Wintrobe’s Clinical Hematology, 12th Edition

Chapter 16

T Lymphocytes and Natural Killer Cells

Frixos Paraskevas

Morphology

The T lymphocyte, under routine staining procedures (Giemsa or Wright), is 5 to 8 μm in diameter, with a high nucleocytoplasmic ratio. The nucleus is purple with densely packed chromatin, and the cytoplasm forms a narrow light-blue rim.

By transmission electron microscopy, the nucleus shows shallow indentations with dense heterochromatin along the nuclear membrane and euchromatin occupying most of the remaining nuclear surface. One or two nucleoli are visible (Fig. 16.1). The cytoplasm shows a few organelles, such as mitochondria and a small Golgi apparatus. By scanning electron microscopy, T lymphocytes show either smooth surface or only small numbers of microvilli, depending on the method of preparation as well as the state of activation (1,2).

Some T lymphocytes present in normal subjects are characterized by a highly indented nucleus and are known as cerebriform mononuclear cells(3). These cells are not detected in T-lymphocyte–depleted fractions and constitute approximately 3 to 4% of the unfractionated T lymphocytes. They possess scanty cytoplasm, and the degree of their nuclear indentation is expressed as a nuclear contour index (nuclear perimeter/area) (4). Their structural similarity to the cells present in cutaneous T-cell lymphomas suggests that they represent the normal equivalent of Sézary cells, which are derived from T lymphocytes.

Ontogeny of the Thymic Microenvironment

The ontogeny of thymus and its structure are discussed in Chapter 14.

The thymic anlage develops from epithelial structures of the third branchial complex (Fig. 16.2). Neural crest cells (Hoxa3-positive) invade the epithelial cluster to form the thymic rudiment. Several studies support the view that the cortical and medullary epithelia originate from a common precursor, and, in the early stages, the epithelium coexpresses markers that are later segregated to the cortical or medullary compartments. After completion of thymic organogenesis, the cortical epithelia are cytokeratin 8⁺, whereas medullary epithelia are cytokeratin 5⁺ (5), except for a small subpopulation in the corticomedullary junction, which is cytokeratin 5⁺/8⁺ (5). Furthermore, neoplastic human thymomas often express both cortical and medullary epithelial markers (6).

The best evidence of the origin of thymic epithelia from a common progenitor is the identification by monoclonal antibodies of a cell that, on differentiation in vivo, generates the full thymic microenvironment (7,8). The monoclonal antibody, MTS24, detects a glycoprotein with mucinlike characteristics and a peptide backbone (7). The antigen is detected during the early embryonic stages in the anterior endodermal epithelium, the pharyngeal endoderm, and a portion of intermediate mesoderm, which develops the urogenital epithelium. The MTS24⁺ cells are major histocompatibility complex (MHC) class II positive and express cytokeratin 5 and cytokeratin 8, which are markers of the medullary and cortical epithelium, respectively. Highly purified MTS24 cells, inserted under the kidney capsule, are able to develop into a complete thymus and, furthermore, provide the complete milieu necessary for T-cell development.

Because primordial epithelial cells normally need the cooperation of mesenchymal cells for thymus organogenesis (9), it is conceivable that, in the case of the ectopically placed MTS24 cells, this is provided by comparable cellular elements from the kidney capsule. The MTS24 antibody completely blocks T-cell development in fetal thymic organ cultures, supporting the conclusion that the molecule recognized by the MTS24 antibody regulates normal epithelial function.

During this early stage in the development of the thymus, the cell interactions are regulated by a number of transcription factors, such as Hoxa3 and Pax1, which initiate the formation of the thymic primordium in mice (10,11). An important step in our understanding of the genetic control of thymic development came from the study of the “nude” mouse and the cloning of the gene that confers the nude phenotype, designated whnfor winged helix nude (12), and more recently renamed Foxn1. The Foxn1 gene encodes a transcription factor with a DNA-binding domain of the forkhead/winged helix class. The defect responsible for the nude phenotype is a single base pair deletion in the third exon of the Foxn1 gene. It results in hair loss, the arrest of the thymic epithelial cell expansion, and inability to attract the hematopoietic precursors into the epithelial rudiment (13,14). Formation of the epithelial primordium is not affected by the loss of whn function, but subsequent differentiation of the primitive epithelial precursors into subcapsular, cortical, and medullary epithelial cells is arrested (15), and the epithelial rudiment becomes cystic.

In normal development, the epithelial rudiment is invaded by mesenchymal cells from the neural crest (16), which stimulate epithelial progenitor differentiation into cortical and medullary subpopulations (17,18,19,20). These mesenchymal–epithelial interactions are mediated by growth factors and their receptors, such as the fibroblast growth factor 10 (FGF10) and its specific receptor, FGFR2IIIb, on thymic epithelial cells. Deficiency of either the factor or the receptor results in severe thymic hypoplasia (21,22). When immature lymphoid cells begin to arrive in the thymic rudiment, another wave of cellular interactions takes place between the developing lymphoid cells and the stromal epithelial cells. This second stage of thymic development establishes thymic microenvironments conducive to thymocyte differentiation and repertoire selection.

Further differentiation and maturation are under the regulation of interactions with thymocytes. Experiments in animals and experiments of nature in humans (disease processes) have clearly shown the symbiotic relationship between epithelial cells and lymphocytes. Prothymocytes regulate differentiation of cortical epithelial cells (23), whereas mature thymocytes organize the medullary microenvironment (24,25,26).

The epithelial cells in the cortex and medulla differ by ultrastructure, by immunophenotype, and by functional characteristics. Ultrastructurally, three subsets have been detected on the basis of cytoplasmic processes, secretory organelles, and desmosome connections (27). Type I cells, located in the cortex, have stellate cytoplasmic processes and make contact with their neighbors, forming a syncytium. Type II cells are found in the medulla, are voluminous with many secretory intracellular vesicles, and have short cytoplasmic processes. Type III cells are rare, may contain vacuoles (pseudocysts), and are located in the medulla.

Figure 16.1. Ultrastructure of T cells. The normal T cell has a small rim of cytoplasm with only a few organelles. Nucleus shows dense heterochromatin. (From Zucker-Franklin D, Greaves MF, Grossi CE. Atlas of blood cells, 2nd ed. Philadelphia: Lea & Febiger, 1988, with permission.)

By immunophenotype, six clusters of thymic epithelial staining (CTES) have been identified. Ultrastructurally, type 1 epithelium (CTES II) (28) produces thymic hormones. Type 2 (pale), type 3 (intermediate electron lucency), and type 4 (dark with oval nuclei) are in the cortex proper (CTES III). These cells extend fine cytoplasmic processes, establishing contacts with neighboring cells, and are connected by desmosomes forming a syncytium, the interstices of which are filled with thymocytes. Some tend to engulf up to 20 to 40 thymocytes in a lymphoepithelial cluster, known as nurse cells, detected in human thymus (29,30) (see Chapter 14, Fig. 14.11). The internalized thymocytes are located within caveolae lined by plasma membranes. The lack of penetration by certain dyes indicates that the nurse cells are not an artefact, are completely sealed from the rest of the thymus, and may play a role in T-cell selection. Their formation is not dependent on interactions of T-cell receptors (TCRs) with MHC because they are present in knockout mice deprived of TCR-αβ (31).

A better understanding of the thymocyte–epithelial cell interdependence for survival came from experiments with a variety of TCR-transgenic mice (34). The final organization of the cortical and medullary epithelium depends on interactions of the TCR on the thymocytes with ligands on stromal cells. Thymocytes expressing a transgenic TCR that triggers strong positive selection, resulting only in maturation of CD4⁺/CD8^- T cells, lose the normal reticular pattern of cortical epithelial cells, and the epithelium in the medulla forms small scattered groups of cells surrounded by macrophages and dendritic cells (33).

Several other transgenic models demonstrated that maintenance of a balance between positive selection, negative selection, and nonselection is necessary to preserve normal compartmentalization and architectural integrity of the thymic epithelia (34,35). It is the diversity of signals emanating from these physiologic processes within the thymus during thymocyte maturation that are critical for the maintenance of epithelial organization. These signals are generated from the endogenously rearranged TCRs, which regulate positive and negative selection.

Figure 16.2. Morphogenesis of thymus. Elements from the third pharyngeal cleft form the epithelial rudiment, joined by mesenchymal cells from the neural crest. The common epithelial progenitor differentiates into cortical and medullary epithelium under the influence of the mesenchyme. The presence of endodermal cysts and tissue-specific “antigens” and “organoids” suggests some contribution from the endoderm. FGF, fibroblast growth factor; KGF, keratinocyte growth factor.

In addition to thymocytes, the thymic vasculature is also an important epithelial organizer (36). In RAG2^-/- mice, the medullary epithelium forms cuffs around intermediate-sized vessels, particularly the postcapillary venules. This anatomic arrangement may have important functional implications. The medullary epithelial cells have been linked to negative selection and tolerance induction (37,38,39) and are therefore strategically located around the postcapillary venules, where the concentration of autoantigens would be high. The possible functional importance of this peculiar anatomic arrangement of the medullary epithelium is also suggested from the detection of several molecules and structures considered to be tissue specific within the medulla (i.e., parathyroid hormone, thyroglobulin, insulin, and even organized epithelial “organoids” with ultrastructural features of respiratory epithelium and thyroid follicles) (40). The ectopic location of these molecules and structures is supported by the detection of the expression of the appropriate genes (40). It has also been postulated that, because these “ectopic” tissues within the medulla have their origin from primordial endoderm, the precursors of the epithelial cells in the medulla also may arise from pharyngeal endoderm. Further support of this view comes from the histologic appearance of medulla in athymic mice, in which further differentiation of these cells fails and the epithelium remains in the primordial condition of endodermal cysts.

Whereas epithelial development depends on FGF, its functional integrity is maintained by other growth factors contributed by the mesenchyme. The keratinocyte growth factor (KGF), a member of the FGF family, is a paracrine growth factor produced by mesenchymal cells. It acts on epithelial cells that express a splice variant of FGF (i.e., FGFR2IIIb). In fetal thymic organ cultures, exogenous KGF expands the medullary epithelium, and in RAG^-/- animals deprived of thymocytes, which normally produce KGF, the administration of KGF restores the normal medullary epithelial architecture (41).

The architectural integrity of the thymus is maintained throughout life, but eventually, thymus atrophies with age (42). Changes in signals between epithelial cells and thymocytes may determine thymic involution (43). Stat3 has been identified as an important signaling molecule between epithelial and mesenchymal cells in the thymic microenvironment (44). Stat3 gene disruption in mice results in severe thymic atrophy and enhanced susceptibility of the thymus to environmental stress, such as glucocorticoids or γ-irradiation.

The extracellular matrix (ECM) is the second important component of thymic stroma after the cells (45). It consists of multiple collagens, reticulin fibers, glycosaminoglycans, and glycoproteins, including laminin and fibronectin. Heterotrimeric laminin molecules consist of at least 15 naturally occurring isoforms, which are formed by five α, three β, and three γ subunits. In the human thymus, laminin with α₂-chains (LN-2/4) or α₅-chains (LN-10/11) are detected in the subcapsular epithelium and blood vessels (46). The CD4^-/CD8^- (double negative) thymocytes are located in the subcapsular area by strong attachment to LN-10/11 through their α₆β₁ integrin. The CD4⁺/CD8⁺ (double positive) thymocytes, however, lose their capacity to adhere to LN-10/11 and move down to the cortex.

These thymocyte–stroma interactions facilitate intrathymic migration and regulate positioning of the developing thymocytes to appropriate microenvironments during differentiation. The ECM proteins support the growth of thymocytes and epithelial cells and facilitate cell–cell interaction, especially migration of thymocytes in and out of nurse cells. Receptors for ECM proteins are highest in the double-negative precursors but gradually decrease with maturation.

T-Cell Progenitors

The T-cell progenitors in fetal life derive from the liver, whereas in adult life, they come from the bone marrow. The difference in stem cell origin has implications in subsequent lymphoid development, apparently as a result of precommitment or restriction of developmental options at the level of stem cell (50).

Human thymus becomes fully differentiated by approximately the 15th week of gestation (i.e., approximately 7 to 8 weeks after colonization of the thymic rudiment). These early migrants contribute to the development of the thymic microenvironment. Cells with the CD34⁺/CD38^weak/CD90⁺ phenotype contain T-cell progenitors when they grow in fetal thymic organ cultures. In human adult bone marrow, CD34⁺/CD38^-/HLA-DR⁺ stem cells have the potential to differentiate toward lymphoid and myeloid lineages (48). These cells are CD45RA⁺, lack Thy1 (CD90) antigen, and may represent an intermediate oligopotent stem cell with T-cell–reconstituting ability (49). By phenotype, there are three subpopulations in the bone marrow, which can differentiate to T cells in the thymus: CD34⁺/CD2⁺, CD34⁺/CD7⁺, and CD34⁺/CD2⁺/CD7⁺, and all three are negative for CD3/CD4/CD8 (50). The thymus is populated by hematopoietic multipotent progenitors from the bone marrow (51,52), corresponding to a common lymphoid progenitor subset (53).

The most primitive progenitor is Lin^-, c-Kit^high, L-selection⁺, TdT⁺, and RAG-1^- (54).

Recruitment involves the chemokines CCL21 (or SLC, secondary lymphoid tissue chemokine) and CCL25 produced by the fetal thymus, which attract CD4^-/CD8^-/CD25^-/CD44⁺ thymocytes (55). At the site of entry P-selection of the thymic vascular endothelium binds arrests precursors expressing the P-selectin ligand (rolling). Chemokine CCL25 from endothelial cells activates precursors through the CCR9 receptor, while intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) of the endothelial cells bind the cells firmly through the integrins α_Lβ₂ and α₄β₁, respectively (56). The multipotent T-cell progenitors entering the thymus trigger Notch activation, which is critical for T-cell development (see below) for later stages of TCR-dependent selection within the thymus as well as during peripheral T-cell differentiation (56). There is some overlap of regulatory programs between those required for T-cell specification and those needed for stem cell maintenance or self-renewal, such as Notch pathway, GATA-3, Ikaros, and PU.1 (57). The genes that need to be turned on for a successful entrance in the early stage of T-cell development include RAG-1 and RAG-2 and CD3-γ and CD3-∊ of the pre-TCR and TCR. While certain options of differentiation are eliminated relatively early after T-cell lineage commitment, others such as differentiation into dendritic cells and natural killer (NK) cells remain through some stages of the T-cell development—B-cell potential is lost early within the thymus probably as a result of Notch signaling.

Additional factors, such as GATA-3 (T-cell specific among the hematopoietic cells), E2A (shared with B lineage), and even non–T-cell specific such as Runx and Ikaros, are also used in T-cell development.

Notch and T-Cell Commitment

Signaling through the Notch receptor is a key factor for T-cell commitment (58,59,60,61). Notch plays a pivotal role in determining T/B-lineage choice, and signaling through Notch drives commitment of lymphoid precursors to T lineage (62,63,64). Notch belongs to a family of conserved proteins that function as cell-surface receptors and direct regulators of gene transcription (65). It was first isolated as a gene involved in chromosomal translocations with the TCR-β gene in a subset of cases of human T-cell acute lymphoblastic leukemia. The extracellular domain of Notch contains a variable number of tandem epidermal growth factor–like repeats and three Lin/Notch repeats, which function for ligand binding and Notch activation (66). The intracellular region contains six Cdc10/ankyrin repeats characteristic of protein–protein interactions and essential for signal transduction. The Notch protein initially is synthesized as a single-polypeptide chain, but, as a result of proteolytic processing, it is split into two parts. The extracellular region is separated and forms a noncovalent heterodimer with the remaining portion consisting of the transmembrane and the cytoplasmic regions.

Ligands for Notch are Delta, Serrate, and several other molecules corresponding to these two classes. In general, those homologous to Delta are referred to as Delta, and those homologous to Serrate are called Serrate or Jagged (66). These ligands are transmembrane proteins with an extracellular domain with a variable number of epidermal growth factor–like repeats and the unique domain for this family, the DSL (Delta/Serrate Lag-2) domain, which mediates binding to Notch and activation. Which of the two groups of ligands is important for T-lineage commitment remains controversial (67). In cells of the immune system, there are two Notch receptors, Notch 1 and Notch 2, and four signals (68). The pleiotropic signaling by Notch regulates differentiation, proliferation, and cell death, but it is not yet clear which function most precisely determines cell fate and ultimately directs T-cell commitment (69). With Notch inactivation, the double-negative T cells diminish in the thymus, whereas B-cell precursors increase, probably from a more efficient production of B cells within the thymus (70). On the other hand, transgenic expression of Notch in the bone marrow permits the accumulation of CD4⁺/CD8⁺ T cells. Notch 1′s functional role seems to be in developmental specification, driving T-cell precursors at the expense of B-cell precursors and perhaps directing the choice of a common precursor between these two fates. Notch is also needed at later stages for TCR-β gene rearrangements, positive selection, and CD4/CD8 lineage choices (57).

Phenotypic Differentiation

The first migrants from the bone marrow to the thymus settle in the corticomedullary junction and are large, dividing cells expressing CD34⁺/CD45RA⁺, CD2+, and CD7⁺, with the TCR genes in the germline configuration (Fig. 16.3). The new immigrants potentially are able to differentiate to other lineages, such as NK cells, dendritic cells, and monocytes. The CD34⁺ cells then coexpress other markers such as CD38 and CD71 (transferrin receptor) associated with proliferating cells, and a portion of them are CD10⁺. They are subdivided into three categories, double negative 1 (DN1), DN2, and DN3, depending on the expression of CD44 and CD25 (71,72). Recruitment involves interactions between P-selectin expressed by thymic endothelium and P-selectin glycoprotein ligand-1 (PSGL-1) detected on the bone marrow progenitor cells (73). Upon arrival they migrate to the subcapsular cortex (DN3 cells), where they acquire expression of CD1α and CD4 and CD8, and as a result become double positive (DP). In the cortex they are submitted to positive selection (vide infra) and those considered “useful” return back to the medulla, while they are separated into two distinct phenotypes, CD4⁺ and CD8⁺ single-positive (SP) cells. These migrations across the thymic parenchyma are mediated by several chemokine receptors, CXCR4, CCR7, and CCR9. Adherence of thymocytes to epithelial cells is also mediated by CD2 and the lymphocyte function molecule-3 (LFA-3, CD58), ICAM, and LFA-1. The SP thymocytes are submitted to another screening in the medulla.

Figure 16.3. The first migrants from the bone marrow arrive and settle in the corticomedullary junction. They lack expression of CD4 and CD8 and are known as double negatives (DN). These new immigrants begin to move to the outer cortex and, depending on certain markers, are distinguished in three stages: DN1, DN2, and DN3. The migration is supported through interactions between P-selectin (thymic epithelium) and its ligand (progenitor cells). In the outer cortex they become double positive (DP) (i.e., CD4+/CD8+), and at the DN3 stage, undergo gene rearrangements with expression of T-cell receptor (TCR)-β gene (β selection). Signals provided by the CCR-7 chemokine receptor guide the positively selected thymocytes that have already separated into the CD4 or CD8 lineages back to the medulla. A final screening for the “affinity” of their TCR binding to autoantigens takes place in the medulla (negative selection). It is triggered by contacts of their TCR with tissue-restricted self-antigens (TRAs) on medullary thymic endothelial cells (mTECs). Promiscuous gene expression (pGE), for autoantigens, regulated by the AIRE gene (see text for details), controls central tolerance. The thymic medullary epithelium ultimately allows survival of the “useful,” while it ignores the “useless” and “destroys” the harmful (129). Exit of mature T cells from the thymus is regulated by the G-protein–coupled receptor sphingosine-1-phosphate receptor 1 (S1P1) as well as the very late antigen (VLA-) and lymphotoxin-β receptor.

Those with high-affinity TCR against self-antigens are deleted (negative selection). After this checkpoint, the surviving thymocytes exit thymus for the secondary lymphoid organs. Emigration of thymocytes requires the sphingosine-1-phosphate receptor 1 (S1P1).

In the process of differentiation, CD34 is progressively lost while the intensity of CD7 decreases. Myeloid and NK cells have been detected in various in vitro systems arising from thymocytes (74,75), and the thymic microenvironment is able to support myeloid differentiation.

Evidence of direct cellular communication between various thymic cells was provided by the demonstration of the existence of gap junctions formed by connexin 43 between two epithelial cells or between epithelial cells and thymocytes (76).

Of the cytokines that have been implicated in T-cell differentiation, interleukin-7 (IL-7) is essential (77). It is produced constitutively by epithelial cells, and it induces proliferation of DN thymocytes (78) or maintains their viability (79). CD34⁺ thymocytes cultured with IL-7 start to express CD8 and CD4 but remain CD3 and TCR negative, indicating that other stimuli from stromal cells are essential for generating CD3⁺/CD4⁺/CD8⁺ cells. IL-7 also induces TCR-β gene rearrangements (81). Mice genetically deficient in IL-7 receptor have a profound reduction of T and B lymphocytes, and thymocyte development is blocked at a very early stage before the induction of CD25 and TCR-β gene rearrangements (81). IL-7 regulation is tightly coordinated during T-cell maturation within the thymus for thymocyte survival, as IL-7 is essential for postselection expansion of positively selected thymocytes (82).

Other cytokines, such as IL-1, IL-2, and IL-4, have also been shown to play some role in thymocyte differentiation (77).

The thymus is continuously colonized by hematopoietic progenitors that have the genes for the TCR in germline configuration. Approximately 100 to 1,000 such progenitors enter the thymus daily, and it takes approximately 3 weeks to undergo complete differentiation to mature functional and self-tolerant T cells.

The biologic processes within the thymus are highly complex, but for a better understanding, we divide them, somewhat arbitrarily, into three areas: (a) lineage determination (i.e., TCR-αβ vs. TCR-γδ), (b) separation of the two main T-cell subsets (i.e., CD4 vs. CD8), and (c) selection for survival of those cells with a TCR able to recognize foreign antigens (positive selection) and elimination of those possessing autoreactive configurations (negative selection). Understanding of these events will be facilitated by a prior description of the genes encoding the TCRs.

T-Receptor Genes

Knowledge of the structure and patterns of expression of the various TCR genes is essential to our understanding of antigen recognition by T cells. The TCR gene, “a needle in the haystack,” was isolated by the technique of subtractive hybridization (83). Four human and murine TCR genes have been identified: α (84), β (85), γ, and δ (86,87) (Fig. 16.4).

Figure 16.4. Organization of the genes for human T-cell receptor chains (see text for details). (Adapted from Kronenberg M, Siu G, Hood LE, et al. The molecular genetics of the T-cell antigen receptor and T-cell antigen recognition. Annu Rev Immunol 1986;4:529–591; and Raulet DH. The structure, function, and molecular genetics of the γ/δ T-cell receptor. Annu Rev Immunol 1989;7:175–207.)

The human α gene is located on chromosome 14 (bands q11-12) (88), as are the immunoglobulin (Ig) genes (band q32). Rearrangement involving the region of the chromosome containing the α genes has been detected in patients with T-cell malignancies (89,90). The δ gene is located on chromosome 14 within the α gene (see below) (91). The β and γ genes are located on chromosome 7 (92,93). Translocations and inversions of chromosomes 14 and 7 are often seen in association with ataxia telangiectasia, probably involving fragile sites that normally are used during the TCR gene rearrangements.

All TCR genes display an overall organization similar to that of the Ig genes. They are composed of variable (V) and constant (C) genes. The V gene is made of three segments (V, J, and D) in the β and δ genes, but only two segments (V and J) in the α and γ genes (87). Each V-gene family is divided into subfamilies, like the Ig V genes, on the basis of sequence similarity (more than 75%).

In humans, the TCR genes group consists of one Cα gene (94), two Cβ genes (95), two Cγ genes, and one Cδ gene. With the exception of Cγ, which is composed of three exons, all of the genes have four exons. The δ locus is located between the Cα and Jα gene segments. Diversity in the TCR genes is generated through rearrangements with each of the V and J segments and D in the case of β and δ, forming the complete V gene. The TCR genes, however, contain a large number of J segments as compared to the Ig genes.

Associated with the Cα gene are approximately 50 Vα and 61 Jα segments that spread over 100 kb of DNA. The β-chain gene complex spans around 600 kb (96) and incorporates 57 Vβ segments. Each of the two Cβ genes possesses a set of Jβ segments and one Dβ segment. The Vβ segments are all located upstream from the two clusters of Cβ genes. During rearrangements when a Vβ segment forms a VDJ complex, the transcript is committed to use the same Cβ segment genes. The γ gene locus contains two Cγ genes, each associated with its own set of Jγ segments. There are approximately 14 Vγ segments, all located upstream separately from the two Cγ-Jγ clusters. The Cδ gene complex lies between the V and J segments of the α gene complex. There are eight Vδ, three Dδ, and three Jδ segments.

Like the Ig genes, the TCR gene segments are flanked by heptamer-spacer-nonamer sequences, which serve as recognition sites for the recombinase, the enzyme that initiates recombination of the segments. The same RAGs 1 and 2 (RAG1 and RAG2) regulate the V(D)J recombinations for both T and B cells. But RAG1/RAG2 act in a lineage-specific manner (i.e., Ig genes are assembled only in B cells and TCR genes only in T cells). The recombinations are also developmentally regulated. For T cells, the β genes are recombined before α genes; as for B cells, the IgH is recombined before IgL. These differences are explained by differential accessibility of V genes during development. Mutations of either one of these genes in mice result in complete arrest of maturation of both B- and T-cell lineages (97,98). The mechanism of V(D)J recombination has been described in detail in Chapter 15.

