Декодирование связи между микроорганизмами и автоиммунизацией: многоаспектный обзор данных по конкретным патогенам, которые вызывают аутоиммунные заболевания.

Introduction

Autoimmune diseases (AIDs) are a group of chronic disorders caused by the immune system losing tolerance to self-antigens and subsequently attacking its own tissues and organs. The global prevalence of AIDs is approximately 5% to 8% of the world’s population, making it a major public health concern [ 1 ]. The exact etiology and pathogenesis of AIDs have not been fully elucidated. Currently, it is generally accepted that their occurrence is the result of complex interactions between genetic susceptibility, immune regulatory dysfunction, and environmental triggers. Among various environmental factors, infectious pathogens (including viruses, bacteria, etc.) have received the most extensive attention.

As early as the late 19th century, Louis Pasteur’s germ theory and Robert Koch’s Koch’s Postulates laid the foundation for the microbiological etiology of diseases, and also paved the way for subsequent investigations into the link between infection and autoimmunity. Over a century of research, abundant evidence has revealed intricate connections between microbial infections and AIDs. For example, studies have found that there is a common antigen between human intestinal mucosal epithelial cells and Escherichia coli O14; following infection, this may lead to the development of ulcerative colitis through cross-immune reactions [ 2 ]. In patients with ankylosing spondylitis (AS), certain Gram-negative bacteria (such as Klebsiella ) share homologous segments with the human HLA-B27 molecule, and antibodies against this homologous short peptide can be detected in patient sera [ 3 ]. Additionally, molecular mimicry between the nucleoprotein of influenza virus and the hypothalamic secretin receptor 2 is considered an important mechanism triggering narcolepsy [ 4 ]. These findings strongly suggest that microbial infections are key drivers of AID occurrence.

However, correlation does not equate to causation. To confirm that a specific microbial infection is the initiating factor of a specific AID, this article argues that evidence from the following three dimensions is crucial: (1) Epidemiological association: There is a clear temporal and spatial correlation between specific microbial infection and the occurrence of AIDs in the population, with infection preceding disease onset; (2) Immunological cross-reaction and pathogenicity: Cross-reactive antibodies or T cells that recognize both microbial antigens and autoantigens can be detected in patients. These cross-reactive antibodies or T cells exhibit pathogenic potential in in vitro or in vivo models; (3) Animal model validation: Immunization of susceptible animals with this microorganism or its components can recapitulate key characteristics of human AIDs.

Although a single AID may be associated with multiple pathogens, and the same microorganism may be involved in the development of different AIDs, this review will focus on microbe-disease combinations that meet or approach the above criteria. We will at the macro level elaborate on the inductive and initiating roles of microbial infections in the following nine AIDs (Guillain-Barré syndrome, systemic lupus erythematosus, rheumatoid arthritis, myocarditis/dilated cardiomyopathy, type 1 diabetes, multiple sclerosis, autoimmune hepatitis, rheumatic fever, herpetic stromal keratitis, and chlamydia-related heart disease) from three aspects: epidemiology, clinical and experimental research, and animal models. The aim is to provide comprehensive and in-depth evidence integration and theoretical analysis for this field.

Guillain-Barré syndrome

Guillain-Barré syndrome (GBS) is an autoimmune peripheral neuropathy characterized by demyelinating lesions of peripheral nerves and nerve roots, and infiltration of inflammatory cells in small blood vessels. The typical subtypes are acute inflammatory demyelinating polyneuropathy (AIDP) and acute motor axonal neuropathy (AMAN). Notably, approximately two-thirds of GBS patients have a reported history of infection prior to disease onset. Among these, Campylobacter jejuni ( C. jejuni ) is by far the most common infectious pathogen. In addition, infections such as Mycoplasma pneumoniae ( M. pneumoniae ), Haemophilus influenzae ( H. influenzae ), cytomegalovirus (CMV), herpes simplex virus (HSV), Epstein-Barr virus (EBV), and hepatitis E virus have also been reported to be associated with GBS occurrence [ 5 ]. The following section focuses on the relationship between C. jejuni infection and GBS.

Epidemiological correlation

A case-control study conducted at hospitals in England and Wales revealed that C. jejuni infection was more prevalent among GBS patients (26%) than in family control groups (2%) and age-matched hospital control groups (1%). The study enrolled 45 patients with GBS (mean age 45 years, male-to-female ratio 1.2:1), with a control group consisting of family controls ( n  = 50) and age-matched hospital controls ( n  = 50). These data provide the first epidemiological evidence for the role of C. jejuni in GBS occurrence and progression [ 6 ]. Further studies have also indicated that C. jejuni infection precedes the development of AIDP and AMAN [ 7 ].

Clinical and experimental research findings

Studies have found that Campylobacter jejuni infection can induce patients to produce high-titer autoantibodies against monosialotetrahexosylganglioside (GM1) and disialoganglioside GD1a (GD1a), which are considered specific serological markers of the AMAN subtype [ 8 ]. Mechanistically, Gregson et al. confirmed that lipooligosaccharides (LOS) in the cell wall of C. jejuni have highly similar epitopes to human ganglioside components GM1, Ganglioside GM1b (GM1b), or GD1a [ 9 ]. This molecular mimicry causes the immune response against bacterial LOS to mistakenly recognize and attack peripheral nerve axons rich in these gangliosides, thereby triggering cross-reactive autoimmune damage. In some GBS patients, epitope spreading from initial GM1 to GD1a, GD1b, or GQ1b gangliosides has been observed, correlating with clinical subtype conversion or disease relapse [ 10 ].

