Immunology of the Fetus and Newborn

Complement

Central to the innate immune response is the complement system that consists of >40 plasma, cell surface, and regulatory proteins that interact to regulate multiple physiologic functions, including resistance to pyogenic infections, interaction between innate and adaptive immunity, and elimination of immune complexes, products of inflammatory injury, and apoptotic self cells. Components of the complement system recognize and lyse bacteria, opsonize microorganisms, release anaphylatoxins, solubilize immune complexes, and induce B-cell proliferation and differentiation.

Activation of the complement cascade occurs via three pathways—classical, lectin, or alternative. Several characteristics of the complement cascade are important for fetal–neonatal immunity. First, the complement system features both antibody-dependent specificity via the classical pathway activation triggered by interaction of antigens with antibodies, and antibody-independent activation of the alternative and lectin pathways initiated by pathogen-associated structures such as endotoxin and polysaccharides. Thus for the fetus or infant who has not received from the mother or has not yet produced antigen-specific immunoglobulin (Ig) G for immunologic recognition, the alternative and lectin pathways may be critical for triggering the effector functions of the complement cascade. Second, the enzymatic activation of the complement cascade enables rapid functional amplification: deposition of a single Ig molecule or C3b fragment can generate enzymatic cleavage of thousands of later-acting components and thus multiple complement activities. In addition, the alternative pathway can be amplified via a positive feedback activation mechanism, because C3b, an activation product of the alternative pathway C3 convertase, is a component of this convertase. As the fetus and newborn are particularly dependent on antibody-independent pathogen recognition for immunologic responsiveness, the positive amplification loop of the alternative pathway may be particularly critical for rapid generation of complement effector functions in early life in the absence of antibody-based recognition. Third, the continuous activation of the alternative pathway requires rigorous regulation in the fetus to avoid tissue damage during organ remodeling. Finally, the contributions of the lectin pathway to fetal–neonatal complement activation and fetal well-being are still under investigation.

Studies of fetal and neonatal complement have focused on quantification of serum concentrations of individual components, examining maternal–fetal transport of these proteins, assessing specific effector functions of the classical and alternative pathways, and investigating contributions of complement activation to common neonatal diseases. Detectable concentrations of C3 (1% of adult levels) and C1 inhibitor (20% of adult levels) can be measured as early as 5 to 6 weeks’ gestation. By 26 to 28 weeks’ gestation, both C3 and C1 inhibitor concentrations increased to 66% of adult levels. Functionally and immunochemically measured classical and alternative pathway protein concentrations in UCB increase with advancing gestational age, such that impairment in CH50 is particularly evident in the preterm, and at full-term gestation the concentrations are only approximately 50% to 75% of adult concentrations. Although neonatal UCB lectin pathway component concentrations are lower than those in older children and adults, the correlation between mannose-binding lectin (MBL) and gestational age has not been consistently observed. Of note, on the basis of studies of genetically determined, structurally distinct complement variants in maternal and umbilical cord serum, no transplacental passage from the mother to the fetus of C3, C4, factor B, or C6 has been observed.

Much remains to be learned regarding regulation of complement effector functions in the fetus and newborn. Activation of the alternative pathway or the lectin pathway enables opsonization of invading microorganisms without specific Ig recognition. Accordingly, for preterm infants or those without organism-specific maternal IgG, alternative or lectin pathway activation provides a critical mechanism for engaging complement effector functions. The functional contribution of the classical pathway to effector functions has been assessed through the use of blood-mediated opsonophagocytosis by polymorphonuclear leukocytes of group B streptococci (GBS) type Ia. This GBS serotype may be opsonized by classical pathway components in the absence of specific antibodies and thus enables characterization of classical pathway function. In 8 of 20 neonatal serum samples examined, decreased bactericidal activity was detected and correlated with significantly lower functional activity of C1q and C4. These studies did not determine whether this decrease was mediated by an inhibitor of function or by an intrinsic change in functional activity of these components in neonatal sera. Studies of MBL concentrations and pathway activity suggest a contribution of the lectin pathway to neonatal susceptibility to infection. Complement regulatory proteins (e.g., C4b-binding protein and factor H) also contribute to neonatal susceptibility as suggested by the failure of neonatal serum to reduce invasion by GBS and Escherichia coli into human brain microvascular endothelial cells. In vitro experiments in which killing of E. coli by neonatal serum samples was limited by C9, but not by other classical pathway components, suggest that this terminal complement component is apparently important for cytolysis of this pathogen. Although relatively lower concentrations of complement components likely contribute to poor control of bacterial replication, these complement concentrations are nevertheless sufficient, via C3- and factor B-dependent activity in the alternative pathway, to enhance GBS-induced production of TNF-α by monocytes in human newborn UCB tested in vitro.