The mechanisms generating diversity are combinatorial associations of different V, D, and J segments and combinatorial pairing of TCR proteins. However, in contrast to the Ig genes, fewer V segments are available to the TCR genes. The diversity of the TCR is mainly junctional (i.e., a result of additions of nucleotides at the DNA cleavage site). Additions that depend on template are known as P nucleotides, and random additions are called N nucleotides added by terminal deoxynucleotidyl transferase (TdT) (see details in Chapter 15). TdT is not expressed during fetal life, and, in TdT knockout mice, the T-cell repertoire is of fetal type (i.e., less diverse). TdT is expressed only in immature lymphocytes and is responsible for the transition from fetal to adult repertoire by contributing enormously to the lymphocyte antigen receptor repertoire. A striking difference between Ig and TCR is the lack of somatic hypermutations, which is very important in generating high-affinity antibodies in the germinal centers (Chapter 17). Assembly of V genes of β, γ, and δ genes occurs during the DN stage. If the β-chain rearranges successfully, it forms a heterodimer with the pre-Tα chain and differentiates along the T-αβ lineage entering the DP phenotypic stage.

TCR gene rearrangements occur in two discrete stages of thymocyte development. The first occurs during the DN3 stage with the δ gene rearranging first, followed by the γ and β genes. The α gene rearranges during the DP stage. During the β gene rearrangement, a Dβ segment joins a Jβ segment, and, as with the Ig gene rearrangement, a Vβ joins the DβJβ complex. If the rearrangement is not productive, the cell may have one more chance at rearranging a second Cβ cluster, as there are two Cβ genes each with its own D and J segments.

T Cell-αβ versus T Cell-γδ and the Pre–T-Cell Receptor

At this point, the decision needs to be made in choosing a TCR-γδ versus a TCR-αβ lineage by any one of three proposed models. According to one model, each lineage starts from a separate precursor (stochastic or independent). A second model proposes that there is a common precursor for both lineages (instructive). The third model (competitive) postulates that rearrangements for the genes of the respective lineages start concurrently and that those finishing successfully first determine the fate of the cell. Some evidence appears to support a variant of the stochastic model (97). It is agreed that the TCR-δ gene rearrangement occurs in thymocytes that can adopt either the γδ or αβ fates, whereas during the DP stage, activation of the TCR-α gene rearrangement seals the fate of the T cell for the αβ lineage.

Determination of lineage is regulated by transcription factors that act on promoters and enhancers and make appropriate genes accessible to the recombinase (98,99). Another regulatory mechanism involves the pre-TCR. At this stage of T-cell development, the TCR-β chain forms a heterodimer with the pre-Tα (pTα) chain, and the complex is referred to as pre-TCR (117). The pTα is a 33-kd type I transmembrane protein of the Ig superfamily, with a single Ig-like domain. Two cysteines form the intrachain disulfide bond, whereas a third cysteine just above the transmembrane region forms the disulfide bridge with the TCR-β chain (101). The human pTα gene is located on the short arm of chromosome 6, in the vicinity of the HLA locus. The pTα chain is not essential for CD3 expression, but has a major role in TCR-αβ versus TCR-γδ commitment. It generates large numbers of CD4⁺/CD8⁺ T cells with productive TCR-β rearrangements and directs these cells to the TCR-αβ lineage (102). At the DN3 stage, thymocytes undergo extensive DNA rearrangements at the β, γ, and δ gene loci and make a choice for selection, between the TCR-αβ and TCR-γ receptors. Expression of the TCR-β chain in combination with the pre–TCR-α chain results in a process known as β selection, which leads to rearrangements of the TCR-α genes and formation of the complete TCR-αβ. If the TCR-δ or TCR-γ genes are rearranged, the cells follow the TCR-γδ lineage (γδ selection) (103,104,105).

The selection for TCR-β is initiated at a stage phenotypically characterized by expression of CD4⁺/CD8α⁺β^- (106). The productive Vβ gene rearrangement in one allele prevents rearrangements of the second allele, a process called allelic exclusion that is regulated by the pre-TCR. Signaling by the pre-TCR requires the CD3 chains γ, ∊, ζ, and η, and the Lck kinase (107), which is associated with the CD3-∊ and -γ chains and is indispensable for the pre-TCR function. Other functions mediated by pre-TCR signaling are cell survival and phenotypic changes. Pre-TCR promotes thymocyte survival, whereas signaling by TCR induces apoptosis (108,109). Because pre-TCR is expressed on DN T cells, the antiapoptotic function of pre-TCR is crucial for their survival and differentiation of DN T cells into the DP stage when the TCR is needed for positive or negative selection.

Both receptors follow initially similar signaling pathways, such as tyrosine phosphorylation and so forth, but in the apoptotic pathway, they diverge at the level of Fas ligand (FasL) induction, which requires induction of Nur77 and transcription factors from the nuclear factor of activated T cells (NFAT). These factors can be induced only by the TCR-αβ but not by the pre-TCR. The pTα possesses a palmitoyl moiety that spontaneously targets the pTα chain to the cell membrane (lipid rafts) (140). This may offset the requirements for a ligand because the pre-TCR from such a location is able to signal constitutively. Signaling through pre-TCR results in expression of certain transcription factors required for the differentiation of precursors to αβ T cells (i.e., β selection [see above]). Some of these factors are the E-proteins of the basic helix-loop-helix transcription activators required for expression of CD4, TCR-α and TCR-β chains, and so forth (111).

CD4/CD8 Lineage Commitment

The DP thymocyte (CD4⁺/CD8⁺) differentiates to two phenotypically and functionally distinct lineages of αβ T cells: CD4⁺ and CD8⁺. The TCR of the CD4⁺ cells interacts with peptides bound to class II MHC molecules, whereas the TCR of the CD8⁺ cells recognizes peptide–class I MHC complexes. The CD4 and CD8 proteins are not clonally distributed and are known as coreceptors because they recognize the same ligands as the TCR (112). The mechanism by which the separation of the lineages from DP thymocytes is achieved remains unresolved (113,114).

According to one theory, known as instructive, thymocytes carrying an MHC class I–restricted TCR differentiate to CD8⁺ lineage, whereas engagement of TCR with class II MHC induces commitment to CD4. An alternative model, known as stochastic (or selective), accepts that the DP thymocytes are already committed randomly to a lineage; they make a choice that is unrelated to the MHC specificity of their TCR. Data collected from a variety of approaches do not agree with any particular mechanism in lin-eage commitment. The instructive model proposes that the CD4 or CD8 transduces differentiation-specific signals, but no lineage-specific signals have ever been identified.

It is also possible that the cytoplasmic region of CD4 directs CD4 lineage commitment. A chimeric construct, for example, made of the cytoplasmic region of CD4 and the extracellular and transmembrane region of CD8a supported the development of cells with the CD4⁺/CD8a^-phenotype but with a class I–restricted TCR (115). CD4 is preferentially associated with its cytoplasmic tail with the tyrosine kinase Lck. Therefore, CD4 is likely to deliver stronger signals than the coengagement of TCR and CD8 (116,117).

The strength of signal model suggests that strong signals dictated by the frequency of the available ligand induce CD4 differentiation, whereas weak signals induce CD8. Another aspect of signaling that was evaluated was the duration of signals as an important parameter. When the interaction is limited to a few hours, the cells become CD8⁺, but exposure for longer periods of time results in CD4⁺ lineage commitment (118). Accordingly, TCR of a DP thymocyte initiates down-regulation of CD8 and produces a CD4⁺/CD8^- intermediate cell. At this stage, the duration of signaling determines the final outcome. That is, short signaling produces CD8⁺ T cells, whereas persistence of signaling in the CD4⁺/CD8^-intermediate cell causes CD4⁺ differentiation (119).

Variation of Lck function seems to be the single most important parameter in CD4/CD8 lineage decision. Constitutively active Lck promotes CD4 differentiation, even in the presence of class I MHC–restricted TCR. When, on the other hand, Lck is catalytically inactive, all thymocytes, including those with class II MHC–restricted TCR, become CD8⁺ (120,121). In the absence of Lck, cross-linking of CD3 induces CD8 differentiation (122). The Lck-dependent regulation of lineage commitment is not only phenotypic, but also functional, because the class II MHC–restricted CD8⁺ cells behave as killer cells, whereas the class I–restricted CD4⁺ cells up-regulate CD40 ligand, a function characteristic of helper T cells (120). Signaling initiating from Lck is channelled through the Ras-Erk pathway (123).

The various experimental approaches used in these studies make it clear that the signals required for CD4 differentiation are promiscuous, which is believed to indicate that CD4 differentiation is a default pathway (124).

The multiplicity of the models entertained and the ambiguity of some of the results are also a testimony that the precise mechanism of T-cell lineage commitment remains still elusive and certainly complex (114,125).

More recently a new transcription factor Th-POK (T-helper–inducing POK factor) was found to be necessary and sufficient for CD4 lineage commitment, and absence of Th-POK results in the development of CD8⁺ T cells (126). The existence of this factor was suspected as a result of studies of the spontaneous recessive mutant mouse with helper deficient (HD) cells (i.e., CD4⁺ T cells) (127). Th-POK belongs to the POK family of transcription factors characterized by two motifs, a regulatory POZ/BTB domain (for interaction with other transcription factors) and a Zn finger DNA-binding domain.

In summary, Lck is a key regulator, but other signals originating from the TCR in the absence of a coreceptor or in the absence of a significant recruitment of Lck provide sufficient Src activity for a response. Such signals depend on the nature of the peptides involved in positive versus negative selection, which is linked to lineage selection (122). Although the prevailing opinion accepts that lineage commitment and the selection of thymocytes with useful TCRs are linked, some of the models proposed point to the opposite (128). The latest view that CD4/CD8 lineage commitment is transcriptionally regulated will put an end to this question hotly debated for a long time.

Positive and Negative Selection

The random nature of rearrangements of TCR genes generates specificities directed not only against foreign antigens, but also against self-antigens. Self-reactive T cells are harmful if they have the opportunity of exiting the thymus. Mechanisms have therefore been developed that allow the thymus to “select the useful, neglect the useless and destroy the harmful” (129). The decision on which pathway each cell will ultimately follow is made by the TCR. The time for selection appears to begin at the DP stage when the TCR is expressed at low levels (130,131,132,133). The TCR of the thymocytes recognizes self-peptides presented by the MHC molecules. The nature of the criteria used by the TCR for directing individual DP thymocytes to three different fates has been, and still is, one of the most challenging questions in immunology.

It has been widely accepted that the options for each DP thymocyte are determined by some fine qualities and properties of the TCR and MHC–peptide (pMHC) ligand interaction. If the affinity of interaction is “weak,” the cell is positively selected, and if the binding is of high affinity, the cell is negatively selected (134). The term avidity, however, incorporates the affinity of the individual TCR, the level of occupancy of all TCRs (i.e., the receptor-ligand densities). Low avidity, therefore, triggers positive selection, and high avidity triggers negative selection.

Positive selection represents a specialized or partial activation, whereas full activation (like that of a mature T cell) is lethal for that thymocyte. Cells with TCRs that cannot bind with sufficient avidity are neglected and die by apoptosis.

When the cell is positively selected, the expression of recombination-activating genes (RAG), which encode the proteins for V(D)J recombination, and the PTα gene are turned off, and the CD4 and CD8 are partially down-regulated. As the cell at this stage moves from the cortex to the medulla, one of the coreceptors is re-expressed and the cell lineage is defined (123,135).

Just as for lineage differentiation, signaling for positive and negative selection largely depends on Lck, and in the absence of Lck, both selection mechanisms are compromised (136). Lck interacts with the coreceptors CD4 and CD8 through two cysteines in the cytoplasmic region of each coreceptor and two cysteines in Lck (137).

The binding of TCR to the ligand induces phosphorylation of the three immunoreceptor tyrosine-based activation motifs (ITAMs) of the ζ-chain and the single ITAM in each of the CD3 ∊-, δ-, and γ-chains. The ζ-chain is constitutively associated in thymocytes with the ZAP-70 kinase, suggesting that a low level of activation takes place continuously (138). Receptor clustering from greater ligand engagement results in further phosphorylation and Lck involvement. The fate of the immature thymocyte depends on the type of APC that it encounters in the various histologic compartments of the thymus. The cortical thymic epithelial cells usually mediate positive selection (PS) (139,140), while the medullary hematopoietic cells mediate negative selection (NS) (141), although under certain circumstances they may also contribute to PS (142). Interactions among thymocytes may also support PS if the selecting thymocyte carries MHC class II molecules (143). In this case the α-chain of the TCR, and particularly the highly diverse CDR3 loop of the α-chain, plays an important regulatory role (144). Signals provided by the CCR-7 chemokine receptor (145) guide the positively selected thymocyte to the medulla. The CCR-7 signals provide guidance through the corticomedullary maze for the selected thymocyte, but it is essential, in preparation of the central tolerance in the medulla, to not simply usher them to exit to the periphery (145). An important selective force in these cellular interactions is the strength of the interaction, which determines the outcome. Strong interactions lead to deletion, while intermediate strength induces positive selection. This may explain the increase of positive selection following blocking of CD28/B7 interactions (146). The removal of strongly self-reactive cells is the sole purpose of the negative selection. Such clones potentially may initiate autoimmunity if they exit to the periphery. An estimated 2,000 to 3,000 tissue-specific antigens are expressed in human or murine medullary thymic endothelial cells (mTECs) (147,148)—that is, 5 to 10% of the known mouse genes, in addition to their normal tissue expression. Of these, about 500 may be AIRE (autoimmune regulator) dependent. Using hen egg lysozyme as reporter for the insulin promoter, it was shown that the expression of the antigen in the thymic stroma was necessary and sufficient for deletion (149).

Expression of tissue-restricted self-antigens (TRAs) by mTECs is the result of promiscuous gene expression (pGE) by the mTECs (147). Some of the genes are expressed by both mTECs and cortical thymic endothelial cells (cTECs) (pool 1), while others are expressed only by mTECs (pool 2) and finally others are expressed only in the more mature mTECs (i.e., strongly positive for MHC class II) (pool 3). Depending on the stage of differentiation of mTECs, two models have been proposed for pGE: (a) in the developmental or progressive restriction model, pGE is detected in immature and perhaps pluripotent progenitor cells; and (b) in the terminal differentiation model, pGE is mutated in mTECs of a mature phenotype (150). The pGE is regulated by AIRE, the gene mutated in the rare autoimmune disorder known as autoimmune polyglandular syndrome type 1 (APS-1) (151).

AIRE has a nuclear localization signal and several potential DNA-binding and protein interaction domains. In cooperation with transcription factor CREB-binding protein (CBP) it transactivates the transcription of other genes, although it is still a mystery how as a single molecule it controls the transcription of such an array of genes.

The pGE has been conserved across species barriers during evolution as a study recently with human pure populations of thymic cTECs and mTECs has shown (152). The AIRE gene showed the highest enrichment in mTECs and the promiscuously overexpressed genes are those remarkably well conserved among species. A set of 443 genes have been detected overexpressed in mTECs (comparable to 555 in mouse). This is an underestimate as most were detected by polymerase chain reaction (PCR). These genes show no preference for any chromosome. Another interesting aspect is the finding of clustering of nonhomologous genes. This clustering has been suggested to result from the juxtaposition during evolution of genes involved in the formation of a particular tissue. Since the promiscuously expressed genes do not all share such a function, the clustering is the result of epigenetic mechanisms of regulation. They become accessible to mTECs as a result of belonging to the same “gene neighborhood.” pGE not only is sufficient for self-tolerance, but also likely has been essential for survival of the species since infertility as a result of gonad-specific autoimmunity is highly prevalent in APS-1 (153) and in AIRE^-/- mice (154). In addition to genetic control, self-tolerance may be regulated also by epigenetic mechanisms. The lymphotoxin β receptor (LTβR) directs the three-dimensional organization of mTECs (155), and TRAF6, a cytoplasmic adaptor molecule, which does not bind to LTβR, was still shown to direct mTEC development (156,157).

Self-tolerance is mediated primarily by negative selection or clonal deletion (158), but some of the self-reactive T cells are submitted to nondeletional central tolerance and give rise to immunosuppressive CD4⁺ T cells, also known as natural regulatory T cells (T reg) cells (159). T reg cells are T cells with medium to high affinity for self-antigens. However, they escape negative selection and are positively selected by a subset of dendritic cells that have been educated by a thymic stromal lymphopoietin produced by Hassall corpuscles (160). Other evidence, on the other hand, indicates that CD4⁺ T reg cells are positively selected by thymic epithelial cells expressing self-antigens but not by bone marrow–derived dendritic cells (161).

ZAP-70 is indispensable for positive and negative selection because both processes are abrogated in ZAP-70–deficient mice (162). The Src-like adaptor protein (SLAP) down-regulates the TCR expression during the DP stage of development and rescues T-cell development in the absence of ZAP-70. Overall, SLAP acts as a negative regulator and probably “marks” activated receptors for retention and degradation (163). The main downstream signaling pathway for positive selection is the Ras/map kinase (MAPK) cascade (164). TCR signaling alone is not sufficient to induce selection for survival or death (165). CD2 expression exerts a strong influence on TCR repertoire. In the absence of CD2, the thymocytes with high affinity for peptide MHC escape negative selection (166). CD2 also influences pre-TCR function in the usage of Vα genes, which is substantially altered in CD2-deficient mice.

The costimulatory interaction of CD40/CD40L is a master regulator of negative selection, usually acting in the regulation of the ligands of other costimulatory molecules, such as CD80 and CD86 (167). CD40 may also induce other costimuli required for thymocyte deletion, such as CD54 (ICAM-1), FasL, or tumor necrosis factor (TNF) (168,169,170). These molecules could regulate negative selection separately or in combination with CD5 and CD28. The CD28 costimulatory molecule engaged with TCR signals thymocytes to undergo apoptosis or maturation, depending on the intensity of costimulation (171).

Stimulation of maturation of DP thymocytes follows activation of the extracellular signal-regulated kinase (ERK)/MAPK pathway and up-regulation of the antiapoptotic protein Bcl-2. Apoptosis is triggered with the expression of the Nur77 family of transcription factors and occurs only if TCR engagement is accompanied by co-stimulation. Ca²⁺ fluxes in mature T lymphocytes regulate proliferation, differentiation, and survival. Some of the functions of Ca²⁺ are mediated by the Ca²⁺-dependent phosphatase calcineurin. The function of calcineurin is disrupted by cyclosporin A and FK506. Both of these substances form complexes with cellular proteins, termed immunophilins, that bind and sequester calcineurin. Cyclosporin A and FK506 block the initial steps of positive selection (172,173). On the other hand, intracellular chelators reduce deletion (174), which suggests that interference of Ca²⁺ fluxes and function exerts several effects on the selection process.

All signaling pathways end up targeting regulation of transcription factors, and various studies have examined the roles of NFAT and nuclear factor-κB (NF-κB), but only members of the Nur77 family have been clearly shown to be involved in the process of selection, with a major role in apoptosis of massive numbers of thymocytes (175). It has been estimated that 90 to 95% of thymocytes die inside the thymus, and the appearance of very large numbers of dead cells on histologic sections is the reason that histopathologists called the thymus the “graveyard” of T cells.

The timing of positive versus negative selection has been difficult to define. Negative selection was believed to occur relatively late because the TCR affinity increases with time, and it is the high affinity of TCR that triggers negative selection.

Cells and Molecules in the Selection of Thymocytes

Hematopoietic components of the thymic stroma are radiosensitive, whereas the epithelial components are radioresistant. These differences allowed the construction of chimeric thymuses (i.e., thymuses composed of epithelial and hematopoietic elements with a different MHC background). These experiments indicated that the thymic cortical epithelium is responsible for positive selection (176). The hematopoietic stromal cells, on the other hand, are potent inducers of negative selection. The cortical epithelial cells have weak MHC expression and the developing DP thymocytes at that stage have low TCR expression; as a result, the interactions are only of low avidity, required by positive. Similarly, the hematopoietic cells in the corticomedullary junction express MHC molecules at higher density, and the T cells have up-regulated their TCRs. As a result, the interactions in this anatomic location are appropriate for negative selection.

The concentration of the peptide may also play a role in the selection process. A thymocyte bearing a TCR specific for class I–peptide complex is positively selected if the peptide it carries is in a low concentration and is killed if the peptide concentration is high (177). However, the weak interactions required for positive selection and the strong interactions operating in negative selection can also be provided by the half-life of the pMHC complex (i.e., short half-life for positive and long half-life for negative selection).

The original question of affinity versus concentration was better answered with manipulation of the pMHC. Positive selection is driven by peptides with varying affinities; however, by increasing the concentration of the peptide, even low-affinity ligands can positively select (178).

It has been argued that the recognition of self-peptides in positive selection must be relatively degenerate. In other words, a single pMHC can trigger the positive selection of multiple thymocytes (179); therefore, the peptide recognition during positive selection is cross-reactive (180). This apparent degeneracy of the peptide binding to TCR is the result of a significant contribution to the binding by the MHC molecule (181).

Other parameters of a pMHC complex, such as the conformation (182) and the concentration (183) of the peptide, contribute to the repertoire selection. Despite the evidence that each pMHC complex selects more than one TCR, the large diversity of the fully developed T-cell repertoire depends on a large number of pMHCs (184) as well as on the quantity and quality of the stromal cells (185). Peptide diversity during the selection seems to have a greater effect on negative than on positive selection. In thymus organ cultures, testing with a diverse array of peptides, the addition of 1% of dendritic cells reduced the number of CD4⁺ T cells selected by 80% compared to that of the controls in the absence of dendritic cells. Thus, the quantity and quality of the selecting stromal cells have a significant impact on the selected repertoire by multiple peptides (185).

At the end of the selection process, mature T lymphocytes exit the thymus and home to the secondary lymphoid organs (see Chapter 14). There are two types of TCR-αβ T cells, the CD4⁺ (T helper cell) and the CD8⁺ (cytotoxic T cell), and one TCR-γδ cell.

α/βT-Cell Receptor Complex

The TCR complex is composed of two components: One component is unique in each T cell and is involved in antigen recognition (ligand binding), and the second component is the same in all T cells and is involved in signal transduction leading to T-cell activation (186,187,188).

Ligand-Binding Component: αβ and Peptide–Major Histocompatibility Complex Interaction

The αβ receptor is formed by two chains, α and β (Fig. 16.5). Each chain consists of a constant domain, Cα and Cβ, and a variable domain, Vα and Vβ. The Vα domain is encoded by two genes, Vα and Jα, and is homologous to the V domain of the Ig heavy chain. The Vβ domain is formed by three polypeptides, Vβ, D, and Jβ. The constant domains correspond to the IgC domains, but there are certain differences. The Cα–Cβ interface is highly polar, whereas that of C_L–C_H1 is hydrophobic. Cβ has a large loop, which extends out to the side of the domain. It has been proposed that it may interact with the coreceptors. The Cα domain has several structural deviations from the C-type Ig domain.

Figure 16.5. T-cell receptor complex. The T-cell receptor complex consists of two components: Ligand binding (antigen recognition) and signal transduction. The antigen recognition component consists of two polypeptide chains, α and β. Because of their short intracytoplasmic tails, they cannot link themselves to the signal transduction cascade. The signal transduction component consists of the CD3 proteins γ, δ, and ∊, members of the immunoglobulin superfamily, and two other proteins forming either a homodimer (two ζ proteins) or a heterodimer (ζ-η). The γ-, δ-, and ∊-chains have one immunoreceptor tyrosine-based activation motif each, and the ζ-chain has three. See text for details.

The V domains are very similar to the V domains of an antibody molecule. They contribute to the formation of the TCR-combining site, which is made up of hypervariable loops or complementarity-determining regions (CDRs) 1, 2, and 3 from the α- and β-chains and another loop termed HV4, which exhibits some hypervariability. The CDR1 and CDR2 are formed by the V segments, which are less polymorphic in the TCR than in Ig because fewer V segments are available for TCR, whereas the CDR3 is polymorphic as a result of the larger number of J segments available for β-chain contributing to CDR3.

The loops of Vβ that form the expected antigen-binding site of the TCR are similarly placed as in the Ig V region. The chains are linked by a disulfide bond, and the heterodimer is anchored to the cell membrane by the transmembrane region, ending in the cytoplasm by a short (three– to five–amino acid) cytoplasmic tail. Crystallization of the TCR shows that it resembles the Fab fragment of the Ig molecule (188). However, the TCR-αβ is extensively glycosylated with up to seven N-linked sites distributed between the α- and β-chains. The combining site is usually flat, similar to antiprotein antibodies and consistent with the TCR’s function of binding to the generally flat, undulating surface of the pMHC (189). The diversity of the CDR3s is much higher, implying that the function of CDR3s is in peptide discrimination, whereas CDRs 1 and 2 interact with more conserved structural elements of the MHC. The TCRs contain many more J segments and thus are able to increase V-Jα and V-D-Jβ junctional diversity in the CDR3s (190). The residues of the TCR that contact the pMHC are always in the apices of the CDRs (i.e., for CDR1α, residues 27 to 30; for CDR2α, residues 50 to 52; for CDR1β, residues 27 to 30; and for CDR2β, residues 52 and 53).

The TCR interacts with peptides bound to MHC molecules (Fig. 16.6). The aminoterminal domains, α₁ and α₂ of the MHC class I heavy chain, form the binding site for the peptide. The site consists of a floor of eight strands of antiparallel β-pleated sheets, which support two α helices, one contributed by the α₁ domain and the other by the α₂ domain aligned in an antiparallel orientation. The floor of the groove is supported by two Ig domains from below; one is the α₃ domain of the heavy chain and the other the β₂-microglobulin. This arrangement forms a groove, on the floor of which lies the peptide from an antigen to be presented to the TCR. Some of the residues of the peptide are exposed above the groove and interact directly with the TCR, whereas others point to the floor of the groove. Depressions in the floor of the groove, known as pockets A through F, interact with some side chains of the bound peptide (191).

Figure 16.6. T-cell receptor (TCR) interaction with HLA/peptide complex. Crystallographic studies of the HLA class I molecule have shown that the two variable domains (α1and α2) form a groove that binds the peptide that is released as a result of antigen processing. The peptide groove is formed by the helices of the α1 and α2 domains (A). Binding of the peptide is mediated through anchor amino acids near both ends of the peptide (in A). The TCR binds to both the major histocompatibility complex molecule and the peptide. The complementarity-determining region (CDR) 3 of both the α- and β-chains interacts with amino acids of the peptide, whereas the CDR1 and CDR2 interact with the major histocompatibility complex molecule (B).

