Immune Response

Immunoglobulin Structure

The structures of antibody molecules all share remarkable conservation of secondary conformation. The Ig-like fold is a domain shared among diverse proteins, each with signature hydrophobic common cores and common disulfide bridges. The antibody molecule is the prototype member of the Ig supergene family, a group that includes not only Ig but also TCR, the MHC, and a number of costimulatory molecules. Each Ig molecule is composed of two light chains of approximately 25 kd and two heavy chains of approximately 50 or 70 kd ( Fig. 23-3 ); each chain is composed of multiple Ig domains. B-cell antibodies can be subdivided by heavy-chain isotype and further subdivided by subclass ( Table 22-2 ).

There are five classes of antibodies—IgM, IgG, IgA, IgD, and IgE—defined by expression of the heavy chain: µ, γ, α, δ, and ε. The Ig-heavy chains contain sites for binding of Fc receptor (FcR) and complement, and each subtype appears to have a particular role in immune defense. The first class of antibody to be produced during development is IgM, which is expressed on the surface as a monomer and secreted as a pentamer. B cells that have undergone class switching (discussed later) express IgG, IgA, IgD, or IgE on the surface as the BCR or secrete a splice form of the Ig with the same heavy-chain isotype and specificity but lacking a transmembrane region (or both). The function of IgD in humans is not clear, although its expression is a marker for later B-cell development. IgA is secreted as a dimer and is found at mucosal surfaces such as the intestine. The bulk of circulating Ig is in the form of IgG, which is monomeric and further subdivided into four subclasses: IgG1, IgG2, IgG3, and IgG4. Persistently low serum levels of one or more IgG subclasses, in the presence or absence of IgA deficiency, may cause individuals to be susceptible to recurrent infections. Though controversial, prophylactic use of intravenous Ig has been suggested.

There are two subclasses of light chains: κ and λ (see Fig. 23-3 ). Both subclasses appear to serve similar functions. Monoclonal disorders of terminally differentiated mature B cells or plasma cells (e.g., multiple myeloma) express cells of a given light chain—that is, either κ or λ, but not both.

The structure of each Ig molecule is determined by hydrophobic interactions and disulfide bridges between the two heavy chains and two light chains (see Fig. 23-3 ). For each chain the N-terminal domain has a variable or polymorphic amino acid structure (V _H for heavy chain, V _L for light chain), whereas the C-terminal portions are constant within each heavy-chain or light-chain class (termed C _H 1, C _H 2, C _H 3 for heavy chain and C _L for light chain). These domains form pairs, V _H with V _L , C _H 1 with C _L , and C _H 3 with C _H 3. Proteolytic digestion of Ig by papain results in two Fab fragments, each composed of the V _H /V _L , C _H 1/C _L domains, and an Fc fragment composed of the two C _H 2 and two C _H 3 domains. The binding region of Ig that recognizes antigen is composed of portions of the V _H and V _L domains. Within each variable region there are three portions with increased amino acid variation, called hypervariable regions (HV1, HV2, HV3) or complementarity-determining regions (CDR1, CDR2, CDR3). The highest variability is found in CDR3, and all three CDRs from both chains participate in the binding of antigen. Variability within each variable region and each CDR and pairing of a unique heavy chain with a unique light chain all contribute to the combinatorial diversity of Ig specificity.

Immunoglobulin Rearrangement and B-Cell Maturation in Bone Marrow

The genes encoding the heavy chains of Ig are located on human chromosome 14, whereas the genes encoding κ and λ light chains are located on chromosomes 2 and 22, respectively. The heavy- and light-chain loci are each composed of variable (V), diversity (D), and joining (J) gene segments that recombine to yield the V _H and V _L domains, which in turn form the antigen contact surface of the antibody molecule and confer specificity. In addition, the heavy-chain locus contains nine heavy-chain constant (C) region genes (µ, δ, γ1, γ2, γ3, γ4, α1, α2, and ε constant regions) that define the mature Ig isotype.

