Autoimmunity
Why does a woman in her thirties suddenly find her own thyroid gland destroyed? Her immune system had ignored it perfectly well for thirty years. Somewhere, a checkpoint failed. A T cell that should have been deleted survived. Or a B cell that should have been silenced got a helping signal it was never supposed to receive. Autoimmunity is the study of exactly where such checkpoints break down. It also asks why the same immune system built to distinguish self from pathogen sometimes attacks the very tissue it lives in. The stakes are enormous. Autoimmune diseases collectively affect roughly 5–8% of the population in industrialized countries. They disproportionately affect women. They include type 1 diabetes, rheumatoid arthritis, multiple sclerosis, systemic lupus erythematosus, Graves' disease, and Hashimoto's thyroiditis. Unlike infections, these diseases are usually chronic and self-perpetuating: the antigen — the patient's own tissue — never clears. Understanding autoimmunity means understanding T-Cell Development and Selection in reverse. It means asking what happens when the deletion and regulation machinery that normally enforces tolerance is incomplete, evaded, or overwhelmed. The big picture is that self-tolerance is not a single wall. It is a layered set of checkpoints. These are established centrally in the thymus and bone marrow. They are reinforced peripherally by regulatory cells, anergy, and ignorance. Autoimmunity arises when a specific combination of genetic susceptibility, environmental trigger, and stochastic immunological accident occurs. Together these breach every one of those layers at once, for a particular self-antigen. No single hit is usually sufficient. Disease is the endpoint of accumulated failures. This is why autoimmune disease is often described as multifactorial rather than caused by one broken gene. This article works through the core distinction between central and peripheral tolerance. It examines the specific mechanisms that convert broken tolerance into overt tissue damage: molecular mimicry, epitope spreading, bystander activation, and defective regulatory T cells. It also covers the classic experiments that established these mechanisms as fact rather than hypothesis. These include thymectomy studies, the AIRE/APECED story, and experimental autoimmune encephalomyelitis. It closes by connecting autoimmunity to hypersensitivity mechanisms already familiar from Hypersensitivity and Allergy. It also looks forward to why the immune system's tolerance machinery matters just as much when transplanted tissue, rather than self-tissue, is the target.
- Self-tolerance operates in layers: central tolerance (thymic/bone marrow deletion) and peripheral tolerance (anergy, regulatory T cells, ignorance).
- Autoimmune disease requires multiple simultaneous failures — genetic risk alleles, permissive environment, and a triggering event — not one single defect.
- Molecular mimicry occurs when a pathogen epitope resembles a self-epitope closely enough that pathogen-primed lymphocytes cross-react with host tissue.
- Epitope spreading and bystander activation explain why autoimmune attack often broadens over time to hit antigens unrelated to the original trigger.
- Classic experiments — neonatal thymectomy in mice, the discovery of AIRE mutations causing APECED, and experimental autoimmune encephalomyelitis (EAE) — directly proved central and peripheral tolerance mechanisms.
- Autoimmune diseases are classified as organ-specific (type 1 diabetes, Hashimoto's, Graves') or systemic (lupus, rheumatoid arthritis), reflecting how broadly the causal antigen is expressed.
Autoimmune disease results not from one broken tolerance mechanism but from the coincident failure of several redundant checkpoints that normally keep self-reactive lymphocytes silent.