Animal model research findings

Animal models provide the most direct causal evidence for C. jejuni -induced GBS. Yuki et al. established an AMAN model by immunizing rabbits with C. jejuni LOS containing a GM1-like structure. Immunized rabbits not only produced high levels of anti-GM1 antibodies but also exhibited progressive limb weakness and even paralysis. The pathological changes in their peripheral nerves were highly consistent with those observed in human GBS [ 11 ]. Further studies showed that immunization of GM1 knockout mice with such LOS could produce monoclonal antibodies reactive to GM1. Although this antibody did not induce paralysis in in vitro models, it could specifically block signal transduction at the neuromuscular junction, revealing the potential mechanism by which anti-GM1 antibodies directly cause disease [ 12 ]. It was also found that the titer of anti-GM1 IgG in immunized rabbits or infected GBS patients was closely correlated with disease onset [ 13 ].

In conclusion, the association between C. jejuni infection and GBS fully meets the three aforementioned argumentative criteria: a clear epidemiological link, detection of cross-reactive antibodies and their in vitro pathogenicity, and successful disease recapitulation in animals. Therefore, GBS is widely recognized as a paradigmatic model of AIDs caused by microbial infections [ 14 ]. However, the clinical observation that only a small proportion of C. jejuni infections ultimately progress to GBS strongly suggests that the host’s genetic background, immune status, and other potential environmental factors collectively constitute the necessary synergistic conditions for disease occurrence.

Systemic lupus erythematosus

Systemic lupus erythematosus (SLE) is a diffuse, systemic AID that primarily involves the skin and mucous membranes, bones and muscles, kidneys, and central nervous system. It can also affect multiple other organs and systems such as the lungs, heart, and blood, presenting diverse clinical manifestations. A growing body of experimental and clinical data supports the key role of infection in the induction and/or exacerbation of SLE [ 15 ]. Epstein-Barr virus (EBV), cytomegalovirus (CMV), parvovirus B19 (B19V), and human endogenous retroviruses (HERVs) may be involved in SLE pathogenesis [ 16 ]. Among these, EBV has the most abundant evidence as an etiological candidate virus [ 17 ].

Epidemiological correlation

Numerous studies have shown that the EBV seropositivity rate in both pediatric and adult SLE patients is significantly higher than in healthy control populations [ 18 ]. Additionally, EBV viral load in the peripheral blood of SLE patients is abnormally elevated, suggesting increased EBV reactivation [ 19 ]. A meta-analysis involving 33 studies further confirmed that SLE patients had significantly higher levels of IgG antibodies against Epstein-Barr virus capsid antigen (VCA) and early antigen (EA), both of which are indirect markers of active EBV replication [ 20 ]. The most compelling evidence comes from a longitudinal study based on the U.S. Department of Defense Serum Repository (DoDSR), which enrolled over 7 million active-duty military personnel (primarily aged 18–45, with males accounting for approximately 85%). Serum samples were collected during longitudinal follow-up, and the study confirmed a strong correlation between EBV infection and the production of autoantibodies as well as the clinical onset of SLE [ 21 ]. The above epidemiological data support the role of EBV in SLE occurrence and progression.

Clinical and experimental research findings

The core mechanism underlying the association between EBV and SLE lies in molecular mimicry between EBV nuclear antigen 1 (EBNA-1) and autoantigens. Studies have found that the PPPGRRP sequence in the 35–58 amino acid residues of the EBNA-1 protein has a high degree of homology with the PPPGMRPP sequence of the SLE-specific autoantigen Sm, which can trigger cross-immune reactions [ 22 ]. Similarly, anti-Ro antibodies, which can be detected prior to clinical SLE onset, have been found to cross-react with another sequence of EBNA-1 (58–72 amino acids: GGSGSGPRHDGVRR). Although the primary sequence homology between the two proteins is not high, they may mimic the epitope of the Ro antigen in spatial conformation [ 23 ]. Functional experiments confirmed that expression of the full-length EBNA-1 protein in mice can successfully induce IgG autoantibodies against Sm and double-stranded DNA (dsDNA), directly linking EBV antigens to SLE autoimmune epitopes [ 24 ]. Regarding geographical comparisons, a systematic review of EBV seroepidemiology found that EBV is generally acquired at younger ages in Asia compared to Europe/North America [ 25 ], but direct standardized comparisons of antibody titers between SLE patients from different regions are limited. Moreover, persistent EBV reactivation further sustains this process through chronic interferon-α production and self-reactive B-cell immortalization [ 26 ].

Animal model research findings

To verify the in vivo pathogenicity of the above cross-reactions, researchers conducted a series of animal experiments. McClain et al. immunized New Zealand rabbits with 60 kDa Ro antigen peptides or their cross-reactive EBNA-1 peptides. Results showed that autoantibodies binding to Ro proteins were induced in both cases, and ultimately led to lupus-like clinical manifestations such as leukopenia, thrombocytopenia, and renal dysfunction in some rabbits [ 21 ]. Poole et al. immunized rabbits and mice with EBNA-1 peptides containing the PPPGRRP sequence (constructed on a polylysine scaffold) and EBNA-1 fragments containing this sequence, respectively. Both groups exhibited high-frequency autoantibody production. The binding specificity of these antibodies against lupus-associated autoantigens (Sm B’, nRNP A, nRNP C) was highly similar to the binding patterns of early-stage human SLE antibodies against these proteins [ 27 ]. Among anti-EBNA-1 monoclonal antibodies prepared in recent studies, two strains were found to cross-react with dsDNA. When these antibodies were injected into mice, they could form immune complex deposits in the kidneys, inducing proteinuria and histopathological changes consistent with lupus nephritis, further demonstrating the pathogenic potential of these cross-reactive antibodies [ 28 ].