Ig rearrangement occurs serially, and cells failing to rearrange the Ig genes successfully are eliminated and do not develop further ( Fig. 23-4 ). B-cell maturation also results in cell surface protein expression patterns that are used to define the pro–B-cell, pre–B-cell, and immature B-cell stages of development. CD34+ CD19− CD10− progenitors that have not yet committed to the B lineage have Ig genes in the germline configuration. The earliest B-cell precursors express the pan B-cell marker CD19 and the early marker CD10. They also express components of the rearrangement machinery such as recombination-activating genes (RAGs) and thus have early gene rearrangements detectable. Rearrangement begins with one heavy-chain allele; one of the 29 D _H exons is spliced to one of the 6 J _H exons, the intervening D and J sequences are eliminated, and a DJ segment is produced. The pro-B cell then splices one of the approximately 50 V _H exons to the DJ segment to form a V(D)J segment. Finally, the µ constant region of the heavy chain (C _µ ) is spliced onto the V(D)J segments.

Differentiation to the pre-B stage is triggered by expression of intact µ heavy-chain protein or cytoplasmic µ, which occurs only in cells that bear in-frame productive V(D)J-C _µ rearrangements. If rearrangement of the first heavy-chain locus is unsuccessful, the second allele undergoes rearrangement. Because the light-chain genes have not yet been expressed, cytoplasmic µ complexes instead with the nonpolymorphic surrogate light-chain proteins λ5 and VpreB. Cytoplasmic µ, surrogate light chain, and the accessory signaling molecules Ig-α (CD79a) and Ig-β (CD79b) together form the pre–B-cell receptor. Ensuing signals induce proliferation and instruct the cell to begin rearrangement of one of the κ light-chain loci. Pre-B cells lose CD34 expression, continue to express CD19 and CD10, and additionally upregulate CD20. The signaling cascade downstream of the pre–B-cell receptor is highly analogous to that of the mature BCR and activates a number of critical kinases, including Bruton tyrosine kinase (Btk) (discussed later). Boys born with mutations in Btk, manifested as X-linked agammaglobulinemia, have normal numbers of pro-B cells but lack pre-B cells and their progeny. Thus formation of the pre–B-cell receptor along with activation of downstream signaling is a critical checkpoint in this stage of maturation.

Light-chain genes rearrange by a process similar to the recombination of heavy-chain genes. The variable region of the κ gene, located on chromosome 2, is spliced to the J gene segment to produce a VJ region, which then splices to the one constant region (C _κ ). If nonproductive, the second κ locus will rearrange. If both attempts to rearrange the κ locus are unsuccessful, the λ locus, located on chromosome 22, will begin to rearrange. Productive rearrangement of any light-chain locus prevents further B-cell rearrangement and also generates survival signals. Pre-B cells that successfully rearrange a light-chain gene will then express mature IgM protein, which appears as surface IgM (sIgM) complexed to Ig-α and Ig-β, thus forming the BCR. Immature B cells are sIgM+, CD79+, CD19+, CD10+, and CD20+, but they have not acquired IgD expression, which occurs after peripheral maturation.

Expression of the tissue-specific recombination-activating genes RAG-1 and RAG-2 is required for recombination. In addition, diversity is enhanced by the fact that the junctions between the D and J regions and the V and DJ regions are not precise; not only can the nucleotides vary, but one or more nucleotides (termed N regions ), not encoded in the genome, may also be randomly inserted (N-region diversification). The nuclear enzyme terminal deoxyribonucleotidyl transferase (TdT) is the polymerase that adds random nucleotides to recombination junctions and may have other roles. Recruitment of TdT to the junctions appears to depend on the Ku molecule (specifically Ku80), a heterodimer that binds DNA ends and is required for V(D)J recombination and DNA double-stranded break repair. Other proteins critical for nonhomologous end rejoining include Ku70, DNA-dependent protein kinase, DNA ligase IV, XRCC4, and the recently cloned protein Artemis. Defects in the recombination machinery in genes such as RAG1, RAG2, DNA ligase IV, and Artemis all result in severe combined immunodeficiency (see Chapter 24 ).