The core idea
Self-tolerance is the immune system's active, ongoing failure to respond to the body's own antigens, and it is enforced at two distinct anatomical and temporal stages. Central tolerance occurs during lymphocyte development, in the thymus for T cells and the bone marrow for B cells. Developing T cells whose receptors bind self-peptide–MHC complexes with high affinity are deleted by apoptosis — negative selection — a process that depends on thymic medullary epithelial cells displaying a startlingly broad sample of tissue-restricted self-antigens (insulin, myelin proteins, retinal antigens) under the control of the transcription factor AIRE (autoimmune regulator). B cells that bind self-antigen strongly in the bone marrow undergo receptor editing or deletion before ever reaching the periphery. Central tolerance is never complete. Not every tissue antigen is expressed at sufficient levels in the thymus, and some self-reactive lymphocytes with lower affinity escape negative selection and enter the circulating repertoire. Peripheral tolerance exists to control these escapees: T cells that encounter self-antigen without adequate costimulation become anergic (functionally unresponsive), some self-reactive T cells never encounter their antigen because it is sequestered behind tissue barriers (immunological ignorance, relevant to sites like the eye and testis), and a dedicated population of regulatory T cells (Tregs, CD4⁺CD25⁺FoxP3⁺) actively suppresses autoreactive lymphocytes that do get activated, largely through IL-10, TGF-β, and CTLA-4-mediated dendritic cell inhibition. Autoimmunity is the clinical consequence when this layered system fails for a specific self-antigen and the resulting immune response causes measurable tissue damage. Disease is conventionally split into organ-specific autoimmunity, where the target antigen is restricted to one tissue (thyroid peroxidase in Hashimoto's thyroiditis, pancreatic beta-cell antigens in type 1 diabetes, acetylcholine receptor in myasthenia gravis), and systemic autoimmunity, where the antigens are ubiquitous (double-stranded DNA and nuclear proteins in lupus), producing multi-organ disease. The distinction matters clinically and mechanistically: organ-specific disease often traces to a single dominant epitope and a T-cell-driven attack, while systemic disease usually involves immune complex deposition and broad loss of tolerance to nuclear self-antigens.
Mechanism
Breaking self-tolerance is rarely a single molecular event. It is better understood as a sequence of permissive conditions that must co-occur, illustrated by four mechanisms recurring across most autoimmune diseases.
1. Failure of central deletion. If AIRE-dependent expression of a tissue-restricted antigen in the thymic medulla is reduced — by genetic mutation, insufficient promiscuous gene expression, or simple stochastic variation — T cells specific for that antigen escape negative selection at higher-than-normal frequency. This raises the baseline pool of autoreactive T cells available to be activated later. Every healthy person carries some low-affinity autoreactive T and B cells; the question is whether peripheral controls contain them.
2. Loss of peripheral control. Peripheral tolerance depends on Tregs continuously suppressing autoreactive effector cells. If Treg number or function is deficient (as in FoxP3 mutations, IPEX syndrome) or if effector T cells become resistant to suppression (often under inflammatory cytokine pressure, e.g. IL-6), previously silenced autoreactive clones can expand. Costimulation blockade also matters: an autoreactive T cell that sees self-peptide on a resting, non-inflamed antigen-presenting cell (low CD80/CD86) becomes anergic rather than activated; the same T cell seeing the identical peptide on an inflamed, pathogen-activated dendritic cell receives full costimulation and differentiates into an aggressive effector cell. Inflammation itself, therefore, can convert a tolerated self-antigen into an immunogenic one — this is the logic behind adjuvant requirements in experimental autoimmune disease models.
3. Molecular mimicry. Some microbial antigens share structural or sequence similarity with host self-antigens closely enough that T-cell receptors or antibodies raised against the pathogen cross-react with host tissue. Because the initial immune response is primed against a genuinely foreign, highly immunogenic antigen (with full costimulation, danger signals, and inflammation), the cross-reactive lymphocytes are driven to full effector differentiation rather than anergy. Once activated, these cells or their antibody products attack host tissue that merely resembles, rather than is, the pathogen. The canonical example is rheumatic fever: antibodies against Streptococcus pyogenes M protein cross-react with cardiac myosin and laminin in heart valve tissue, producing rheumatic carditis weeks after a streptococcal throat infection has cleared.