The above chain of evidence clearly indicates that EBV infection, particularly through EBNA-1-mediated molecular mimicry, plays an important role in SLE initiation. However, currently confirmed immunological associations mainly focus on the subset of SLE patients who produce anti-Sm and anti-Ro antibodies, accounting for only approximately 40% of all SLE cases. This suggests that EBV-driven molecular mimicry is not the sole pathway, and other microorganisms or endogenous factors may also be involved in SLE initiation.

Rheumatoid arthritis

Rheumatoid arthritis (RA) is a chronic systemic AID characterized by joint synovitis. Clinical and experimental studies have highlighted the role of microorganisms in RA development, with the most important including EBV, Porphyromonas gingivalis ( P.gingivalis ), Proteus mirabilis ( P. mirabilis ) and Escherichia coli ( E. coli ) [ 29 ]. Other microorganisms associated with RA include parvovirus, human immunodeficiency virus (HIV), hepatitis B virus, and hepatitis C virus [ 30 ].

EBV and RA

Epidemiological correlation

The association between EBV and RA has a broad epidemiological basis. Studies have shown that compared with healthy control groups, RA patients have significantly higher antibody titers against EBV nuclear antigen 2 (EBNA-2) and a significantly increased frequency of EBV-infected B cells in circulation [ 31 ]. More importantly, EBV DNA load in the peripheral blood of RA patients has been confirmed to be approximately 10 times higher than in healthy individuals [ 32 ], strongly suggesting abnormally active EBV replication in RA patients.

Clinical and experimental research findings

The molecular mechanism underlying the EBV-RA association involves multiple levels. Firstly, RA genetic susceptibility is closely related to the HLA-DRB1*04:01 allele, which carries the “shared epitope” (QK/RRAA motif) [ 33 ]. Studies have found that the EBV glycoprotein gp110 contains a QKRAA amino acid sequence highly similar to the shared epitope, which can lead to cross-recognition by T cells targeting viral gp110 and subsequent attack on self-tissues expressing the shared epitope [ 34 ]. Secondly, the glycine-alanine repeat sequence in the EBV nuclear antigen EBNA-1 was found to be similar to host proteins (such as αB-crystallin), which may induce cross-reactive antibodies [ 35 ]. Additionally, the production of anti-cyclic citrullinated peptide antibody (ACPA), an important diagnostic marker for RA, may also be associated with EBV infection. Studies have found that citrullinated peptides derived from EBNA-1 and EBNA-2 proteins share homologous sequences with human citrullinated peptides, which may explain the early appearance of ACPA several years before clinical RA onset [ 36 ]. Beyond molecular mimicry, EBV-encoded superantigens may non-specifically activate T cells, contributing to early T-cell repertoire perturbations observed in RA patients [ 37 ].

Animal model research findings

Since EBV cannot naturally infect mice, researchers have adopted humanized mouse models to overcome this limitation. Human cluster of differentiation 34+ (CD34+) hematopoietic stem cells (HSCs) were transplanted into immunodeficient NOD/Shi-scid/IL-2Rγnull (NOG) mice to reconstitute the human immune system, followed by EBV inoculation. Results showed that among 23 NOG mice transplanted with human HSCs and inoculated with EBV, 15 (approximately 65%) developed erosive arthritis. The pathological features included synovial hyperplasia, pannus formation, and bone marrow edema, which closely resemble those of human RA. No erosive arthritis was observed in the control group that only received HSC transplantation without EBV inoculation [ 38 ]. This model has for the first time directly demonstrated in vivo the pathogenic ability of EBV to induce arthritis.

Porphyromonas gingivalis and RA

Epidemiological correlation

The association between the periodontal pathogen P.gingivalis and RA has become increasingly clear. An epidemiological study in the United States involving 69 RA patients (57 males and 12 females) demonstrated that the incidence of periodontitis was significantly higher in RA patients compared to the control group, and was independent of age, ethnicity, smoking, diabetes, and gender [ 39 ]. The association between rheumatoid arthritis and periodontitis is suggested to be linked to the periodontal pathogen P. gingivalis. A cross-sectional study involving 95 confirmed RA patients in the Netherlands found that, compared with a matched non-RA control group, RA patients exhibited a higher prevalence of severe periodontitis (27% versus 12%), with an positive correlation observed across age groups; however, no significant association with gender was detected [ 40 ].

Clinical and experimental research findings

P.gingivalis can express peptidylarginine deiminase (PAD), a key enzyme for protein citrullination in the body. Bacterial PAD may generate new autoantigens by citrullinating host proteins (such as human α-enolase), thereby disrupting immune tolerance. In addition, antibodies against bacterial enolase can cross-react with human α-enolase, further exacerbating the autoimmune response [ 41 ].

Animal model research findings

Animal experiments have provided direct evidence that P.gingivalis promotes RA progression. In the collagen-induced arthritis (CIA) mouse model, infection with P.gingivalis strain W83 significantly exacerbates arthritis severity, manifested by earlier onset, accelerated disease progression, and increased severity. Studies have confirmed that this exacerbating effect depends on the expression of bacterial-specific PAD. P.gingivalis promotes RA progression in susceptible individuals by introducing exogenous PAD and altering host protein citrullination [ 42 ].

Multiple sclerosis

Multiple sclerosis (MS) is a degenerative disease of the central nervous system (CNS) caused by chronic inflammation, primarily leading to neurological dysfunction [ 43 ]. More than 24 types of microorganisms have been associated with MS [ 44 ], among which EBV is considered the main infectious trigger. The following focuses on the correlation between EBV and MS.