For HLA-B27, the side chain of the C-terminal of the bound peptide sits deep in the F-pocket, whereas the N-terminal forms strong hydrogen bonds with the hydroxyl groups of conserved amino acids at the end of the groove (192). (For details of interaction with the pMHC, see Chapter 17.)

For crystallization of TCR-pMHC, the peptide was attached covalently through a linker to the amino terminus of the β-chain of an MHC class II molecule (193) or covalently connected to the amino terminus of the β-chain of the TCR (194) and thus tethered to the MHC groove. Such approaches provided stable complexes of HLA-DR1 and HLA-DR4 that permitted their crystallization. Several TCR-pMHCs have been analyzed to clarify the contributions of Vα for the buried surface area (i.e., the interface between TCR and the pMHC). The contribution of Vα is on the average 57% and that of the Vβ 43% (195). The overall orientation of the TCR over the pMHC is diagonal (196) rather than orthogonal (197). It is more likely that the TCR twists over the MHC molecule by approximately 35 degrees and varies also in its roll (range, 19 degrees) and tile (range, 30 degrees) (197a).

The interactions that position the TCR in a diagonal orientation also place the CDR1 and CDR2 loops over the α₁/α₂ or α₁/β₁ helices for the MHC class I and class II molecules, respectively. The Vα domain is critical in setting up these orientations and the read-out of the peptide sequence. Thus, the CDR1 and CDR2 interactions of Vα are conservative (187) and provide the basic affinity of the TCR, whereas the CDR3 loop is positioned to primarily contact the peptide in the peptide-binding groove and the Vβ interaction is more variable in the C-terminal half of the peptide. The contribution of the individual CDR loops to the interaction varies. CDRs 1β and 2β actually contribute little or nothing to the interaction in those studied. In general, the CDR3 loops are centrally located and usually dominate the interactions (195).

The contributions of the peptide to the interactions with the TCR also vary. Usually, approximately two to five side chains of the peptide make direct contact with the TCR. These contact points, known as hot spots, are peptide residues that bulge out of the groove, and this bulging is more prominent in MHC class I and is sometimes profound (198). In MHC class II–TCR interactions, the contributions of the side chains of the peptide are more uniformly dispersed (P₁, P₂, P₃, P₅, P₈), and the peptides are slightly deeper in the MHC binding groove (198).

No single contact dominates the TCR–pMHC interactions, as is often observed in the antigen–antibody interactions. A small number of amino acids dominate the energy landscape in antigen–antibody interaction, as shown by somatic mutations that result in higher-affinity binding (199).

Based on these considerations, it appears likely that the CDR1 and CDR2 loops interacting with the helices of the MHC are responsible for positive selection, whereas the CDR3 loops probably play a more important role in negative selection. As an extension of these considerations in delineating the roles of V-region loops of the TCR, the CDR1 and CDR2 loops provide the basic affinity in the interaction, whereas the CDR3 provides the specificity (195). The residues of the peptide that protrude highest from the groove provide the basis for discrimination of peptide and for altering affinity or half-life of the TCR–pMHC interaction.

An important point that came out of these crystallographic studies is the role of the water that fills the TCR–pMHC interfaces. This water provides additional complementarity by filling the cavities in the interface, and some of these molecules mediate contact between TCR and pMHC (200).

Overall, the consistent feature of TCR–pMHC interaction is that the peptide contributes a smaller portion of the binding interface (21 to 34%) and a smaller proportion of contacts (26 to 47%) than the MHC. The central positions of the peptide play the critical role, and those define the peptides as agonists, partial agonists, and antagonists (201). The binding of TCR to the pMHC results in T-cell activation. Usually, there is a broad correlation between affinity, half-life, and the functional outcomes (202). Mutational analysis of the role of the centrally located residues indicates that in some systems, the biologic activity increases (203), yet in others, the peptide is converted from an agonist to an antagonist (204); nevertheless, the affinities of the pMHC for TCR change only marginally.

Contacts on certain hot spots are very sensitive in changing the functional read-out but are not based on changes in the affinity of binding or half-life differences in the TCR–pMHC complexes. It has been argued that affinities usually have been measured in isolated TCR–pMHC complexes, and true affinity measurements may require the presence of coreceptors and signaling components and need to be measured with cellular assays (204).

Because changes in the TCR–pMHC complementarity interface are important for triggering biologic reactions, the conformational changes that have been observed in the TCR-complex crystal structures have been considered for initiation of signaling, but their role is not clear (205). Conformational changes contribute to an increase in the binding of a number of other peptides (i.e., an expansion of the repertoire) (206).

Coreceptors in Peptide–Major Histocompatibility Complex Interaction

The TCR is not alone in its interaction with the pMHC but is associated with coreceptors (i.e., CD4, CD8, and the CD3 chains). The monomorphic CD3 γ-, δ-, ∊-, and ζ-chains, together with the αβ heterodimer, form the TCR complex. The CD4 or CD8 acts as assistants to the TCRs of the helper or cytotoxic function of the cells, respectively. Therefore, they have been known as coreceptors (112). In the current model, the CD4 binds to the same MHC II as the TCR of the CD4⁺ T cell, and similarly, the CD8 binds to the same MHC I as the TCR of the CD8⁺ T cell. The binding of the coreceptors occurs with another site of the MHC molecule than that involved in the TCR binding. The TCR binds to the pMHC surface at an angle between 45 and 80 degrees relative to the axis of the two α helices of MHC (207), which excludes the possibility for direct association of the coreceptors with the TCR that binds the same pMHC. One possibility is that the coreceptors could be linked with TCR in the cytoplasm through signaling molecules (i.e., ZAP-70 and Lck) (208). Another possibility is that the coreceptor associated with a TCR binds to a different pMHC to which a second TCR binds, forming a pseudodimer (209).

The CD8 acts as a dimer that includes either two α-chains or one α- and one β-chain, whereas CD4 is a monomer with four Ig-like domains. Both CD8 chains consist of an Ig-like V domain and a long mucinlike stalk. The Ig-like domain binds to the α₃ domain of MHC (210), forming also some hydrogen bonds with the α₂ domain and the β₂M-chain, away from the peptide interface. In CD4, only the fourth Ig-like domain binds directly to pMHC. The ternary complex (TCR-pMHC-CD4/CD8) within the T-cell–APC interface of the immunologic synapse has a V shape (211) (Fig. 16.7). Although the CD8α stalk is longer than the CD8β-chain, even in its full extension, it reaches only 50 to 60 Å, which is not long enough to traverse the distance of approximately 100 Å to the TCR–pMHC complex. As a result, the TCR–pMHC has to tilt for the CD8 to reach the MHC. The stalk of the CD8 chains is heavily glycosylated, and changes in glycosylation (which occurs after T-cell activation) result in a decrease of binding to MHC (212).

Figure 16.7. T-cell receptor (TCR)–major histocompatibility complex (MHC)– coreceptors interaction. The CD4 and CD8 are coreceptors of the TCR (i.e., “assistants” in the interaction of TCR with peptide–MHC complex [pMHC]). In this function, the N-terminal domain of CD4 (D1) makes contact with the α2 and β2 domains of MHC II. In the CD8, the CD8α subunit of the CD8αβ heterodimer contributes the binding energy, interacting with both the α2 and α3 domains of pMHC I, and CD8β binds only to α3. Therefore, both coreceptors recognize different sites of MHC from the TCR. The length of each receptor is not long enough to reach the MHC binding site, traversing alongside the TCR-MHC complex. It is proposed that TCR-MHC has to tilt for the coreceptor to reach the MHC, forming a V-shaped ternary complex. The CD3 components of the TCR (γ, δ, ∊) probably are located inside the V-shaped structure. APC, T cell–antigen-presenting cell. (Courtesy of Dr. G. F. Gao.)

The coreceptors enhance TCR signaling by strengthening the stability of the TCR–pMHC complex (213). The V-shaped ternary complex accepts that the coreceptors bind to the same pMHC as the TCR (211), whereas the CD3 components probably lie inside the “open angel” of the V-shaped structure (Fig. 16.7). Such an arrangement makes possible the association of signaling molecules, such as ZAP-70, Lck, and Src kinases, with CD4 and CD8. With this topologic model, direct interaction of the coreceptors with TCR is not likely as was previously indicated (214).

The CD2 binds to CD58 (LFA-3) in humans based on electrostatic complementarity, with the CD2 surface heavily populated by basic residues, whereas the CD58 is acidic. The interactions span approximately 134 Å—very similar to that of TCR–pMHC. Therefore, the CD2/CD58 interaction in the contact zone between T cell and the APC would facilitate the scanning of pMHC by TCR and lowers thresholds for TCR triggering in vitro and T-cell activation in vivo (215).

CD28 and CTLA-4 (CD152) are type I membrane proteins consisting of one moderately to heavily glycosylated V Ig-like domain and are expressed as disulfide-linked homodimers. Their counterreceptors B7-1 (CD80) and B7-2 (CD86) consist of two Ig-like domains, a membrane-proximal C2 type, and a membrane-distal V type. B7-2 binds CD28 more effectively than CTLA-4 and, as a result, enhances costimulatory effects when CD28 and CTLA-4 are coexpressed.

In contrast, B7-1 binds preferentially CTLA-4, and its inhibitory effect would not be affected in the presence of CD28. Delayed expression of B7-1 on APCs appears to be timed to enhance the inhibitory function of CTLA-4. The CTLA-4 periodic arrays of crystal lattices enhance the avidity of interaction, whereas the CD28/B7-2 interaction does not have this potential. As a result, during an immune response, the CD28/B7-2–activating complexes are 10,000-fold less stable than the inhibitory complexes formed later by CTLA-4/B7-1 (216).

Signal-Transducing Component of T-cell Receptor Complex

The TCR-αβ is accompanied by five other polypeptide chains, collectively known as CD3 proteins: γ, δ, ∊, ζ, and η. They form disulfide-linked heterodimers γδ, ∊δ, ζη, or a homodimer, ζζ (217) (Fig. 16.5). The γ, δ, and ∊ proteins show a significant degree of similarity to one another and consist of an Ig-like domain similar to a C domain with an intrachain disulfide bond. The extracellular region of the ζ-chain is only nine amino acids long and contains the only cysteine of the molecule, which forms the disulfide bond with another ζ chain or with an η-chain. In the transmembrane region, the γ, δ, and ∊ proteins have a negatively charged amino acid complementary to a positively charged amino acid of the transmembrane region of the TCR-αβ chains. The cytoplasmic regions of γ-, δ-, and ∊-chains are long, ranging from 40 to 80 amino acids, whereas that of the ζ-chain is even longer with 113 amino acids. The η-chain is a splice variant of the ζ-chain and, as a heterodimer with the ζ-chain, exists only in a small number of T cells.

The CD3 proteins have a dual mission in the function of the TCR: Escort the receptor to the cell membrane and mediate the signals generated by the TCR–pMHC complex. The complex is assembled in the endoplasmic reticulum, transported to the Golgi apparatus, and then transferred to the plasma membrane. The amount of the ζ-chain is rate limiting because it is synthesized at only 10% of the level of the other chains, which results in the degradation of a vast majority of the newly synthesized α, β, or CD3 components within 4 hours of their synthesis. The remaining undegraded chains are long lived and form complete TCR–CD3 complexes with the limiting ζ-chain. The TCR–CD3 complex lacking the ζ-chain migrates through the endoplasmic reticulum and Golgi intact and then is transported to the lysosomes, where it is degraded. A lysosome-targeting motif has been identified in the δ- and γ-chains and consists of a dileucine-based motif (DKQTLL) and tyrosine-based motif (218) in the carboxyterminal region. In the completed complex, the TCR α-chain pairs with CD3 δ- and ∊-chains, and the TCR β-chain pairs with CD3 γand ∊. The ζ-chain joins the TCR and the CD3 chains in the last stage of the assembly. The topology of the TCR–CD3 complex is shown in Figure 16.5. Two TCR αβ heterodimers are associated with one each of γ∊ and δ∊ heterodimers and one ζζ homodimer. The signal transduction function of the CD3 proteins is based on the presence of one ITAM in each of the γ, δ, and ∊ proteins and three in the ζ-chain. An ITAM consists of two YXXY/L sequences separated by six to eight amino acids (in the one-letter code for amino acids: Y, tyrosine; L, leucine; and X, any amino acid) (219).

Phosphorylation of the tyrosines turns the ITAM motifs into docking sites for protein tyrosine kinases, which bind to the ITAM through their SH₂domain (see below). An important protein tyrosine kinase in T-cell activation, the ZAP-70, is recruited to the ζ-chain ITAM motifs. The multiplicity of ITAM motifs in the cytoplasmic tails of the CD3 proteins results in signal amplification and increases the sensitivity of the TCR to ligand stimulation. Triplication of a single ITAM motif significantly enhances Ca²⁺ mobilization, association with ZAP-70, and transcriptional activity in the NFAT complex involved in IL-2 gene regulation (220). Cross-linking of a single isolated ITAM results in approximately threefold induction in NFAT-regulated activity, and cross-linking of a triplicated motif results in approximately eightfold increase in NFAT-regulated activity (i.e., comparable to the intact ζ-chain).

Coreceptors to T-Cell Receptor: CD4 and CD8

CD4

CD4 is the characteristic marker of αβ T cells with helper activity and cytokine secretion. The molecule consists of four Ig-like domains, with domains D1 and D3 similar to a V domain, and D2 and D4 similar to C domains (C-2 type) with patches of sequences similar to V domain. In D1 and D3, the nine β strands form two β sheets (ABED and GFCC’’’). The D1 shares all the features conserved and characteristic of the V antibody domain. In the D3, the intersheet disulfide bond and the salt bridge are absent. As a result, the two sheets move apart and “slide” relative to each other. D2 also has no intersheet disulfide bond. The domains are linearly arranged, forming a rod with limited flexibility between the domains (221). The binding of CD4 to MHC II molecules and to HIVgp120–envelop protein was studied by mutational analysis. The binding is mediated by a sequence of two β strands, C′C″, within the D1 domain that form a ridge (residues 35 to 46) (222). Phe43 within this sequence is critical for binding to gp120, as well as for CD4 coreceptor function (223) because it provides the major binding energy, whereas the surrounding charged amino acids facilitate specificity (224). Crystal structure of the D1/D2 domains from CD4 complexed with class II molecule shows that the CD4 N-terminal V domain is directed and reaching into the two membrane-proximal domains of MHC II molecule (225). Both TCR and CD4 are tilted rather than oriented vertically, forming a V-shaped CD4-pMHCII-TCR ternary complex (Fig. 16.7).

In this complex, the antigen-binding groove of pMHC II is no longer parallel to the cell surface but makes an approximately 45-degree angle with the membrane. However, despite the V shape of the ternary complex, the membrane-proximal domains of each of the components are all roughly vertical, including domain 4 of CD4 and the α₂ and β₂ domains of pMHC II as well as the Cβ domain of TCR. Only the Cα domain of TCR hangs almost parallel above the membrane with its lengthy stalk bridging the space.

The extracellular fragment of CD4 crystallizes as a dimer associated via the D4 domains (226). Oligomers of CD4 have been extracted from isolated lymphocytes (227). The CD4 dimers are associated with superdimers or dimers of dimers of the MHC II molecules (228). Each CD4 molecule interacts with one dimer of MHC II.

Amino acid substitutions in the faces of the MHC II molecules that participate in the formation of the superdimers block activation of CD4⁺ T cells, implying that superdimer formation is a prerequisite for T-cell activation (228). Stable binding to MHC II requires oligomerization of CD4, and this is facilitated by the D3/D4 domains. Collectively, data gathered from a variety of approaches suggest that CD4 oligomerizes once it binds to one MHC II molecule, forming tetramers or even larger oligomers and cross-linked lattices (229). In this process, CD4 oligomerization is influenced not only by MHC II, but also by the TCR.

When CD4 and TCR colocalize to interact with the same pMHC II molecule, the CD4 brings p56^lck, which is associated with its short cytoplasmic tail, to the site of immune recognition. This may constitute one of the main contributions of CD4 to TCR signaling (230).

CD8

CD8 consists of an Ig-like ectodomain, a membrane-proximal stalk (or hinge) region, a transmembrane region, and a cytoplasmic region. The Ig-like domain is involved in the binding to MHC, as shown by crystallographic evidence (231) and mutational analysis (232). The CDR loops of the A and B strands of the CD8αα molecule contact the α₂, α₃, and β₂M domains of MHC I (233). The β subunit of the CD8 interacts only with the α₃domain, whereas the α subunit interacts with the α₂ and α₃ domains (234). In these interactions, the α₃ domain is shifted to better accommodate the CD8 binding (235). Multiple contacts between the coreceptor and MHC I promote the functional contributions of the coreceptor to T-cell activation (236). CD8 enhances cytotoxic T lymphocyte (CTL) activation by enhancing the stability of interaction between the APC and the T cell (237).

In addition, the α-chain of the CD8 is associated with the p56^lck, which is brought closer to the ZAP-70 kinase associated with the TCR, a role that has also been assigned to the CD4 coreceptor. The contributions, therefore, of both coreceptors in ligand recognition and enhancement of TCR signaling are multiple and involve facilitation of TCR clustering, stabilization of the TCR–pMHC complex, and promotion of signaling by bringing together signaling molecules attached separately on the cytoplasmic tails of the TCR and coreceptors (238). The availability of pMHC tetramers has encouraged studies on the function of CD4 and CD8 coreceptors in TCR binding to pMHC. The CD4 is critical for signal transduction with pMHC tetramers (239,240,241), but the CD8 coreceptor also contributes to the initial phase of interaction (binding), the duration of interaction (stability), and the delivery of signal transduction (242,243).

The final result of CD8 function depends on the epitope of CD8 involved in the interactions, which are more pronounced with low-affinity ligands (264). Blockade of CD8β may affect TCR–CD8 rather than CD8–MHC interaction. The CD8β is more efficient than CD8α at association with the TCR (244).

An important new function of the coreceptors in TCR-mediated activation is the demonstration that CD4 and CD8 associate with LAT (linker for activation of T cells), a 36- to 38-kD membrane-associated adaptor protein that plays a central role in TCR signaling (245,246). As a result of LAT association with surface coreceptors and the coengagement of the TCR with the coreceptors, LAT is phosphorylated and recruits downstream signaling molecules (247). In conclusion, the contribution of the coreceptors in TCR signaling results from (a) the physical approximation of the Lck, which is associated with the coreceptors; (b) phosphorylation of the ITAMs of the CD3 chains of TCR; (c) recruitment of ZAP-70; and (d) phosphorylation of LAT, also associated with the coreceptors. Individual TCR molecules are probably associated with either Lck or with LAT (247).

T-Cell Activation

Topology of Immune Recognition

The αβ-TCR recognizes short peptides (i.e., eight to ten residues long) for class I–restricted TCRs or 13 to 25 amino acids for class II–restricted TCRs. However, only approximately nine amino acids contact the TCR. The peptide is held by the MHC molecule embedded in a groove formed by the α₁-α₂ domains of MHC I or by the α₁-β₁ domains of the MHC II (see Chapter 17). Interaction of TCRs initiates signal transduction leading to transcription of genes encoding cytokines in CD4⁺ T cells or assembling and mobilizing the lytic machinery in CD8⁺ T cells for killing cells infected by viruses, transformed to a malignant state, or being “strangers,” such as in transplanted tissues. The crystal structure of αβ-TCR has been solved and, for several of them, in complex with their cognate pMHC ligand (236,248,249,250,251,252). The data support the docking model of the TCR ligand interaction, in which the TCR approaches the peptide held on an MHC platform. In this model, the Vα domain of the TCR is closest to the N-terminal residues of the peptide, whereas the Vβ domain of the TCR is closest to the C-terminal end of the peptide. In this orientation, the TCR Vβ domain contacts the MHC I α₁ domain, whereas the TCR Vα domain interacts with the α₂ of MHC I. The TCR orientation relative to the long axis of the MHC platform varies between 45 and 80 degrees (i.e., the angle formed between a line passing through the centers of the Vα and Vβ domains of the TCR and a second line defined by the peptide on the MHC platform) (253). The peptides in all MHC I structures have their N- and C-termini anchored into two fixed pockets approximately 20 Å apart in the MHC I platform. Longer peptides usually bulge in the center. In MHC II, the peptides assume an extended conformation, and the middle portion is smooth and concaves away from the TCR. Residues 1 and 5 of the peptide (p-1, p-5) in MHC II point toward the TCR and are critical in TCR recognition. The coreceptors bind to the same pMHC ligand as the TCR, but the binding of the coreceptor is independent of the TCR binding and is severalfold weaker (230,254).

The TCR, the pMHC, and the coreceptor form a ternary complex that assumes a V shape with the pMHC II at the apex and the TCR and CD4 as the two arms.

Overall, the data from crystallized complexes of TCR with pMHC suggest that TCR docking on the MHC platform involves residues that are conserved within the CDR1 and CDR2 of the TCR and provide a basal level of stabilizing energy in the formation of the complex (255). Variations within the CDR3α and CDR3β loops affect the way the TCRs footprint on the pMHC surface.

The Gathering Storm: Lipid Rafts and the Immunologic Synapse

A model for T-cell activation proposes that a naïve or resting T cell needs to cross several stages to achieve full activation (256). During the first stage, which lasts a few seconds, contacts need to be established by adhesion molecules that overcome charge repulsions between cells. In the next stage, active cytoskeletal rearrangements bring together accessory molecules and the TCR and concomitantly exclude unligated molecules, such as CD45 (a protein tyrosine phosphatase) as well as CD43. CD43 is a highly glycosylated protein with strong repulsive forces forming a thick glycocalyx “cloud” approximately 45 nm thick around the cells. Because the distance spanned by the TCR–MHC complex is a mere 15 nm, for TCR to sample the pMHC ligand, clearing the area of contact between TCR and MHC from the interfering tall molecules, like CD43 and CD45, is mandatory. This takes minutes. Finally, the fundamental signaling units of several signaling molecules are “loaded” on a lipid raft, sustained by the cytoskeleton. This takes hours.

Lipid rafts are evolutionarily conserved structures that gather receptors involved in signaling in various cell types (257,258,259,260). The lipid rafts are defined as cholesterol-dependent microdomains resistant to solubilization in nonionic detergents at low temperatures (261). The lipids contained within the cellular plasma membrane include glycerophospholipids, glycosphingolipids, and sterols. Rafts consist of sphingolipids and cholesterol, which can move through the more liquid-disordered phase of the membrane containing glycerophospholipids (262). Because of their lipid consistency, lipid rafts have also been referred to as glycosphingo-lipid-enriched membrane microdomains or GEMs. Lipid rafts are anchored by filamentous actin, and actin polymerization causes their coalescence. Phosphatidylinositol-(4,5) biphosphate (PIP₂) activates proteins involved in actin–membrane interactions, such as binding of the ezrin-radixin-moesin family to CD44, a protein broadly expressed on detergent-resistant membranes. Detergent-resistant membranes are presumed to be isolated rafts. PIP₂ activates actin nucleation and polymerization of raft domains. Raft size is poorly understood because they cannot be visualized by light microscopy without clustering (263). In resting T cells, the TCR and GM1 are homogeneously distributed. Cross-linking, however, produces large enough aggregates to be visualized using epifluorescence microscopy. Fluorescence resonance energy transfer is sensitive to distances in the order of a few nanometers. Because fluorescence resonance energy transfer was detected between glycosylphosphatidylinositol (GPI)-anchored transferrin receptors interacting with natural ligands, it was concluded that rafts exist but that they are smaller than 70 nm in diameter (263). Others could not find microdomains enriched in GPI-anchored proteins or GM1. Some MHC II molecules loaded with a select set of peptides are detected on microdomains made up of tetraspan proteins, such as CD9, CD63, CD81, or CD82 (261). These tetraspan microdomains are recognized by CDw78 antibody (mAb FN1), which originally was believed to recognize an epitope of HLA–class II molecules.

A central feature of lipid rafts is that they allow for the lateral segregation of proteins within the plasma membrane. This provides a mechanism for compartmentalization of signaling components within the membrane, concentrating certain components in lipid rafts and excluding others (Fig. 16.8). Proteins with GPI linkage, such as CD14, CD16, CD48, and CD58, are associated with the outer leaflet of lipid rafts. Cytoplasmic proteins, on the other hand, associate with the inner leaflet of lipid rafts through acylation. Most of the Src signaling proteins are acylated and raft associated. The vast majority of transmembrane proteins are excluded from rafts constitutively and cannot be induced to partition into rafts on cross-linking.

Some proteins important in T-cell activation, such as LAT, CD4, and CD8, reside on rafts. Other proteins reside constitutively outside of the rafts but, when activated, become translocated to rafts. The multichain immune recognition receptor family is an example (264,265). Translocation into rafts after ligand binding appears to be immediate, occurring within seconds of engagement of the receptors, and is selective. For example, CD45, α₄ integrin, and IL-1 receptor are excluded from rafts and do not translocate into rafts, even on cross-linking (266,267). The transmembrane domains of the receptors appear to have a significant influence on translocation, as exchange or mutation of the transmembrane regions alters the translocation properties of the receptors. Translocation of the receptor into the lipid rafts does not need actin reorganization. However, stable residency into the rafts depends on interaction with actin cytoskeleton.

The constitutive presence of receptors in rafts may be related to their role in cell survival. For example, the pre-TCR is constitutively associated with rafts and, with its signaling, instructs lin-eage commitment. TCR/CD28 is excluded from rafts in immature thymocytes, and signaling leads to apoptosis (268).

In addition to their role in signaling, lipid rafts serve as platforms for B-cell receptors (BCRs) to transport bound antigen for processing and presentation. The internalization of BCRs with captured antigen is initiated from rafts.