4. Epitope spreading and bystander activation. Once an initial autoimmune attack damages tissue, it releases intracellular self-antigens that were never previously visible to the immune system in inflammatory context (sequestered antigens, or antigens simply not abundant enough to prime a response before). These newly liberated antigens are captured by activated, already-inflamed local antigen-presenting cells and presented with full costimulation, recruiting new autoreactive T- and B-cell clones against additional self-epitopes — epitope spreading. Separately, bystander activation describes how the inflammatory cytokine milieu created by a genuine infection (type I interferons, IL-12, TNF) can lower the activation threshold for nearby autoreactive lymphocytes that are not specific for the infecting pathogen at all, allowing them to respond to self-antigen without any structural mimicry. Both mechanisms explain why autoimmune diseases so often broaden over the years, attacking more epitopes and more tissue than the original trigger would predict. The outcome of these breakdowns is executed through the same effector pathways used against real pathogens, discussed alongside hypersensitivity in Hypersensitivity and Allergy: autoantibody-mediated cytotoxicity and receptor blockade/stimulation (Type II), immune-complex deposition and complement-mediated inflammation (Type III, as in lupus nephritis), and direct cytotoxic T-cell or delayed-type inflammatory attack on tissue (Type IV, as in type 1 diabetes and multiple sclerosis).
The evidence
1. Neonatal thymectomy experiments (Miller, 1961; Nishizuka and Sakakura, 1969). The dependence of tolerance on thymic education was established when J.F.A.P. Miller showed that removing the thymus from neonatal mice abolished normal T-cell-mediated immunity, directly demonstrating the thymus's role in generating a functional T-cell repertoire. Extending this line, Nishizuka and Sakakura found that thymectomy performed on a very specific day after birth (around day 3) in mice caused spontaneous autoimmune oophoritis (ovarian destruction) rather than immune deficiency. The explanation, worked out over the following decades, was that thymic output at that developmental window is enriched for regulatory T cells; removing the thymus at day 3 depletes the regulatory population disproportionately while some autoreactive effector clones have already seeded the periphery, unmasking autoimmune attack. This was one of the first clean experimental demonstrations that a dedicated regulatory cell population, not just deletion, is required to prevent autoimmunity — foreshadowing the later molecular identification of FoxP3⁺ Tregs by Sakaguchi and colleagues in the 1990s.
2. APECED and the discovery of AIRE (Aaltonen et al., 1997; Nagamine et al., 1997). Human genetics provided the decisive proof that central tolerance depends on ectopic self-antigen expression in the thymus. Patients with autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy (APECED), a rare autosomal recessive syndrome causing simultaneous autoimmune attack on the parathyroid, adrenal cortex, and other endocrine organs, were found to carry loss-of-function mutations in a single gene, named AIRE. Subsequent work (Anderson et al., 2002, Science) showed that AIRE drives promiscuous expression of hundreds of tissue-restricted self-antigens (insulin, thyroglobulin, and more) in thymic medullary epithelial cells; AIRE-knockout mice fail to delete T cells specific for these antigens and develop multi-organ autoimmune infiltrates strikingly similar to human APECED. This experiment converted central tolerance from a theoretical postulate into a molecularly defined, testable mechanism: remove one transcription factor, lose ectopic thymic antigen display, and autoimmunity across multiple organs follows predictably.
3. Experimental autoimmune encephalomyelitis, EAE (Rivers, Sprunt, and Berry, 1933; refined through the 20th century). EAE is the paradigm animal model proving that autoimmune disease can be deliberately induced by immunizing an animal against its own tissue antigen — the cleanest possible demonstration that self-tolerance is a broken checkpoint, not an impossibility. Rivers and colleagues first showed that repeated injection of brain and spinal cord tissue (initially studied because rabies vaccines of that era were prepared from infected brain tissue and occasionally caused neurological complications) could produce a demyelinating brain disease in monkeys. Later work purified the responsible antigen to myelin basic protein (MBP), and showed that injecting MBP peptide emulsified in complete Freund's adjuvant (killed mycobacteria plus oil, providing the inflammatory danger signal and costimulation that convert a tolerizing self-antigen encounter into an immunizing one) reliably produces a paralytic, multiple-sclerosis-like disease in susceptible mouse and rat strains. Crucially, EAE can be transferred to a naive, unimmunized animal purely by injecting CD4⁺ T cells isolated from a diseased animal — proving the disease is T-cell mediated, not antibody mediated, and directly demonstrating that autoreactive T cells alone are sufficient to cause tissue-specific autoimmune damage. EAE remains the primary animal model for testing multiple sclerosis therapeutics today.