Epidemiological correlation

EBV has the most compelling epidemiological evidence for association with MS to date. Case-control studies have demonstrated that EBV-seronegative individuals have an extremely low risk of developing MS [ 45 ]. A longitudinal study based on samples from over 7 million individuals (Primarily aged 18–45 years, with males account for approximately 85%) in the U.S. Department of Defense Serum Repository ultimately confirmed that all individuals who developed MS during follow-up became EBV-seropositive prior to disease onset, and MS clinical symptoms only began to appear several months after seroconversion [ 46 ]. This nearly establishes EBV infection as a necessary condition for MS development.

Clinical and experimental research findings

Analysis of autoimmune responses in MS patients indicates the presence of myelin basic protein (MBP)-specific T cells that cross-react with EBNA-1 [ 47 ]. Additionally, structural homology between MBP and the EBV peptide presented by DRB5*01:01 leads to DRB5-restricted cross-reactions [ 48 ]. Interestingly, CD4 + T cells isolated from the CNS of MS patients can recognize autologous EBV-transformed B cells, providing conclusive evidence for cross-reactions [ 49 ]. The above research results indicate that immunity and cross-reaction are potential mechanisms for EBV-induced MS. In addition to cross-reaction, EBV-infected B cells within the CNS may act through bystander activation, amplifying the demyelinating cascade [ 50 ].

Animal model research findings

Since EBV cannot infect mice, researchers often use the Theiler’s murine encephalomyelitis virus (TMEV) model to investigate the relationship between viral infection and demyelination. To verify the initiating effect of molecular mimicry, researchers inserted the epitope or mimic sequence of the myelin proteolipid protein (PLP) peptide PLP 139−151 (HSLGKWLGHPDKF) into a non-pathogenic TMEV vector. Results showed that mice infected with these recombinant viruses could initiate a PLP139-151-specific CD4 + T cell response and develop autoimmune demyelinating diseases [ 51 ]. More interestingly, insertion of Haemophilus influenza epitopes with only partial homology to PLP can also induce cross-reactive CD4 + T cell responses and demyelinating lesions in the context of TMEV infection [ 52 ]. The above findings indicate that innate immune signals activated by viruses may be indispensable synergistic factors for molecular mimicry to initiate disease.

EBV as a shared trigger: differential outcomes in SLE, RA, and MS

The aforementioned studies demonstrate that EBV is associated with the pathogenesis of various autoimmune diseases, including SLE, RA, and MS. The differential clinical phenotypes observed following EBV infection or EBV antigen immunization are determined by a combination of virus antigen-specific molecular mimicry patterns, tissue-specific epitope distribution, host genetic background, and targeted effects on immune cell subsets [ 53 , 54 ]. Cross-reactivity with the 60 kDa Ro protein can drive early SLE-associated autoimmune responses [ 21 ], whereas multiple EBNA1 mimetics targeting neural antigens are more likely to induce MS [ 55 ]. These initial immune responses are further regulated by tissue-specific immune cell behaviors, such as systemic B-cell reprogramming in SLE [ 56 ] and B-cell subsets infiltrating the central nervous system in MS [ 57 ], with the final phenotypic outcomes determined by a combination of host genetic factors and immune regulatory variations.

Myocarditis/dilated cardiomyopathy

Myocarditis is a localized or diffuse inflammatory lesion of the myocardial layer caused by various etiologies. Viral infections are associated with myocarditis development; viral myocarditis is considered the main cause of dilated cardiomyopathy (DCM) and can lead to heart failure. Viruses associated with myocarditis include picornaviruses, orthomyxoviruses, and herpesviruses, among which Coxsackievirus B3 (CVB3) and adenoviruses are the most common suspected pathogens [ 58 ]. Recently, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has also been implicated in myocarditis etiology [ 59 ]. Here, we focus on the correlation between Coxsackievirus B3 and myocarditis.

Epidemiological correlation

Serological studies have shown that virus-reactive neutralizing antibodies can be detected in approximately 50% of DCM patients [ 60 ]. Furthermore, the titer of anti-CVB3 neutralizing antibodies in the serum of patients with chronic myocarditis or DCM is significantly higher than in patients with other types of heart disease [ 61 ], and CVB3-specific IgM antibodies can even be detected persistently in some patients [ 62 ].

Clinical and experimental research findings

Early studies found that neutralizing antibodies against CVB3 can cross-react with adenine nucleotide translocases in myocardial mitochondria [ 63 ]. Subsequent research revealed a more complex cross-reaction network: CVB3 neutralizing antibodies can react not only with myocardial myosin but also cross-react with antigens such as Streptococcal M protein [ 64 ]; antibodies against myosin can also cross-react with CVB3 capsid protein and the β1-adrenergic receptor [ 65 ]. CVB3 infection-induced myocyte necrosis releases cardiac myosin and other sequestered autoantigens, amplifying the inflammatory cascade through bystander mechanisms [ 66 ]. These findings suggest that viral infection can trigger widespread autoimmunity against multiple cardiac antigens.

Animal model research findings

The CVB3-infected mouse model provides sufficient causal evidence for viral myocarditis. In terms of humoral immunity, specific autoantibodies against cardiac tissue antigens and myocardial myosin can be detected in CVB3-induced chronic myocarditis mouse models [ 67 ]. More compelling evidence is that passive transfer of monoclonal antibodies targeting different epitopes of the CVB3 capsid protein to normal mice can directly induce cardiac pathological changes. Some of these monoclonal antibodies can bind to cardiac fibroblasts in vitro and enhance macrophage chemotactic activity, demonstrating the pathogenic potential of these cross-reactive antibodies [ 68 ]. In terms of cellular immunity, CD4 + T cells specific to cardiac alpha-myosin heavy chain (αMYHC) have been confirmed to be pathogenic. Studies have found that CVB3 and αMYHC share mimetic epitopes. Following infection of susceptible mice, αMYHC-specific CD4 + T cells (which contain large numbers of IL-17-producing Th17 cells) are generated and infiltrate the heart, inducing myocarditis [ 69 ]. These studies have comprehensively demonstrated the entire process from CVB3 infection to autoimmune myocarditis from both humoral and cellular immunity perspectives.

