T-Cell Development and Selection

T-cell development is the process by which bone-marrow-derived lymphoid progenitors travel to the thymus and are converted into a functional, self-tolerant repertoire of T lymphocytes. Two problems must be solved simultaneously. First, T cells must be MHC-restricted: their receptors must be able to physically engage the host's own MHC molecules, since T cells never recognize free antigen — only peptide fragments displayed in the groove of MHC class I or class II. Second, T cells must be self-tolerant: despite recognizing self-MHC, they must not react destructively to the self-peptides displayed on it, or they would attack every cell in the body. The thymus resolves both problems through sequential screening of an initially random receptor repertoire. Random V(D)J recombination during early thymocyte development generates an enormous diversity of TCR specificities — theoretically more than 10^15 distinct αβ receptors — without regard to whether any given receptor is useful or safe. Positive selection then tests whether a thymocyte's receptor can engage self-MHC at all (weak affinity is a pass; no engagement is a fail, and the cell dies by neglect, the fate of the vast majority). Negative selection subsequently tests whether the surviving receptor binds self-peptide-MHC complexes too avidly (strong binding is a fail, triggering apoptosis; weak binding is a pass). The window between "just enough" and "too much" affinity is narrow, and this narrow window is what makes central tolerance possible. This two-step logic is often summarized as the "affinity model" of thymic selection: T-cell fate is determined largely by the strength (avidity/affinity and duration) of TCR engagement with self-peptide-MHC ligands, not simply by whether binding occurs. The same signal — TCR engagement — is interpreted differently depending on its intensity and the developmental context (cortex vs. medulla) in which it occurs, producing survival at low signal strength and death at high signal strength, with subtle variations in signal duration also influencing CD4 versus CD8 lineage choice.

Step 1 — Seeding the thymus and commitment. Common lymphoid progenitors arise in bone marrow and travel via blood to the thymus, entering at the corticomedullary junction. These early thymocytes lack CD4 and CD8 co-receptors ("double-negative," DN, further subdivided into DN1–DN4 stages) and have not yet rearranged a functional TCR.

Step 2 — TCR gene rearrangement (V(D)J recombination). At the DN stage, the TCRβ locus rearranges first: a Variable (V), Diversity (D), and Joining (J) gene segment are stochastically joined by the RAG1/RAG2 recombinase complex, which recognizes conserved recombination signal sequences flanking each segment and introduces double-strand breaks that are resolved by non-homologous end-joining repair machinery. Terminal deoxynucleotidyl transferase (TdT) adds non-templated "N-nucleotides" at the junctions, a major source of diversity beyond combinatorics alone. A successfully rearranged β chain pairs with an invariant pre-Tα chain to form the pre-TCR, which signals the cell to proliferate, suppress further β-chain rearrangement (allelic exclusion — ensuring one cell expresses one β chain), and turn on CD4 and CD8 together, becoming a "double-positive" (DP) thymocyte. The TCRα locus then rearranges (V and J segments only, no D segment), and because there is no allelic exclusion checkpoint analogous to pre-TCR signaling, a substantial fraction of DP cells transiently express two different α chains — later resolved by selection, since only one αβ combination need succeed.

Step 3 — Positive selection in the cortex. DP thymocytes, now expressing a complete αβ TCR alongside both CD4 and CD8, migrate through the densely packed cortex and probe cortical thymic epithelial cells (cTECs), which display self-peptides on both MHC class I and MHC class II. cTECs use a specialized proteasome (the "thymoproteasome," containing the catalytic subunit β5t) and distinctive peptide-loading machinery, producing a peptide repertoire different from that seen in peripheral tissue — deliberately favoring low-avidity interactions.

• If the TCR fails to engage self-MHC/self-peptide at all, the cell receives no survival signal and dies by apoptosis ("death by neglect") — this is the fate of roughly 90–95% of DP thymocytes.
• If the TCR engages self-MHC/peptide weakly, the cell receives a survival signal (mediated through Lck/ZAP-70 downstream of TCR engagement, promoting expression of the pro-survival protein Bcl-2 and the transcription factor Bcl-11b) and survives.

Step 4 — CD4/CD8 lineage commitment. During positive selection, the co-receptor bound (CD4 to MHC-II, CD8 to MHC-I) determines which lineage the cell adopts, in a process still actively studied but well described by the kinetic signaling model: sustained TCR/co-receptor signaling (persistent because the correct co-receptor stabilizes the interaction) favors CD4⁺ helper lineage and induces the transcription factor ThPOK, while a shorter, interrupted signal (because the "wrong" co-receptor cannot stabilize it) allows an interleukin-7 signal to dominate, inducing the transcription factor Runx3 and CD8⁺ cytotoxic lineage commitment. The result is that MHC-II-restricted cells become CD4⁺ helpers and MHC-I-restricted cells become CD8⁺ cytotoxic T cells — a matching that is essential, since a CD4⁺ cell that could only see MHC-I would have no functional co-receptor engagement during activation.

