Loading field map
For more than a century, the central puzzle of autoimmunity has been deceptively simple: what normally prevents the immune system from damaging its own tissues? The answer has never been one thing. Early immunologists thought self-attack was almost impossible; later ones saw it as a consequence of a few rogue cells; still later ones argued that it is a constant possibility only averted by complex layers of regulation, environmental context, and genetic predisposition. The history of autoimmunity research is therefore not a straight line from ignorance to understanding but a series of frameworks that each reframed the question itself.
Horror Autotoxicus (1900–1950) Paul Ehrlich observed that animals rarely produced antibodies against their own tissues and concluded that the immune system is intrinsically wired to avoid self-reactivity—a condition he called "horror autotoxicus." This idea hardened into a dogma, effectively squelching research into autoimmune mechanisms for decades. When autoimmune phenomena were observed (as in hemolytic anemia), they were dismissed as laboratory artifacts or attributed to some form of infection that allowed self-tolerance to break. For half a century, the dominant assumption was that autoimmunity could not happen in a healthy organism.
Clonal Selection Theory (1950–1970) Frank Macfarlane Burnet’s clonal selection theory preserved Ehrlich’s intuition but gave it a mechanistic basis. Burnet proposed that self-reactive lymphocyte clones are eliminated during development—a process later called central tolerance. In this view, autoimmunity arose only if a forbidden clone escaped deletion and was subsequently activated. The theory narrowed the explanation from a global inability to an accident in clonal purging. For a time, it seemed that self/nonself discrimination was simply a matter of deleting self-reactive cells early in life. But clinical observations of organ-specific autoimmunity (e.g., Hashimoto’s thyroiditis) suggested that self-reactive T and B cells could exist in the periphery in healthy individuals, forcing the field to look beyond central deletion.
Idiotypic Network Theory (1970–1990) Niels Jerne’s idiotypic network theory offered a radically different answer. Rather than a clean separation of self and nonself, Jerne envisioned the immune system as a network of interacting antibodies and lymphocytes constantly modulating each other through idiotype–anti-idiotype interactions. Autoimmunity, in this framework, was not a failure of deletion but a disruption of network equilibrium—a kind of regulatory imbalance. The theory accounted for peripheral regulation that clonal selection could not explain. However, its abstract formalism made it difficult to test experimentally, and as molecular tools advanced, it gradually gave way to more concrete mechanisms. Yet its emphasis on regulation—an immune system that actively prevents self-attack—remained influential.
Th1/Th2 Paradigm (1980–2000) Around the same time, the discovery that CD4+ T helper cells could differentiate into two functional subsets—Th1 cells driving cellular immunity and Th2 cells driving humoral immunity—opened a new avenue for classifying autoimmune diseases. Researchers found that certain autoimmune conditions were dominated by Th1 responses (e.g., multiple sclerosis, type 1 diabetes) while others were Th2-driven (e.g., systemic lupus erythematosus, allergies). The framework shifted attention from whether self-reactivity occurs to how the effector response is directed. But as more subsets were discovered (Th17, T follicular helper cells, etc.), the Th1/Th2 dichotomy was revealed as an oversimplification. The paradigm was not wrong but narrowed: it organized a vast literature and provided testable predictions, even as later work absorbed and expanded its categories.
Genetic Susceptibility Framework (1980–Present) While these immunological theories developed, a parallel line of inquiry arose from human genetics. The strong association between certain HLA alleles and autoimmune diseases—such as HLA-DR4 in rheumatoid arthritis—established that autoimmunity has a heritable component. The genetic susceptibility framework treats autoimmune disease as a polygenic risk, where multiple alleles each confer small effects that collectively predispose an individual. Genome-wide association studies (GWAS) have since identified hundreds of risk loci, many of which implicate immune regulatory pathways. Unlike earlier frameworks that focused on single mechanisms, this one emphasized the role of individual genetic background. It does not replace immunological theories but adds an infrastructure of risk stratification. Today, it remains an active research program, integrated with mechanistic frameworks to explain why some individuals—but not others—develop autoimmunity.