Figure 16.8. Rafts: Elementary units of the immunologic synapse? In the resting state, the multichain immune recognition receptors (MIRR) (i.e., T-cell receptor, B-cell receptor, and so forth) “float” free on the lipid membrane, or transiently, some may get on board on certain membrane “microdomains,” also known as “lipid rafts.” Interaction with a multivalent ligand aggregates several receptors that accumulate within the rafts, forming multimolecular clusters. On prolonged exposure to the ligand, the MIRR clusters result in raft coalescence with the formation of a large aggregate, the immunologic synapse. Cytoskeletal reorganization from multichain immune recognition receptor and chemokine signaling promotes immunologic synapse formation. (Adapted from a model proposed by Dykstra M, Cherukuri A, Sohn HW et al. Location is everything: lipid rafts and immune cell signaling. Annu Rev Immunol 2003;21:457–481.)

Retention of receptors in rafts is a mechanism for augmentation of cell activation. The CD19/CD21 complex is excluded from rafts, but its colligation with BCRs by antigen–antibody–complement complexes causes translocation of BCR and CD19/CD21 complex into rafts, where the complex prolongs the retention of BCR within the rafts (269).

T-cell activation causes rearrangement of the actin cytoskeleton and polarization of the cell toward the site of activation, such as the APC. This polarization is reflected by reorientation of the microtubule-organizing center toward the APC (270,271). The TCR is clustered together with associated molecules (CD4, CD8, CD2, CD28, and so forth), a process known as capping. Capping depends on actin reorganization regulated by phosphoinositides, which activate Vav (a guanosine diphosphate–guanosine triphosphate [GTP] exchange factor) and the Wiskott-Aldrich syndrome protein (WASp). In patients with Wiskott-Aldrich syndrome, WASp is lacking or markedly reduced, and they have defects in actin polymerization, capping, and antigen-induced proliferative responses (272). WASp regulates the Arp2/3 complex, which mediates actin branching and polymerization (see Chapter 13).

PIP₃ is a strong activator of Vav, which is targeted to detergent-resistant membranes after TCR activation. In Vav-deficient mice, the capping is severely disrupted, and T-cell proliferation is reduced. Raft aggregation is disrupted by the negative regulator of T-cell activation, Cbl (Casitas B-cell lymphoma-b), which is a molecular adaptor and part of the ubiquitin ligation machinery involved in the degradation of phosphorylated proteins. Cbl inhibits TCR clustering and sustained tyrosine phosphorylation. In Cbl-deficient mice, TCR/CD3 stimulation alone can activate receptor clustering without the need for CD28 costimulation (273).

Immunologic Synapse in Three Stages

The TCR interactions with pMHC take place in an intercellular junction between the T cell and the APC. In this junction, signal 1 (TCR) and signal 2 (costimulation) are processed. This interface reveals a dramatic reorganization of signaling components, forming what is called the immunologic synapse (IS), a term borrowed from Sherrington’s turn-of-the-century definition of the interconnections of neurons as synapses (from the Greek synapsis, meaning joining, linking, connecting) (274). The IS relays information across the cell junction in both directions (275).

The IS is organized into two major compartments: The central supramolecular activation cluster (cSMAC), enriched in TCRs and CD28, and the peripheral supramolecular activation cluster (pSMAC), which contains the LFA-1 molecule and talin (276) (Fig. 16.9). On the side of the APC, correspondingly, the cSMAC contains the pMHC and CD80 (ligand for CD28) and the pSMAC contains the ICAM-1 (counterreceptor for LFA-1) and CD58, ligand for CD2. The IS develops over a period of minutes after interactions of the T cell and the APC. Formation of the synapse depends on an intact cytoskeleton (277). T-cell activation is accompanied by a dynamic reorganization of cortical actin with increase of filamentous actin. These cytoskeletal changes are accompanied by progressive morphologic changes of the T cells, which first become round, followed by spreading of the cell (278). In the absence of antigen, the T cell maintains its motility, continues to crawl around the APC, and may even leave for another partner. Some receptors on the APC may convey a “danger signal” for the T cell to pay particular attention and explore the APC in search of antigen (279).

In the first stage, the contact is antigen independent and is mediated by CD28 on the T cell with CD80/CD86 on the APC, which are more abundant (>10⁴) than the pMHC (approximately 100 to 200). The CD28 affinity for CD80 is at least two orders of magnitude above that determined for the TCR–pMHC interaction. These interactions of CD28 preceding the TCR encounters are actually contrary to the original definition of the costimulatory function of CD28, believed to parallel or even follow TCR signaling.

The second stage of IS formation is antigen dependent. The T cell extends large cytoplasmic, pulsatile protrusions toward the APC (280,281). Tyrosine phosphorylated Vav-1 and tyrosine phosphorylated SLP-76 assemble with the p21-activated kinase (Pak) via the adaptor protein Nck (282,283,284).

The T cell–APC complex is stabilized in the third stage, which is regulated by increases of intracellular Ca²⁺. By the end of the third phase, SMAC is in place with all the receptors and the signaling molecules, held together with clusters of GEMs (lipid rafts) on the surface and an elaborate cytoskeleton scaffolding underneath.

The kinetics of the IS from its early beginnings are coordinated and organized by the cytoskeleton (277). The TCR–pMHC can only interact at a distance of 15 nm, which is significantly below the thick glycocalyces of the two cells that separate their membranes by a 50- to 100-nm distance. Adhesion molecules, such as L-selectin, located on the tip of microvilli, may very well initiate the T-cell–APC interaction, until LFA-1, lying on flat surfaces of the membrane, is released from its inhibitory state by activating signals delivered by chemokines.

These signals also result in the formation of myosin II thick filaments, which disrupt and pull the thick network of polymerized actin away. In this clearing, new actin polymerization pushes forward new filopodia and lamellipodia (i.e., the cell becomes motile). Now that the T cell is polarized, the long interfering molecules of the glycocalyx, such as CD43 and CD45, are pulled to the rear end, or uropod, of the cell.

In the meantime, activated high-affinity LFA-1 released from inhibition stabilizes the contact area between the cells, moving laterally and forming strong bonds with ICAM-1 across the cell junction (285). In the membrane clearing created and stabilized by LFA-1 adhesion, the TCRs sample MHC on the APC surface for complementary peptides. It has been noted that paradoxically, a large cluster of TCRs has been pulled to the rear of the cell but, through the mediation of myosin II, is brought to the front, reinforcing the frontal cluster (286). Nascent IS is further stabilized by TCR links with the cytoskeleton. These links are mediated by some of the components of CD3 (i.e., ζ-chain), which, through its phosphorylated ITAMs, induces actin polymerization (291).

If pMHCs are present on the APC surface, the T cell stops moving (281,287). The central area of T cell closest to the APC includes the bulk of the TCRs, which are surrounded by the further away integrins (276,288).

The final arrangement defines the mature IS or the bull’s eye (276) (i.e., the cSMAC, a central area 1 to 3 μM in diameter that contains TCR, CD28, and CD2, surrounded by pSMAC, an adhesion ring that contains the LFA-1 and talin). Formation of mature IS with APCs (instead of artificial lipid bilayers) shows that the TCR signaling precedes the completion of the mature IS (289). Bull’s eye IS also has been observed with CD8⁺ T cells during recognition and killing of target cells (290). Granule secretion occurs after the microtubule organizing center (MTOC) polarization in cSMAC, where membrane fusion occurs. Cellular functions between T cells and the target cells have been demonstrated by transmission and scanning electron microscopy several years ago. These junctions are followed by disruption and blebbing of the target cell membrane (291).

Synapses with dendritic cells are formed even in the absence of antigen or MHC (292,293). Encounters with dendritic cells are relatively short as compared to B cells (294). This may be due to chemokine secretion by dendritic cells, which stimulate T-cell migration. There also seems to be other differences between dendritic cells and B cells, such as the length of time of the encounter, which is short for dendritic cells but longer for B cells. The role of the dendritic cell in the formation of IS is active with full involvement of its cytoskeleton, whereas the B cell remains passive and the T cell makes the major contribution (295). The functional consequences of the formation of a mature IS are believed to be primarily related to polarized secretion (296,297) and signaling (298). IS may help to retain secreted substances close to the targeted cell. Although this is true for the synapses formed by killer lymphocytes such as the CD8⁺ cytotoxic T cell or the NK cell, it may not apply to the CD4⁺ T cell, which uses the APC to pull the trigger for its activation. In this case, the IS concentrates crucial molecules, such as CD28, that provide costimulatory signals and enhance TCR-mediated signaling (299,300).

Figure 16.9. The structure of the immunologic synapse. The immunologic synapse is formed between T cell and an antigen-presenting cell (APC) (in the case of CD4+ T cell) or with a target cell (in the case of CD81 T cell). Immunologic synapse formation is a multistep process. A thick glycocalyx on both cells, made predominantly from the mucin molecule CD43, comes in conflict with the approaching T cell and APC. This distance (50 to 100 nm) is too long for the T-cell receptor (TCR)–peptide-major histocompatibility complex (MHC), which interacts at 15 nm. The integrin lymphocyte function molecule (LFA)-1 and its counterreceptor, intercellular adhesion molecule (ICAM), which interact at 40 nm, may bring the cells for an initial contact. More important, chemokine signaling that activates heterotrimeric G proteins activates myosin II, and the cortical cytoskeleton collapses, disanchoring CD43 by the ERM (ezrin/radixin/moesin) adaptor proteins. With loss of the cell rigidity, a new F-actin network creates a pseudopod that propels the leading edge of the T cell toward the APC. This approach at an intercellular distance of 15 nm prevents CD43 re-entry in the central area of T cell–APC contact. Concomitantly, talin (a large cytoskeletal protein with attachment to integrins) maintains LFA-1 immobilized in a ring around the central part of the synapse. Although LFA-1/ICAM-1 interacts initially at a distance of 40 nm (extended LFA-1 form), after activation, it assumes a bent form (high affinity) that brings the cell membranes closer (“ratchet”-like effect). Multiple other adhesion molecules of low affinity, such as between CD2 and CD58, contribute to the alignment of the two cell surfaces at 15-nm distance, allowing the TCR to sample the small numbers of peptide–MHC (“proofreading”). The final mature immunologic synapse (“bull’s eye”) consists of a central supramolecular activation cluster and the peripheral integrin-rich zone. (From Dustin ML, Cooper JA. The immunological synapse and the actin cytoskeleton: molecular hardware for T cell signaling. Nat Immunol 2000;1:23–29; and Delon J, Stelon S, Germain RN, et al. Imaging of T-cell interactions with antigen presenting cells in culture and in intact lymphoid tissue. Immunol Rev 2002;189:51–63.)

Table 16.1 T-Cell Receptor-αβ Activation Machinery

Receptors/Ligands Antigen Recognition Signal Transduction Adhesion Ligand/Counterreceptors α/β + – – Major histocompatibility peptide complexa ´/γ–´/δ – + – — ∊/γ – + – — CD4 – + – HLA class II CD8 – + – HLA class I CD2 – + + CD58 (LFA-3) CD28/CTLA-4 – + – CD80/CD86 CD40L – + – CD40 CD11a/CD18 (LFA-1) – + + CD54 (intercellular adhesion molecule-1) LFA, lymphocyte function molecule. aSee text for details.

T-Cell Receptor Signaling

Two models have been proposed for sustained T-cell signaling: (a) serial engagement (migration), when the T cell migrates from one APC to the next and thus renews its signaling capacity (294); and (b) signaling based on formation of IS. Both models achieve similar results but require different molecular mechanisms. The mature IS forms a specialized mode of signaling and enables T cells to remain responsive to antigen while still with the initial APC. For a better understanding of the complexities of signaling, we organize it into three phases: (a) initiation, (b) generation of phosphoinositides, and (c) the Ras pathway (Fig. 16.10, Table 16.1).

Initiation Phase

Signaling is initiated by activation of Lck, which is regulated by two tyrosines: Tyr 394 in the activation loop and Tyr 505 in the C-terminus. Lck is kept inactive or “closed” by two intramolecular bonds: One is between Tyr 505, which is phosphorylated by the C-terminal Src, Csk, and binds to the SH₂ domain of Lck; a second bond is formed between the SH₃ domain of Lck and a sequence connecting the SH₂ and the kinase domains. For activation of Lck, the Tyr 505 needs to be dephosphorylated by CD45, a protein tyrosine phosphatase (301), whereas Tyr 394 is autophosphorylated and activates the kinase domain. The large-size CD45 isoforms are excluded from the IS (302,303), but some move back to cSMAC adjacent to the TCRs (304). Lck, recruited by CD4, is maintained in the activated state by CD28 (300) and phosphorylates the ITAMs of the ζ-chain of TCR in a sequential and ordered manner, establishing thresholds of T-cell activation (305). This mechanism determines whether a sufficient number of tyrosines are phosphorylated for full activation and supports the kinetic proofreading model of T-cell activation, which examines the relationship between kinetics of TCR–ligand interaction and intensities of T-cell activation (306,307,308,309).

ZAP-70 is recruited to the phosphorylated ITAMs of the ζ-chain and in turn activates the adaptor protein LAT, which then is localized in the rafts. LAT has a short extracellular and long intracellular region and possesses a central position in T-cell activation because it assembles other adaptor molecules and signaling proteins.

There are two groups of adaptor proteins: Transmembrane adaptor proteins and cytosolic adaptor proteins (310). LAT (a transmembrane adaptor protein) is located in the rafts and is phosphorylated by ZAP-70. As a result, it recruits PLC-γ1, phosphoinositide 3-kinase (PI3-K), IL-2–inducible T-cell kinase (Itk), adaptor proteins Grb2 and Gads, and, indirectly, Vav and Slp-76 (311,312,313). SLP-76 is a cytosolic adaptor protein that has three protein-binding motifs and plays an essential role in signaling pathways required for IL-2 secretion (314). It is expressed on thymocytes, T cells, mast cells, NK cells, and platelets. Through Gads, it binds indirectly to LAT after TCR ligation. So LAT and SLP-76 function as mutually dependent intermolecular scaffolds, together recruiting crucial signaling regulators to sites of raft aggregation. In mice deficient in SLP-76 (or LAT), thymocyte development is arrested at the stage at which the TCR β-chain is coupled to the pre–Tα-chain. SLP-76 recruits Itk to lipid rafts and allows for optimal phosphorylation of PLC-γ1, which also associates with ZAP-70.

Phosphoinositide Metabolism

Phosphoinositides are produced by the action of PLC-γ1 and PI3-K. PLC-γ1 binds to LAT and is activated as a result of phosphorylation of multiple tyrosines by ZAP-70. PLC-γ1 hydrolyzes inositol phospholipids generating diacylglycerol and inositol (1,4,5)-triphosphate (IP₃). Diacylglycerol contributes to activation of protein kinase C (PKC), whereas IP₃ increases Ca²⁺ released from intracellular sources. There are multiple isoforms of PKC serine kinases, which are regulated by Ca²⁺, diacylglycerol, phospholipids, the classic PKC (α, β, γ), and the novel PKC (δ, ∊, η, and θ) (315). PKCθ is recruited to the plasma membrane and is the only isoform detected in cSMAC together with TCR (316). PKCθ’s targets are nuclear factor-κB activation, IL-2 production, regulation of integrin function, and control of cytoskeleton through association with Vav. Both Ca²⁺ and PKC synergize in the increase in transcriptional activity of NFAT. Ca²⁺ acts through calcineurin, a calmodulin-dependent phosphatase that contributes to induction or function of NFAT. The immunosuppressive drugs cyclosporin A and FK506, bound to their binding protein, immunophilin, inhibit Ca²⁺-mediated signaling of T lymphocytes by sequestering cytosolic calcineurin. This is the mechanism of inhibition of IL-2 secretion.

The PI3-K produces phosphatidylinositol 3,4,5-triphosphate (PI-3,4,5-P₃), which acts as a second messenger. It binds to proteins that contain a pleckstrin homology domain and recruits them to the inner leaflet of the cell membrane (317). PI3-K overall is involved in survival and cytoskeletal signaling processes and is essential for adhesion signals. Important targets for PI3-K products are the GTPases, Rac and Rho, stimulated by GEF (guanine nucleotide exchange) proteins promoting transition from the inactive guanosine diphosphate–bound state to the active GTP-bound conformation. Rac and Rho regulate several functions in the life of the T cell, and during T-cell activation, they regulate cytoskeletal rearrangements. The PI-3,4,5-P₂ product of PI3-K is converted to PI-3,4-P₂ by the phosphatidylinositol-5 phosphatase (SHIP). This product binds to the pleckstrin homology domain of the protein kinase B or Akt.

Figure 16.10. T-cell activation. T cells are activated by two signals: Signal 1 is delivered by the T-cell receptor (TCR) interacting with peptide–major histocompatibility complex (pMHC), and signal 2, or costimulatory signal, by CD28 interacting with CD80/CD86. A number of adaptor proteins (i.e., proteins acting as scaffolding) assemble a supramolecular signaling complex. Foremost among them are LAT (linker for activated T cells) and SLP-76 (SH2-domain–containing leukocyte-specific phosphoprotein of 76 kD). LAT expression is limited to T cells, natural killer (NK) cells, platelets, and mast cells and is not expressed on B cells or monocytes. LAT is a membrane adaptor protein as compared to SLP-76, which is cytoplasmic. Engagement of TCR activates Lck, which is associated with the coreceptor (CD4 or CD8). Lck phosphorylates ZAP-70 (ζ-chain–associated protein). ZAP-70 in turn phosphorylates LAT, which at this point makes the transition between proximal and downstream signaling events initiated by TCR. LAT is also associated with the coreceptor, competing in the binding with Lck. LAT and Lck are linked to individual coreceptors, rather than both of them being linked to the same molecule. LAT as an adaptor protein is a scaffolding that is associated with several downstream molecules: Phospholipase (PLC)-γ generates phosphoinositides and increases Ca2+, which activate protein kinase C (PKC) and calcineurin, respectively. Another cluster is formed with Gads (Grb-related protein), SLP-76, Vav, and so forth that regulates the cytoskeleton together with the PIP3 product of phosphoinositide-3-kinase (PI3-K). The other signaling pathway linked to LAT is through the Ras activation, linking to the activation of MAPK/ERK kinase (MEK) and extracellular signal-regulated kinase (ERK). A central position in T-cell signaling is occupied by the novel PKC isoform, PKCθ, which is selectively expressed in T lymphocytes and is recruited to the immunologic synapse. It induces essential activation signals for interleukin-2 synthesis in cooperation with calcineurin. It is a master inducer of NFκB activation and its translocation to the nucleus. PI3-K associated with CD28 generates PIP3 that recruits Vav and PKCθ to the membrane. APC, antigen-presenting cell; DAG, diacylglycerol; SLPP, serum lipophosphoprotein. (Important information from Bosselut R, Zhang W, Ashe JM, et al. Association of the adaptor molecule LAT with CD4 and CD8 co-receptors identifies a new co-receptor function in T cell receptor signal transduction. J Exp Med 1999;190:1517–1525; Myung PS, Boerthe NJ, Koretzky GA. Adapter proteins in lymphocyte antigen-receptor signaling. Curr Opin Immunol 2000;12:256–266; Koretzky GA, Myzeng PS. Positive and negative regulation of T cell activation by adaptor proteins Nat Rev Immunol 2001;1:95–107, and Cantrell DA. T-cell antigen receptor signal transduction. Immunology 2002;105:369–374.)

Akt moves to the nucleus, where it acts on several substrates regulating cell survival, NF-κB activation, metabolism, and energy generation (318). It also regulates transcription factors (i.e., NFAT and the Forkhead family), promoting cell survival and progression through the cell cycle. All signals from PI3-K products are eventually terminated by the inositol phosphatase SHIP.

Ras Pathway

The adaptor protein Grb2 is a bifunctional molecule having an SH₂ domain with which it binds to phosphorylated tyrosines of LAT, whereas its SH₃ domain binds other proteins that have proline-rich sequences. These proteins are cytosolic, but through association with LAT, they are translocated to the membrane (312). Grb2 functions in many cells and associates with a large number of proteins (319). In T cells, it is associated with the proline-rich domain of SOS, the mammalian homolog of the Drosophila “son of Sevenless” protein, which is a critical activator of the small G protein Ras (320). The Grb2-SOS binding is also mediated by another small linker protein, Shc, while Cbl binding to Grb2 inhibits activation.

The Ras pathway activates Erk kinase and MAPK. MAPK pathway may induce integrin activation, and, in reverse, the integrin may induce MAPK pathway through PI3-K. There are other bidirectional signaling pathways that, in general, send signals to the nucleus and activate membrane molecules by a feedback pathway.

T-Cell Activation and the Cytoskeleton

The cytoskeleton plays an important role in IS formation and T-cell activation. T cells polarize toward their target, focusing the signa-ling on the secretory apparatus at the APC or target cell. Cytochal-asin D, which disrupts the cytoskeleton assembly, blocks T-cell activation. Actin dynamics are controlled by a biphasic model at the IS. TCR engagement initiates actin solubilization through changes in the phosphorylation of ezrin, radixin, and moesin that allow certain molecules to move away from the points of contact with APC (321). With the formation of new actin filaments, LFA-1 is anchored and stabilizes the adhesion between T cell and APC (322). Polarization of the MTOC requires TCR and involves members of the Rho family of GTPases Cdc42, Rac, and Rho.

Disruption of the Rho family members perturbs survival of T cells, proliferation, differentiation, migration, and effector functions (278,323). Cdc42, like all members of the Rho family of GTPases, functions as a binary switch regulated by nucleotide binding. When Cdc42 binds GTP, it is converted from the inactive to the active form and binds effector molecules with high affinity. One of its important substrates is WASp (see Chapter 13). Mutations of the WASp gene are responsible for the clinical manifestations of Wiskott-Aldrich syndrome (i.e., thrombocytopenia, eczema, and recurrent infections). T cells from Wiskott-Aldrich syndrome patients lack microvilli and have abnormal cell shape. Mutants with inactive Cdc42 prevent efficient conjugation of T cells with APC. In contrast, T cells with constitutively active Cdc42 mutants form extensive filopodia rich in F-actin, which interfere with the formation of IS because Cdc42 accumulates in the T-cell–APC contact area. Activation of WASp is achieved by a complex formed by SLP-76 adaptor protein, Vav, and Nck (phosphorylated by ZAP-70). In this trimolecular complex, SLP-76 binds Vav, which converts Cdc42–guanosine diphosphate to the GTP form. Cdc42-GTP binds to the GTPase binding domain of WASp. At the same time, the Nck binds by its SH₃ domain to the polyproline sequence of WASp. These interactions release WASp from its autoinhibitory state and enable it to activate the Arp2/3 complex for actin-branching polymerization (324,325,326,327) (see Chapter 13). Other studies suggest that WASp may not be the critical or even the sole regulator of actin polymerization, but other molecules, such as its homolog N-WASp, are involved. However, mice with a WASp deficient in the VCA domain have defects in T-cell development (328).

A significant contribution to the T-cell interaction with the APC is made by the ERM cytoskeletal proteins (i.e., ezrin-radixin-moesin) (329). These proteins act as linkers between cortical actin and plasma membrane, connecting the F-actin to cytoplasmic tails of several transmembrane proteins (i.e., ICAMs, CD43, Fas, and so forth). In the resting T cell, the plasma membrane is rigid because of the thick actin network kept by phosphorylated moesin. With TCR engagement, moesin is dephosphorylated, the actin network “thaws,” and the proteins linked to the cytoskeleton are freed to move. The tall ones, CD43, are “squeezed” out of the narrow junction between T cell and APC (330). At the same time, new F-actin networks create pseudopod extensions for contact with the APC. Moesin in the back of the cell is rephosphorylated and keeps the excluded molecules actively out of the IS. CD43 and other molecules that are excluded to the uropod form a cluster called the distal pole complex. This cluster may not be simply a negative contributor to T-cell activation by collecting tall molecules interfering in the formation of the IS, but it has its own signaling mechanism in some aspects of T-cell activation, such as cytokine secretion (331).

Regulation of T-Cell Activation: Costimulation and Inhibition

T-cell activation depends on signals delivered by the TCR engaged with pMHC. However, additional signaling is needed, and this function is known as costimulation. Costimulatory signals are delivered to the T cell through the CD28 molecule reacting with ligands (counterreceptors) on APCs (i.e., CD80 [B7-1] and CD86 [B7-2]). These receptors do not act independently but modify the responses mediated by TCR. The CD28 consists of one Ig-like domain of V-type, whereas the two ligands contain two Ig domains, one V-type and one C-type. CD28 is constitutively expressed on T cells (all CD4⁺and approximately 50% of CD8⁺). Another receptor, the CTLA-4 (CD152), shares approximately 30% identity with CD28; it is not detected in naïve T cells but is induced after T-cell activation.

Both CD28 and CTLA-4 share the same ligands. In costimulation, the critical event is up-regulation of the B7 molecules on the surface of the APCs (332). The major role of CD28 function is to stimulate cell cycle progression and prevent apoptosis. It also enhances production of various cytokines, such as IL-1, IL-2, IL-4, IL-5, and interferon (IFN)-γ, and plays a fundamental role in Th1-Th2 differentiation. CD28/B7 interactions also play a critical role in B-cell stimulation. The importance of CD28/B7 interaction was established in transplantation. The importance of the CD28/B7 costimulation pathway was established with studies in transplantation in mice in which blockade of the pathway by CTLA4-Ig prolonged cardiac graft survival and prevented development of vascular lesions associated with chronic rejection (333). Another costimulatory receptor homologous to CD28 and CTLA-4, termed ICOS (inducible costimulator), is a disulfide-linked homodimer (334). ICOS lacks the extracellular motif present in CD28, which is implicated in binding with the B7 molecules. It is an inducible molecule expressed in activated, but not resting, T cells. ICOS augments T-cell proliferative responses and cytokine secretion, particularly IL-10 (335). The ligand for ICOS (ICOS-L or B7h) is a B7-like molecule expressed constitutively on B cells and macrophages. Costimulation by ICOS promotes germinal center reaction and isotype switching. ICOS and CD28 regulate Th2 responses, but whereas CD28 is critical in the priming stage to induce Th2 differentiation, ICOS plays a role in regulating Th2 effector functions (336). Activation of the ICOS pathway of costimulation initiates acute and chronic graft rejection, which indicates that the ICOS costimulatory pathway also regulates Th1 responses (337).