Peripheral B-Cell Maturation and Differentiation

Immature B cells must undergo further maturation in the spleen before acquiring the phenotype and functions of a naïve mature B cell. Naïve mature B cells are resting cells in G ₀ phase; express the pan B-cell marker CD19, the marker CD20, and sIgM; and in contrast to immature bone marrow B cells, have lost CD10 expression and gained IgD expression. A similar population in the mouse is believed to be the common precursor for both follicular and marginal-zone B cells, named for their location within the spleen, discussed later.

The development of early B precursors into immature B cells is a process that requires Ig rearrangement and expression of the pre–B-cell receptor and BCR, but it does not require recognition of antigen. Maturation to the naïve resting B cell involves selection of B cells that recognize antigen (positive selection) while eliminating B cells whose BCR recognize self-antigens with high affinity (negative selection). It has been proposed, on the basis of murine data, that determination of mature B-cell fate relies on the relative signal strength and specificity of the BCR, with higher and lower signal strength driving to the follicular and marginal-zone lineages, respectively. Negative selection in particular is important for the deletion of potentially autoreactive clones and induction of B-cell tolerance. Indeed, it is estimated that 55% to 75% of immature B cells in humans recognize self-antigens, and thus a minority of developing B cells pass the negative-selection test.

T-Cell Receptor Subtypes and Specificity

The TCR is a cell surface receptor capable of recognizing discrete antigens bound to the MHC or to MHC-related molecules expressed on the surface of APCs. TCRs are disulfide-linked heterodimers of either αβ or γδ proteins. TCR αβ−bearing T cells constitute the vast majority of circulating T cells, most of which recognize peptide antigens bound to polymorphic MHC class I or class II molecules. A subset of TCR αβ T cells, representing approximately 1% of all T cells in humans, in contrast recognize lipid antigens bound to the MHC class I–related nonpolymorphic molecule CD1 and are termed CD1-restricted or NK T cells, discussed in a later section. TCR γδ–bearing T cells account for a tiny fraction of the circulating T cells in humans but are very prevalent in intestinal tissue and are thought to recognize nonpeptide antigens (discussed later). The functions of CD1-restricted TCR αβ T cells and TCR γδ T cells are somewhat distinct from those of conventional TCR αβ T cells. The discussion of intrathymic T-cell development in a subsequent section will focus on the major conventional subset of T cells.

Early T-Cell Commitment and Seeding

The molecular events associated with and responsible for commitment to the T lineage have been increasingly elucidated in both murine and human models over the last 10 years. The question of whether T-lineage commitment occurs within the thymus or the bone marrow has been actively investigated. In addition to pluripotent CD34+ HSCs, a number of other more committed progenitors in bone marrow with specific cell surface marker characteristics have been identified with T-lymphoid potential, primarily in the mouse, but these progenitors in general also retain the capacity to differentiate into other lymphoid lineages, including the B and NK lineages. That various subsets identified in the human thymus similarly retain bilineage potential (i.e., NK/T potential or NK/dendritic cell potential) has led to the notion that the final steps in T-cell commitment occur in the thymus rather than the bone marrow.

In addition to losing the capacity to differentiate into non–T-lineage cells in vitro, T-lineage commitment is characterized by the upregulation of certain cell surface markers ( Fig. 23-5 ) and T-cell–specific genes important for the execution of downstream developmental programs. CD34+ cells in human thymus can be further subdivided by the expression of CD38 and the MHC-like molecule CD1a, with progression from the CD34+ CD38− CD1a− stage to the CD34+ CD38+ CD1a− stage to the CD34+ CD38+ CD1a+ stage. Acquisition of CD1a+ is largely associated with the loss of NK, dendritic cell, and plasmacytoid dendritic cell differentiation in vitro. Likewise, CD1a+ cells express the recombinase-activating gene products RAG1 and RAG2, and early TCR rearrangements are detectable after the acquisition of CD38 and CD1a. Transcription factors upregulated during these stages of early human T-cell commitment, whose requirement for T-cell development have also been demonstrated in gene-deficient mice, include GATA3 and Notch.