Danger Model (1994–Present) In 1994, Polly Matzinger challenged the bedrock assumption that the immune system is primed to distinguish self from nonself. Instead, she proposed that the key decision is whether to mount a response to a dangerous situation—tissue stress, cell death, or microbial products—regardless of the antigen’s origin. In the danger model, autoimmunity occurs when self-antigens are presented in a context that the immune system interprets as dangerous. This framework directly contested the clean Self-Nonself model inherited from Burnet. It was absorbed into the broader field of innate immunity, but its core insight—that context matters as much as antigen—persists. However, the danger model never fully supplanted the regulatory view; instead, it forced a reexamination of what “danger signals” actually are.
Regulatory T Cell Framework (1995–Present) Just as the danger model gained traction, a more concrete cellular mechanism for preventing autoimmunity emerged: regulatory T cells (Tregs). Sakaguchi and colleagues showed that a distinct subset of CD4+ T cells expressing FoxP3 actively suppresses self-reactive lymphocytes in the periphery. The regulatory T cell framework provided the cellular basis for the kind of peripheral tolerance that the idiotypic network theory had only abstractly described. Where the danger model emphasized environmental signals, the Treg framework emphasized an active, dedicated suppressor population. The two frameworks are not contradictory but complementary: danger signals may override Treg suppression, and Treg deficiency may allow self-reactivity even without danger. Today, Treg biology is a cornerstone of autoimmune research, with therapies aiming to expand Tregs to treat autoimmune disease.
Innate Immune Priming Framework (2000–Present) The danger model opened the door for a deeper appreciation of innate immune contributions. The innate immune priming framework argues that autoimmunity often begins with innate immune activation—through pattern recognition receptors (e.g., TLRs) and the phenomenon of trained immunity, where innate cells develop a long-term memory-like state. Chronic inflammatory stimuli, such as those from the microbiome or sterile inflammation, can prime innate cells to produce cytokines that break T cell tolerance. This framework moves beyond the adaptive-centric view of earlier models, placing innate cells as active participants rather than passive relays. It overlaps with the danger model (both emphasize activation signals) but extends it by describing specific molecular pathways (e.g., inflammasome activation, type I interferon production) that can initiate or perpetuate autoimmune inflammation.
Systems Immunology Framework (2005–Present) The most recent framework is not a single hypothesis but a methodological shift. The systems immunology framework aims to integrate data across genomics, transcriptomics, proteomics, and clinical phenotyping to build comprehensive models of autoimmune disease. Instead of championing one mechanism (deletion, network, danger, Tregs, genetics), it treats autoimmunity as a systems-level property emerging from the interaction of multiple components. For example, high-dimensional analyses of blood from autoimmune patients have revealed clusters of gene expression signatures that correlate with disease activity, often pointing to previously unappreciated pathways. This framework does not replace earlier ones; rather, it provides the tools to test how they interact in individual patients. It is leading today because it accommodates the complexity that earlier frameworks individually oversimplified.
Today, the most active frameworks are the genetic susceptibility framework, the regulatory T cell framework, the danger model (now subsumed into innate immune signaling), the innate immune priming framework, and the systems immunology framework. They agree on several points: autoimmunity is a multifactorial process in which genetic predisposition sets the stage, regulatory mechanisms normally maintain tolerance, and environmental triggers (infections, tissue damage, dysbiosis) can disrupt that balance. They disagree on which of these elements is most decisive. The regulatory T cell framework emphasizes the failure of active suppression; the danger model and innate immune priming framework emphasize the presence of activating signals; the genetic susceptibility framework points to the polygenic lottery; and the systems immunology framework argues that the field should not choose among them but must consider their interplay. The unresolved tension—whether autoimmunity is primarily a failure of tolerance, an excess of activation, or a specific genetic vulnerability—continues to drive research forward.