In addition to relatively low serum concentrations of classical, alternative, and lectin pathway complement proteins, additional complement functions important for fetal and neonatal well-being can contribute to reduced capacity to activate the classical and alternative pathways. For example, fetal and neonatal serum demonstrates reduced concentration of C4b-binding protein, a critical regulator of classical pathway C3 convertase activity. Lower C4b-binding protein concentration increases the functional anticoagulant activity of protein S, with which it complexes and thereby contributes to decreased coagulation function of the fetus and newborn. Consideration of complement components that also express nonimmunologic functions will likely be important in characterizing developmental regulation of complement component production.

Complement activation contributes to tissue injury in several common neonatal diseases, including neonatal hypoxic–ischemic encephalopathy, NEC, meconium aspiration syndrome, and intrauterine growth restriction and fetal loss. Unregulated complement activation may occur in selected infants undergoing extracorporeal membrane oxygenation therapy, raising concern for inflammatory injury on that basis. C5a is present in the cerebrospinal fluid of human newborns, at especially high concentrations in those born preterm, raising the possibility that complement activation in the neonatal brain may contribute to preterm brain injury.

Overall, study of the complement system in early life, including characterization of the developmental and genetic regulation of this important group of plasma and cell surface proteins, promises to shed further light on immune ontogeny in relation to health and disease.

Antimicrobial Proteins and Peptides

The human body expresses natural antibiotics, including APPs that act alone and in combination with endogenous (e.g., complement) and exogenous (e.g., conventional antibiotic) systems to prevent infection and/or eliminate invading microorganisms. APPs are expressed by a range of cells, including epithelial cells and leukocytes, especially neutrophils (see Fig. 32.4 ), and are found associated both with cells and in plasma. Plasma levels of APPs vary with gestational age such that preterm plasma is relatively deficient in multiple APPs, likely contributing to reduced microbicidal capacity (see Fig. 32.3 ). Examples of APPs include (1) lactoferrin, an 80-kDa protein with iron-binding and direct membrane perturbing properties found in tear fluid, saliva, and neutrophil secondary granules, (2) the 5-kDa bactericidal/permeability-increasing protein (BPI), expressed on certain mucosal epithelia as well as neutrophil primary granules, with high affinity for LPS that enables it to neutralize the inflammatory activity of endotoxin and targets its microbicidal activity toward Gram-negative bacteria, (3) 14-kDa phospholipase A ₂ , an acute-phase reactant expressed in liver with ability to enzymatically kill a range of gram-positive pathogens, and (4) 4-kDa disulfide-rich defensin peptides of neutrophil primary (azurophilic) granules with broad microbicidal activity. Ongoing efforts are aimed at developing congeners of APPs as novel antiinfective agents for individuals who are relatively APP deficient, including preterm infants and those undergoing chemoradiotherapy. For example, oral administration of lactoferrin to human preterm newborns has shown promise in reducing the incidence of sepsis and NEC.

Innate Lymphoid Cells, Including Natural Killer Cells

Innate lymphoid cells (ILCs) are derived from a common lymphoid progenitor, are defined by the absence of antigen-specific B-cell receptors (BCRs) or TCRs, and do not express myeloid or DC markers. ILCs are divided into subgroups based in part on the cytokine profile they produce: (1) group 1 ILCs produce IFN-γ and are functionally dependent on the transcription factor T-bet; (2) group 2 ILCs produce type 2 cytokines (e.g., IL-4, IL-5, IL-9, and IL-13) in response to helminth infection and are dependent on RORα and GATA3; and (3) group 3 ILCs produce IL-17A and/or IL-22 and are dependent on the transcription factor RORγt.