Step 5 — Negative selection in the medulla. Positively selected thymocytes downregulate the chemokine receptor CCR9 and upregulate CCR7, which draws them toward the medulla along a CCL19/CCL21 gradient produced by medullary thymic epithelial cells (mTECs). mTECs display an extraordinary breadth of self-peptides, including tissue-restricted antigens (insulin, thyroglobulin, myelin proteins, and thousands of others normally confined to peripheral organs) that they express "promiscuously" thanks to the transcription factor AIRE (autoimmune regulator) and a related factor, Fezf2. Thymic dendritic cells (some of which migrate in from the periphery carrying peripheral self-antigens) and B cells also contribute antigen-presenting capacity in the medulla.

• If a thymocyte's TCR binds a self-peptide-MHC complex with high avidity, it receives a strong signal that triggers apoptosis (upregulating the pro-apoptotic protein Bim, among other mediators) — clonal deletion.
• Some high-affinity self-reactive cells escape deletion but are instead diverted into the regulatory T cell (Treg) lineage, marked by the transcription factor FoxP3; these cells leave the thymus and actively suppress self-reactive immune responses in the periphery rather than mounting them — a safety net layered on top of deletion.
• Cells with only weak-to-moderate residual self-reactivity survive negative selection and are exported as mature naive T cells.

Step 6 — Export. Mature single-positive (CD4⁺ or CD8⁺) thymocytes exit the thymus via blood vessels at the corticomedullary junction, a process gated by the lipid-sensing receptor S1P1, and enter the recirculating pool of naive T cells that populates lymph nodes and spleen, awaiting activation as described in T-Cell Activation. Of every thymocyte that begins this process, on the order of 95–98% die within the thymus — only a small minority, commonly cited as 1–5%, complete the gauntlet successfully.

Thymus-grafting experiments proved MHC-restriction is imposed by the thymic epithelium itself. In classic studies by Bevan and Zinkernagel (late 1970s and early 1980s), radiation bone-marrow chimeras were constructed: bone marrow from strain A mice (MHC haplotype a) was transplanted into irradiated, thymectomized strain B mice (MHC haplotype b) that had been given a thymus graft from a third, MHC-different strain, or the reverse combinations. The key finding was that the MHC restriction of the mature T cells that developed was dictated by the MHC type of the thymic epithelium the precursor cells matured in, not by the MHC type of the bone-marrow donor. T cells from strain-A bone marrow that matured in a strain-B thymus were restricted to strain-B MHC when tested for cytotoxic killing, even though the T cells themselves were genetically strain A. This directly demonstrated that positive selection — the acquisition of MHC restriction — is imposed by contact with thymic stromal cells during development, not fixed by the lymphocyte's genotype alone.

TCR-transgenic mice provided a clean, traceable system that separated positive from negative selection. Kisielow, von Boehmer, and colleagues (1988) engineered mice carrying a transgenic TCR specific for the male HY antigen presented on MHC class I (H-2Db). In female transgenic mice (no HY antigen present), the transgenic thymocytes underwent normal positive selection and matured efficiently into CD8⁺ single-positive cells expressing the transgenic receptor — proving that a single, defined TCR specificity, in the absence of its cognate self-antigen, is sufficient to drive maturation. In male transgenic mice (HY antigen present, since males express it), by contrast, the double-positive thymocytes bearing the high-affinity HY-specific receptor were massively and specifically deleted at the DP stage, before maturing — proving that encounter with the antigen the receptor recognizes, at high affinity, triggers deletion rather than maturation. Because the only variable between males and females was the presence of the antigen, this elegantly isolated negative selection as a distinct, antigen-driven process rather than a side effect of positive selection.

AIRE-knockout mice and human AIRE mutations revealed the mechanism of tissue-antigen display in the medulla. Mice engineered to lack AIRE fail to express many tissue-restricted self-antigens in the thymic medulla and develop multi-organ lymphocytic infiltration and autoantibodies against organs including the retina, ovary, and adrenal gland. In humans, loss-of-function mutations in AIRE cause Autoimmune Polyendocrinopathy–Candidiasis–Ectodermal Dystrophy (APECED/APS-1), a rare Mendelian autoimmune disease featuring autoimmune destruction of the parathyroids, adrenal cortex, and other endocrine organs, together with chronic mucocutaneous candidiasis. Because APECED is a single-gene disease whose entire phenotype traces to a failure of thymic self-antigen display, it stands as strong human genetic evidence that negative selection against tissue-restricted antigens is a real, necessary, and AIRE-dependent process — not merely a mouse phenomenon or theoretical model.

RAG-knockout mice showed the requirement for gene rearrangement itself. Mice lacking RAG1 or RAG2 cannot rearrange TCR (or immunoglobulin) genes and produce no mature T or B cells at all, arresting at the double-negative stage; this confirmed that V(D)J recombination is an obligatory, not merely convenient, step generating the receptor before any selection can act on it (Mombaerts et al., 1992; Shinkai et al., 1992).