Interactions between thymocytes and epithelial cells are critical for events in later thymic development, such as positive and negative selection, as detailed later, as well as for early commitment and development. The so-called nude mouse , which lacks expression of the epithelial transcription factor Foxn1 , is born athymic, and humans lacking FOXN1 have severe T-cell deficiency, alopecia, and nail dystrophy. Unlike B cells, which can be induced in vitro to mature from precursors by co-culture on bone marrow stromal cells, T cells generally cannot be differentiated in vitro by this means, only when seeded into a three-dimensional source of primary thymic stroma, such as whole thymic organ culture. This distinction implies that specific soluble or cell-cell contact factors present only in the three-dimensional primary organ are required to engage developing thymocytes and induce their differentiation. The Notch signaling pathway, which is activated by the interaction of Notch ligands on epithelial cells and Notch family members on thymocytes, has emerged as one such factor. Expression of activated Notch1 in human HSCs is sufficient to induce T-cell development in vitro and in immunodeficient mice, and deletion of Notch1 from mouse thymocytes severely impairs early T-cell development. Interestingly, primary three-dimensional thymic stroma expresses Notch ligands such as Delta-like 1 and 4, and this expression is lost when the architecture is disrupted. Indeed, expression of Delta-like 1 or 4 in the OP9 bone marrow stromal cell line is capable of directing mouse and human uncommitted HSCs and even embryonic stem cells into the T lineage when co-cultured in a two-dimensional format in vitro. This powerful system underscores the importance of thymocyte-epithelial interactions during intrathymic development and raises the intriguing possibility of differentiating and expanding T cells in vitro for use in human therapeutics, which has been pursued in murine systems and recently in expansion of umbilical cord blood units to expand access to this source of therapeutic stem cells.

Stages of Intrathymic T-Cell Receptor αβ T-Cell Development and Spatial Considerations

T-lymphoid precursors that enter the thymus from the blood mature through a series of ordered developmental stages characterized by changes in cell surface markers, sequential expression of the TCR genes, and predictable migration from the cortex to the medulla (see Fig. 23-5 ). The earliest thymocytes reside in the cortex, are CD34+, express pan-T-cell markers such as CD2 and CD7, but lack the TCR-associated αβ heterodimer and CD3-associated subunits of the TCR complex and the helper and cytotoxic mature T-cell coreceptors CD4 and CD8. Hence, these thymocytes are CD3− CD4−CD8−, or triple negative. During αβ T-cell development, triple-negative thymocytes that succeed in rearranging a functional TCRβ chain express TCRβ complexed to the invariant pre-TCRα protein. Pre-TCR signaling in thymocytes promotes rapid expansion and differentiation into double-positive (DP) thymocytes expressing both the CD4 and CD8 coreceptors and initiates rearrangement of the TCRα gene. In humans, in contrast to mice, there is an additional intermediate stage in which thymocytes express only CD4 and not CD8 (intermediate single positive [ISP]). DP thymocytes account for approximately 80% to 90% of the total thymocyte number and can be found in the cortex and at the corticomedullary junction. Although the majority of DP thymocytes die by apoptosis, those that successfully undergo positive and escape negative selection survive, commit to either the CD4/helper or CD8/cytotoxic lineage, and migrate to the thymic medulla before terminal maturation and export to the periphery.