CD4 T-Cell Differentiation

Activated naïve CD4⁺ helper T cells (Th), in response to signals from TCR engagement with pMHC, proliferate and differentiate into cytokine-secreting effector cells, which have been distinguished into two major categories (338). Th1 cells produce primarily IFN-γ, IL-2, and TNF-β, whereas Th2 cells secrete IL-4, IL-5, IL-6, IL-10, and IL-13. Both types of cells produce IL-3, TNF-α, and granulocyte-macrophage colony-stimulating factor. Once the initial stimulus from TCR is received, the cells proliferate in response to the autocrine growth factor IL-2. At this stage, they are called pT helper cells (pTh) because they have not yet differentiated enough to secrete cytokines. Both Th1 and Th2 cells derive from a single precursor, and several factors regulate their differentiation (339,340,341). The Th1 cells induce predominantly inflammatory immune reactions and control intracellular bacterial infections (cellular immunity). They are also associated with some autoimmune diseases. The Th2 cells provide defense against extracellular pathogens, regulating the humoral antibody-mediated immune response. They are also the mediators of allergic reactions.

Th1 versus Th2 differentiation takes place in two stages. During the first stage, the activation signals delivered by TCR precondition the naïve T cell for one or the other pathway of differentiation. In the second stage, final development of effector cells is dependent on IL-12 or IL-4 (Fig. 16.11).

Stage I: Transcriptional Regulation

In the early stages of Th cell differentiation, TCR stimulation activates a number of downstream signaling pathways. Differential signaling of MAPKs, PKC, and calcineurin preconditions Th cells toward one or the other pathway in the absence of cytokines from the environment (342). A number of parameters related to antigenic stimulation and costimulation influence the final outcome. Low doses of antigen or low affinity of binding favors Th2 development, whereas large doses of antigen or high affinity of binding support predominantly Th1 differentiation. Costimulatory signals modulate the outcome. CD28 ligation favors Th2 development, perhaps by enhancing IL-4 production (343) or by direct activation of signal transducer and activator of transcription 6 (STAT-6) (344).

The CD40/CD40L interactions selectively induce Th1 cells, but this is the result of the production of IL-12 from APCs (345). During the early stage of Th activation, key genetic and epigenetic events take place that lead to accessibility and activation of specific genes. Expression of certain transcription factors is critical in the early regulation of differentiation (346,347).

Th1 Regulation

The most prominent factor for Th1 differentiation is IL-12. In naïve T cells, the IL-12 receptor is not functional but is induced by IFN-α/β. Binding of IL-12 to its receptor induces phosphorylation of Janus kinases Jak2 and Tyk2, which phosphorylate STAT-4. STAT-4 translocates to the nucleus and activates target genes. The GTPase Rac 2, which is selectively expressed on Th1 cells, activates the IFN-γ promoter via NF-κB and MAPK (348). During the early Th1 polarizing signaling, the key transcription factor for Th1 development is T-bet (for T-box expressed in T cells) induced in naïve T cells (349,350). T-bet transactivates IFN-γ promoter, induces chromatin remodeling of the gene that encodes INF-γ (351), and induces expression of the β₂ subunit of the IL-12 receptor (350). The importance of T-bet in Th1 differentiation is underscored by the susceptibility of T-bet knockout mice to challenge with Leishmania major (352) and their predisposition to allergic disease.

Figure 16.11. CD4+ T-cell differentiation. After their differentiation in the thymus into the two main lineages, CD4+ and CD8+, the CD4+ or Th cells differentiate further in the peripheral lymphatic organs into effector cells with distinct patterns of cytokine secretion. The initial stimulus is delivered by the T-cell receptor (TCR) and results in the proliferation of the naïve T cell. These activated T cells (pTh) have only a limited potential of interleukin (IL)-2 secretion, and they acquire the propensity to respond to additional signals for further differentiation. Primary stimuli via the TCR (dose of antigen, intensity and duration of TCR triggering) influence downstream signaling (Ca2+, protein kinase C, map kinase, and so forth) that regulates transcription factor expression. Expression of GATA-3 leads to Th2, whereas T-bet leads to Th1 differentiation. In the second stage, exposure to cytokines (i.e., IL-4 or IL-12/interferon [IFN]-γ) drives the final steps of differentiation to the Th1 or Th2 pattern of cytokine secretion. NKT, natural killer T cell; TNF, tumor necrosis factor. (Adapted from Noble A. Review article: molecular signals and genetic reprogramming in peripheral T-cell differentiation. Immunology 2000;101:289–299; Murphy KM, Reiner SL. The lineage decisions of helper T cells. Nat Rev Immunol 2002;2:933–944; and Diehl S, Rincón M. The two faces of IL-6 on Th1/Th2 differentiation. Mol Immunol 2002;39:531–536.)

Th2 Regulation

The key transcription factor for Th2 differentiation is GATA-3 (346), induced in the early stages under Th2 polarizing signaling. STAT-6 activation enhances expression of GATA-3. GATA-3 induces expression of another Th2-specific factor, c-maf. C-maf is a basic region/leucine-zipper transcription factor that binds to and transactivates the IL-4 promoter (353,354).

Epigenetic Modification of Cytokine Genes

Detailed description of these early events is beyond the scope of this brief review. The reader is referred to reviews of the topic (346,355). Transcriptionally inactive gene loci have a condensed chromatin with DNA tightly packed around the nucleosomes. For gene transcription, “open” chromatin is associated with acetylated histones and hypomethylation of DNA. Accessibility of IL-4 and IFN-γ genes is initiated promptly after TCR and CD28 activation, but for sustained transcription, STAT-6 or STAT-4 signaling, and induction, GATA-3 and T-bet are required. Within the first few days of Th1 and Th2 differentiation, signs of accessibility of the cytokine gene loci are observed in Th-activated cells. These signs are hypersensitivity to DNAase I and DNA demethylation.

Stage II: Maturation of Th1/Th2 Cells

In the second stage, IL-12 and IL-4 play a major role in the maturation of Th1 and Th2 cells, respectively. IL12 and IL18 act synergistically to produce IFN-γ from terminally differentiating Th1 cells. IL-23 is composed of the p40 subunit of IL-12 paired with the IL-23 α-chain related to one of the chains of IL-12. It binds to IL-12 β-chain but interacts with its own IL-23R. It activates STAT-4 and may act during the induction of Th1 and the production of IFN-γ in cooperation with IL-18.

The IL-27 is produced by APCs and induces proliferation of naïve T cells. It acts together with IL-12 in promoting IFN-γ production and is the ligand for the T-cell cytokine receptor in the early development of Th1 cells (356). Th2 maturation is promoted by IL-21, which is produced from Th2 cells, and specifically inhibits IFN-γ production and decreases responsiveness of T cells to IL-12, thus amplifying Th2 development (357).

IL-6 is produced by several types of cells, especially APCs. It activates NFAT, leading to production of IL-4, which promotes Th2 differentiation. However, IL-6 also inhibits Th1 development because it up-regulates the suppression of cytokine signaling (SOCS) 1. SOCS, also known as STAT-induced STAT inhibitor, belongs to a family of regulators of cytokine production. SOCS1 inhibits IFN-γ production and the development of Th1 cells. Thus, IL-6 plays a dual role in Th1/Th2 differentiation through induction of IL-4 and SOCS-1 (358). IL-10 has also been reported to promote Th2 differentiation, but its main effect is in suppressing Th1 cells.

Another cytokine with double regulatory function is IL-18. IL-18 is produced from macrophages and synergizes with IL-12 for IFN-γ production from NK cells and T cells. In collaboration with IL-2, it promotes Th1 differentiation in activated T cells (359). Overproduction of both cytokines induces severe inflammatory disorders. In addition to its function as a Th1 inducer and as a proinflammatory cytokine, under certain experimental conditions, IL-18 stimulates Th2 cell differentiation, increase of IgE, and allergic manifestations (360). The role of dendritic cells in humans is not clear as the production of cytokines varies depending on signals received from T cells.

CD8 T-Cell Differentiation

CD8⁺ T cells produce primarily type 1 cytokines because CD8⁺ T cells have no requirement for STAT-4 signaling via IL-12 to develop into Tc1 effectors. The Tc1/Tc2 regulation is mediated by transforming growth factor (TGF)-β with IL-4 promoting Tc1 development and cytotoxicity in the presence of TGF-β (361). There is some kind of cross-regulation between CD4⁺ and CD8⁺ T cells. CD8⁺T cells produce relatively high levels of IFN-γ and, as a result, enhance Th1 immunity. On the other hand, Th2 cell–derived IL-4 stimulates development of Tc2 cells in allergic states.

Genomic View of Type 1 and Type 2 Differentiation

Application of the new technology of gene expression by gene microarrays on polarized type 1 and type 2 CD4 and CD8 T cells identified similarities in the broad pattern of gene expression in both CD4 and CD8 T cells for type 1 and type 2 polarization, but differences were also identified between the two lineages (362). Large numbers of apoptosis-related genes were expressed, particularly in Th1 cells, which correlates with the propensity of these cells to undergo activation-induced cell death (363). A large number of cytokine and growth factor genes are preferentially expressed by either type 1 or type 2 cells. Th2 cells resemble Tc2 cells in their cytokine gene profile. Differences are also noted in genes involved in synthesis of cell migration molecules, such as CCR1, CXCR4, and β₇integrin in Th2 cells and α₄ integrin in Th1 cells. This is something that should be expected because the two Th populations home to different locations (364).

TH1 and TH2 Paradigm: Strong Defense System

The adaptive immune system evolved to recognize, discriminate, and memorize foreign antigens and pathogens. It has developed specialized cells that are able to capture anything that manages to cross into the body’s interior and present it to lymphocytes possessing specific receptors.

The “supreme commander” in this system is the T lymphocyte that regulates all defense operations. It became clear in recent years that T lymphocytes constitute a diversified population. One of them (CD4⁺, Th2) activates the B cells for antibody production, and a second (CD4⁺, Th1) activates macrophages for cell-mediated immunity. The first provides defense against extracellular and helminthic infections. The second acts against intracellular bacterial infections, fungi, and protozoa (Fig. 16.12).

Figure 16.12. Th1 and Th2 cells and effector mechanisms against pathogens. See text for details. ADCC, antibody-dependent cell-mediated cytotoxicity; CTL, cytotoxic T lymphocyte; IFN, interferon; Ig, immunoglobulin; IL, interleukin; PMN, polymorphonuclear cells.

Th1 cells coordinate the activation of macrophages and constitute the most important cellular defense mechanism against intracellular pathogens. Macrophage activation is mediated by IFN-γ, the principal cytokine produced by Th1 cells. Macrophages activated by IFN-γ rapidly kill susceptible intracellular bacteria. They produce TNF-α, which synergizes with IFN-γ. IFN-γ has been used successfully as adjunct to chemotherapy in the treatment of leprosy, tuberculosis, and atypical mycobacteriosis. In induction of Th1 immunity, IL-12 is rapidly produced by infected macrophages, which activates NK cells and stimulates production of IFN-γ, which subsequently induces Th1 differentiation. However, overproduction of IL-12, by a positive feedback mechanism triggered by IFN-γ, results in inflammatory responses. IL-12 production is inhibited by IL-10.

There is compelling evidence that the Th1 cells have an essential role in protection during mycobacterial infection. Mice deficient in IL-4, a major Th2 cytokine, have a normal response to Mycobacterium tuberculosis, whereas increased production of IL-4 correlates with disease progression (366). Mice infected acutely with M. tuberculosis are protected by adoptive transfer of Th1 cells with a 10-fold reduction in bacterial counts, whereas recipients of Th2 cells suffer from weight loss and lung fibrosis (367).

Leishmanial disease currently affects some 12 million people in 88 countries. The annual incidence is approximately 2 million new cases. Experimental studies have clearly documented that Th1 response is associated with restriction of the disease and cure, whereas a Th2 response is associated with progressive systemic disease. Balb/c mice are susceptible to Leishmania infection because they are unable to generate a Th1 response, whereas C57Bl/6 mice are resistant to the infection as a result of strong Th1 response. In Balb/c mice, the draining lymph nodes show elevated transcripts for IL-4 but not for IFN-γ (368).

The most critical unifying effect able to induce resistance has been the successful attenuation of IL-4 expression in the draining lymph node of infected susceptible animals during the first 24 hours of infection. The extent of IL-12 responsiveness is also a critical determinant for the development of a curative immune response because it induces a Th1 response (369).

Another example of the importance of a balanced Th1/Th2 response is related to the allergic inflammation (370). The pathophysiologic mechanisms of asthma seem to be based on dysregulation of the Th1/Th2 balanced response with a preponderance of Th2 cytokines. Asthma affects 8 to 10% of the population in the United States and is the leading cause of hospitalization among children younger than 15 years of age, causing an exorbitant financial burden on society. Th2-dominant responses stimulate antibody-mediated responses, activate mast cells, and elicit tissue eosinophilia (i.e., the predominant response in the asthmatic airway) (371). IL-13 is one of the Th2 cytokines, and overexpression of IL-13 in transgenic mice induces an inflammatory response with an infiltrate rich in eosinophils and macrophages. Furthermore, it causes airway fibrosis, mucous metaplasia, and airway hyperresponsiveness (372). It is likely that asthma is the result of a dysregulated mucosal immune system and pathologic T-cell response in genetically susceptible individuals (373).

CD8 T Lymphocytes

The CD8⁺ T lymphocyte is one of the two professional cytotoxic lymphocytes, the other being the NK cell. The CD8⁺ T lymphocytes, also known as CTLs, differ from NK cells in the basic mechanism of target recognition. CTL expresses an αβ-TCR, recognizing processed peptides presented by MHC class I molecules (pMHC), whereas the NK cell cytotoxicity is regulated by the C-lectin type of NK cell receptors (CD94) or members of the killer cell Ig-like receptors (KIRs), which recognize class I HLA allotypes rather than pMHC (374).

Two fundamentally different groups of methods of evaluation of target cell lysis have been developed: One evaluates disruption of cell membrane and release of tracers incorporated into the target, and the other evaluates DNA fragmentation resulting from apoptotic nuclear damage. In the first category, the most widely used method has been the release of radioactive chromium, ⁵¹Cr, preloaded into the target (375). More sensitive techniques using fluorescent impermeant dyes have been developed. DNA fragments released from the nuclei are harvested, and a “ladder” pattern is identified (376). Direct comparisons of the cytotoxic potency of different cell populations cannot be made by these methods because the target cell death is not linearly related to the cytotoxic input. Usually, the number of cytotoxic cells required to achieve a given level of target cell lysis is expressed in lytic units, which are inversely related to the effector cell number.

Activation of Cytotoxic T Lymphocytes

For the CTL to become an active effector, the precursor cell must be stimulated by antigen to undergo proliferation and differentiation (377). The activation or priming results from the interaction of the naïve CD8⁺ T cell with professional APCs. Granules are not always visible before activation, but the killing machinery (i.e., perforin, granzymes, and FasL) is delivered immediately upon priming (378,379). The granule by electron microscopy is 0.5 to 1.0 μm in diameter and is heterogeneous in its structure (380). The core is homogeneous and sometimes it is surrounded by double membranes containing the perforin enclosed by a thin membrane (381). Multiple small vesicles surround the core toward the periphery of the granule. Depending on the preponderance of these two components, granules have been distinguished as type I (dominated by the cores) or type II (with dominant multivesicular component but no cores), while other granules in terms of content are intermediate between types I and II. The granules are similar to late endosomes and have the properties of two usually separate organelles: Those of the secretory type and those of the lysosomes (382). Similarities with lysosomes include the acidic pH, the mannose-6-phosphate receptor (MPR), and the lysosomal marker, lysosome-associated membrane protein. Endocytic components carrying CD3/TCR, CD8, and MHC molecules reach the perforin-containing granule and are displayed in the outer leaflet of the membrane.

Granule Contents

Perforin (Cytolysin)

Perforin is a 65- to 75-kD glycoprotein with patchy homology to C9 complement component. It is synthesized as an inactive precursor, which is cleaved to yield a 60-kD active form (383). The protein consists of two regions: One has homology to complement proteins (C₆ to C₉), and the other is a C2 domain related to Ca²⁺-binding proteins (384). The C-terminal portion is cleaved by proteolytic enzymes activating the C2 domains for phospholipid binding (385). The N-terminal is involved in interaction with the membrane and polymerization. However, the central portion contains four membrane-spanning domains, potentially capable of forming amphipathic α helices of β sheets. At the carboxy terminal, a short peptide (propiece with a bulky glycan attached) is removed, and the remaining perforin monomer undergoes conformational changes in the presence of Ca²⁺, inserting itself in the membrane. Interaction with other perforin monomers forms the polyperforin pores (386,387,388). At least three to four monomers are needed to form a functional channel, whereas 10 to 20 aggregated monomers are needed to produce a pore visible by electron microscopy.

Granulysin

Granulysin is a member of the saposin (389,390) family of lipid-binding proteins, related functionally to defensins and other bacterial peptides, but is structurally different. It is active against Gram-positive and Gram-negative bacteria, fungi, and parasites. It disrupts artificial liposomes, damages mitochondria, and activates caspase 9 to induce apoptosis. It probably plays an important role in innate and acquired antimicrobial defenses. It kills extracellular M. tuberculosis and decreases their viability inside the cell (391).

Granzymes

Granzymes are serine proteases of the chymotrypsin family (392,393). The crystal structure of granzyme B has been solved, and its structural similarity with chymotrypsin has been verified (394). On the basis of the gene structure, proteolytic specificity, and biologic function, these enzymes are divided into three subfamilies. They are produced as proenzymes, with an acidic inactivating peptide. During their transport through the endoplasmic reticulum and Golgi apparatus, they are processed so that they are targeted to the secretory pathway. The activation peptide is removed by dipeptidyl peptidase I (DPPI), and a sequence motif interacts with proteoglycan in the granule to maintain proper conformation for activation. Granzyme B has a unique specificity among mammalian serine proteases in that it requires aspartic acid as P1 amino acid (i.e., the cleavage leaves a carboxy-terminal aspartic acid). Granzymes are highly positively charged proteins at neutral pH and form complexes with proteoglycans in the granule and extracellularly with polyanionic components.

Calreticulin

Calreticulin is a Ca²⁺ storage protein and carries a sequence that retains it in the endoplasmic reticulum. It colocalizes with perforin and is released together with perforin, which binds to the P-domain of calreticulin. Calreticulin functions as a chaperone protein for perforin and protects the CTL during biogenesis of the granules (395). Fragments of individual calreticulin domains used in lytic assays showed that the Ca²⁺-binding C-domain, which does not bind perforin, has the strongest capacity for inhibitory activity (396). However, lysis is independent of calreticulin’s ability to sequester Ca²⁺. It is suggested that calreticulin stabilizes membranes and thus prevents polyperforin pore formation.

Other Components

Chondroitin sulfate proteoglycans are negatively charged and are exocytosed during target lysis. They probably regulate delivery of the positively charged granzymes (397). The multivesicular domain of the granule is rich in MPR, which is normally absent in mature lysosomes but present in early endosomes.

The dipeptidyl peptidase (cathepsin C) is a lysosomal cysteine protease responsible for posttranslational processing in the generation of activated myeloid and lymphoid granule serine proteases.

Mechanism of Target Cell Lysis

Secretory Synapse

As we have seen with the CD4 T lymphocyte, recognition of the antigenic determinants by TCR is associated with the formation of the IS. In the synapse, the SMAC is organized by TCR and adhesion molecules. CTL also forms a synapse with the target cells, and within the synapse, there is a defined secretory domain (398). LFA-1 and talin form an outer ring of adhesion proteins with a distinct secretory domain in the center and separate from the cluster of the TCR and signaling molecules (399). Electron microscopy shows granules on the point of degranulation. The CTL acquires the hand-mirror configuration during movement, with the nucleus leading in front and cytoplasmic organelles trailing behind. The Golgi is apposed tightly to the membrane at the point of contact, and organelles appear to be “streaming” toward the contact site. Confocal microscopy shows that the granules initially cluster just behind the MTOC and then go around the MTOC to reach the secretory domain. Intimate interdigitations are visible over a large area, but a thin extracellular space separates the two cells that are held together by gap junctions (400) (Fig. 16.13). Gap junctions exist normally between cells in various tissues and probably serve the function of cellular communication. The nucleus moves away, and the granules take up position next to the area of adherence with the target.

Confocal microscopy identifies regions of the secretory synapse where granules are secreted between lck and talin or CD11a (Fig. 16.9). The granules stream around the nucleus along microtubules and accumulate behind the MTOC and the Golgi apparatus. Then they move around the Golgi and reach the synapse, where secretion occurs between the adhesion ring and the signaling domain. They fuse with the membrane within 4 minutes after contact with the target (401).

This fusion marks the beginning of the Ca²⁺-dependent second stage characterized by striking intracellular changes. The most remarkable of these changes is the reorientation of the MTOC, which, together with the Golgi, takes a position facing the area of contact with the target (402,403,404). The Ca²⁺ requirements for lysis may also be due to the Ca²⁺ dependency of the MTOC reorientation that is a prerequisite for CTL killing (405). Granules attach to and then move along microtubules toward the MTOC and finally are secreted at the MTOC. Disruption of microtubules by certain drugs severely impairs killing (404).

For the last stage of granule secretion, a GTP-binding protein, Rab27a, is critical for moving the granules from the MTOC to the synapse (406). In a rare autosomal recessive disease known as Griscelli syndrome, Rab27 is defective. These patients have albinism because melanocytes require Rab27a to secrete melanosomes. WASp, which activates the Arp2/3 complex for actin polymerization, is also required for cytotoxicity (407).

Lymphocyte-mediated killing can be confined to two pathways: The perforin-granzyme–mediated and the Fas-mediated pathways (408). Independent of the importance of the contribution of each pathway in target cell lysis, the fact is that lysis absolutely requires exocytosis of granules and their contents. Exocytosis requires signaling from PI3-K and ERK. The importance of exocytosis is emphasized by markedly decreased cytotoxicity of CTLs and NK cells in Griscelli syndrome and the ashen mouse. The ashen mice have a profound decrease of cytotoxicity, even though they have normal FasL expression and FasL cytotoxicity (409). Patients with Griscelli syndrome and the ashen mouse have a loss of function mutation in the RAB27A gene that abrogates the expression of Rab27a GTPase (one of approximately 50 GTPases). Rab27a affects the functions of the dense granules of platelets, melanosomes of melanocytes, and secretory lysosomes of CTLs. In platelets, Rab27a regulates secretion only of the dense granules. Increase of Ca²⁺ is critical for cytotoxicity, which results from extracellular sources, because it does not occur if extracellular Ca²⁺ is removed (410).

Role of Perforin

Perforin was the name given to a protein within the granules that perforates the cell membrane and opens pores, which originally were believed to be the cause of lysis and cell death. The C-terminal domain of perforin is the Ca²⁺-binding site that initiates the insertion of the molecule into cell membrane (385). The insertion is mediated by exposure of several aspartate residues after cleavage of the C-terminus to yield a 60-kD active form. These residues are presumed to become approximated in three dimensions and bind Ca²⁺, and the molecule becomes highly reactive for lipids from exposure of amphipathic domains. Some data suggest that perforin actually is inserted into the lipid bilayer with the help of a receptor. NK cells release a lysolipid, the platelet-activating factor, which binds to its receptor and forms a bridge between the platelet-activating factor receptor and perforin (411).

Figure 16.13. Interaction between a cytotoxic T cell and its target. A–C: Cytotoxic T cell attaches to its target over broad areas of the cell membrane. The formation of the conjugate involves interdigitations of microvilli between the two cells. D: Lanthanum nitrate fills the gap between cytotoxic T lymphocyte and its target and reveals junctions (arrows) that stretch between the two cells. The function of these junctions remains unknown. E, EL-4 tumor cell; L, lymphocyte; P, peritoneal exudate cell (i.e., cytotoxic T cell); Tu, tumor. (A, B, and D from Grimm E, Price Z, Bonavida B. Studies on the induction and expression of T cell-mediated immunity. VIII. Effector-target junctions and target cell membrane disruption during cytolysis. Cell Immunol 1979;46:77–99; C from Kalina M, Berke G. Contact regions of cytotoxic T lymphocyte-target cell conjugates. Cell Immunol 1976;25:41–51, with permission.)

At 37°C, perforin inserts into the membrane, and approximately 20 perforin monomers form a tubular structure (16 nm wide) with a torus in the upper ring (412), similar to that formed by the C9 component of complement (Fig. 16.14). Purified perforin causes cell lysis but not the DNA fragmentation and condensation associated with apoptosis, which is a hallmark of target cell lysis by CTLs (413). Furthermore, nuclear changes occur before cell membrane damage (414).

Target cell death requires combined action of perforin and granule-associated granzymes. However, mice deficient in perforin suffer more serious consequences of lack of or diminished cytolytic functions (415) as compared to mice deficient in granzymes A and B (416). It has been assumed that granzymes enter passively through perforin pores (Fig. 16.15). Large pores that allow passive diffusion of the granzymes are formed only with large concentrations of perforin. The pore size formed by small concentrations of perforin does not permit diffusion of proteins larger than 8 kD. However, even under these conditions, granzymes (32 to 65 kD) have access to the cytosol, although evidently not by direct diffusion through perforin pores. A lysin from Listeria monocytogenes also permits granzyme access to cytosol even in the absence of any measurable plasma membrane damage (417).