Example 1 — Predicting the fate of a thymocyte given relative TCR affinity. Suppose a researcher generates three retrogenic mouse lines, each expressing a different fixed TCR on an otherwise RAG-deficient background (so every thymocyte expresses only the introduced receptor). Line 1's TCR has essentially no measurable affinity for any self-peptide-MHC complex in the mouse's own tissues. Line 2's TCR binds a ubiquitously expressed self-peptide-MHC-I complex with intermediate (micromolar) affinity. Line 3's TCR binds a different self-peptide-MHC-I complex with very high (low-micromolar to nanomolar) affinity, and that peptide happens to derive from a protein expressed broadly in the medulla under AIRE control. Reasoning it through: Line 1 thymocytes receive no engagement signal in the cortex and die by neglect — this mouse will show a nearly empty peripheral T-cell compartment, dominated by cells with severely reduced thymic output. Line 2 thymocytes receive just enough signal to pass positive selection; because the affinity is intermediate rather than extreme, and assuming the antigen is not additionally amplified to high local concentration in the medulla, many of these cells survive negative selection too, and the mouse develops a robust CD8⁺ T-cell compartment monospecific for that receptor. Line 3 thymocytes pass positive selection in the cortex (any binding at all is initially permissive) but then encounter the same or a related high-affinity ligand displayed abundantly by AIRE⁺ mTECs in the medulla; the strong signal triggers clonal deletion, and the mouse's periphery is nearly devoid of that T-cell clone — unless some cells escape into the Treg lineage instead, in which case the periphery shows a population of FoxP3⁺ regulatory cells bearing that specificity rather than conventional effector cells. This thought experiment mirrors real HY-TCR transgenic data and shows how a single affinity parameter, read against the cortex/medulla staging of selection, predicts population-level outcomes.

Example 2 — Quantifying thymic output and repertoire loss. A simplified but pedagogically standard estimate: a typical thymus in a young mouse produces on the order of 10^8–10^9 DP thymocytes per day as candidates for selection. Suppose:

If 10^8 DP thymocytes enter selection on a given day, roughly 10^7 pass positive selection (10%). Applying a further 40% loss to negative selection leaves 0.6 × 10^7 = 6 × 10^6 cells, or about 6% of the original DP cohort, exiting as mature single-positive thymocytes. This back-of-envelope calculation, consistent with commonly cited estimates that only 1–5% of thymocytes are ultimately exported, illustrates why the thymus is often described as running a strikingly wasteful — but tolerance-preserving — production line: the "cost" of guaranteeing self-tolerance is discarding the vast majority of randomly generated receptors.

Much of the conceptual framework for thymic selection was built by immunologists working in the 1970s–1990s who combined mouse genetics, radiation chimeras, and (later) transgenic technology to dissect a process that had previously been observed only as a puzzling fact: mice thymectomized shortly after birth failed to develop functional cell-mediated immunity and wasted away, a phenomenon first documented by Jacques Miller in 1961, which established that the thymus is not vestigial but essential for T-cell development. Building on this, Rolf Zinkernagel and Peter Doherty's discovery (1974) that cytotoxic T cells recognize antigen only in the context of self-MHC (for which they received the 1996 Nobel Prize) raised the question this article answers: how does the immune system ensure T cells are restricted to self-MHC in the first place? Michael Bevan's and Zinkernagel's own subsequent thymus-transplantation chimera experiments supplied the answer — the thymic stroma imposes MHC restriction during development. Harald von Boehmer's laboratory then built the TCR-transgenic mouse systems (including the HY-specific model) that cleanly separated positive selection from negative selection as mechanistically distinct, independently manipulable processes.

Understanding thymic selection underlies modern approaches to autoimmune disease, transplantation, and cancer immunotherapy. In APECED/APS-1, direct genetic evidence that AIRE loss causes multi-organ autoimmunity has motivated research into whether subtler variation in AIRE function or mTEC antigen display contributes to more common autoimmune diseases such as type 1 diabetes and Addison's disease. In thymic transplantation for DiGeorge syndrome (a congenital condition in which the thymus fails to develop, leaving infants without functional T-cell immunity), cultured thymic tissue grafts are used clinically, and their success depends on the same cortex/medulla selection logic engineered ex vivo. In cancer immunotherapy, engineering T cells with chimeric antigen receptors (CAR-T therapy) or TCR-engineered T cells deliberately bypasses normal thymic selection, which is precisely why such engineered receptors must be screened carefully for off-target cross-reactivity with normal self-tissue — a risk that natural negative selection would otherwise have filtered out. Conversely, understanding negative selection informs checkpoint-inhibitor toxicity: drugs that block inhibitory receptors (like CTLA-4 or PD-1) to unleash anti-tumor T-cell responses often unmask self-reactive T-cell clones that survived thymic selection at the edge of tolerance, producing autoimmune-like side effects in patients. In organ transplantation, knowledge of thymic MHC restriction informs strategies for inducing mixed chimerism and transplant tolerance by manipulating thymic education of new T cells. Finally, in vaccine and repertoire engineering, deliberate modulation of thymic selection (for example in mouse models expressing defined self-antigens) is used to probe how much of the natural T-cell repertoire is shaped by particular self-proteins, informing rational vaccine antigen design.

Ask the tutor: How does the kinetic signaling model explain why a thymocyte with sustained TCR engagement becomes CD4+ rather than CD8+?