NK cells are the most studied of the ILCs. These group 1 ILCs constitute approximately 10% to 15% of all peripheral blood lymphocytes. They are present in the spleen, lungs, and liver and are also rarely found in lymph nodes and thoracic duct lymph. NK cells represent up to 70% of all lymphocytes in the maternal decidual tissue. They demonstrate distinct morphology, function, and surface molecule expression, including expression of CD16 (Fc gamma receptor FγRγIII) and CD56 (nerve cell adhesion molecule 1). Mature NK cells appear larger and more granular than T or B cells and express both activating and inhibitory receptors that are used to selectively identify and kill virally infected cells and tumors. The presence of MHC class I molecules on potential target cells induces signals that suppress NK-cell function. MHC class I–deficient target cells activate NK-cell function, triggering release of lysosomal granules containing serine proteases, perforin, and transforming growth factor beta (TGF-β), thereby disrupting the target cell membrane and inducing an inflammatory response. Fetuses and neonates demonstrate reduced NK-cell activity compared with adults.

NK cells are derived from a common hematopoietic progenitor that retains T-cell and B-cell developmental potential. NK cells first make their appearance in fetal liver as early as 6 weeks’ gestation. Committed CD34 ⁺ CD56 ⁻ NK cell progenitors have been identified in the fetal thymus, bone marrow, and liver. In the human neonate, the NK-cell population is immature: only half of all NK cells express CD56, and the NK-cell cytolytic activity is lower. This functional reduction in NK-cell activity may contribute to the severity of neonatal herpes simplex virus (HSV) infections. Profound defects in NK-cell activity result in familial hemophagocytic lymphohistiocytosis (HLH), a disease characterized by fever, hepatosplenomegaly, cytopenia, hyperferritinemia, and hemophagocytosis. Familial HLH arises from mutations in genes that encode proteins involved in the granule-exocytosis pathway and can be fatal without bone marrow transplant.

NK-cell receptors are fundamentally different from TCRs and BCRs. NK-cell receptor gene expression does not require gene segment rearrangement, and the receptors are not clonally distributed. Instead, NK cells use an array of stimulatory and inhibitory receptors to regulate their cytolytic functions. A cluster of 10 or more genes encoding killer-cell Ig-like receptors (KIRs) is located on human chromosome band 19q13.4. Each of these type I glycoproteins recognizes a different allelic group of HLA-A–, HLA-B–, HLA-C–, or HLA-G–encoded proteins, and each KIR is expressed by only a subset of NK cells. Another family of Ig-like receptor genes termed ILT is present near the KIR locus at 19q13.3. These receptors are not as restricted as the KIRs and bind multiple HLA class I molecules. A third inhibitory receptor gene locus is on chromosome band 12p12-p13, encoding a C-type lectin inhibitory heterodimeric receptor called CD94/NKG2 that binds HLA-E. Those KIRs, ILT receptors, and CD94/NGK2 molecules with long cytoplasmic tails and two immunoreceptor tyrosine-based inhibitory motifs (ITIMs) function as inhibitory receptors. On phosphorylation, the two ITIMs recruit and activate the Src homology domain 2 (SH2)-containing phosphatases, which turn off the kinase-driven activation cascade. The KIR family member KIR2DL4 is distinct from other KIRs in structure and distribution. KIR2DL4 binds HLA-G and has a single ITIM in the cytoplasmic tail and a lysine in the transmembrane region, enabling association with adaptor proteins. This inhibitory receptor was found on all decidual NK cells in the placenta at term but not on circulating maternal NK cells, suggesting that expression of KIR2DL4 is induced during pregnancy.