Type 1 diabetes

Type 1 diabetes (T1D), also known as insulin-dependent diabetes mellitus (IDDM), is an endocrine and metabolic disorder caused by pancreatic beta cell destruction, which impairs insulin secretion and leads to disturbances in glucose, fat, and protein metabolism. This condition poses a significant threat to children’s health. In 1899, Harris published the first report on the potential relationship between viral infection and T1D [ 70 ]. Subsequently, multiple viruses have been reported to be associated with T1D occurrence, such as rotavirus, adenovirus, retrovirus, reovirus, cytomegalovirus, EBV, mumps virus, and rubella virus [ 71 ]. However, enteroviruses, particularly Coxsackievirus B, are the most extensively studied environmental triggers associated with T1D.

Epidemiological correlation

Multiple retrospective studies have demonstrated that the detection rate of anti-enterovirus antibodies and/or enterovirus genomes is higher in newly diagnosed T1D patients than in healthy control groups [ 72 ]. Prospective studies have provided more convincing temporal evidence: in genetically susceptible children, enterovirus infections are more frequently concentrated in the period preceding the first appearance of islet autoantibodies, strongly suggesting that infection events trigger the initiation of autoimmunity [ 73 ]. The above epidemiological studies have revealed the association between enterovirus infection and T1D occurrence in genetically susceptible individuals.

Clinical and experimental research findings

Studies have found that the enterovirus protein VP1 shares partial sequence (PALTAVETGA/HT) homology with the islet β-cell autoantigen tyrosine phosphatase, enabling cross-reactions [ 74 ]; the protease 2 C (P2C) of Coxsackievirus also has cross-reactive epitopes (PEVKEK) with another key autoantigen, glutamate decarboxylase 65 (GAD65), which may be involved in autoimmunity resulting from molecular mimicry [ 75 ].

Animal model research findings

Animal models have confirmed the potential of coxsackievirus B4 (CVB4) to induce diabetes. Both SJL/J mice infected with the CVB4 JVB strain and CD-1 mice infected with the E2 strain developed pancreatitis, β-cell destruction, and hyperglycemia [ 76 ]. However, these models also reveal more complex mechanisms than simple molecular mimicry. Studies have found sequence similarities between the P2C protein of group B Coxsackieviruses and the cellular autoantigen GAD65, and cellular and humoral immunity targeting common epitopes of GAD and group B Coxsackieviruses exists in T1D patients [ 77 ]. However, key experiments indicated that CVB4 infection did not accelerate diabetes progression in Non-obese diabetic (NOD) mice carrying GAD cross-reactive epitopes. Conversely, CVB4 infection can rapidly induce diabetes in T-cell receptor transgenic mice targeting non-GAD islet antigens [ 78 ]. This discovery has shifted the core of the pathogenic mechanism from “molecular mimicry” to “bystander activation”. Specifically, local viral infection in the pancreas causes inflammation and tissue damage, leading to the release of large amounts of sequestered islet antigens, which subsequently activate pre-existing and diverse autoreactive T cells, ultimately resulting in β-cell destruction.

Autoimmune hepatitis

Autoimmune hepatitis (AIH) is a progressive chronic inflammatory lesion of the liver. Hepatitis viruses, Epstein-Barr virus (EBV), varicella-zoster virus (VZV), and cytomegalovirus (CMV) are all implicated in increasing the risk of AIH, among which hepatitis C virus (HCV) has the strongest correlation [ 79 ].

Epidemiological correlation

A specific serological marker of AIH is autoantibodies against liver-kidney microsomes type 1 (LKM1), whose target antigen is cytochrome P450 2D6 (CYP2D6). A pivotal epidemiological finding is that LKM1 antibodies can be detected in the serum of approximately 10% of chronic HCV-infected individuals [ 80 ], suggesting that HCV infection may unexpectedly trigger an autoimmune response against liver autoantigens.

Clinical and experimental research findings

Studies have shown that the association between HCV and AIH stems from molecular mimicry. Antibodies recognizing HCV proteins NS3 and NS5A can specifically identify a conformational epitope on amino acids 254–288 of the CYP2D6 protein [ 81 ]. Additionally, HCV-specific CD8 + T cells were found to cross-react with cytochrome P450 2A6 and 2A7 subtypes containing HCV homologous sequences (LLALLSCLTV) [ 82 ]. Importantly, LKM1 antibodies in patients with type 2 AIH can not only recognize CYP2D6 (amino acids 193–212) but also cross-react with homologous regions of HCV, CMV, and HSV-1 viral proteins [ 83 ], revealing that multiple pathogens may contribute to AIH initiation through shared mimetic epitopes.

Animal model research findings

Researchers successfully established a virus-induced autoimmune liver disease model by infecting mice with a hepatotropic adenovirus (Ad-2D6) expressing human CYP2D6. Infected mice developed persistent hepatitis, fibrosis, and necrosis, and produced LKM1-like autoantibodies similar to those in human type 2 AIH patients [ 84 ]. A key finding was that wild-type mice infected with Ad-2D6 exhibited more severe liver damage than transgenic mice stably expressing the same human CYP2D6 protein in their livers [ 85 ]. This indicates that pathogen infection is more likely to break immune tolerance and induce autoimmunity in hosts carrying similar target antigens rather than identical ones.

