer T (NKT) cells, elicited due to unique cell wall antigens of Novosphingobium, explains the liver-specific pathology, although the potential T- and B-cell antigen, the E2 subunits of the pyruvate dehydrogenase complexes (PDC-E2), is ubiquitously expressed.
The division/phylum of proteobacteria and the class of alpha-proteobacteria therein have been particularly often associated with autoimmunity (Figure 1). Infections with Brucella spp., for example, have been reported to involve multiple organ systems  and to induce (chronic) inflammation and hepatitis, immune-induced thrombocytopenia and/or thrombocytopenic purpura [37–47]. Emerging pathogens such as Ehrlichia spp. have also been associated with similar disease characteristics [48–51]. Although the pathologic mechanisms are not understood to date, disease symptoms and inflammation often persist even though microbial copies are hardly recovered from the respective tissues, suggesting a potential autoimmune component elicited by the bacteria – a phenomenon that has also been observed with other infections, such as Chagas or Lyme disease [40,52–54].
Even though multiple organ systems can be involved, the liver often becomes the target of these aberrant immune reactions during or after infection with alphaproteobacteria, potentially due to the preferential persistence of microbes at this organ site [55–65]. Having a unique and distinct cellular composition, with predominant abundance of Kupffer cells (KCs), natural killer (NK) cells, and NKT cells, the liver is considered an organ with innate immune features. Being exposed to microbial products, toxic environmental substances and food antigens due to the blood circulation from the intestine, plays a critical role in the induction of immune tolerance [66,67]. However, the liver may be involved in many systemic diseases and can become the target of adverse immune reactions in (putative) autoimmune diseases  such as PBC. Bacteria and/or bacterial products translocating from the intestine to the liver may be involved in these processes.
The pathogenic mechanisms of biliary injury in PBC are poorly defined. The stimuli that trigger autoreactivity are unknown but include both genetic and environmental factors [69,70]. Both factors are implicated by the clinical uniformity of PBC and the presence of a highly specific serologic marker, antimitochondrial antibodies directed against PDC-E2 (Box 2) [26,71] . These subunits are both the dominant autoreactive B-cell epitopes and the main antigens recognized by liver-infiltrating, autoreactive T cells [72,73]. However, it is not understood to date why these mitochondrial antigens are specifically targeted and why the immune reaction is restricted to the liver, although the autoantigen is ubiquitously expressed in all tissues.
Molecular mimicry has been proposed as a mechanism for the development of autoimmunity in PBC . Environmental factors implicated in PBC pathogenesis include tobacco, reproductive hormones, exposure to nail polish or toxic waste, xenobiotics and repeated urinary tract infections [75,76]. Vaccination studies in mice that use 2-octanoic acid (2-OA) for the induction of histologic lesions as well antibodies to PDC-E2 provide an argument in favor of an environmental origin for human PBC and the induction of disease due to molecular mimicry . Some bacteria and viruses have also been suggested as causative agents. One particular alphaproteobacterium, however, N. aromaticivorans, stands out because it has a 100–1000-fold greater homology with the immunodominant region of human PDC-E2 than any microorganism studied thus far.
N. aromaticivorans is a Gram-negative bacterium belonging to the Sphingomonodaceae family (Figure 1 & Box 3) that is found ubiquitously in the environment [78–81], at human mucosal surfaces and in human feces . Strains of Sphingomonas/Novosphingobium exhibit xenobiotic-metabolizing properties [31,78–81] and degrade a wide variety of environmentally hazardous compounds, including polycyclic aromatics , dioxine compounds  and chlorinated phenols .
The biochemical identification of glycuronosylceramides as substitutes for lipopolysaccharide was originally made in Sphingomonas capsulata [100,102], although the presence of these glycolipids has been extended to all other genera tested so far, including Novosphingobium aromaticivorans . The unique cell wall of these bacteria elicits distinct innate immune pathways that may be responsible for triggering liver-specific pathology.
Sphingomonas/Novosphingobium spp. [85,86], preferentially Sphingomonas paucimobilis (formerly Pseudomonas paucimobilis), have been implicated in a variety of community-acquired and nosocomial infections, including bacteremia, catheter-related sepsis, meningitis, peritonitis, pneumonia, cutaneous infections, visceral abscesses, urinary and biliary tract infections, adenitis and diarrheal disease [87–97]. The lack of a typical lipopolysaccharide constituent of the cellular membrane of Sphingomonas/Novosphingobium, accompanied by the deficiency of endotoxin activity, may explain the lack of deaths attributed to this organism [98–100].
Administration of antibiotics resulted in the resolution of infection [87–97]. Nosocomial Novosphingobium infection can be as resistant as Pseudomonas infection, therefore combination therapy of a third-generation cephalosporin and an aminoglycoside is recommended .