Other KIRs or members of the C-type lectin receptor superfamily are activating receptors. These receptors lack the long cytoplasmic tail of the inhibitory receptors and therefore do not contain ITIMs. Instead, they have a charged amino acid in the transmembrane region that enables receptor association with the adaptor molecule DAP12. This adaptor contains an immunoreceptor tyrosine-based activation motif (ITAM) that allows these receptors to activate NK cells. The physiologic role of these HLA class I–specific activating receptors remains unknown. NK cell–activating receptors also include natural cytotoxicity receptors (NKp46, NKp30, NKp44), proteins that are Ig superfamily members with little similarity to one another or to other NK-cell receptors. These receptors are highly specific for NK cells and apparently interact with non-HLA molecules.

CD244 (2B4) is a member of the signaling lymphocyte activation molecule (SLAM) family of receptors expressed on all human NK cells. On interaction with the ligand CD48 on target cells, NK-cell signaling proceeds via interactions between the immunoreceptor tyrosine-based switch motif (ITSM) (switch motif) in the cytoplasmic tail of CD244 and one of two SH2 domain-containing adaptor proteins, SLAM-associated protein (SAP) and Ewing sarcoma–associated transcript 2 (EAT-2). SAP interactions trigger activation, as evidenced in humans with X-linked lymphoproliferative disease, caused by loss-of-function mutations in the SAP linker. In the absence of SAP, interactions with EAT-2 may be inhibitory.

Recent studies have explored the potential contribution of ILCs beyond NK cells. A role for lung group 2 ILCs in mediating respiratory syncytial virus (RSV)-induced IL-33–driven T _h 2-biased immunopathology has been demonstrated in neonatal mice. IL-23–responsive group 3 ILCs played a role in the pathogenesis of neonatal intestinal inflammation in a murine model. Much remains to be learned regarding the ontogeny of ILC function with respect to both the quantity and the quality of these cells in the very young.

Polymorphonuclear Neutrophils

Neonatal polymorphonuclear neutrophils (PMNs) are present at early stages of gestation, but their functional capacities are different from those of adult PMNs. Progenitor cells that are committed to maturation along granulocyte or macrophage cell lineages (granulocyte–macrophage colony-forming units) are detectable in the human fetal liver between 6 and 12 weeks’ gestation in similar proportions as in adult bone marrow. Human fetal blood has detectable granulocyte–macrophage colony-forming units from 12 weeks’ gestation to term. Although these progenitor cells are detectable in the fetus and newborn, developmental differences between adult and mature neonatal PMNs have been demonstrated—in signal transduction, cell surface protein expression, cytoskeletal rigidity, rolling adhesion, microfilament contraction, transmigration oxygen metabolism, intracellular antioxidant mechanisms, and neutrophil extracellular trap formation. The magnitude of PMN functional differences correlates with the maturity of the infant and begins to decrease within the first few weeks after birth.

In addition to intrinsic age-dependent differences in PMN function, age-dependent cell extrinsic soluble factors may developmentally regulate induction of specific functions and maturation of these cells. For example, low concentrations of the chemoattractant complement component C5a in neonatal sera might impair establishment of chemoattractant gradients at sites of inflammation. In addition, elevated concentrations in human neonatal blood plasma of adenosine, an endogenous purine metabolite that acts via seven-transmembrane adenosine receptors to inhibit inflammatory leukocyte responses, could contribute to inhibition of newborn neutrophil function.

Systemic bacterial infection in newborns is frequently accompanied by profound neutropenia, prompting investigation of neutrophil kinetics in infected infants. These studies have suggested diverse, developmentally specific regulatory mechanisms required for mobilization of the neutrophil response to infection. The absence of detectable neutrophil precursors in bone marrow aspirates of infected infants and systemic neutropenia motivated studies of neutrophil replacement therapy in neutropenic, infected infants. Although this approach has been successful in some cases, the results have not been uniformly beneficial, and a Cochrane review suggests a need for adequately powered multicenter trials of granulocyte transfusions in neutropenic septic neonates. Metaanalysis suggesting that granulocyte colony-stimulating factor or granulocyte–macrophage colony-stimulating factor may reduce mortality in newborns when systemic infection is accompanied by severe neutropenia requires confirmation in adequately powered trials. Heterogeneity in the impact of neutrophil modulation may reflect the importance of individualizing immunologic interventions for the genetic background, developmental stage, and pathogenic microorganism being treated.