T-Cell Recognition of Peptide-Major Histocompatibility Complex Complexes

A hallmark of the cellular immune response is antigen specificity, which is conferred by the rearranged and selected TCR on the surface of T lymphocytes. The TCR recognizes processed fragments of foreign proteins embedded in MHC molecules. There are two forms of MHC molecules: MHC class I (the major determinants are HLA-A, HLA-B, and HLA-C in humans) and MHC class II (the major determinants are HLA-DR, HLA-DQ, and HLA-DP in humans) ( Fig. 23-6 ). MHC class I molecules are composed of a 42-kd transmembrane, polymorphic α chain encoded in the MHC and noncovalently associated with a 12-kd soluble, nonpolymorphic (non-MHC) β chain termed β ₂ -microglobulin. X-ray crystallographic analysis of HLA molecules in which peptide was embedded has demonstrated that the first two domains of the α chain both form α helices that together form a cleft ( Fig. 23-7 ). This cleft forms the binding domain for peptide antigen and accommodates a 9– to 11–amino acid fragment in an extended conformation; residues that interact with the MHC and those that interact with the TCR can be defined. Polymorphism within the MHC itself serves to ensure variation in the affinity of peptide binding; the extraordinary MHC polymorphism within the species further ensures that any single microbe is unlikely to mutate such that it is unable to bind all MHC molecules in the population and therefore escape T-cell recognition.

MHC class II molecules are formed by the noncovalent association of two transmembrane glycoproteins, αβ, both of which are polymorphic and encoded in the MHC. Solution of the structure of MHC class II molecules has shown that a peptide-binding cleft is formed by the first domains of each of the αβ chains; however, the ends of the cleft are open, unlike the situation in MHC class I molecules (see Fig. 23-7 ). The peptide-binding cleft of MHC class II accommodates processed peptides that are 10 to 30 (mean of 14) amino acids in length. As with MHC class I, the genetic polymorphism of MHC class II determines the affinity and specificity of peptide binding and T-cell recognition. Although MHC class I molecules are expressed constitutively on all human nucleated cells, MHC class II molecules are constitutively expressed only on B cells, monocyte-macrophages, and dendritic cells. Expression of MHC class II proteins may be induced on the surface of monocyte-macrophages, fibroblasts, endothelial cells, and certain mesenchymal and epidermal cells by a variety of inflammatory mediators and cytokines such as IFN-γ. In humans but not in mice, MHC class II proteins can be inducibly expressed on T lymphocytes as well, thus rendering T cells capable of antigen presentation to other T cells. These two classes of MHC proteins generally interact with different classes of T lymphocytes: CD4+ T cells recognize peptide antigen in association with MHC class II molecules, whereas CD8+ T cells recognize peptide antigen in association with MHC class I molecules. CD8 binds to invariant portions of the α3 domain of MHC class I and CD4 to a hydrophobic crevice between α2 and β2 domains (see Fig. 23-7 ).

A growing number of drug allergies have been reported and shown to be a consequence of their drug binding to MHC molecules. In general, peptides associate with HLA molecules by inserting parts of their amino acid residues into a set of six binding pockets in the HLA. The structure of these pockets is highly allele-specific, thereby dictating peptide-binding preferences for each HLA molecule. Remarkably, a recent study demonstrated that life-threatening drug-induced immune responses occur during abacavir treatment of HLA-B*57:01+ individuals as well as carbamazepine treatment of HLA-B*15:02+ patients through their binding to a specific pocket of the respective HLA molecules. Drug binding shifts the repertoire of self-peptides normally complexed with that HLA molecule for surface display. As a consequence, self-peptides never before arrayed on the surface of cells within the patient's body are expressed and perceived by the T cell immune system as foreign. These led to the CTL-mediated destruction of such peptide arrayed target cells. It is likely that other drugs, environmental chemicals and toxins, and bacterial products function in a similar way, including interactions with class II MHC molecules. Given the immunologic basis of many cases of severe aplastic anemia (SAA), it is conceivable that the etiology of some or most SAA cases might be linked to such a mechanism.

T-Cell Receptor Rearrangement and Selection in the Thymus

Effective adaptive immunity relies on the random generation of a diverse repertoire of antigen-specific T cells, along with appropriate quality control measures to ensure that nonfunctional T cells are not needlessly allowed to mature fully. To achieve this goal, the TCR genes undergo somatic rearrangement in a serial fashion, and the growth and proliferation of developing T cells are tightly linked to signaling and transcriptional events downstream of functional TCR or TCR intermediates.