The cell wall of the Sphingomonodaceae is unusual because it contains glycosphingolipids (GSLs) instead of LPS [98–100]. NKT cells specifically recognize Sphingomonodaceae GSLs and, in the absence of TLR-4 activation by LPS, dominate the innate immune response [102,103]. A latent, unrecognized infection with N. aromaticivorans or related Novosphingobium spp./bacteria may therefore account for the increased NKT cell numbers that may react to Novosphingobium GSLs, as suggested by our mouse model (see later), and the enhanced expression of CD1d that is found in the livers of PBC patients [104–107].
Molecular mimicry between mitochondrial enzymes and phylogenetically related antigens of Novosphingobium, which share conserved PDC-E2 epitopes, has been suggested [31–34]. Analysis of mitochondrially encoded genes and their genomic organization/distribution imply that mitochondrial genomes are derived from an alphaproteobacterium (-like) ancestor that invaded an Archea-type host more than 1.5 billion years ago . This evolutionary association of alphaproteobacterial and mammalian mitochondrial antigens might be critical for the development of PBC and the target specificity of the immune attack to mitochondrial enzymes. Targeting of these enzymes and subsequent interference with the metabolism of the bacterium may reflect an ancient defense mechanism of the immune system against the invading agent, although we do not know why the inner mitochondrial oxoacid dehydrogenase complex – and not other mitochondrial proteins – are targeted by the autoimmune assault, as several other diseases are associated with mitochondria; however, this pathologic pattern and evolutionary relationship may have a broader impact and translate to those as well.
Patients with PBC express antibodies against lipoylated Novosphingobium proteins, including the bacterial homolog of the mitochondrial enzyme PDC-E2, the major PBC antigen [31–34]. In fact, seropositivity for Novosphingobium is found in seronegative PBC patients and is therefore more closely associated with disease than the presence of anti-PDC-E2 antibodies. Seropositivity for Novosphingobium is highly specific to PBC and is not found in healthy subjects or in other diseases . Although these antibodies reflect the hallmark of PBC, their role in the pathogenesis of PBC has remained unknown. As in other autoimmune diseases, there has not been established an association between the titers and the severity/progression of disease. We are currently investigating the role of B cells and antibodies in our mouse model.
As Novosphingobium/Sphingomonas are found at mucosal surfaces and are part of the bacterial flora of the gut, bacterial translocation may occur. Genetic predisposition may predispose for this translocation. Therefore, genetic susceptibility may not only reflect the predisposition of an individual to develop responses against self, but also the inability to prevent systemic infection with a Gram-negative bacterium that does not cause apparent clinical symptoms in otherwise healthy (not severely immunocompromised) individuals due to the lack of LPS. An infection with Novosphingobium may therefore remain undetected. Subsequent studies now need to analyze the port of bacterial entry from the intestine into the liver. Possible alternatives are an ascending bile duct infection, or a barrier defect allowing bacteria access to the bloodstream and subsequently to the liver.
Based on these clinical reports, we have established a model where infection of mice with Novosphingobium induces anti-PDC-E2 IgG responses and liver lesions resembling PBC in humans (Figure 2) . Onset and severity of liver disease in this model was dependent on the mouse genetic background, hepatic persistence of Novosphingobium and hepatic presence of NKT cells activated by Novosphingobium GSLs. The greater persistence of Novosphingobium in the liver than in other organs and the activation of NKT cells, which are particularly abundant in the liver of mice, by Novosphingobium GSLs explain the liver-specific pathology. Therefore, this model does not only unleash a mechanism for the organ-specific development of (auto-)immune responses to ubiquitously expressed antigens, but also reflects the situation in humans that is characterized by the redistribution of NKT cells from the blood to the liver. In contrast to many other disease models of infection-triggered autoimmunity, no ectopic expression of host antigens by the pathogen and/or the mouse is required in our animal system.
Induction of autoimmunity in this model is strictly dependent on bacterial persistence and chronic infection and could not be induced by the injection of heat-killed bacteria alone. Selective application of antibiotics (e.g., ampicillin and streptomycin) during these early stages also abolished the development of liver lesions and long-term autoimmunity. Although persistent bacterial infection is critical for the induction of full-blown autoimmunity in the mouse model in the initial stages of disease, liver lesions persist and even advance after bacterial clearance, suggesting an autoimmune component. Once disease is established in the mouse model, liver lesions can be adoptively transferred by conventional CD4+ and CD8+ T cells from wild-type, but not NKT-deficient, mice in the chronic phase. This illustrates the importance of early microbial activation of NKT cells in initiating autonomous, organ-specific autoimmunity. However, the antigens these T cells are reacting to are unknown and need to be evaluated (Figure 2).