Monocytes, Macrophages, and Dendritic Cells

Cells committed to phagocyte maturation, including granulocyte or monocyte–macrophages, are detectable in the human fetal liver by 6 weeks’ gestation and in peripheral fetal blood by 15 weeks’ gestation. Unlike granulocytes, whose tissue half-life is hours to days, macrophages migrate into tissues and reside for weeks to months. In a tissue-specific fashion, these cells regulate availability of multiple factors, including proteases, antiproteases, prostaglandins, growth factors, reactive oxygen intermediates, and a range of cytokines and chemokines. Importantly, monocytes can migrate from the bloodstream to tissue sites, becoming tissue-based DCs. Monocytes, macrophages, and DCs share the ability to present antigens to T lymphocytes, thereby triggering the classic adaptive immune responses.

Compared with their adult counterparts, newborn monocytes, macrophages, and DCs demonstrate reduced chemotaxis and phagocytosis as well as distinct TLR signaling that is polarized toward T _h 2 and antiinflammatory cytokine production. The distinct function of newborn APCs reflects both intrinsic characteristics, including reduced nucleosome remodeling for IL-12 p70 production as well as the modulatory effects of age-specific extrinsic factors such as the antiinflammatory purine metabolite adenosine, the level of which is relatively elevated in human newborn UCB plasma.

Stimulation of monocytes results in a change in innate “setpoint” such that responses to subsequent stimuli are altered. This phenomenon, reflecting adaptive features of the innate immune system that are mediated by epigenetic changes, has been termed trained immunity and may contribute to the heterologous beneficial (“nonspecific”) effects of live attenuated vaccines (Goodridge et al., 2016). Recent studies indicate that trained immunity varies by age, with human neonatal monocytes demonstrating distinct immunometabolic BCG-induced training resulting in tolerogenic cytokine responses to subsequent LPS stimulation in vitro. Much remains to be learned regarding the scope, ontogeny, and mechanisms underlying innate training/innate memory.

T Lymphocytes

Most T cells develop in the thymus, which includes cell types of nonhematopoietic origin, such as epithelial cells, as well as multiple cell types of hematopoietic origin, including the developing immature T cells or thymocytes, DCs, mononuclear phagocytes, and small numbers of B cells. The thymic epithelial cells are derived from the third branchial cleft and the third or fourth branchial pouch, a process that is perturbed in DiGeorge syndrome, resulting in thymic epithelial hypoplasia. Thymic lobes can be divided into four regions, which, proceeding from outward to inward, are the subcapsular region, cortex, corticomedullary junction, and medulla. Prothymocytes, which are bone marrow–derived CD34 ⁺ CD38 ⁺ CD62L ⁺ lymphoid cells, have the capacity to commit to the T-cell or other lymphocyte lineages depending on their receipt of instructive signals. Circulating prothymocytes enter the thymus via vessels at the cortical–medullary junction. The prothymocyte becomes committed to the T-cell lineage by the engagement of its surface notch 1 receptor by ligands displayed on the thymic epithelium, such as delta-like ligand 4. This engagement results in an early T-cell progenitor that acquires expression of CD1a, CD2, CD7, and progressively loses its initial capacity for B-cell, myeloid, or NK-cell differentiation and becomes a fully T-lineage committed pro-T cell. The pro-T cell migrates to the subcapsular region just below the outer capsule.