The four TCR gene clusters, α, β, γ, δ, are each composed of germline genes encoding discontinuous variable regions (Vα, Vβ, Vγ, Vδ), diversity regions (Dβ, Dδ), joining regions (Jα, Jβ, Jγ, Jδ), and constant regions (Cα, Cβ, Cγ, Cδ). The TCRγ locus is situated on human chromosome 7, and the remaining three are located on human chromosome 14, with the TCRδ locus embedded between the V _α and J _α regions of the TCRα locus. The TCRβ germline genes consist of approximately 50 Vβ, 2 Dβ, 13 Jβ, and 2 Cβ genes, whereas the TCRα germline genes consists of at least 70 Vα, 60 Jα, and a single Cα gene. The TCRβ and TCRα loci have many more genes than the TCRδ or TCRγ loci do; however, the diversity of TCR γδ T cells is increased by the splicing of 2 D _δ regions to each other (VDDJ). Similar to the Ig genes in B cells, rearrangement of the TCR genes is carried out by the RAG1 and RAG2 proteins, with serial somatic rearrangements between the V, D, J, and C regions. Although N-region diversification is found in both the TCRα and TCRβ genes, T cells do not undergo somatic mutation as B cells do and instead undergo limited further rearrangements after emerging from the thymus.

Several major checkpoints regulate the development of TCR αβ and TCR γδ thymocytes. First, serial rearrangement of the TCR gene clusters in an ordered fashion ensures that the TCRβ and TCRα genes are not rearranged in developing TCR γδ T cells. Thus in human thymocytes TCRγ and TCRδ loci rearrange first, during the triple-negative stage. That human TCR γδ T cells generally have unrearranged TCRβ genes and that TCR αβ T cells generally have nonproductively rearranged TCRδ genes implies that only thymocytes that do not generate a functional γδ TCR go on to rearrange the TCRβ gene. The molecular mechanisms controlling this process are still being elucidated. The second or so-called beta selection checkpoint occurs after TCRβ rearrangement. Cells that undergo productive rearrangement of the TCRβ locus express functional TCRβ protein that complexes to the invariant pre-TCRα protein and CD3 signaling complex, thereby resulting in the expression of pre-TCR. In gene-deficient mice, where beta selection is known to occur during a precise phase of the triple-negative stage of development, loss of a number of signaling molecules or transcription factors downstream of pre-TCR results in defective beta selection, failure of differentiation into DP stage, and lack of proliferation or apoptosis (or both). Beta selection in humans begins during the CD34+ CD38+ CD1a+ stage and continues through the CD4 ISP and early DP stage of development (see Fig. 23-5 ). Thus only thymocytes that express functional TCRβ are allowed to proliferate and proceed to TCRα locus rearrangement.

Before TCRα gene rearrangement, the TCRδ locus is excised by a nonproductive rearrangement between the δRec and ψJα regions that generates an episomal circle of DNA or T-cell receptor excision circle (TREC), which is increasingly being used on both a research and a clinical basis to quantify thymic activity ( Fig. 23-8 ). The utility of TREC as a marker of new T cell generation has recently been used to detect severe combined immunodeficiency (SCID) in newborn dried blood spots, and universal screening for SCID with TREC was endorsed for addition to the Recommended Universal Screening Panel in the United States by Secretary of Health and Human Services Kathleen Sebelius in May 2010. Unlike the TCRβ locus, rearrangement of the TCRα locus occurs processively; that is, Vα-Jα recombination of 5' J segments that are nonproductive or out of frame is followed by further rearrangement on the same allele to more 3' J segments. This process increases the chances of generating a complete TCR αβ protein from any single developing DP thymocyte. Successful rearrangement of the TCRα gene results in expression of the mature TCR αβ heterodimer and noncovalent association with CD3εγ, εδ, and ζζ dimer. This TCR complex appears during the late DP stage. These cells then undergo positive and negative selection (see the next section) and simultaneously commit to either the CD4 or the CD8 lineage to generate CD4 single-positive (SP) and CD8 SP thymocytes (see Fig. 23-5 ).