Rheumatic fever

Epidemiological correlation

Rheumatic fever (RF) and its most serious complication, rheumatic heart disease (RHD), are classic examples of autoimmune reactions following Streptococcus pyogenes (group A Streptococcus , GAS) infection. RF typically occurs in approximately 3–4% of children after untreated GAS pharyngitis [ 86 ], while RHD develops 4–8 weeks or later after GAS infection, affecting 30% to 45% of RF patients [ 87 ].

Clinical and experimental research findings

In 1966, Zabriskie first discovered that cross-reactive antigens and antibodies are associated with acute rheumatic fever and group A Streptococcus [ 88 ]. Subsequent studies confirmed that anti-GAS antibodies in RF patients can be absorbed by human cardiac tissue, and vice versa [ 89 ]. Mechanistically, the M protein of Streptococcus shares α-helical coiled-coil structures with human myocardial proteins (such as myosin and laminin). T cell clones isolated from the myocardium and valves of RHD patients can recognize both M proteins and these cardiac autoantigens, directly demonstrating that cross-reactive T cells are involved in the pathophysiological processes of the disease [ 90 ].

Animal model research findings

Immunization of Lewis rats with recombinant Streptococcal M proteins (such as RM5 and RM6) can successfully induce carditis characterized by valvulitis and focal myocardial lesions [ 91 ]. Further research identified specific pathogenic peptide epitopes in the M protein. Passive transfer of these epitope-specific T cell lines to infant rats could directly cause valvulitis [ 92 ]. The model also recapitulated another key feature of human disease: repeated Streptococcal infections can exacerbate cardiac damage. In Lewis rats, multiple immunizations with M protein enhanced anti-myosin antibody levels and T-cell responses, and resulted in more severe cardiac lesions [ 93 ]. Recent studies have confirmed the pathogenic role of antibodies and T cells produced in response to group A Streptococcus M protein in inducing cardiac inflammation through co-adoptive transfer of serum and M protein-stimulated T cells [ 94 ].

Herpetic stromal keratitis

Herpetic stromal keratitis (HSK) is a disease caused by herpes simplex virus type 1 (HSV-1) infection, characterized by T cell-mediated corneal tissue destruction and inflammation, and is a major cause of human blindness [ 95 ].

Epidemiological correlation

In the United States, approximately 50,000 new cases of HSV corneal infection are reported annually. Disease onset peaks much later than primary infection, occurring most commonly in individuals aged 30 to 40 years, suggesting that it is an autoimmune disease following infection [ 96 ].

Clinical and experimental research findings

A peptide sharing 16 amino acid residues with host IgG2ab (amino acids 435–451) was identified in the unique long 6 (UL6) protein of HSV-1, which is considered a molecular mimic causing HSK [ 97 ]. In addition to the molecular mimicry between HSV-1 UL6 and host IgG2ab, bystander activation has been established as an alternative or complementary mechanism in HSK pathogenesis [ 98 ].

Animal model research findings

Mouse models confirmed that HSK is mediated by CD4 + T cell clones recognizing the IgG2ab cross-reactive peptide. Adoptive transfer of T cells activated against UL6 viral peptides to immunodeficient mice can induce severe HSK. The key causal evidence is that HSV-1 mutant viruses lacking this specific UL6 peptide segment neither induce T cell clone activation nor cause HSK damage in susceptible mice [ 99 ]. The above perfectly confirms that a single molecular mimicry event is sufficient to initiate a specific AID.

Evidence gaps and future directions

Although the aforementioned sections integratively reviewed the association between microbial infections and nine AIDs from three dimensions: epidemiology, clinical/experimental, and animal models, and the overall evidence supports microorganisms as important environmental triggers, this field still faces several common limitations and unsolved issues. Addressing these challenges not only facilitates a prudent interpretation of existing evidence but also provides guidance for future research.

Role of host factors in infection-triggered autoimmunity

It is estimated that only 1 out of every 1,000 C. jejuni infections results in GBS [ 100 ]; EBV infection is nearly universal, yet only a very small proportion of carriers develop SLE, MS, or RA [ 101 ]. This phenomenon suggests that host genetic background (specific HLA alleles significantly increasing risk), immune status, and other environmental factors (such as coinfections and gut microbiota) play critical roles, although current understanding of these factors remains incomplete. Future large-scale, multicenter prospective cohort studies integrating host genome sequencing, pathogen whole-genome sequencing, and immune repertoire analysis before and after infection are required to establish risk prediction models.

Limitations of animal models and future perspectives

Animal models provide the most direct causal evidence, but existing models also have certain limitations [ 102 ]: mouse models rarely progress spontaneously to the chronic disease stage in humans, and some models rely on active immunization rather than natural infection, thereby overlooking other immunomodulatory signals during the infection process. Human-specific viruses such as EBV cannot directly infect conventional animals, and humanized mouse models still exhibit gaps compared to humans [ 103 ]. Future efforts should focus on developing next-generation humanized mice and promoting the use of natural infection models.

Reverse causality in infection-autoimmunity associations

In cross-sectional or retrospective studies, the association between infection and AIDs may be inverse. The autoimmune state itself may increase susceptibility to specific pathogens; for example, cohort studies have shown that EBV infection precedes the appearance of autoantibodies, although EBV reactivation may also occur secondary to immune dysregulation associated with SLE [ 101 ]. Treatment for AIDs can also elevate the risk of infection [ 104 ]. Future Mendelian randomization (MR) studies are required to determine the causal direction. The MR method has been increasingly applied to infer causality among risk factors in diseases such as SLE, RA, MS, and T1DM [ 105 ].

Unresolved questions in pathogenic mechanisms

Although molecular mimicry are widely cited, there is currently a lack of quantitative standards defining the level of homology required to trigger pathogenic cross-reactivity [ 106 ]. How do the relative contributions of molecular mimicry, bystander activation, and epitope expansion vary across different disease stages (initiation, progression, and chronic phase)? Future research could employ single-cell technologies to isolate pathogenic clones from patient lesion tissues and develop structure-based epitope prediction algorithms for large-scale screening microbial-autoantigen mimetics with pathogenic potential [ 107 ].