4. Molecular mimicry proof — rheumatic fever cross-reactivity (Kaplan and Meyeserian, 1962; subsequent monoclonal antibody studies). Kaplan first showed by immunofluorescence that antisera raised against group A streptococcal cell membranes bound to human cardiac tissue, providing early serological evidence of cross-reactivity. Later work using monoclonal antibodies derived from cardiac tissue of rheumatic fever patients confirmed these antibodies bind both streptococcal M protein and human cardiac myosin/laminin, and that T-cell clones from affected heart tissue are cross-reactive between the two antigens — establishing molecular mimicry as a mechanistic cause of a specific human autoimmune disease, rather than an inference from sequence similarity alone.
Worked examples
Example 1 — reading a family/genetic risk scenario for type 1 diabetes. A patient's identical twin has type 1 diabetes. Population studies (Redondo et al., 2001, and subsequent cohorts) show the concordance rate for type 1 diabetes in monozygotic twins is roughly 30–50%, compared to roughly 6–7% in dizygotic twins and about 0.4% in the general population. Interpreting this: if type 1 diabetes were purely genetic (single-gene, fully penetrant), monozygotic concordance would approach 100%; if purely environmental with no genetic contribution, monozygotic and dizygotic concordance would be equal. The observed pattern — monozygotic concordance far above dizygotic, but far below 100% — quantitatively demonstrates the multifactorial model: genetic susceptibility (largely HLA-DR3/DR4 haplotypes affecting thymic negative selection efficiency and peptide presentation) substantially raises risk, but a non-genetic trigger (proposed candidates include enteroviral infection, e.g. coxsackievirus B, acting via molecular mimicry with the beta-cell autoantigen GAD65) is required to convert susceptibility into disease. A student should conclude: genetic risk alleles narrow the population who can get the disease; environmental exposure determines who, among the susceptible, actually does.
Example 2 — distinguishing organ-specific from systemic autoimmune disease from a lab panel. A patient presents with joint pain, a facial rash, and a positive anti-nuclear antibody (ANA) test at high titer, plus antibodies against double-stranded DNA. A second patient presents with an enlarged thyroid, fatigue, and antibodies against thyroid peroxidase (TPO) only. Reasoning through the antigen distribution: dsDNA and nuclear histones are expressed in every nucleated cell in the body, so an immune attack on them cannot be confined to one organ — immune complexes form systemically and deposit wherever blood flow and vessel geometry favor filtration (skin capillaries, renal glomeruli, joint synovium), producing the classic multi-organ picture of systemic lupus erythematosus. TPO, by contrast, is expressed almost exclusively in thyroid follicular cells; even a vigorous, fully activated autoimmune response against TPO has no other tissue to attack, producing the organ-restricted picture of Hashimoto's thyroiditis. The rule to extract: the anatomical distribution of an autoimmune disease is dictated primarily by the tissue distribution of the target autoantigen, not by the intensity of the immune response.