The entrance of granzyme B into the cell at low perforin concentrations is suggested to occur, probably as a result of endocytosis (“facilitated access” hypothesis). Perforin endocytosed together with granzyme disrupts the endocytic pathway and releases granzyme for delivery to the nucleus. Support for this interpretation comes from the observation that brefeldin, which interferes with redistribution of proteins out of the endosomal system, inhibited perforin-induced release of granzyme B, blocked its translocation to the nucleus, and inhibited cell death (417). Granzyme B is therefore able to enter into the interior of the cell autonomously in the absence of perforin. However, apoptotic death does not occur unless perforin is added (418). Granzyme binds to MPR when it is trafficking within the cell at the time of synthesis but also on the surface of the target cell (419). However, MPR is not critical for transportation of granzyme B within the target cell because cells lacking MPR are still subject to apoptosis by granzyme B, which enters the cell by constitutive fluid-phase micropinocytosis (420,421) or some other, probably specific, receptor.

Figure 16.14. Lesions (arrows) inflicted on its target by cytotoxic T lymphocytes. Cytotoxic T lymphocytes form punched-out lesions on the membrane of the target similar to those formed by complement. (From Dennert G, Podack ER. Cytolysis by H-2-specific T killer cells. Assembly of tubular complexes on target membranes. J Exp Med 1983;157:1483–1495, with permission.)

Figure 16.15. What is the role of perforin in cell lysis? The perforin lesion used to be considered the cause of cell death by osmotic lysis (as with complement) (A). When the granzymes were implicated in the cause of cell death by the apoptotic pathway, it was believed that the pores of perforin allow the entrance of the granzymes into the cell (B). Granzymes, however, can still enter the cell without perforin, but by themselves, they cannot cause cell death (C). Because granzymes enter the cell by endocytosis and are within endocytic vesicles, it is argued that perforin is needed to release them in the cytosol by punching holes in the vesicles (endosomolytic mechanism) (D). At this point, it is known that cytotoxic T lymphocytes kill their targets, and for this function, they need at least two of the contents of the granules: The granzymes and the perforin. The exact mechanism, however, is still strongly debated.

Endocytosis follows the binding to the receptor, and the granzyme B is detected first within Rab5-positive endocytic vesicles and subsequently in Rab5-negative, novel endocytic compartments that are not identifiable by any of the known endocytic markers (422). The granzyme B is released to the cytoplasm by a second signal provided by perforin or replication-deficient adenovirus (Ad2). From the cytoplasm, the granzyme B reaches the nucleus, initiating the apoptotic pathway. The localization in the nucleus occurs before the nuclear events of apoptosis, suggesting that nuclear translocation of the granzyme B transmits an apoptotic signal that is communicated to the nucleus (423).

Role of Granzymes

Independent of the role played by each of the constituents of the granules, it is absolutely clear that exocytosis is crucial for target cell death. In T cells, the granules are synthesized when the cells receive activation signals, whereas in NK cells, the granules are preformed. At least four granzymes are present ubiquitously in human cytotoxic cells (i.e., A, G, H, and K). After their synthesis, the granzymes undergo posttranslational modifications and, as a result, assume an active conformation. First, the signal peptide is removed and, subsequently, a short propeptide, which for granzymes A and B is DPPI (424). Subsequently, they are glycosylated and then sorted by the MPR in the Golgi apparatus on the way to the granules (425). Granzyme B is a serine protease originally defined as an aspase because it cleaves after aspartic acid in the P1 position and is the only granzyme with the preference for proteolytic cleavage after aspartate residues. In this respect, it has a specificity similar to caspases and has an extended substrate specificity with nine amino acids making contact with the substrate. The first substrate of biologic significance of granzyme B was found to be a member of the caspase family (426). Cleavage of target-cell caspases (427) results in the activation of the cellular apoptotic cascade (Fig. 16.16). Granzyme B activates apoptosis by two distinct pathways (i.e., by directly cleaving its substrates, caspase 3 or caspase 8 [428], and by a caspase-independent pathway through mitochondria). Mitochondria have a central role in the execution of apoptosis, involving disruption of electron transport and energy metabolism, production of reactive oxygen radicals, and the release of apoptotic proteins, such as cytochrome c (429,430).

Figure 16.16. Cytotoxic T-lymphocyte killing of target cells. The cytotoxic T-lymphocyte killing of its targets is mediated by both the granzyme(s) (Gr), especially GrB, and the perforin, and because both of them share the same intracellular residence (i.e., the granule), the killing mechanism is known as the granule exocytosis mechanism. GrB activates caspase (Casp)-3, either directly or, most likely, in vivo by cleavage of the proapoptotic member of the Bcl2 family, Bid. The active Bid acts on the mitochondrion and causes opening of the permeability transition (PT) pore of the inner mitochondrial membrane that causes (or is the result of) the collapse of the ΔΨm (mitochondrial transmembrane potential). ΔΨm normally results from the asymmetric distribution of protons and other ions on both sides of the inner mitochondrial membrane. It is essential for normal mitochondrial function. ΔΨm disruption occurs before cells exhibit nuclear DNA fragmentation or aberrant exposure of phosphatidyl serine on the outer cell membrane; therefore, it constitutes probably the earliest common event of the apoptotic cascade. Mitochondrial disruption activates a factor not yet well identified (X) and contributes to amplification of activation of Casp-3 and other Casps subsequently (Casp-6, -7, -8, -9, -10). The factor X may be Diablo (direct inhibitor of apoptosis proteins–binding protein), which facilitates processing of Casps through inhibition of inhibitor of apoptosis proteins. The Bid pathway of Casp-3 activation provides a greater lethal threshold of amplification of activation of Casps than the direct GrB activation. In addition to the Casp-dependent pathway (apoptosis), disruption of mitochondria by GrB-activated Bid leads to cell death by necrosis. HSP, heat-shock protein. (From Kroemer G, Zamzami N, Snesin SA, et al. Mitochondrial control of apoptosis. Immunol Today 1997;18:44–51; Heibein JA, Barry M, Molyca B, et al. Granzyme B-induced loss of mitochondrial inner membrane potential [Delta Psi m] and cytochrome c release are caspase independent. J Immunol 1999;163:4683–4693; Sutton VR, Davis JE, Cancilla M, et al. Initiation of apoptosis by granzyme B requires direct cleavage of bid, but not direct granzyme B-mediated caspase activation. J Exp Med 2000;192:1403–1413; and Barry M, Bleackley RC. Cytotoxic T lymphocytes: all roads lead to death. Nat Rev Immunol 2002;2:401–409.)

Mitochondrial factors enhance the extramitochondrial caspase activation. Bcl-2 can rescue cells from granzyme B–mediated cell death, specifically blocking the pathways that operate directly through mitochondrial perturbations. Bcl-2 suppresses the mitochondrial pathway because it prevents loss of mitochondrial membrane depolarization and inhibits the release of cytochrome c and apoptosis-inducing factor into the cytosol (431). The mitochondrial apoptotic pathway is triggered by direct cleavage of Bid (432), which results in the translocation of tBid to mitochondria, where it interacts with its receptors, Bax and Bak, to cause cytochrome c release. Cytochrome c then activates the apoptosome, which activates caspase 9 and ultimately caspase 3 (433).

Bcl-2 apparently blocks granzyme B–induced apoptosis by acting at an upstream point of the granzyme B pathway (i.e., blocking the translocation of the granzyme to the nucleus) (434).

Granzyme A

Granzyme A is a tryptase and induces caspase-independent cell death. It concentrates in the nucleus of the targeted cells and degrades histone H1 into small fragments (435). Histone H1 plays a critical role in chromatin hypercondensation, which protects genomic DNA from endonuclease digestion. Histone digestion provides a mechanism for unfolding compacted chromatin and facilitating endogenous DNAase access to DNA during T-cell granule–mediated apoptosis. Another target for granzyme A is protein HMG2 (high-mobility group protein 2). HMG2 is a nonhistone protein that binds to the internucleosomal linker region of DNA and to core histones and is involved in critical steps in DNA replication and transcription. It binds preferentially to distorted DNA and unwinds damaged DNA for its repair. It facilitates the assembly of higher-order nucleoprotein structures by bending and looping DNA or by stabilizing underwound DNA. Granzyme A cleaves HMG2 protein and thus opens up chromatin and blocks the de novo transcription required for cellular repair responses. Opening up chromatin probably contributes to the observed synergy of granzyme A with granzyme B in the induction of oligonucleosomal DNA fragmentation during CTL lysis (436). Both granzyme A and B directly cleave lamin B (437), a member of the lamin family of proteins that maintain the integrity of the nuclear envelope.

Granzyme A, bound to proteoglycans, has been detected in the blood of patients with viral diseases and rheumatoid arthritis (438). In complexes with proteoglycans, it is protected against inactivation by protease inhibitors (α₂-macroglobulin and so forth). However, its role in the blood in these conditions remains unknown. The entry of granzyme A into the nucleus requires the signal from perforin, and once inside the nucleus, it binds to insoluble factors because it does not leak out, even after the nuclear membrane is permeabilized (439).

Death Receptor Pathway

Cytotoxic lymphocytes use two pathways for killing their targets: The exocytosis pathway (perforin-granzyme) and a death receptor pathway. Although there are multiple receptors on the cell surface that can initiate an apoptotic cascade, they converge at one point downstream to a common final pathway. The point of confluence is the adaptor molecule FADD (Fas-associated death domain). These alternate apoptotic pathways may be considered as the FADD pathway (440). The pathways that converge to FADD are initiated by Fas (CD95), the most physiologically important receptor in the family of TNF receptors. FADD binds and recruits caspase 8, which stands at the apex of the cascade of all caspases (441) and forms the DISC (death-inducing signaling complex) (442). Caspase 8 may target the mitochondria through Bid or caspase 3, depending on the cell type (443,444).

In the FADD pathway, the FasL is not stored even in activated cells, and as a result, it requires the induction of a new ligand after TCR stimulation, which requires 1 to 2 hours after stimulation.
The half-life of the ligand is long (2 to 3 hours), and the CTL can kill innocent bystanders (as long as they express the appropriate receptor, Fas) without the need for TCR signaling (445). In this respect, the FADD pathway is much more promiscuous than the perforin pathway. The death receptor pathway is important for CD4⁺ Th1 cells (446).

Membrane Morphology of Target Cell Lysis

In the early stages of cell death induced by CTL, plasma membrane components are translocated to intracellular membrane structures, including nuclear envelope and mitochondria. Membrane-bound perforin and the granzymes are internalized at the same time, and it is postulated that, subsequently, perforin releases the granzymes to the cytosol by an endosomolytic action for activation of the caspase pathway (Figs. 16.15 and 16.16). In the early stage, the target does not undergo extensive permea-bilization during perforin-dependent CTL lysis.

The first detectable change in the target cell–surface morphology consists of the formation of small dilatations of the surface microvilli, forming small vesicles that eventually expand into large blebs even before ⁵¹Cr release can be detected. Some of the detached vesicles attach to neighboring CTLs, suggesting that they carry some target-specific antigens on their surface (Fig. 16.17).

Functions Mediated by Lymphocyte Toxicity

Deficiencies of Granule Contents

Virus-specific cytotoxic CD8⁺ T cells are induced on infection against many viruses (Fig. 16.12). Lysis of an infected cell may occur in the eclipse phase before any infectious virus has been produced, and CTLs kill virus-infected cells before new viral antigens can be detected on the surface with antibodies (447). Fas deficiency has no effect on the role of viral clearance, but elimination of both Fas and perforin leads to uncontrolled infection (448). Normal mice easily survive a high inoculum (10⁶ PFU) of the ectromelia virus, a murine-specific poxvirus. However, perforin-deficient mice succumb with a dose as small as 10 PFU.

Figure 16.17. Cytotoxic T-lymphocyte–target cell interaction. A–C: Scanning electron microscopy of the formation of the killer cell synapses. Intimate contact is established between the microvilli of the two cells. Compare with the electron microscopic view in Figure 17.15. (B and C from Kalina M, Berke G. Contact regions of cytotoxic T lymphocyte-target cell conjugates. Cell Immunol 1976;25:41–51; and A from Grimm E, Price Z, Bonavida B. Studies on the induction and expression of T cell-mediated immunity. VIII. Effector-target junctions and target cell membrane disruption during cytolysis. Cell Immunol 1979;46:77–99, with permission.)

Perforin deficiency in mice is associated with increased susceptibility to a variety of infections by viruses, protozoa (plasmodium, Trypanosoma cruzi), and bacteria (i.e., Mycobacteria, Salmonella, Chlamydia, Listeria, and so forth) (449). Defense involves not only cytolytic mechanisms but also cytokines (i.e., IFN-γ, TNF, and so forth), as well as microbicidal molecules. In the experimental model of lymphocytic choriomeningitis virus infection in mice, a nonlytic viral infection, perforin-dependent cytotoxicity is crucial in the control of the acute stage (450). During the chronic stage of this infection, perforin down-regulates CD8⁺ T-cell expansion and prevents immunopathology (451). In perforin-deficient mice, however, the lack of this immunoregulatory control results in expansion of CD8⁺-activated T cells, with approximately 50% of the animals dying within 2 to 4 weeks as a result of the immunopathologic damage from uncontrollable CD8⁺ proliferation and activation. Infusion of normal CD8⁺T cells fully reverses pathology and the associated mortality. This example clearly differentiates between defense and immunopathology, which normally is averted from an “exocytosis”-mediated mechanism of regulation of CD8⁺ cytolytic cells.

In another model involving the lpr/lpr mice, which are deficient in the Fas receptor, deficiency of perforin markedly accelerates the spontaneous lymphoproliferative disease, which normally occurs at a slower pace in these mice. This evidence also supports the notion that perforin plays a role in immune regulation, prevention of immunopathology, and autoimmunity.

Recently, a homozygous loss of function defect in the human perforin gene has been detected that is associated with several clinical manifestations, mainly due to uncontrolled T-cell and macrophage activation with overproduction of inflammatory cytokines. The syndrome is known as familial hemophagocytic lymphohistiocytosis (FHL) (452). The disease is mapped in chromosome 10q21-22 (i.e., the location of the perforin gene). Overall, the incidence of the mutation is approximately 20% in all FHL patients. For the development of FHL syndrome, a viral infection or a defect of an additional pathway that controls lymphocyte homeostasis is required. Perforin-deficient mice, for reasons unknown at the present time, do not develop symptomatology similar to human FHL.

Mice deficient in granzyme B lose the ability to induce DNA fragmentation, even though perforin causes membrane damage. Deficiency of both granzymes A and B causes susceptibility to ectromelia infection (453), although cytotoxic lymphocytes with the same deficiency and inability to cause DNA fragmentation are still able to exert a potent antitumor effect (454).

DPPI, also known as cathepsin c, is a lysosomal cysteine protease expressed in most tissues, and in CTLs, it is found in the secretory compartment. It belongs to the papain superfamily of proteases and shares similarities to lysosomal cysteine proteases cathepsins B, H, and L.

Granzymes A and B become enzymatically active only after cleavage of the N-terminal dipeptide by DPPI and are stored in the lysosomal granules in the active form along with perforin. DPPI knockout mice resemble the perforin-deficient mice: Both fail to cause apoptosis. DPPI deficiency results in failure to activate the granzymes, whereas perforin deficiency fails to deliver the granzymes into target cells (455,456). Deficiency of DPPI has been detected in humans with Papillon-Lefèvre syndrome, which is characterized by keratosis palmoplantaris with periodontopathia (457). There is premature tooth loss due to periodontal disease and thickening of the skin.

Graft versus Host and Graft versus Leukemia Reactions

Lymphocyte cytotoxicity has been implicated in graft versus host disease after allogeneic bone marrow transplantation, with contribution to the pathogenesis by both the perforin and Fas pathways (458,459,460). It is agreed that cytotoxic lymphocytes contribute to the development of the disease, but there is no agreement about the underlying mechanism. The graft versus leukemia effect, however, seems to be mediated predominantly by the perforin pathway.

Granzyme Inhibitors

An important characteristic of the function of cytotoxic cells is that they avoid successfully the damaging effect of the lethal weapons they deliver to kill their targets. As a matter of fact, the CTLs, after a successful hit, disengage and are directed against another target.

Some recent observations provided a molecular explanation for their protection from “suicide.” The effector cells contain a potent inhibitor of granzyme B, known as proteinase inhibitor 9 (461,462). The inhibitor is a serpin, is found in both the cytoplasm and the nucleus, and forms tight complexes with granzyme B but does not inhibit most of the caspases. Serpins are a large family of intracellular and extracellular protease inhibitors. Many viruses encode serpins that block caspases, the enzymes of apoptosis. Inhibitors of granzyme B can be encoded by several viruses. Best described is the poxvirus-encoded cytokine response modifier A (Crm A). Overexpression of Crm A in target cells inhibits CTL-mediated killing, but predominantly through a Fas-mediated pathway. Crm A inhibits granzyme B and several caspases. With several of the steps of CTL cytotoxicity now understood, therapeutic interventions are possible for several of the steps of lymphocyte cytotoxicity in a number of human conditions.

How Many Roads Lead to Death?

Despite the wealth of information on the mechanism of this important pathway, there are several areas that are not well understood. Perforin has an indispensable role in the delivery of granzyme B, but certainly not simply as a pore-forming molecule. Granzymes, on the other hand, induce the nuclear changes affecting the DNA (i.e., by apoptosis), but even if they enter the cell in the absence of perforin, they cannot be translocated to the nucleus without perforin (463). The mechanism of granzyme delivery by perforin is not clear. It has been postulated that intracellular delivery of granzymes is through an endosomolytic mechanism (464). Translocation of a fluorescent probe from the target cell membrane to interior membranes, including the nuclear envelope and mitochondria, is supportive of this prediction (465).

The relationship of the roles of perforin and granzymes in cytotoxicity was examined in the gld/gld mice, which have a FasL deficiency and therefore cannot have Fas-mediated cytotoxicity. The additional deficiency of granzyme B in gld/gld mice (i.e., mice with the phenotype GrB gld/gld) leads to CTLs with residual cytotoxicity, which can only be perforin dependent. This can be concluded also from the fact that perforin^-/-gld/gld CTLs have no residual cytotoxicity (466). It is reasonable to conclude that the cytotoxicity, which is independent of granzyme B, is normally accounted for partially by a Fas pathway and partially by a second perforin-dependent mechanism. Finally, inhibitors of caspases block the Fas death pathway, but in CTL granule exocytosis, the target cell lysis is not detectably blocked, although the accompanying apoptotic nuclear damage is efficiently blocked. Thus, caspase inhibitors prevent the hallmark phenotype of apoptosis without affecting cell death, as evidenced by lysis (467). At this point, not all the roads used by the CTL that may lead to cell death have been fully explored.

Regulatory T Cells (T reg)

T reg cells constitute about 5 to 10% of the peripheral CD4⁺ T cells and express the characteristic phenotype CD4⁺/CD25⁺ (the α-chain of the IL-2 receptor). They are also CTLA-4⁺ (costimulatory molecule cytotoxic T-lymphocyte antigen-4) and GITR⁺ (glucocorticoid-induced TNF receptor family-related protein) (468). Differentiation of CD4⁺ T reg cells is mediated by interactions of CD4⁺thymocytes with stromal cells. Some thymocytes acquire MHC II molecules (469), which are derived from APCs (470). T reg cells are enriched in autoreactive cells (471), but are resistant to deletion by negative selection, as compared to conventional CD4⁺ thymocytes (472). The TCRs of T reg cells are of relatively high affinity for autoantigens, and this may facilitate their reactivation in the periphery by self-antigens. However, their affinity is still below the level required to trigger negative selection. These fundamental differences between the TCRs of T reg cells and the other thymocytes, in terms of specificity (MHC II vs. peripheral tissue antigens), affinity, and the nature of the selecting cell, provides an additional mechanism of protection against autoimmunity (i.e., active peripheral suppression in addition to central tolerance). Selection of T reg cells is mediated by the bone marrow–derived APCs and not the epithelial cells (473). It has been suggested that the TCR affinities of T reg cells are just below the deletion level, permitting them to follow a unique differentiative pathway with regulatory function (474). The development and function of natural T reg cells depends on the transcription factor Foxp3, which encodes a new member of the forkhead/winged helix family of transcription factors (475). Mutations of the human gene FOXP3 is the cause of the IPEX syndrome (immune dysregulation polyendocrinopathy enteropathy and X-linked syndrome) (476), an X-linked immunodeficiency syndrome associated with autoimmunity involving multiple endocrine organs, inflammatory bowel disease, and atopic dermatitis. TGF-β maintains a central role in the generation and function of T reg cells (477). CD4⁺/CD25⁺ T reg cells may arise from CD4⁺/CD25^- T cells following stimulation of TCR and costimulation with TGF-β (478). Once they exit the thymus, the T reg cells depend for their survival, preservation, and function on dendritic cells, which present them with autoantigens (479). The continuous exposure to self-antigens will maintain the T reg cells in a state of activation.

A portion of T reg cells are “anergic,” defined as a state of partial or total unresponsiveness induced by partial activation. Anergy may follow TCR stimulation without subsequent costimulation. Anergic T cells are unable to produce IL-2 and induce a proliferative response. A portion of T reg cells arising in the thymus are anergic (480), and energy is their default state. Naturally arising anergic T cells upon stimulation suppress effector T cells but are prone to apoptosis, probably as a result of a continuous antigenic stimulation from self-antigens. Anergic T cells exert their suppressor function by targeting APCs. The APC brings the anergic-suppressive cell into proximity with the target T cell, forming a tricellular complex and setting the stage for a T–T-cell interaction (481). T–T-cell interactions form the foundation for induction and spreading of the anergic state, otherwise known as infectious tolerance.

γδ T Cells

γδ T-Cell Repertoire

Commitment to the γδ T-cell lineage takes place in the thymus, and the operating mechanisms have been discussed earlier in this chapter. (See T Cell-αβ versus T Cell-γδ and the Pre–T-cell Receptor.)

Rearrangements of the human γ and δ loci appear to occur in a developmentally ordered fashion (482). Initially, the γδ TCR repertoire is small because rearrangements involve a small number of V segments, and the junctional diversity is limited.

In human embryos between 8.5 to 15.0 weeks of gestation, the most common V fragments are Vδ2 joined to Dδ3 and Vγ1–8 or Vγ9 with Jγ1. These cells are referred to as the Vδ2 cells. Rearrangements after birth at approximately 4 to 6 months of age involve joining other γδ segments such as Vδ1 to Dδ1 and Dδ2 and the Vγ1 family with the Jγ2 cluster. Postnatally, in the thymus, the Vδ2 subset represents 15% and the Vδ1 85% of the γδ cells, and these proportions remain relatively constant throughout adult life. However, in the blood, the Vδ2 subset increases with age from 25% in cord blood to more than 70% in adult blood, whereas the Vδ1 subset decreased from 50% in cord blood to <30% in adult blood (483).

Although intrinsic or genetic factors generate subsets of γδ T cells, extrinsic or environmental factors act further to shape and select specific clones. An enormous selective pressure is exerted on the development of γδ T cells throughout life to produce populations of cells that express antigen receptors that are encoded by specific gene segments. The predominance in adult human blood of the Vδ2 to Vγ9 population is explained by such antigen-mediated expansion. These expansions create oligoclonal populations due to selection pressures from environmental microbes and certain edible plants. In contrast to the Vγ2/Vδ2 (Vγ2 is the same as Vγ9) T cells that are a major circulating population, the Vδ1 cells account for the vast majority of the γδ T cells in tissues such as intestine and spleen (484).

γδ T-Cell Receptor Structure and Antigen Recognition

Antigen recognition by the γδ TCR resembles recognition by antibodies (485). The V and C domains are organized into “Ig folds” (i.e., approximately seven β strands packed face to face in two antiparallel β sheets, constrained by intradomain disulfide bonding). The V regions are subdivided into framework and hypervariable regions, which have three CDRs. The orientations of Vγ and Vδ are similar to the relative orientations between the V domains in the Fab Ig fragment or the αβ TCR. However, the CDR3 of Vδ is diverse in length and composition (8 to 21 amino acids), a range similar only to the IgH (3 to 25). Furthermore, the CDR3 loops of the γδ TCR protrude above the rest of the molecule and create clefts between them, which strikingly distinguishes them from the equivalent loops of the αβ TCR, which are flat and bind to pMHC, and from the antibodies that bind large proteins (486).

In general, the γδ TCR is broadly conserved but with unique structural features. Antigen recognition by γδ TCRs is fundamentally different from that of αβ TCRs. Antigens are not required to be processed and there is no MHC restriction; as a result, the γδ TCR is allowed to recognize a wide array of antigens (485,486,487,488). The antigens recognized are those with a wide distribution that are constitutively expressed by host cells and by microbial pathogens and those that are inducibly expressed or might be restricted to certain cell types. Examples of the former category are nonprotein substances such as pyrophosphates and alkyl-amines that are found in bacteria, plant, or animal cells (489,490,491), and bacterial and mammalian homologs of heat-shock protein 60 kD (HSP-60) (492).

Human peripheral blood Vγ2Vδ2 cells are present in large numbers in lepromatous lesions reactive with monoalkyl compounds of mycobacterial cell walls. The most potent compound is monoethylphosphate, which stimulates cytotoxic activity of these cells. The phosphate group is very important in the recognition of this substance by the γδ TCR. Several other phosphorylated compounds could be shown to have reactivity with the γδ TCR. Alkyl-amines, such as isobutylamine and so forth, derived from plant food products or from bacteria also stimulate Vγ9Vδ2⁺ T cells (491).

A third class of antigens (i.e., the aminobiphosphonates) that are used to inhibit osteoclastic bone resorption, particularly in cancer patients, can also stimulate Vγ9Vδ2 cells (493). Antigens recognized by γδ TCR are also found in other infectious agents (i.e., Listeria, Plasmodium coccidia, and so forth). Vγ9⁺ T cells also respond to superantigens, such as staphylococcal enterotoxin A (494). Superantigens interact with the MHC class II molecules and TCRs in a way distinct from that of normal peptide antigens. They bind independently to MHC class II and to TCRs. They interact with the Vβ domain of the TCR outside of the CDRs and with the outer faces of the MHC molecule, outside the peptide-binding groove.