Conclusions and prospects

This review collates evidence for the association between nine AIDs and specific microbial infections, from epidemiological associations and the discovery of cross-reactive immune responses to causal confirmation in animal models (Table  1 ). These collectively strongly demonstrate that microbial infections are crucial environmental initiators of AIDs. Accumulating evidence indicates that in genetically susceptible individuals, one or more specific microbial infection events can break immune tolerance to autoantigens through multiple intertwined immunological mechanisms, thereby “pulling the trigger” of the autoimmune response (Fig.  1 ).

Table 1 Summary of evidence that microbial infections trigger autoimmune diseases Autoimmune Diseases Major Pathogens Other related pathogens Epidemiological links Identified clinical autoantibodies Animal Model Proposed mechanisms Refs. Guillain-Barré syndrome (GBS)
C. jejuni
M . pneumoniae , H . influenzae , CMV, HSV, EBV, HEV C. jejuni more prevalent in GBS (26%) vs. controls (2%, 1%); Infection precedes the disease Anti-GM1 and anti-GD1a ganglioside IgG antibodies Rabbits immunized with C. jejuni LOS (GM1-like structure) → limb weakness, paralysis; pathology similar to humans molecular mimicry (LOS-gangliosides); epitope spreading [5–14] Systemic lupus erythematosus (SLE) EBV CMV, B19V, HERVs Increased EBV seropositivity in SLE patients; DoDSR cohort shows EBV infection precedes autoantibodies and symptoms Anti-Sm, anti-Ro, anti-dsDNA (cross-reactive with EBNA-1) Rabbits immunized with Ro or EBNA-1 cross-reactive peptides → lupus-like manifestations; mice injected with cross-reactive mAb → immune complex deposition, proteinuria Molecular mimicry (EBNA-1 ↔ Sm/Ro/dsDNA); persistent EBV reactivation [15–28] Rheumatoid arthritis (RA) EBV, P . gingivalis P. Mirabilis , E. coli , PV、HIV、HBV、HCV 10-fold higher EBV DNA load in RA patients; 2-fold increased RA risk in periodontitis patients Anti-CCP/ACPA (cross-reactive with EBV citrullinated peptides); elevated anti-EBNA-2 antibodies Humanized NOG mice (human HSC) inoculated with EBV → 65% develop erosive arthritis; P. gingivalis infection in CIA mice → exacerbated arthritis (PAD-dependent) Molecular mimicry (gp110 ↔ shared epitope); citrullination-induced autoantigens; superantigen [29–42] Multiple sclerosis (MS) EBV TMEV Extremely low MS risk in EBV-seronegative individuals; DoDSR cohort: all MS cases convert to EBV-positive before onset Anti-MBP antibodies (cross-reactive with EBNA-1) TMEV vector inserted with PLP139-151 epitope → post-infection induction of CD4 + T cells and demyelination Molecular mimicry (EBNA-1 ↔ MBP); virus provides innate immune signals; bystander activation [43–52] Myocarditis / DCM CVB3 Picornaviruses, Orthomyxoviruses, Herpesviruses, AdV Approximately 50% of DCM patients have viral neutralizing antibodies; elevated anti-CVB3 antibodies in chronic myocarditis patients Anti-cardiac myosin, anti-ANT, anti-β1-adrenergic receptor antibodies CVB3-infected mice → cardiac autoantibodies; passive transfer of anti-CVB3 capsid mAb → cardiac pathological changes; αMYHC-specific CD4 + T cells mediate myocarditis Molecular mimicry + bystander activation [58–69] Type 1 diabetes (T1D) CVB RoV, AdV, Retro, ReoV, CMV, EBV, MuV, RuV Increased enterovirus detection rate in new-onset T1D patients; prospective studies show infection precedes islet autoantibodies Anti-GAD65, anti-IA-2 (cross-reactive with CVB P2C/VP1) CVB4-infected SJL/J mice → pancreatitis, β-cell destruction, hyperglycemia; TCR transgenic mice → rapid diabetes (bystander activation) Bystander activation predominant, molecular mimicry auxiliary [70–78] Autoimmune hepatitis (AIH) HCV EBV, VZV, CMV Approximately 10% of chronic HCV-infected individuals are serum LKM1 antibody-positive Anti-LKM1 (anti-CYP2D6) Mice infected with adenovirus expressing human CYP2D6 → persistent hepatitis, LKM1-like antibodies; more severe lesions in wild-type mice Molecular mimicry (HCV NS3/NS5A ↔ CYP2D6) [79–85] Rheumatic fever (RF) GAS 3–4% develop RF after untreated GAS pharyngitis; 30–45% of RF progress to RHD Anti-cardiac myosin, anti-laminin antibodies (cross-reactive with M protein) Lewis rats immunized with M protein → valvulitis, myocardial lesions; passive transfer of T cell lines → valvulitis in young mice Molecular mimicry (M protein ↔ cardiac proteins) [86–94] Herpetic stromal keratitis (HSK) HSV-1 50,000 new cases annually in the US; disease peaks at 30–40 years (long after primary infection) Anti-UL6 cross-reactive antibodies (cross-reactive with IgG2ab) Adoptive transfer of CD4 + T cell clones recognizing UL6 peptide → severe HSK; HSV-1 mutant lacking this peptide does not induce disease Molecular mimicry (UL6 ↔ IgG2ab); bystander activation [95–99]

Summary of evidence that microbial infections trigger autoimmune diseases

A unified model for microbe-triggered autoimmune diseases. Schematic illustrating how microbial infections initiate and propagate autoimmune responses in genetically susceptible hosts. Key mechanisms include molecular mimicry, bystander activation/epitope spreading, and persistent infection with innate immune activation

The aforementioned studies demonstrate that autoimmune initiation cannot be summarized by a single mechanism, but rather is a cascade process driven by microbial infection and involving the coordination of multiple mechanisms.