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "Autoimmune disease means the immune system is simply 'too strong.'" | Severity is not the issue; specificity is. A perfectly normal-strength immune response directed at the wrong (self) target causes disease; immunosuppression treats it by reducing that response, not by "calming down" an inherently overactive system. |
| "One faulty gene causes autoimmune disease, like a Mendelian trait." | Almost all common autoimmune diseases are polygenic and require an environmental trigger; APECED (AIRE) and IPEX (FoxP3) are rare monogenic exceptions that prove the rule by showing what happens when a single checkpoint is fully removed. |
| "Autoantibodies are always the cause of tissue damage." | Many organ-specific autoimmune diseases (type 1 diabetes, multiple sclerosis) are driven primarily by autoreactive T cells; autoantibodies present in the blood can be markers/diagnostic tools without being the direct mediators of damage. |
| "Molecular mimicry means the pathogen and self-antigen are genetically related." | Mimicry requires only enough three-dimensional structural similarity in the bound peptide–MHC complex to be recognized by the same receptor; the underlying gene sequences can be otherwise unrelated. |
| "Once autoimmunity starts, it targets only the original antigen forever." | Epitope spreading routinely broadens the response to new self-antigens over the course of chronic disease, which is why disease severity and antigen targets can evolve years after onset. |
The scientist behind it
Autoimmunity's modern mechanistic framework rests substantially on the work of Shimon Sakaguchi, whose experiments in the 1980s–1990s identified the cellular basis of dominant peripheral tolerance. Working initially in Japan and later at Kyoto University, Sakaguchi built directly on the thymectomy paradox established by Nishizuka and Sakakura: why does removing the thymus at a precise neonatal time point cause autoimmune disease rather than immunodeficiency? Sakaguchi's answer, developed over more than a decade, was that a specific, transferable subpopulation of CD4⁺ T cells — later shown to be defined by high CD25 expression and, crucially, expression of the transcription factor FoxP3 — actively suppresses autoreactive cells in the periphery. Transferring this population back into thymectomized or Treg-deficient mice prevented autoimmune disease, providing a rigorous, cell-transfer-based proof of concept comparable in logic to classic cell-transfer experiments elsewhere in immunology. His work reframed autoimmunity from a story about deletion alone to a story about active, ongoing suppression, a shift that underlies current Treg-based immunotherapy research.
Origins · 1 minShimon Sakaguchi and how this was discovered
Where this matters today
Autoimmunity research directly drives modern drug development. Biologic therapies targeting TNF-α (infliximab, adalimumab) and IL-6 (tocilizumab) treat rheumatoid arthritis by blocking the inflammatory cytokines that lower the activation threshold for autoreactive and bystander lymphocytes. Low-dose IL-2 therapy and engineered regulatory T-cell infusions are in active clinical trials for type 1 diabetes and lupus, aiming to directly restore the Treg-mediated peripheral tolerance checkpoint identified by Sakaguchi's experiments. Costimulation blockade drugs such as abatacept (a CTLA-4-Ig fusion protein) exploit the exact anergy-versus-activation logic described above, preventing autoreactive T cells from receiving the second signal required for full activation. Diagnostically, autoantibody panels (ANA, anti-dsDNA, anti-TPO, anti-CCP) remain first-line clinical tools precisely because the antigen-distribution logic worked out above predicts disease pattern from autoantibody identity. In vaccine safety science, molecular mimicry concerns motivate careful sequence-homology screening of candidate vaccine antigens against the human proteome, a direct legacy of the rheumatic fever mimicry story. Looking forward, the same tolerance concepts — central deletion, peripheral anergy, regulatory suppression, and the danger of unwanted costimulation — govern why a transplanted organ is rejected and why tumor-reactive T cells are often suppressed rather than activated, connecting autoimmunity mechanistically to Transplantation and Tumor Immunology.
Ask the tutor: How does epitope spreading during chronic autoimmune disease compare mechanistically to how the immune response evolves against a persistent, mutating virus?
Is autoimmune disease inherited, and can I predict it from family history?
Most common autoimmune diseases are polygenic, meaning many risk alleles (often HLA variants affecting antigen presentation and thymic selection) each contribute a small amount of risk. Family history raises probability but does not determine outcome: monozygotic twin concordance for type 1 diabetes is roughly 30–50%, well above the general population risk of about 0.4%, but far short of 100%. This gap is exactly what proves an environmental trigger is also required. Rare monogenic exceptions exist — APECED (AIRE mutations) and IPEX (FoxP3 mutations) are highly penetrant single-gene autoimmune syndromes — but these are clinically distinct from the polygenic diseases most patients and families encounter.
Why are most autoimmune diseases more common in women?