Positive Selection, Negative Selection, and CD4/CD8 Lineage Commitment

The massive proliferation of DP thymocytes results in TCR αβ thymocytes with an immense range of specificities, including many that are incapable of binding host MHC, cannot recognize host MHC complexed to endogenous peptide ligands, or have inappropriately high affinity for MHC–self-peptide and hence are autoreactive. Selecting only thymocytes with intermediate MHC–self-peptide affinity (so-called Goldilocks or just-right conditions) is critical for maintaining diversity of the repertoire while promoting self-tolerance.

Positive selection is the term used to describe the process of selecting thymocytes capable of recognizing host MHC complexed to self-peptides generated from the processing of endogenous proteins. DP thymocytes are exquisitely prone to cell death in response to γ-irradiation, corticosteroids, exposure to anti-CD3 antibodies, or simply removal of the thymus from the body, and thus a positive signal via TCR-MHC/peptide interactions is required to rescue the small number of DP thymocytes destined to become SP thymocytes from cell death. Gene-deficient mouse experiments have demonstrated that multiple elements of TCR signaling are required for optimal positive selection, including TCR components (TCRα, CD3δ), early signaling molecules (GADs, ZAP70), late signaling proteins (RasGRP, calcineurin, ERK), and transcription factors (nuclear factor of activated T cells [NF-AT], Egr family members). Interestingly, humans deficient in CD3δ and ZAP70 have been described who have discrepant development when compared with the knockout mouse counterparts. Deficiency of CD3δ in humans causes an earlier block in development at the CD4− CD8− stage, well before the development of DP thymocytes. ZAP70 deficiency results in selective deficiency of CD8 SP thymocytes and mature CD8+ T cells. Likewise, although in the mouse interaction of DP thymocytes with MHC on radioresistant cortical thymic epithelium is probably the dominant requirement for positive selection, the source of MHC and the relative contributions of MHC from thymic epithelium, thymic dendritic cells, or other thymic APCs to positive selection in humans are less clear.

Negative selection is the term used to describe the process of removing thymocytes that are self-reactive, also termed central tolerance, by clonal deletion. This is the first and major mechanism for control of autoimmunity, although clones that escape deletion in the thymus are subject to peripheral tolerizing influences (discussed later). In mice negative selection can be mediated by cortical epithelium, medullary epithelium, and medullary dendritic cells. However, it was never clear how thymocytes reactive against self-antigens expressed only in peripheral tissues, such as islet cells, were selected against in the thymus. In other words, how are these tissue-specific proteins expressed in the thymus? More recently, murine and human studies have revealed that all three cell types, medullary thymic epithelial cells (mTECs) in particular, exhibit promiscuous gene expression and transcribe a broad range of tissue-specific, nonthymic antigens. This insight helps explain how developing thymocytes reactive against nonthymic proteins can be centrally deleted during negative selection. A newly discovered gene, AIRE (autoimmune regulator), is a factor shown to control the transcription of tissue-specific antigen expression in murine mTECs. Simul­taneously, this gene was shown to be responsible for the autoimmune polyendocrinopathy–candidiasis–ectodermal dystrophy (APECED) syndrome, characterized by chronic mucocutaneous candidiasis, hypoparathyroidism, adrenal insufficiency, and other organ-specific autoimmune manifestations. Thus this disease may be the first example of failure of negative selection underpinning a human monogenic autoimmune disorder.

Maturation of CD4 and CD8 SP thymocytes from DP thymocytes occurs concurrently with positive selection and results in “matching” of coreceptor to MHC restriction with downregulation of the opposite coreceptor. The mechanisms controlling the CD4/CD8 lineage decision are separable from those governing positive selection and have been the subject of active investigation in murine models. Patients with defects in a number of genes that control MHC class II expression, including RFXANK, RFX5, RFXAP, and CIITA, all manifest the so-called bare lymphocyte syndrome , a rare autosomal recessive disorder characterized by decreased numbers of CD4+ T cells, deficient helper function, and absence of specific antibody production because of lack of CD4-mediated B-cell maturation.