Molecular mimicry is a central mechanism: when microbial antigens share similar epitopes with host autoantigens, the efficient immune response against pathogens mistakenly attacks host tissues—such as C. jejuni LOS and gangliosides GM1/GD1a, Streptococcal M protein and myocardial myosin, and EBNA-1 and Sm/Ro antigens [ 108 , 109 ]. This cross-reaction represents one of the best characterized molecular foundations between infection and autoimmunity .

Bystander activation and epitope spreading are amplifiers: simple molecular mimicry may not be sufficient to trigger full-blown disease. Local inflammation and tissue damage caused by microbial infection (“bystander injury”) can lead to the release of large amounts of sequestered autoantigens, thereby activating a broader repertoire of autoreactive T and B cells (i.e., “epitope spreading”) [ 110 ]. For example, in the CVB4-induced T1D model, the virus itself does not have strong cross-reactivity with GAD65, but the islet inflammation it triggers is sufficient to activate resting T cells targeting other islet antigens, ultimately leading to diabetes [ 78 , 111 ].

Persistent infection and innate immune activation are sustained engines: Certain viruses (such as EBV and CMV) can establish lifelong latent infections. Their periodic reactivation not only continuously provides antigenic stimulation but also may persistently activate the innate immune system through pattern recognition receptors, maintaining a chronic pro-inflammatory state and fostering the chronicity of autoimmune responses—particularly prominent in SLE and MS [ 26 , 112 ].

Genetic susceptibility is the stage for mechanisms: The effective operation of all the above mechanisms relies on the stage provided by the genetic background [ 113 ]. For example, individuals carrying specific HLA alleles (such as HLA-B27, HLA-DRB1*04:01) have antigen-presenting cells that are more likely to efficiently present cross-reactive epitopes between microorganisms and self to T cells, thereby significantly increasing disease susceptibility [ 114 , 115 ].

Growing evidence suggests that dysbiosis and alterations in microbial metabolites may serve as critical co-factors determining whether infections can breach immune tolerance and progress to clinical AIDs. Numerous cross-sectional studies have confirmed the presence of characteristic gut microbiota dysbiosis in AIDs patients. For instance, multiple studies have demonstrated reduced α-diversity of the gut microbiota and decreased abundance of specific butyrate-producing bacteria, alongside increased abundance of pro-inflammatory microbiota in RA patients [ 116 ]. Mechanistically, gut microbiota dysbiosis contributes to joint inflammation by releasing pro-inflammatory metabolites, enhancing intestinal permeability, and impairing Treg function [ 117 ]. Short-chain fatty acids (SCFAs), metabolites of the gut microbiota, are essential for the establishment of immune tolerance. SCFAs promote the differentiation and function of regulatory T cells (Tregs) while suppressing the pathogenic response of Th17 cells, thereby exerting a protective effect in various autoimmune disease models [ 118 ]. Additionally, the commensal microbiota itself may participate in triggering autoimmune responses through molecular mimicry mechanisms. For example, the human gut commensal Parabacteroides distasonis carries a mimic peptide hprt4-18 that replicates the epitope of insulin B-chain 9–23, which activates human T-cell clones and NOD mouse T-cell hybridomas; colonization of this bacterium in female NOD mice accelerates T-cell infiltration into islets and increases the incidence of T1D [ 119 ].

A deeper understanding of the initiating role of microorganisms is driving earlier and more precise strategies for AID prevention and control. Future risk assessment models should combine genetic risk (HLA typing), microbial serological markers (such as antibodies against specific pathogens), autoantibody profiles, and immune status, while comprehensively documenting infection history to aid diagnosis. Targeted interventions—including microbe eradication (such as periodontitis treatment in RA) and biological agents or tolerance vaccines that block cross-reactive antibodies or T cells—hold promise for primary prevention.

In summary, although substantial evidence supports microbial infections as key environmental triggers in genetically susceptible individuals, the relationship is neither linear nor universal. The outcome depends on a delicate interplay among host genetics, pathogen characteristics, immune context, and tolerance mechanisms. Through interdisciplinary collaboration across microbiology, immunology, genetics, and clinical medicine, precision prevention based on individual risk and early intervention at disease onset can be achieved, bridging the gap from mechanistic exploration to clinical application.

Author contributions

Haixiang Zhang: Conceptualization, Formal analysis, Investigation, Writing-original draft, Writing-review & editing. Jun Hu: Formal analysis, Resources, Visualization. Jingying Sun: Investigation, Formal analysis, Funding acquisition. Lijun Sun: Conceptualization, Methodology, Writing-original draft, Writing-review & editing. Cuixiang Xu: Conceptualization, Funding acquisition, Project administration, Writing-review & editing.

Funding

This work was supported by the Talent Support Program of Shaanxi Provincial People’s Hospital (grant No. 2023JY-27), Shaanxi province innovation ability support plan (grant No. 2023-CX-TD-74), and Leading Talents Program of Shaanxi Provincial People’s Hospital (grant No. 2021LJ-10), Shaanxi province innovation ability support plan (grant No. 2024RS-CXTD-84).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.

Clinical trial number

Not applicable.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Lijun Sun, Email: lijun200206@163.com.

Cuixiang Xu, Email: xucuixiang@spph-sx.ac.cn.

References

Associated Data

Data Availability Statement

No datasets were generated or analysed during the current study.