The mechanistic explanation is incomplete but several contributing factors are established. X-linked genes include several immune regulators (including FOXP3 itself), and incomplete X-inactivation can create a population of cells expressing two active X-linked immune alleles, potentially increasing autoreactive clone diversity. Estrogen has documented immunostimulatory effects on B-cell survival and antibody production, while androgens tend to be immunosuppressive, which may explain sex-differential incidence patterns and disease onset around hormonally active life stages (pregnancy, postpartum, menopause transitions). No single mechanism fully explains the roughly 9:1 female predominance seen in lupus, so this remains an active research area rather than settled fact.
Can an infection really cause a lifelong autoimmune disease after it's gone?
Yes, and rheumatic fever following streptococcal throat infection is the clearest proven example. The triggering infection does not need to persist because molecular mimicry means the cross-reactive lymphocyte population, once activated and expanded, continues attacking self-tissue independent of the original pathogen's presence. Epitope spreading can then broaden the response to additional self-antigens over time, further decoupling ongoing disease from the initial trigger. This is mechanistically different from simple ongoing infection and explains why treating or clearing the original pathogen after autoimmune disease has been established does not reverse the disease.
Do all autoimmune diseases involve autoantibodies?
No. Type 1 diabetes and multiple sclerosis are primarily driven by autoreactive CD4⁺ and CD8⁺ T cells attacking pancreatic beta cells or myelin directly; autoantibodies (like anti-GAD65 in diabetes) are present and diagnostically useful but are largely markers of the underlying autoreactive process rather than the direct cause of tissue destruction. By contrast, diseases like myasthenia gravis (anti-acetylcholine receptor antibodies) and Graves' disease (anti-TSH receptor antibodies) are directly antibody-mediated — the antibody itself blocks or aberrantly stimulates a receptor. Classifying a disease by its dominant effector mechanism (T-cell versus antibody) matters clinically because it determines which therapies (e.g., plasmapheresis versus T-cell-targeted immunosuppression) are likely to help.
What is the difference between anergy and immunological ignorance?
Anergy occurs when an autoreactive lymphocyte actually encounters its self-antigen, typically presented by a resting antigen-presenting cell lacking costimulatory molecules, and becomes intrinsically, durably unresponsive as a result — the cell is alive but functionally silenced. Immunological ignorance describes a different situation: the autoreactive lymphocyte never encounters its antigen at all, because the antigen is physically sequestered behind a tissue barrier (the blood-retinal barrier, blood-testis barrier) or present at levels too low to trigger a response. Ignorance is fragile: if the barrier is breached by injury or infection, previously ignorant autoreactive cells can suddenly encounter antigen and become activated, which is one proposed mechanism for sympathetic ophthalmia, where injury to one eye triggers autoimmune attack on both.
Why can't the immune system just "learn" to tolerate a transplanted organ the same way it tolerates self-tissue?
Central tolerance is established during T- and B-cell development specifically against antigens the thymus and bone marrow can display, which for an adult recipient does not include a donor organ transplanted later in life. Peripheral tolerance mechanisms (Tregs, anergy) can sometimes be induced against transplanted tissue, which is the basis of tolerance-induction protocols in transplant medicine, but this requires deliberate clinical intervention rather than occurring automatically. The donor organ's cells display foreign MHC molecules directly, which are intrinsically highly immunogenic and provoke a much stronger baseline T-cell response than a typical self-antigen ever would, making spontaneous tolerance far less likely without immunosuppression. This connects directly to the logic covered in transplantation immunology.
If everyone has some autoreactive lymphocytes, why doesn't everyone get autoimmune disease?
Because the presence of autoreactive clones is necessary but nowhere near sufficient for disease. Redundant peripheral controls — Treg suppression, anergy induction on resting APCs, physical sequestration of some antigens — normally keep these clones permanently silent regardless of their existence. Disease requires a specific, usually rare combination: genetic variants that weaken one or more of these checkpoints, an environmental trigger (often infection) that provides the inflammatory costimulation needed to activate rather than tolerize the autoreactive clone, and enough time for epitope spreading to sustain a self-perpetuating attack. This layered-redundancy model is precisely why autoimmune disease incidence, even for genetically susceptible individuals, remains well below 100%.