NKT cells also provide help for PDC-E2-reactive B cells, as indicated by reduced anti-PDC-E2 IgG titers in NKT-cell-deficient mice. Interestingly, direct cognate interactions between NKT and B cells are required to produce in particular anti-PDC-E2 IgG2a responses, the subclass of autoantibodies that is considered pathogenic. Future studies need now to evaluate the role of B cells and these autoantibodies in the pathogenesis of liver disease, by using B-cell depletion approaches and adoptive transfer studies.
Although our data suggest that molecular mimicry may initiate disease itself (Figure 2), additional studies need to be performed to confirm this hypothesis, as results in a diabetes model suggested that disease can only be accelerated, but not initiated, due to molecular mimicry . However, the unique innate signals associated with Novosphingobium infection may promote the expansion of PDC-E2-reactive B and T cells (that may even only be present at low frequencies) as NKT cell-promoted help for coadministered protein antigens is a well-established concept [111–116].
The observation that PBC might be triggered by an intestinal alphaproteobacterium opens new routes and possibilities for treating and diagnosing this devastating disease. Activation of NKT cells by Novosphingobium GSLs in the liver, where NKT cells are abundant and Novosphingobium persists, explain the organ-specific break of tolerance to an ubiquitous autoantigen and the liver-specific pathology of PBC. Translation of these observations may reflect basic pathogenic mechanisms in PBC patients who display a striking redistribution of NKT cells from the blood to the liver, which may be provoked by a latent infection with Novosphingobium continously triggering NKT cell activation. Therefore, signs of infection need to be more carefully examined as they may not be immediately apparent. The elevation of the biliary alkaline phosphatase may, for example, reflect a mechanism to defend bacteria and/or to detoxify bacterial components, as shown for its intestinal isoforms . Tissue specimens, especially liver biopsies from PBC patients, need to be analyzed for the presence of bacteria by a 16S rRNA-specific qPCR, also at different time points. In particular, patients in early disease stages need to be identified, as the bacterial infection may be cleared at the time of overt disease, as indicated in our mouse model. As anti-PDC-E2 antibodies generally precede the onset of liver lesions in PBC patients [118–121] and may reflect the onset of an antibacterial immune response, this may be the most suitable time window to apply antibiotics. However, to prove the presence of bacterial infection, a bacterial screening test needs to be developed that does not only show cross-reactivity of sera to bacterial PDC-E2 antigens, but also to other bacterial structures that do not cross-react, as suggested in previous studies in humans [118–121] as well as in our mouse model . Using ELISA to screen for bacterial flagellin in these early stages may provide a useful tool in this respect. These clinical tests will help to identify if the cross-reactivity to bacterial antigens correlates with a persistent bacterial infection and to define the time-frame in which the application of antibiotics might be useful for the treatment of disease.
Given the emerging role of microbial pathogens in autoimmune diseases, defining their control and regulation by the host immune response will improve our understanding of pathogenic mechanisms in these disorders. As anti-PDC-E2 responses can precede the development of liver lesions (reported even for two decades in advance) [118–121], the presence of anti-PDC-E2 antibodies in the serum might be a good indicator for the administration of antibiotics to patients. However, prescreening tests would be necessary as patients generally only present at stages of disease where liver lesions are already present.
Although we do not consider PBC as an infectious disease, (persistent) bacterial infection may drive the early stages and/or the initiation of disease. The data obtained in our mouse model together with the clinical data in PBC patients strongly imply this possibility and suggest that intensive prescreening for patients at risk, for example family members, should be carried out. Antibiotic administration, especially those that are secreted by the bile like β-lactam antibiotics such as ampicillin should be considered in clinically asymptomatic patients who exhibit anti-PDC-E2 responses or reactivity against Novosphingobium antigens. However, the resistence profile in each individual situation has to be included in the treatment options and considered for the choice of the respective antibiotic(s).
The discovery of infectious triggers of autoimmunity and the elucidation of the mechanisms by which they induce tissue damage will change our current concepts about the etiology of various autoimmune syndromes and may suggest new and simpler ways to diagnose and treat these debilitating diseases.
The authors thank M Wills-Karp and all members of the Mattner laboratory for critical reviewing the manuscript and helpful suggestions.
Financial & competing interests disclosure
The project described was supported by Award Number R01DK084054 from the National Institute of Diabetes and Digestive and Kidney Diseases to Jochen Mattner. Jochen Mattner is also supported by the Lupus Research Institute and in part by PHS grant P30 DK078392. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Javid P Mohammed, Division of Immunobiology, Cincinnati Children’s Hospital, Cincinnati, OH 45229, USA.
Jochen Mattner, Division of Immunobiology, Cincinnati Children’s Hospital, Cincinnati, OH 45229, USA Tel.: +1 513 803 0768, Fax: +1 513 636 5355, Email: firstname.lastname@example.org.
Papers of special note have been highlighted as:
• of interest
• • of considerable interest