The subcapsular pro-T cell expresses all of the internal proteins required for V(D)J recombination, including the recombinase activating gene (RAG) 1 and RAG2 endonucleases that make double-stranded breaks in DNA; the proteins involved in nonhomologous end joining repair (e.g., Artemis, XLF [Cernunnos], and DNA ligase IV); and those that are essential for generating junctional diversity at complementarity determining region (CDR) 3 (e.g., terminal deoxynucleotidyl transferase [TdT]). CDR3, which is the most variable in amino acid sequence, is located at the center of the antigen-specific binding site of both TCRs and BCRs. The pro-T cell lacks most cell surface proteins characteristic of mature peripheral T cells, including CD3, CD4, and CD8, and is therefore also referred to as a triple-negative thymocyte ( Fig. 32.6 ). The pro-T cell is the first stage in which there is VDJ rearrangement of TCR gene loci, with the TCRγ gene rearrangement occurring most frequently. If this rearrangement is productive (i.e., capable of expressing a full-length TCRγ chain protein) and the thymocyte subsequently undergoes a productive TCRδ gene arrangement, a γδ TCR heterodimer is expressed on the cell surface, allowing the thymocyte to differentiate into a mature γδ T cell that emigrates from the thymus into the periphery. More frequently (>95% of the time) these TCRγ and/or TCRδ gene rearrangements are nonproductive, and the pro-T cell attempts TCRβ chain gene rearrangement. If this arrangement is productive, the TCRβ chain is expressed on the cell surface in association with an invariant pre-T alpha chain forming the pre-TCR complex, which defines the pre-T-cell stage of development. Like the mature TCR, the pre-TCR is associated with the CD3 complex of proteins, which includes CD3γ, CD3δ, CD3ε, and CD3ζ (also known as CD247) chains, all of which have cytoplasmic tails containing specific amino acid sequences called ITAMs. These ITAMs serve as molecular targets for tyrosine phosphorylation and binding by tyrosine kinases, such as Lck and zeta chain-associated protein of 70 kDa (ZAP-70), which generate intracellular signals leading to the induction of target genes. These signals direct the pre-T cell to (1) proliferate, (2) upregulate expression of CD4 and CD8 and become a double-positive (CD4 ⁺ CD8 ⁺ ) thymocyte, (3) migrate from the subcapsular area to the thymic cortex, and (4) start rearrangement of the TCRα chain gene locus (see Fig. 32.6 ).

Rearrangement of the TCRα chain gene by CD4 ⁺ CD8 ⁺ thymocytes is a two-step process in which there first is an internal deletion of a ψδ rec segment that brings the unrearranged Vα segments in close proximity with J _α segments and the C _α constant region. The intervening DNA, which is excised as a circular product with fused signal joint sequences, referred to as a signal joint TCR excision circle (sjTREC), is highly stable within the cell ( Fig. 32.7 ). The sjTREC content of the T-lineage cell subsequently decreases mainly as a result of cell proliferation, which under normal conditions is minimal until the mature T cell undergoes antigen activation-induced clonal proliferation. Thus, the measurement of the sjTREC content by peripheral blood T cells is an indirect but useful assessment of the adequacy of production of new T cells by the thymus. The measurement of sjTRECs in neonatal blood spots is routinely used for newborn screening in all 50 United States for identifying infants with impaired production of T cells by the thymus as occurs in most forms of severe combined immunodeficiency (SCID) (see Specific Immunologic Deficiencies of the Newborn and Their Diagnosis).