The functional γδ TCR exists as a complex with the CD3 polypeptide chains. Some mucosal γδ T cells interact with MHC-encoded proteins, MICA and MICB (495). Recognition is through the activating NKG2D C-type lectin receptor (496), but the contribution of the γδ TCR is not clear. MICA and MICB class I molecules identify stressed cells and have a very restricted pattern of expression, primarily limited to intestine. MICA and MICB do not present peptides because the peptide-binding groove is of limited size (497) (see Chapter 18). These molecules may function in innate immunity as important targets for Vδ1⁺ cells for elimination of stressed cells (484). Some Vδ1⁺ cells recognize the nonpolymorphic CD1c member of the CD1 family of molecules, expressed on APCs, that presents lipid and glycolipid foreign antigens to T cells (498). These γδ cells activated in response to CD1c produce IFN-γ and direct αβ T cells to Th1 differentiation. Furthermore, they are cytotoxic and express granulysin, and they could lyse infected dendritic cells via the perforin pathway and kill released bacteria by granulysin. Therefore, their role is significant in host defense before antigen-specific T cells have differentiated (499).

Direct recognition of CD1c may represent a bridge between innate and adaptive immunity in a similar fashion to recognition of CD1d by murine and human NK⁺ αβ T cells (500), which polarize T cells to a Th2 phenotype. The CD1c-restricted γδ T cells promote the maturation of myeloid-derived dendritic cells, which, at the immature stage, express high levels of CD1a, b, and c antigens. When these dendritic cells mature, they are able to present antigens to CD4⁺ T cells. This function of γδ T cells is important because they rapidly provide mature dendritic cells early during microbial invasion and, at the same time, secrete IL-12 to drive T-cell polarization to the Th1 type (501).

γδ T-Cell Function

The recognition of antigens by the γδ TCR is more akin to that of antibodies rather than the αβ TCR. It does not need antigen proc-essing by APC and presentation by MHC. The available crystal structure of human Vγ9/Vδ2 shows that the functional similarities are, as expected, based on structural similarities with the heavy as well as the light chains of Igs, as discussed earlier.

Subsets of γδ T cells are characterized by the expression of distinct sets of Vδ genes, the Vδ1 and Vδ2. They are believed to represent separate lineages with different developmental pathways (502) and different tissue distribution. Most of the γδ T cells with intraepithelial localization (i.e., nasal mucosal, small intestine, and colon) are Vδ1, whereas the Vδ2 are detected in the peripheral blood, where they constitute approximately 5% of all T cells.

In general, the function of the γδ T cells in infectious diseases includes an immediate response to invasion by pathogens and a long-term regulation of the inflammatory response. The Vγ9/Vδ2 T cells release proinflammatory chemokines (503), and provide protection against mycobacteria by directly killing infected macrophages (504) and extracellular bacteria by granulysin with the help of perforin (505). The γδ T cells represent the largest population of mycobacteria-reactive T cells in the blood of humans, and more than 85% of them are Vγ9/Vδ2 and proliferate vigorously in response to the infection.

Persistent chronic disease correlates with a decline and eventual disappearance of mycobacteria-reactive γδ T cells (506). This loss is due to Fas-FasL–mediated apoptotic death as a result of activation-induced cell death from chronic stimulation by mycobacterial antigens. After clearance of the mycobacteria at a late stage of the disease, the γδ T cells contribute to the resolution of the response and prevention of chronic inflammatory disease by directly killing activated macrophages (507). This may be facilitated by the expression of CCR5, the receptor for the chemokine RANTES (MIP-Iα) produced by macrophages (508). γδ T cells participate in responses against several other infectious agents (488). The evidence of a role of γδ cells against cancer is still circumstantial (509).

A role of γδ T cells in autoimmune disease has been demonstrated in some experimental models. As has been discussed earlier, their mucosal localization places the γδ T cells in a unique position for defense against infectious agents, but also in induction of mucosal tolerance. Induction of oral tolerance is a very old observation (510). Antigens given by mouth induce specific systemic unresponsiveness, an observation that has been documented repeatedly in animals and humans.

Nonobese diabetic mice develop spontaneous insulitis and diabetes, a disease that shares a number of immunologic and pathogenetic features with human type I diabetes. Conformationally intact but biologically inactive insulin administered in these animals intranasally induces immunoregulatory γδ T cells, which secrete IL-10, accounting for the antidiabetic suppressor effect (511).

γδ T cells accumulate in the synovium of patients with rheumatoid arthritis (512). In an animal model of collagen-induced arthritis, the role of γδ T cells is controversial; whether γδ T cells promote or reduce exacerbation of inflammation may depend on the stage of the disease (early or late, respectively) (513). The evidence so far for the role of γδ T cells in defense, immune regulation, and surveillance as a result of their broad reactivity places this cell in a critical position as a “sentinel” between innate and adaptive immunity (514). This function is supported by its receptors, which share broad structural features of the Ig domains. A population of TCR-γδ cells resides within the intestinal epithelium (intraepithelial lymphocytes [IELs]) some of which may have an extrathymic origin (bone marrow) (515) and express a CD8αα homodimer (516). These intraepithelial TCR-γδ cells are important in maintaining self-tolerance (517). The cutaneous intraepithelial TCR-γδ lymphocytes are dendritic in morphology with monospecific γ- and δ-chains (Vγ5⁺/Vδ1⁺). They are “educated” by thymic stroma and are positively selected on self-agonists. Their selection involves a thymic stromal determinant not encoded by MHC (518).

Natural Killer Cells

NK cells were originally described on a functional basis according to their capability of killing certain tumor cells of hematopoietic origin in the absence of prior stimulation. Subsequently, they were identified by monoclonal antibodies as a discrete population of cytolytic lymphocytes and were implicated in several activities in vivo, such as activity against tumor cells, resistance to viral infections, and regulation of hematopoiesis (519,520).

NK cells are a heterogeneous population with respect to phenotype and target specificity. Although the majority of the CD56⁺ NK cells are CD3^-, small numbers of CD45⁺/CD3⁺ cells have been detected, and large granular lymphocyte (LGL) leukemias with the same phenotype have been reported (521).

Morphology, Cytochemistry, and Surface Markers

LGLs are large cells with pale blue cytoplasm and a high cytoplasmic-to-nuclear ratio. Their main histologic feature is the presence of azurophilic granules. These cells constitute 2 to 6% of the peripheral white cells and approximately 10 to 15% of the peripheral blood lymphocytes. They are larger than the typical lymphocytes (10 to 12 μm) with a larger amount of cytoplasm that contains peroxidase-negative granules (Fig. 16.18). The α-naphthyl acetate esterase distribution is similar to that found in monocytes, with a diffuse membrane-associated pattern in the cytoplasm, and is different from the dotlike pattern of T lymphocytes. LGLs do not adhere to surfaces and have no phagocytic activity. Ultrastructurally, they are heterogeneous in terms of size and the density of the granule matrix (522). The granules are located next to the Golgi apparatus, which also contains several smooth membranes and coated vesicles. The Golgi apparatus apparently is involved in the packaging of the granule contents. The granules have an electron-dense center surrounded by a layer of lesser opacity. Like the granules of CTL, the NK-cell granules contain perforin and granzymes, which are both important for their cytotoxic function (523). The granules may be present in various forms, such as smooth or coated, depending on the stage of cell activation.

Figure 16.18. Natural killer cell morphology and ultrastructure. A: Natural killer cell is characterized by large amounts of pale blue cytoplasm and the presence of azurophilic granules. B: By electron microscopy, the granules in the cytoplasm appear heterogeneous in terms of size and density (arrows). A few scattered mitochondria are present. C: The granule consists of an electron-dense center that contains the perforin and may be enclosed by a thin membrane. Surrounding the core is a layer of lesser opacity containing the granzymes. The granules of the cytotoxic cells are probably derived from two separate organelles—those of secretory granules and those of lysosomes—and are sometimes called granulosomes. (From Zucker-Franklin D, Greaves MF, Grossi CE. Atlas of blood cells, 2nd ed. Philadelphia: Lea & Febiger, 1988, with permission.)

LGLs are phenotypically and functionally heterogeneous (i.e., CD56⁺/CD3^-/CD8^-), and the majority of them are CD16⁺ (approximately 80 to 90%); others are CD57⁺/CD3⁺/CD8⁺/TCR-α/β⁺. Both populations are cytolytic, but the CD57⁺ cells are T (or NK-like) cells that are non–MHC restricted. The CD56⁺ population is sometimes known as NK-LGL, whereas the CD57⁺ population is known as T-LGL. Both populations are CD2⁺and CD7⁺. The separation of the two cell types is not absolute, and we often detect an intermediate cell population that is CD56⁺/CD57⁺. The relationship of these cells to those expressing only one of the two markers is not known. Clonal diseases have been described from both the NK-LGL and T-LGL lineages, with distinct clinical syndromes.

Ontogeny of Natural Killer Cells

Phenotypic Studies

The presence on NK cells of some markers (CD2, CD7) characteristic of T cells raised the question of common ontogenetic origins between these two lymphocytes. However, interference with T-cell development, as in athymic or severe combined immunodeficiency (SCID) mice (524) or in mice with targeted disruptions of RAG, has no impact on NK-cell development. It is conceivable that the two lineages derive from a common progenitor with separation of their developmental pathways.

Some evidence for a common T/NK precursor has been obtained (525,526) and relates to expression of cytoplasmic CD3γ, δ, and ∊. Cytoplasmic expression of CD3γ, δ, and ∊ proteins has been detected in fetal NK cells (527), and overexpression of CD3∊ blocks T/NK-cell development without affecting other hematopoietic lineages, including B lymphocytes (528). Progenitors of NK cells have been identified in the thymus based on the expression of CD56 within the triple-negative (TN) thymocyte population (529), and a common T/NK-cell precursor was detected among the CD34⁺ TN population (530). The common progenitor has a CD45/CD5 phenotype and becomes an NK-committed cell with loss of the capacity for T-cell differentiation with the expression of CD56. The common precursor is present among TN thymocytes that are CD34⁺ and become CD34^-as the cell is committed to NK lineage. NK-cell differentiation requires a combination of stem cell factor, IL-7, and IL-2 in the presence of a stromal feeder layer.

Precursors of NK cells have also been identified in fetal thymuses of mice (531) that can differentiate into DP (CD4⁺/CD8⁺) thymocytes if they remain in contact with the thymic stroma or into NK cells if they are removed from the thymic microenvironment. This again supports the notion that the thymus is not essential for NK-cell development. The common T- and NK-cell progenitor in fetal mouse thymus is CD117⁺/CD44⁺/CD25^-/NK1.1⁺ and shows commitment to NK lineage with loss of CD117 expression (532). The overall scheme of differentiation in fetal mouse thymus suggests that multipotent precursors entering the thymus first lose B-lymphoid potential as they up-regulate NK1.1 and commit to a T/NK lineage. Then CD25 is up-regulated while CD117 is down-regulated, and cells commit to T- and NK-cell lineages, respectively.

Bone marrow is also an important location of NK-cell progenitors. They are CD34⁺/CD7⁺ and differentiate to mature NK cells either in the presence (533) or absence of stromal cells (534,535,536). The progenitors express CD34 and CD7 but are CD33^-, and cytokines such as stem cell factor, IL-2, and IL-7 are essential, especially at the early stage of their differentiation. Depending on the conditions of culture, other cytokines, such as IL-7 and IL-3, are needed.

Cytolytic activity against NK-cell targets is detected at the time of CD56 expression. Mature NK cells can be stimulated by IL-2 for further enhancement of cytotoxic activity, increase of cytotoxic granule content, expression of adhesion molecules, and acquisition of properties attributed to lymphokine-activated killer cells. Other cytokines, such as IL-7 and IL-12, have similar effects, albeit to a lesser degree than IL-2. For example, IL-12 potentiates suboptimal concentrations of IL-2.

The common lymphocyte precursor in the bone marrow with the CD34⁺/CD10⁺/CD45RA⁺ phenotype is able to develop into NK cells in vitro (537). Uncommitted hematopoietic progenitors are CD34⁺/C38^-, but up-regulation of CD38 indicates enrichment of cells committed to a particular hematopoietic lineage. CD34⁺/ CD38²⁺ progenitors in fetal thymus develop into T cells, but CD34⁺/CD38²⁺ progenitors in fetal liver have no T-cell precursor activity. This population has no TCR-δ rearrangements and no pre-TCR-α chain expression and develops in vitro into NK cells through an intermediate stage of CD3^-/NKRP-1⁺/CD34^-/CD56^- (538,539). A similar NKRD-1⁺/CD56^- population has been detected in cord blood and develops into mature CD56⁺ NK cells in the presence of IL-12 (540).

Functional Studies and Differentiation

The cytokine environment regulates NK-cell maturation for the two distinct human NK-cell populations, one IFN-γ and a second IL-13 producing. IL-4 regulates the size of the IL-13 population primarily by inducing their proliferation, whereas IL-12 has minimal effects on the proliferation of IFN-γ⁺ NK cells. Cells with the CD161⁺(NKRP-1)/CD56^- phenotype produce IL-13 but do not produce IFN-γ and have no perforin-mediated cytotoxicity (541). CD161 is encoded by the NK-gene complex, which includes CD69 and CD94. The CD161⁺56^-, IL-13–producing cells differentiate to phenotypically mature IFN-γ–producing cells in the presence of IL-12 and feeder cells. As cells mature, they pass through an intermediate stage of IL-13⁺/IFN-γ⁺ NK cells and eventually acquire the mature irreversible phenotype IFN-γ⁺/IL-13^-/CD56⁺ (542).

The molecular defect of X-linked SCID conclusively demonstrated that cytokines are critical for T- and NK-cell development (543,544). These X-SCID patients present with a severe block of T- and NK-cell development, whereas normal or even elevated numbers of B cells are present. The cause of the disease is the lack of the common γ_c-chain shared by several interleukin receptors (i.e., IL-2, IL-4, IL-7, IL-9, and IL-15). The most important among these interleukins is IL-7, which promotes development of human thymocytes (and, in the mouse, B lymphocytes also). Deficiency of Jak-3, a tyrosine kinase associated with the γ_c-chain, also blocks NK-cell development (545). Other data suggest a critical role for IL-15 (546,547). Of the transcription factors, the Ikaros family is required for transcriptional regulation of NK-cell development (548,549).

Natural Killer Cell Receptors

NK cells are one of the important cellular components of innate immunity, with the mission to defend the body immediately against pathogens or in the early stages of tumor development. As a consequence, the recognition molecules or receptors of the NK cells are displayed on the cell surface without the need of assembly (i.e., rearrangements from multiple DNA segments after the antigenic encounter) (550,551,552,553,554,555). Another important difference from other receptors such as BCRs or TCRs is that NK receptors do not directly recognize pathogens or their products but the quantitative change of MHC molecules induced as a result of the infection. In humans, three distinct families of genes have been defined that encode for receptors of HLA class I molecules. One family belongs to the Ig superfamily and is composed of killer Ig-like receptors (KIRs). The second family is structurally Ig-like, named Ig-like transcripts (ILTs). Ig-like transcripts are expressed mainly on B, T, and myeloid cells, but some members are also present on NK cells. They are also called LIRs for leukocyte Ig-like receptors. The third family consists of C-type lectin receptors. C-type lectins are a superfamily with homologous modular carbohydrate-recognition domains (CRDs) that bind carbohydrates in a Ca²⁺-dependent manner. The proteins of this family of NK receptors form group V (outside of a total of seven groups) of the C-type of lectins superfamily (556). The C-type lectin receptors, however, have structural differences from the other (more than 200) members of the superfamily, and it has been proposed that it be renamed as a new family of C-type lectin–like NK-receptor domains (CLTDs) (557). The human lectinlike receptor gene complex is on chromosome 12p13.1 and the genes for the Ig-like receptors are in chromosome 19g13.4.

Immunoglobulinlike Receptors: The Killer Ig-Like Receptors and Leukocyte Ig-Like Receptors

The KIRs have two or three Ig-like domains and hence are designated KIR2D or KIR3D receptors, respectively (558) (Fig. 16.19 and Table 16.2). The cytoplasmic domains of the KIRs can be either long (L) or short (S), corresponding to their function, either inhibitory or activating, respectively. The inhibitory receptors contain one or two immunoreceptor tyrosine-based inhibitory motifs, or ITIMs (I/V/L/S)-X-Y-XX-(L/V) (where X denotes any amino acid) (559). When tyrosine (Y) is phosphorylated, it recruits and activates SHP-1 phosphatase, leading to inhibition of signaling. Activating receptors do not signal directly but must associate noncovalently (via a salt bridge linking the transmembrane regions) with other signaling adaptor molecules that have ITAMs in their cytoplasmic domain (consensus sequence, -Y-X-X-L-X_6-8-YXXL/I) (560). NK cells express three ITAM-bearing transmembrane proteins: ζ, Fc∊RIγ, and DAP12. The first two are present as homo-dimers or heterodimers, whereas DAP12 is exclusively a disulfide homodimer. (The KIRs in the CD designation are known CD158a-m and CD158z. See Table 16.2 and Appendix for details.)

Figure 16.19. Natural killer (NK)-cell receptors. The NK-cell receptors in humans structurally belong to two different families: One family has immunoglobulin (Ig) domains in the extracellular region and are known as killer Ig receptors (KIRs) and the other family has a C-type lectin–like domain. The C-type lectin domain recognizes oligosaccharides (and sometimes polypeptides), but the binding is directly mediated by Ca2+, hence the term C-type lectin. However, the NK-cell receptors with the C-type fold lack the Ca2+-ligating elements and thus have been termed C-type lectin–like. (See Kogelberg H, Feizi T. New structural insights into lectin-type proteins of the immune system. Curr Opin Struct Biol 2001;11:635–643.) The NK-cell receptors in both families are functionally divided into inhibitory and activating. The labyrinthine jargon of the NK-cell receptors becomes simpler with the acquaintance of some rules: The inhibitory KIRs have either two or three Ig domains in the extracellular region (i.e., 2D or 3D) and a long or short cytoplasmic tail (i.e., 2DL or 2DS and 3DL or 3DS). Those with long cytoplasmic tails are inhibitory, whereas those with short tails are activating. Thus, the receptor KIR 2DL is an inhibitory receptor with two Ig domains, whereas the receptor KIR 3DS is an activating receptor with three Ig domains. (See more details in text and in Table 16.2.)

A fundamental difference between recognition by KIRs and TCRs is that the KIRs recognize more than one MHC allele. They do this by recognizing conserved residues within the polymorphic regions of MHC, whereas TCR recognizes the polymorphic residues. In addition, KIRs display a precise specificity for a particular MHC allotype. This is achieved by variations in single amino acids of KIR molecules (561).

The ligands for KIRs are alleles of all three MHC class I molecules, HLA-A, HLA-B, and HLA-C, which can confer protection from lysis by NK cells. Generally, KIR3D receptors recognize HLA-A and -B, whereas KIR2D receptors recognize HLA-C alleles. A number of techniques have been used to study the binding and specificity of the receptors. No other molecule is necessary for binding except the HLA alleles (562,563,564). Crystal structures of some KIR receptors with their ligands have been solved. The KIR2DL2 in complex with HLA-Cw3 and peptide shows that KIR binds in nearly orthogonal orientation across the α₁ and α₂ helices of HLA-Cw3. It contacts positions 7 and 8 of the peptide, but most contacts are between the KIR and conserved HLA-C residues (565).

Table 16.2 Natural Killer Cell Receptors and CD Designation

A KIRs B LIRs/ILTs CD158a KIR2DL1 CD85a LIR3 (ILT5) CD158 b1/b2 KIR2DL2/L3 CD85b ILT8 CD158c KIR2DS6 CD85c LIR8 CD158d KIR2DL4 CD85d LIR2 (ILT4) CD158 e1/e2 KIR3DL1/S1 CD85e LIR4 (ILT6) CD158f KIR2DL5 CD85f ILT11 CD158g KIR2DS5 CD85g ILT7 CD158h KIR2DS1 CD85h LIR7 (ILT1) CD158i KIR2DS4 CD85i LIR6 CD158j KIR2DS2 CD85j LIR1 (ILT2) CD158k KIR3DL2 CD85k LIR5 (ILT3) CD158z KIR3DL7 CD85l CD85m ILT9 ILT10 NOTE: See also Figure 16.20 for structural classification. (Designations were assigned during the seventh HLADA Workshop, Harrogate, UK, June 2000.) ILT, immunoglobulin-like transcript; KIR, killer cell immunoglobulinlike receptor; LIR, leukocyte immunoglobulinlike receptor. From Mason D. CD Antigens 2001. Immunology 2001;103:401–406, with permission.

Allotypic specificity is determined by the interaction between Lys44 of KIR2DL2 and Asn80 of the HLA-Cw3. In general, the peptides play a minimal role in the interaction, a point that strikingly distinguishes TCR–MHC from the KIR–MHC interactions (566). LIR-1 is a member of the LIR family (LIR-1 to LIR-8) expressed on monocytes, B cells, dendritic cells, and some NK cells. LIRs have two to four extracellular Ig-like domains and ITIMs in the cytoplasmic region; therefore, engagement with MHC molecules protects target cells from lysis. The first aminoterminal domain of LIR-1 binds to the nonpolymorphic α₃ domain of MHC I, an interaction that is more similar to that of CD4 with the MHC II than to that of KIRs. LIR-1 also recognizes the human cytomegalovirus protein UL18, which has a structure similar to MHC I and associates even with β₂-microglobulin and host peptides. LIR-1 binds to this human cytomegalovirus protein with an affinity that is more than 1,000-fold higher than the host MHC I. This mechanism illustrates an example of viral subversion of host defenses and protection of infected cells from lysis (567) because the binding of LIR-1 to UL18 sends an inhibitory signal.

C-Type Lectin–Like Receptors

The members of the C-type lectin–like receptor family are either homodimers (the large murine family Ly49A-W, CD69, and NKR-P1) or heterodimers, which consist of an invariant chain CD94 and a second subunit from the NKG2 family members A, B, C, and E (568). The function of the heterodimeric proteins depends on the cytoplasmic regions of the variant chains (i.e., whether long [NKG2A,B] or short [NKG2C,E], corresponding to inhibitory or activating functions, respectively). The inhibitory subunits have one pair of ITIMs in the cytoplasmic region, whereas the activating subunits associate with the ITAM-containing adaptor molecule DAP12. The NKG2D protein forms a homodimer (does not pair with CD94) and is an activating receptor. Each subunit is composed of an extracellular C-type lectin–like domain known also as NK-receptor domain or NKD. The NKD, however, binds proteins and not carbohydrates, which are the ligands for the classical C-type lectins, and furthermore, their structure differs from the classical C-type lectins (569). C-type lectin–like NK-receptor domains include other important molecules of innate immunity, macrophage mannose receptor, collectins (i.e., pulmonary surfactants), and so forth. The ligands for CD94/NKG2 proteins are the nonclassical HLA-E molecule, which binds peptides from the leader sequence of the HLA-A, -B, -C, and -G molecules (570,571). Expression of HLA-E depends on the presence of the signal peptides, thus providing a safe strategy for NK cells to monitor the presence of polymorphic HLA molecules (572) because absence of HLA-E would indicate absence of some of the HLA-A, -B, or -C. HLA-E is also capable of binding signal sequence of heat shock protein (HSP)-60. HSP-60 is present in all living cellular organisms (573) and serves as a mitochondrial chaperone. HSP-60 levels are increased in response to stress stimuli (i.e., temperature increase, nutrient deprivation, exposure to toxic chemicals, inflammation, and so forth). HSP-60 protects these cells from harmful stimuli, but at the same time, the HLA-E/HSP-60 peptide is not recognized by the inhibitory receptor, CD94/NKG2A, and these cells are eliminated by NK-cell activation (574).

In humans the ligands for the NKG2D receptor are the MHC class I–related molecules MICA and MICB, which are up-regulated in virally infected cells and many tumors (575,576). They are minimally expressed in normal tissues but up-regulated in stressed cells (577). MICA is composed of two structural domains: One in the form of a platform formed by α₁/α₂ and the second in the form of a C-type Ig-like α₃ domain (578). The NKG2D receptor binds orthogonally to the MICA platform in a way similar to the docking of TCR on pMHC (see above) (579). For signaling, it uses the adaptor protein DAP10, which has no ITAMs but contains the sequence YINM and propagates signaling through PI3-kinase and is thus less susceptible to the SHP-1–coupled receptors. The NKG2D receptor in mice may associate with another adaptor protein, DAP12 (also known as KARAP), which recruits protein tyrosine kinases (580). Through DAP12, it can activate ZAP-70 or Syk protein kinases through the ITAM of DAP12 and the p85 subunit of PI3-K. Thus depending on the availability of the adaptor partners, NKG2D may mediate costimulation in T cells and/or activation in NK cells (580). Furthermore, the functional outcomes of NKG2D stimulation depends on ligand density, cytokines, or the presence of inhibitory ligands (581). Another family of ligands for NKG2D contains proteins that bind the UL16 cytomegalovirus protein, known as UL16-binding proteins (ULBP) (582,583). The UL16-binding proteins possess α₁ and α₂ domains but differ from MIC and MHC I in lacking the α₃domain, and they are GPI-anchored proteins without β₂M protein. Binding to NKG2D receptor has been demonstrated by blocking of ULBP binding by anti-NKG2D antibodies (584). The presence on NK cells of receptors that recognize infected or abnormal (malignant) cells provides a new platform for developing strategies for immunotherapy of malignancy (585).

