provenance: negative-space populate (2026-06-23) — Document B (form-respecting companion) to synthesis/the-immune-self-the-boundary-that-learns.md. This is the immunology gap filled in the empirical register the corpus forecloses (Face C: the operative/empirical/made). Its plainness is the fill. Do NOT add metaphor, consciousness-framing, or "maps to" moves — enriching it toward house style would re-coat the gap it exists to open. State what is true.
Immunology: The Mechanism
A plain account of how the adaptive immune system learns to distinguish what to attack from what to tolerate, and what happens when it fails. No analogy is drawn. This is the companion specimen to a house-style document; here the claims are meant to be simply correct.
Two systems
Immunity has two arms. Innate immunity is fast (minutes to hours), non-specific, and inherited unchanged: phagocytes, natural killer cells, the complement cascade, and pattern-recognition receptors that detect conserved microbial molecules (pathogen-associated molecular patterns) and signals of tissue damage. It has no memory. Adaptive immunity is slow on first exposure (days), highly specific, and capable of memory: B lymphocytes (which secrete antibodies) and T lymphocytes (which kill infected cells or regulate other immune cells). The account below concerns adaptive immunity, where the problem of self-tolerance arises.
Generating the repertoire
Each lymphocyte carries one type of antigen receptor. The diversity of these receptors is produced by V(D)J recombination: during lymphocyte development, enzymes (the RAG-1 and RAG-2 recombinases) cut and rejoin variable (V), diversity (D), and joining (J) gene segments in a combinatorial, partly random fashion, with additional nucleotides inserted at the junctions. A germline of a few hundred segments yields an estimated 10^11 or more distinct receptor specificities. The repertoire is therefore generated before any antigen is encountered, and largely at random. It necessarily includes a large number of receptors that bind the organism's own molecules.
Clonal selection
Antigen does not instruct a lymphocyte what to recognize; it selects lymphocytes that already, by chance, bear a fitting receptor. A pathogen binds the few cells whose receptors match it; those cells proliferate (clonal expansion) and differentiate into effector cells and memory cells. This is the clonal selection principle (Burnet, 1957). Recognition capacity precedes and is independent of the antigen; the antigen merely amplifies the matching clones.
Central tolerance
Because the repertoire is random, self-reactive cells must be removed or restrained, or the system will attack its own tissues. The first checkpoint is central tolerance, in the primary lymphoid organs.
T cells mature in the thymus and undergo two screens. Positive selection: a T cell must bind self-MHC (the molecules that present peptides) weakly enough to be functional; cells that cannot are allowed to die. Negative selection: a T cell that binds self-peptide–MHC complexes too strongly is deleted (clonal deletion) or diverted into a regulatory lineage. Critically, medullary thymic epithelial cells express a broad sample of tissue-restricted self-antigens — proteins normally found only in the pancreas, retina, nervous system, and elsewhere — under the control of the transcriptional regulator AIRE (autoimmune regulator). This allows developing T cells to be screened against self-antigens they would not otherwise meet in the thymus. Loss-of-function mutations in AIRE cause a multi-organ autoimmune syndrome (APECED/APS-1), demonstrating that this thymic presentation of peripheral self is causally required for tolerance.
B cells undergo analogous negative selection in the bone marrow, including receptor editing (further V(D)J rearrangement to replace a self-reactive receptor) and deletion.
Peripheral tolerance
Central tolerance is incomplete; some self-reactive lymphocytes reach the circulation. A second layer, peripheral tolerance, restrains them. Its mechanisms include: anergy (functional inactivation of a lymphocyte that receives an antigen signal without the required co-stimulatory signal); regulatory T cells (Tregs, marked by the transcription factor FOXP3) that actively suppress other lymphocytes; and deletion of chronically activated cells. Mutations in FOXP3 cause IPEX syndrome, a severe systemic autoimmunity, showing that active suppression — not merely the absence of self-reactive cells — is required to maintain tolerance.
Tolerance is therefore not a passive default. It is produced and maintained by dedicated cells and signals, and it consumes resources continuously.
The danger / damage model
The classical framework held that the immune system distinguishes self from non-self. This framework does not account for several observations: the immune system tolerates the commensal microbiota (an enormous foreign load), tolerates dietary antigens (oral tolerance), tolerates the semi-allogeneic fetus, and yet can attack autologous tissue. An alternative model (Matzinger, 1994) proposes that immune activation is driven by signals of damage — molecules released by cells dying by necrosis or under stress (damage-associated molecular patterns) — rather than by foreignness as such. On this view, a foreign entity that causes no damage need not provoke a response, and a self-antigen presented in a context of tissue damage can. The discriminating variable is harm to the tissue, not the genetic origin of the antigen. The danger model and the infectious-non-self model (Janeway, 1989) both locate the decisive signal in context (damage or microbial pattern) supplied to innate cells, which then license or withhold the adaptive response.
Tolerance of beneficial and necessary non-self
Two cases make the point that tolerance is an active achievement directed at specific foreign entities.
The microbiota. The gut contains on the order of 10^13 bacteria of thousands of species. The mucosal immune system actively restrains responses to them — through Tregs, secretory IgA that contains rather than eliminates them, and an epithelial barrier — while remaining able to respond to pathogens. Breakdown of this regulation is implicated in inflammatory bowel disease.
The fetus. The placental mammalian fetus expresses paternal alloantigens and is, immunologically, a semi-allograft. It is not rejected because the maternal–fetal interface enforces local tolerance: invasive trophoblast expresses the non-classical, minimally polymorphic HLA-G rather than classical HLA molecules; regulatory T cells specific for fetal antigens expand during pregnancy; and the decidua restrains effector responses. Failures of these mechanisms are associated with recurrent miscarriage and pre-eclampsia.
Failure: autoimmunity and its converse
When tolerance fails, self-reactive lymphocytes attack autologous tissue, producing autoimmune disease: type-1 diabetes (T-cell destruction of pancreatic beta cells), multiple sclerosis (against central nervous system myelin), rheumatoid arthritis, systemic lupus erythematosus, and others. Causation is typically multifactorial — predisposing HLA alleles, defects in central or peripheral tolerance, and environmental triggers including infection (via molecular mimicry or bystander activation).
The converse failures are misdirected responses against harmless or beneficial targets: allergy (responses to innocuous environmental antigens), transplant rejection (responses to therapeutic non-self), and graft-versus-host disease. In each case the pathology is a misclassification — tissue that should be tolerated is attacked, or a threat that should be attacked is tolerated (immune evasion by tumors and chronic pathogens).
Memory
After an adaptive response resolves, long-lived memory B and T cells persist, enabling a faster and larger response on re-exposure to the same antigen. Vaccination exploits this by presenting antigen (attenuated, inactivated, subunit, or nucleic-acid-encoded) without the pathology, generating memory in advance of natural exposure.
Summary
The adaptive immune system generates a vast, largely random recognition repertoire; selects from it clonally upon antigen encounter; removes or suppresses self-reactive members through central and peripheral tolerance, including the thymic presentation of peripheral self under AIRE and active suppression by FOXP3+ Tregs; gates effector responses on signals of damage and microbial pattern rather than on foreignness alone; actively tolerates specified beneficial and necessary non-self (microbiota, fetus); and retains memory of prior encounters. Its characteristic diseases are failures of these discriminations in either direction.