Double-positive thymocytes that have productive TCRα chain gene rearrangements replace their pre-TCR with a TCR consisting of an αβ TCR heterodimer in association with the CD3 proteins (see Fig. 32.6 ). The next major checkpoint of thymocyte development is positive selection, in which the αβ TCR of the CD4 ⁺ CD8 ⁺ thymocyte is tested for whether it has significant affinity for either the MHC class I alleles (HLA-A, HLA-B, and HLA-C in humans) or the MHC class II alleles (HLA-DR, HLA-DP, and HLA-DQ in humans) expressed by thymic cortical epithelial (TCE) cells. All MHC class I and class II molecules during their biosynthesis have peptides loaded into their peptide-binding grooves. In the absence of infection or vaccination with foreign proteins, these peptides are derived from self-proteins, as is the case for the MHC of TCE cells. A specialized set of MHC class I-binding peptides may be generated by a special type of proteasome expressed by TCE cells (the thymoproteasome) and that these play an important role in increasing the antigen responsiveness of mature peripheral T cells. In positive selection, thymocytes with TCRs that are unable to bind to MHC/peptide complexes on TCE cells with sufficient affinity to generate intracellular signals die by apoptosis, the default pathway. In cases where the TCR binding to MHC generates a relatively weak to moderate signal, the thymocyte is positively selected for survival. A large range of “analog” signals—from relatively weak to medium strength—are converted by the intracellular signaling protein Themis into a single “digital” outcome of thymocyte survival and maturation. In cases where the TCRs of CD4 ⁺ CD8 ⁼ thymocytes receive very high levels of signaling, double-positive cortical thymocytes undergo apoptosis. CD4 ⁺ CD8 ⁺ thymocytes with TCRs that receive MHC class II/peptide survival signals lose CD8 expression and upregulate CD3 expression, thereby becoming CD4 ⁺ CD8 ^– thymocytes that are CD3 ^high . CD4 ⁺ CD8 ^– thymocytes also begin to acquire a gene expression pattern that is characteristic of mature peripheral CD4 T cells that is required for their capacity to carry antigen-induced effector functions, such as IL-2 secretion and CD40-ligand (CD40L) expression. CD4 ⁺ CD8 ⁺ thymocytes with TCRs that receive MHC class I/peptide survival signals lose CD4 expression and upregulate CD3 expression, thereby becoming CD4 ^– CD8 ^– thymocytes; they also begin to express genes that are characteristic of peripheral CD8 T cells and that are required for their antigen-induced capacity to become cytotoxic cells. These CD4 ⁺ CD8 ^– thymocyte-specific versus CD4 ^– CD8 ⁺ thymocyte-specific outcomes of positive selection are directed by the master transcription factors ThPOK (encoded by the Zbtb7b gene) and Runx3, respectively.

A small subset of thymocytes bearing TCRα chains containing the V _α 24J _α 18 segments in association with TCRβ chains containing V _β 11 segments interact with relatively high affinity with CD1d, an MHC class I-like protein that is expressed on double-positive thymocytes. This TCR interaction leads to the thymocyte-positive selection for NKT-lineage cells, which are distinct in function from conventional T cells in having the ability to secrete rapidly large amounts of cytokines, such as IFN-γ and IL-4, during the early phase of innate immune responses to pathogens.

Another subset of thymocytes bearing TCRα chains with V _α 7.2-J _α 33 segments paired with TCRβ chains using either V _β 2 or V _β 13 interact with high affinity with an MHC class I-like protein, MR1, on non-hematopoietic cells, resulting in the positive selection of mucosal-associated invariant T (MAIT) cells. MAIT cells, which are predominantly CD8 single positive, are found in greatest amounts in mucosal tissues, such as the intestine, lung, and liver.

Positively selected single-positive CD4 ⁺ CD8 ^– or CD4 ^– CD8 ⁺ thymocytes move into the medulla, where they undergo a final selection process before emigrating from the thymus as CD4 and CD8 T cells called negative selection . This selection process, which is an important mechanism for maintaining tolerance of T cells to peptides derived from self-proteins, involves the exposure of the mature thymocytes to medullary APCs expressing a highly diverse repertoire of peptides derived from self-proteins. These self-proteins include those that are normally expressed in a tissue-restricted manner (e.g., peptides derived from insulin, which is produced by pancreatic beta islet cells or parathyroid hormone) or that are characteristic of only certain stages of early development. This unusual pattern of protein expression by medullary DCs and B cells is the result, at in part, of a nuclear protein encoded by the autoimmune regulator (AIRE) gene. AIRE stochastically relieves the transcriptional repression of these tissue-specific and developmentally regulated genes and the proteins they encode are transferred to thymic CD11c ⁺ dendritic cells, which play a major role in presenting these self-peptide/MHC complexes to medullary thymocytes (Breed et al., 2018). Thymocytes that have high levels of signaling for peptides derived from self-proteins are induced to undergo apoptosis, whereas intermediate levels of signaling by CD4 ⁺ CD8 ^– thymocytes result in their differentiation into Treg cells. Thus, thymically derived Treg cells have TCRs that have substantial affinity for self-proteins expressed in a tissue-specific or developmental-specific manner.