Conclusion

The major breakthrough in the regulation of NK-cell function came with the formulation of the “missing self” hypothesis (586,587) (Fig. 16.20). This hypothesis states that NK cells can recognize and selectively lyse targets that fail to express self–MHC I antigens. The validity of the hypothesis has been demonstrated in multiple in vivo and in vitro systems as summarized (586). The basic premise is that when appropriate MHC I molecules are expressed, the lysis of target is inhibited, but when the target is deficient in MHC I expression, the target is lysed. Specific inhibitory receptors engaged with normal MHC I molecules prevent activating receptors to kill the target. Cells that have lost a normal MHC I expression (“missing self”) (i.e., tumors or infected cells) are unable to deliver an inhibitory signal to NK cells and become susceptible to lysis. Cells susceptible to lysis by NK cells as a result of deficiency of MHC I molecules could be protected by transfection of MHC I alleles. This experiment provided formal demonstration of the validity of the missing self hypothesis. However, NK cells under persistent stimulatory signals, below the level of full activation, become hyporesponsive or “anergic” (i.e., chronic stimulation induces loss or responsiveness to NK cells [disarming model]) (588,589). Another possibility (arming model) assumes that MHC class I inhibitory receptors induce functional maturation and in their absence the NK cells remain immature (i.e., hyporesponsive).

Figure 16.20. Natural killer (NK)-cell cytotoxicity: The missing self. NK-cell cytotoxicity is a delicate balance between activating receptors and inhibitory receptors. Both (whether of the immunoglobulin SF or C-type lectin) have HLA class I specificity, although some receptors recognize non–major histocompatibility complex molecules. The presence of activating receptors against normal cells causes no problem as long as the NK cells possess at least one inhibitory receptor, the signal of which is always dominant. However, with alterations of the structure of a class I molecule or complete loss of the molecule, the binding of the inhibitory receptor is abolished and the activating signal remains unopposed. This is the essence of the missing self hypothesis. KAR, killer-activating receptor; KIR, killer Ig receptor. (From Kärre K. How to recognize a foreign submarine. Immunol Rev 1997;155:5–9.)

The effector functions of the NK-cell receptors are controlled by the transmembrane and cytoplasmic regions. The inhibitory receptors have ITIMs in their cytoplasmic tails and recruit SHP-1 phosphatase. Activating receptors have a short cytoplasmic region, but for signaling, they borrow the function of ITAMs by associating with adaptor molecules like DAP-12. Association is mediated by a positively charged residue of the receptor and an oppositely charged (i.e., aspartate) residue of the adaptor (590). Further research into the regulation of NK-cell function by the NK receptors is expected to expose new initiatives for investigating tumor immunosurveillance (591,592) and autoimmunity (593). In hematology specifically, the manipulation of receptor–ligand interactions may help in allogeneic bone marrow transplantation to prevent leukemia relapse and graft versus host disease (594,595,596).

Natural Cytotoxicity

Two major mechanisms are used by NK cells to kill target cells and both require contact with their targets. One pathway involves proteins released from granules by the process of exocytosis such as perforin (see section on Cytotoxic T lymphocytes) or serine proteases, the granzymes with various substrate specificities. The other pathway involves engagement of death receptors (e.g., Fas/CD95) on target cells, with their cognate ligands (FasL) on NK cells (597). Engagement of the NK cell with its targets is mediated by the NK-cell receptors (see above).

Natural Resistance

NK cells are one of the main effectors of innate immunity. Their role against viral infections has been well documented (598). NK cells selectively lyse virally infected cells but spare noninfected cells. The mechanism of natural resistance against viral infections may involve direct lysis of the infected cells or production of TNF-α, which stimulates NK-cell activity. A patient with selective deficiency of NK cells has been described who had life-threatening viral infections (599). NK cells also play a role against certain intracellular parasites, such as Toxoplasma gondii. Production of TNF-α and IL-12 activates NK cells for production of IFN-γ, which is important for macrophage activation in the defense against intracellular parasites (600).

Regulation of Adaptive Immunity

NK cells enhance responses of B and T cells during the early stages of an immune response. As a result of stimulation by IL-12, produced by macrophages in response to infectious agents, NK cells produce large amounts of IFN-γ (601). IFN-γ acts on macrophages and enhances their antigen-presenting function by increasing expression of class II antigens. IFN-γ is important in directing T-cell differentiation to Th1 cytokine responses. Induction of predominantly Th1 responses leads to enhancement of cell-mediated immunity. Finally, IFN-γ activates macrophages, enhancing their microbicidal function.

Natural Killer Cells and Malignancies

NK cells provide surveillance against tumor cells and virus-infected cells (602). Circumstantial evidence consistent with the role of NK cells in surveillance against tumors includes the higher incidence of lymphoproliferative diseases in patients with Chédiak-Higashi syndrome, who have profound deficits in NK activity. In patients with X-linked lymphoproliferative disease, NK activity is deficient as well (603), and the same is true for people with high familial incidence of cancer (604). Similarly, beige mice have a defect of NK cells and a high incidence of lymphoproliferative diseases. On the other hand, nude mice that are T-cell deficient but have normal NK activity do not have increased susceptibility to tumor development. Certain carcinogenic substances have been shown to suppress NK-cell activity. Long-term studies of NK-cell activity in patients with solid tumors revealed a positive correlation between NK-cell activity and survival time without metastasis (605). Decrease of NK-cell activity has been detected in patients with malignant lymphoma, in patients before relapse of leukemia, and also in women with breast cancer and metastasis to regional lymph nodes but not in those without involvement of lymph nodes.

Natural Killer Cells and Bone Marrow Stem Cells

When mice from inbred strains are crossed, the F1 progeny can accept organ grafts from either parent. Acceptance of the organ grafts results from the fact that the F1 animals express codominantly the class I antigens from both parents, and therefore, the F1 T cells recognize them as self. In contrast to organ grafts, however, the F1 mice reject bone marrow stem cells from either parent (606,607). This is known as hybrid resistance and is mediated by NK cells but not T cells.

Bone marrow stem-cell rejection is based on another set of antigens known as hematopoietic histocompatibility (Hh) antigens, which are inherited in a recessive pattern. The nature of Hh antigens and the mechanism of rejection of bone marrow stem cells in hybrid resistance are not well understood. NK cells may recognize the Hh antigens, or the lack of an MHC class I antigen on the donor cells may also play a role (missing self).

Natural Killer Cells and Bone Marrow Transplantation

NK cells engraft quickly after bone marrow transplantation and constitute the majority of the peripheral blood lymphocytes during the first few weeks after transplantation (608,609). In approximately one third of the patients after autologous or allogeneic bone marrow transplantation, a normally minor NK subpopulation (CD56⁺/CD16^-) was markedly expanded (up to 40% of peripheral blood lymphocytes) (610). This population may represent a different stage of differentiation from the major CD56⁺/CD16⁺ population.

In human bone marrow transplantation, NK activity may help in graft take by controlling viral infections, eliminating leukemic cells, or stimulating hematopoiesis (604).

Regulation of Hematopoiesis

Existing evidence suggests that NK cells exert both inhibitory and stimulatory effects on hemopoietic progenitors. These effects are mediated predominantly by the release of cytokines from NK cells rather than by cytotoxicity.

Strong evidence exists that an increase in the number of LGLs is associated with anemia or granulocytopenia. NK cells inhibit in vitro granulocytopoiesis in the granulocyte-macrophage colony-forming cell assay (611). NK cells also exert a promoting effect on hematopoiesis, however, and the net effect may actually depend on the stage of maturation of the progenitor cells (612).

In some patients with neutropenia, there is an increase of NK-like cells in the bone marrow but not in the peripheral blood. These cells can be identified only by immunophenotyping (613).

Decidual NK Cells

A unique subset of CD56^bright NK cells accumulate in the maternal decidua in direct contact with fetal trophoblasts (614). These NK cells are not derived from peripheral blood and produce IL-8 and interferon-inducible protein-10 chemokines. They secrete a variety of angiogenic factors, induce vascular growth in the decidua, and possess low or even inefficient cytotoxic abilities and overall orchestrate developmental processes in the fetal–maternal interface.

Natural Killer Cell Proliferations

Expansions of LGLs can be either transient or persistent. The former are reactive, whereas the latter can be either reactive or clonal. Markers for clonal expansions of LGLs are not available in practice and diagnosis of LGL expansions can be made by morphology and then confirmed by phenotype. Most patients have a chronic elevation of LGLs without other signs of malignancy and their clinical course is not progressive, even without treatment. In many cases, an associated disease (such as rheumatoid arthritis, hepatitis, or malignancy) has been observed. Artificially, a level of 2,000/μl of LGLs in the peripheral blood, which is five to seven times the normal value (250 to 450/μl), is believed to represent a lymphoproliferative disease of LGL if it persists for longer than 3 months (615). Some of the most common clinical presentations are fever, infections, neutropenia, anemia, and thrombocytopenia. Mortality was associated with moderately elevated counts of LGLs (2,000 to 3,000/μl) and very high counts (more than 7,000 μl), which indicates that LGL proliferations are, in general, highly heterogeneous.

The clonality of LGL proliferations cannot be demonstrated by immunophenotype. However, cytogenetic abnormalities have been detected (616,617), and in LGL cases that are CD57⁺ (and therefore express TCRs), rearrangements of TCR-β genes are detected in most of the cases (618). Because these cases are also associated with widespread involvement of several organs, such as the spleen, liver, and bone marrow, the term LGL leukemia has been proposed (616).

In some studies, the predominant phenotype of the LGL is CD57⁺ (HNK-1⁺)/CD3⁺, and these patients often have rheumatoid arthritis, neutropenia, and splenomegaly, a combination resembling Felty syndrome (619). Anemia and thrombocytopenia are not uncommon, and serologically, rheumatoid factor, antinuclear antibodies, hypergammaglobulinemia, and immune complexes are detectable.

Two major subtypes have been distinguished in terms of clinical presentation and the phenotype of the cells (620,621,622). The LGLs in type A, also known as T-LGLs, are CD57⁺/CD3⁺/CD8⁺/ CD2⁺, and those in type B, also known as NK-LGLs, are CD56⁺/ CD3⁺/CD8⁺/CD2⁺. Patients with T-LGL often have neutropenia (84%), rheumatoid arthritis, and autoantibodies. However, these clinical and serologic findings in patients with T-LGL (CD57⁺) could not be confirmed in other studies (623). The patients had clonal proliferations, as shown by TCR gene rearrangements, and the pathognomonic finding is neutropenia with an otherwise indolent course. NK-LGL leukemias are observed in younger patients and run an aggressive course (624). The most characteristic clinical presentation is hepatosplenomegaly and involvement of the gastrointestinal system with ascites. The course of the disease is acute, and often the patients die within 1 to 2 months after diagnosis.

Natural Killer T Cells

Natural killer T (NKT) cells have been identified as a novel lymphocyte lineage, which, in humans, express the cell-surface marker CD161 (NKR-PIA) (see Appendix) that is structurally related to several other proteins encoded by the NK-gene complex, including CD94, NKG2, and CD69 (625). These cells are specific for α-galactosylceramide (α-Gal-Cer) that binds to CD1d (626). Because α-glycosphingolipids are not present in mammals, the α-Gal-Cer probably mimics self-antigens that are recognized by NKT cells. They express V regions that are homologous to those that are expressed by mouse Vα14i T cells and have a rearrangement of Vα24 to Jα15, which forms the CDR3α that is highly similar to Vα14i of mice.

The NKT cells express an invariant Vα14 receptor (Vα14iTCP), which is used only by the NKT cells but not by T cells. The expression of an invariant TCR suggests that the selection is mediated by a monomorphic rather than polymorphic MHC molecule (627). Using CD1d/α-Gal-Cer tetramers (see Chapter 2 for details on this technique), Vα14i⁺ cells could not be detected in athymic mice, indicating that maturation and selection of NKT cells take place in the thymus (629) and represent a separate T-cell lineage. In a striking difference from the selection of mainstream T cells (i.e., by thymus epithelial cells), the NKT cells are selected by DP CD4⁺/CD8⁺ thymocytes expressing CD1d (630). After selection, the Vα14i precursor population is exported and acquires the NKT phenotype in the periphery (631). The distribution of the Vα14i T cells in the periphery is unique, as they migrate to the liver and from there to sites of inflammation responding to stimuli (i.e., chemokines and glycolipids).

The cells have a surface phenotype characteristic of recently activated or memory T cells, even when the cells are obtained from cord blood (632) or from germ-free mice (633). This is consistent with the postulated autoreactivity of the NKT cells in vivo and their peripheral expansion in the presence of autologous ligands, presented by CD1d.

Inhibitory receptors prevent unchecked autoreactivity for CD1d (634). The role for these NK receptors in directing the distribution and function of NKT cells in the body remains to be seen.

Natural Killer T-Cell Function

The NKT cells with a TCR expressing Vα24 play a critical role in immune responses. Vα24 cells are important for regulation of immune responses, inhibition of tumor development, and protection from autoimmune disease development (635,636). These cells have a cytolytic function and rapidly induce cytokines after stimulation (636). In response to these cytokines, they recruit and activate other cells (i.e., NK cells, T cells, B cells, and macrophages).

In humans, CD4⁺ Vα24i cells preferably produce IL-4, which is believed to regulate Th1/Th2 differentiation (637,638), but the Th1 versus Th2 polarization is more complex and depends on additional factors such as the number of antigenic stimulations with a shift to Th2 after multiple challenges (639,640).

Shift to Th2 cytokine secretion prevents the development of type I diabetes in mice with genetic predisposition (641), as well as allergic encephalomyelitis in another strain susceptible to the development of this disease (642). In experimental autoimmune diseases that develop spontaneously, there is a direct correlation between the development of the disease and the decline of Vα14 cells (643).

Selective reduction of Vα24 cells has also been shown in patients with systemic sclerosis (644), systemic lupus erythematosus, rheumatoid arthritis (645), and type I diabetes (646). Stimulation of NKT cells by α-Gal-Cer in animals triggers secretion of large amounts of cytokines (IFN-γ, IL-4, Il-2, IL-5, etc.) (647). In patients with systemic lupus erythematosus, these cells exert a protective effect against lupus development. The fact remains, however, that NKT cells promote Th2 responses and suppress Th1-dominant autoimmunity (648). Although their role in autoimmunity is surrounded by controversy, they may still be used as targets for clinical applications.

NKT cells have been implicated in the prevention and metastasis of tumors in mice (650); however, tumor suppression is regulated by more than one subset of NKT cells secreting IL-13 (651). Ultraviolet irradiation induces immunosuppression and skin cancer, and the suppression can be transferred by CD1-restricted NKT cells, which act as suppressor cells (652). Others have also found that NKT cells are not critical in the IL-12–mediated rejection of tumors (653), and it appears that the conflicting results in the literature about the role of NKT in tumor responses are due to complex factors, including the types of tumors and their microenvironment, and other cells interacting with NKT cells and, not least, with the functionally different subsets of NKT cells (654). NKT cells have been implicated in protective immunity against several pathogens (i.e., mycobacteria [655] and malaria parasites [656]).

In conclusion, NKT cells are phenotypically and functionally heterogeneous, endowed with cytolytic and cytokine secretion functions. They regulate several functions in the immune response that are related to host defenses against infections, autoimmunity, and tumor immunity. Their function seems to depend on their “age,” state of activation, regional microenvironments, and, of course, interaction with other cells. They have been called “nonconformist” and “unconventional” and probably for good reason, because they are suited to get involved in such a great variety of homeostatic disequilibrium conditions of the host.

Th17 CD4⁺ Cells

Effector CD4⁺ T cells develop in response to cytokines elicited from cells activated during an immune response. The best example of this response is the development of Th1 and Th2. Th1 cells enhance clearance of intracellular pathogens, while Th2 cells promote clearance of parasites. Another CD4 T-cell lineage recently described, known as Th17, represents an additional CD4⁺ T-cell effector cell, characterized by a distinct profile of effector cytokines such as IL-17, IL-17F, and IL-6 (657). In the absence of IFN-γ and IL-4, IL-23 induces naïve precursors to differentiate to Th17 cells (658). TGF-β is critical for up-regulation of the IL-23 receptor, thus conferring responsiveness to IL-23 (659). IL-6 probably is also needed in cooperation with TGF-β for Th17 cell development from naïve T cells (660). Others, however, have shown that TGF-β alone is a critical factor for differentiation of IL-17–producing effector cells (661). The role of IL-23 is not quite clear; it may not be required for Th17 commitment and IL-17 production, but it may be important for stabilizing the Th17 phenotype in chronic inflammation (662). The Th17 CD4⁺ lineage has evolved to control certain pathogen-induced diseases not covered by Th1 and Th2 (663).

References

1. Pollack A, Lampen N, Clarkson BD, et al. Identification of human B and T lymphocytes by scanning electron microscopy. J Exp Med 1973;138:607–624.

2. Pollack A, Hammerling U, Lampen N, et al. Surface morphology of murine B and T lymphocytes: a comparative study by scanning electron microscopy. Eur J Immunol 1975;5:32–39.

3. van der Loo EM, Gnossen J, Meijer CJ. Morphological aspects of T cell subpopulations in human blood: characterization of the cerebriform mononuclear cells in healthy individuals. Clin Exp Immunol 1981;43:506–516.

4. Meijer CJ, van der Loo EM, van Vloten WA, et al. Early diagnosis of mycosis fungoides and Sezary’s syndrome by morphometric analysis of lymphoid cells in the skin. Cancer 1980;45:2864–2871.

5. Klug DB, Carter C, Drouch E, et al. Interdependence of cortical thymic epithelial cell differentiation on T-lineage commitment. Proc Natl Acad Sci U S A 1998;95:11822–11827.

6. Schleup M, Wilcox N, Ritter MA, et al. Myasthenia gravis thymus: clinical, histological and culture correlations. J Autoimmun 1988;1:445–467.

7. Gill J, Malin M, Hollander GA, et al. Generation of complete thymic microenvironment by MTS24+ thymic epithelial cells. Nat Immunol 2002;3:635–642.

8. Benett AR, Farley A, Blair NF, et al. Identification and characterization of thymic epithelial progenitor cells. Immunity 2002;16:803–814.

9. Suniara RK, Jenkinson EJ, Owen JJ. An essential role for thymic mesenchyme in early T cell development. J Exp Med 2000;191:1051–1056.

10. Manley NR, Cappeechi MR. The role of Hoxa-3 in mouse thymus and thyroid development. Development 1995;121:1989–2003.

11. Su DM, Manley NR. Hoxa3 and Pax1 transcription factors regulate the ability of fetal thymic epithelial cells to promote thymocyte development. J Immunol 2000;164:5753–5760.

12. Nehls M, Pfeifer D, Schorp PM, et al. New member of the winged-helix protein family disrupted in mouse and rat nude mutations. Nature 1994;372:103–107.

13. Itoi M, Kawamoto H, Katsura Y, et al. Two distinct steps of immigration of hematopoietic progenitors into the early thymus anlage. Int Immunol 2001;13:1203–1211.

14. Blackburn CC, Augustine CL, Li R, et al. The nu gene acts cell-autonomously and is required for differentiation of thymic epithelia progenitors. Proc Natl Acad Sci U S A 1996;93:5742–5746.

15. Nehls M, Kyewski B, Messerle M, et al. Two genetically separable steps in the differentiation of thymic epithelium. Science 1996;272:886–889.

16. Bockman DE, Kirby ML. Dependence of thymus development on derivatives of neural crest. Science 1984;223:498–500.

17. Anderson G, Jenkinson EJ. Lymphostromal interactions in thymic development and function. Nat Rev Immunol 2001;1:31–40.

18. Manely NR. Thymus organogenesis and molecular mechanisms of thymic epithelial cell differentiation. Semin Immunol 2000;12:421–428.

19. Blackburn CC, Manley NR, Palmer DB, et al. One for all and all for one: thymic epithelial stem cells and regeneration. Trends Immunol 2002;23:391–395.

20. Bennett AR, Farley A, Blaer NF, et al. Identification and characterization of thymic epithelial progenitor cells. Immunity 2002;16:803–814.

21. Ohuchi H, Hori Y, Yamasaki M, et al. FGF10 acts as a major ligand for FGF receptor 2IIIb in mouse multi-organ development. Biochem Biophys Res Commun 2000;277:643–649.

22. Revest JM, Suniara RK, Kerr K, et al. Development of the thymus requires signaling through the fibroblast growth factor receptor r2-iiib. J Immunol 2001;167:1954–1961.

23. Hollander CA, Wong B, Nichogiannopoulou A, et al. Developmental control point in induction of thymic cortex regulated by a subpopulation of prothymocytes. Nature 1995;373:350–353.

24. Van Ewijk W, Shores EW, Singer A. Crosstalk in the mouse thymus. Immunol Today 1994;16:214–217.

25. Ritter MA, Boyd RL. Development in the thymus: it takes two to tango. Immunol Today 1993;14:462–469.

26. Suhr CD, Ernst B, Sprent J. Growth of epithelial cells in the thymic medulla is under the control of the mature T cells. J Exp Med 1992;176:611–616.

27. Van Ewijk W. T-cell differentiation is influenced by thymic microenvironments. Annu Rev Immunol 1991;9:591–615.

28. Boyd RL, Tucek CL, Godfrey DI, et al. The thymic microenvironment. Immunol Today 1993;14:445–459.

29. van Vliet E, Melis M, van Ewijk W. Immunohistology of thymic nurse cells. Cell Immunol 1984;87:101–109.

30. Kyewski BA. Thymic nurse cells possible sites of T cell selection. Immunol Today 1986;7:374–379.

31. Mombaerts P, Clarke AR, Rudnicki MA, et al. Mutations in T-cell antigen receptor genes α and β block thymocyte development at different stages. Nature 1992;360:225–231.

32. Goverman J, Brabb T, Huseby ES, et al. TCR signaling regulates thymic organization: lessons from TCR-transgenic mice. Immunol Today 1997;18:204–208.

33. Brabb T, Huseby ES, Morgan TM, et al. Thymic stromal organization is regulated by the specificity of T cell receptor/major histocompatibility complex interactions. Eur J Immunol 1997;27:136–146.

34. Matechak EO, Kileen N, Hedrick SM, et al. MHC class II-specific T cells can develop in the CD8 lineage when CD4 is absent. Immunity 1996;4:337–347.

35. Hogquist KA, Jameson SC, Heath WR, et al. T cell receptor antagonist peptides induce positive selection. Cell 1994;76:17–27.

36. Anderson M, Anderson SK, Farr AG. Thymic vasculature: organizer of the medullary epithelial compartment? Intern Immunol 2000;12:1105–1110.

37. Schonrich G, Momburg F, Hammerling GJ, et al. Anergy induced by thymic medullary epithelium. Eur J Immunol 1992;22:1687–1691.

38. Hoffmann MW, Allison J, Miller JF. Tolerance induction by thymic medullary epithelium. Proc Natl Acad Sci U S A 1992;89:2526–2530.

39. Klein L, Klein T, Ruther U, et al. CD4 T cell tolerance to human C reactive protein, an inducible serum protein is mediated by medullary thymic epithelium. J Exp Med 1998;188:5–16.

40. Farr AG, Dooley JL, Erickson M. Organization of thymic medullary epithelial heterogeneity: implications for mechanisms of epithelial differentiation. Immunol Rev 2002;189:20–27.

41. Erickson M, Morkowski S, Lehar S, et al. Regulation of thymic epithelium by keratinocyte growth factor. Blood 2002;100:3269–3278.

42. Mackall CL, Gress RE. Thymic aging and T cell regeneration. Immunol Rev 1997;160:91–102.

43. Mackall CL, Punt JA, Morgan P, et al. Thymic function in young/old chimeras: substantial thymic regenerative capacity despite irreversible age-associated thymic involution. Eur J Immunol 1998;28:1886–1893.

44. Sano S, Takahuma Y, Sugawara T, et al. Stat3 in thymic epithelial cells is essential for postnatal maintenance of thymic architecture and thymocyte survival. Immunity 2001;15:261–273.

45. Savino W, Villa-Verde DM, Lannes-Vieira J. Extracellular matrix proteins in intrathymic T-cell migration and differentiation? Immunol Today 1993;14: 158–161.

46. Kutlesa A, Siler U, Speiser A, et al. Developmentally regulated interactions of human thymocytes with different laminin isoforms. Immunology 2002;105: 407–418.

47. Shortman K, Wu L. Early T lymphocyte progenitors. Annu Rev Immunol 1996; 14:29–47.

48. Huang S, Terstappen LWMM. Lymphoid and myeloid differentiation of single human CD34+HLA-DR+CD38^- hematopoietic stem cells. Blood 1994;83: 1515–1526.

49. Galy AHM, Cen D, Travis M, et al. Delineation of T-progenitor cell activity within the CD34+ compartment of adult bone marrow. Blood 1995;85:270–278.

50. Tjönnfjord GE, Veiby OP, Steen R, et al. T lymphocyte differentiation in vitro from adult human pre-thymic CD34+ bone marrow cells. J Exp Med 1993;177: 1531–1539.

51. Ikuta K, Uchida N, Friedman J, et al. Lymphocyte development from stem cells. Annu Rev Immunol 1992;10:759–783.

52. Weerkamp F, Baert MR, Dik WA, et al. Human thymus contains multipotent progenitors with T/B lymphoid, myeloid, and erythroid lineage potential. Blood. 2006;107:3131–3137.

53. Krueger A, Garbe AI, von Boehmer. Phenotypic plasticity of T cell progenitors upon exposure to Notch ligands. J Exp Med 2006;203:1977–1983.

54. Perry SS, Weiner RS, Kauro T, et al. Primitive lymphoid progenitors in bone marrow, with T lineage reconstituting potential. J Immunol 2006;177: 2880–2887.

55. Liu C, Veno T, Kuse S. The role of CCL21 in recruitment of T-precursor cells to fetal thymi. Blood 2005;105:31–39.

56. Scimone ML, Aifantis I, Apostolou I, et al. A multistep adhesion cascade for lymphoid progenitor cell homing to the thymus. Proc Nat Acad Sci U S A 2006;103:7006–7011.