Signaling Life and Death in the Thymus: Timing Is Everything

See allHide authors and affiliations

Science  21 Mar 2003:
Vol. 299, Issue 5614, pp. 1859-1863
DOI: 10.1126/science.1067833


T lymphocytes are generated in the thymus, where developing thymocytes must accept one of two fates: They either differentiate or they die. These fates are chiefly determined by signals that originate from the T cell receptor (TCR), a single receptor complex with a remarkable capacity to decide between distinct cell fates. This review explores TCR signaling in thymocytes and focuses on the kinetic aspects of ligand binding, coreceptor involvement, protein phosphorylation, and mitogen-activated protein kinase (MAPK) activation. Understanding the logic of TCR signaling may eventually explain how thymocytes and T cells distinguish self from nonself, a phenomenon that has fascinated immunologists for 50 years.

Upon infection with a pathogen, proteolytic fragments of the pathogen's proteins become bound to the major histocompatibility complex (MHC) proteins of the host and expressed on the surface of antigen-presenting cells (APCs). This complex of foreign peptide and self-MHC protein forms the ligand that initiates a T cell response and requires that the TCR recognize contributions from both the host's MHC protein and the pathogen- derived peptide, a situation termed MHC restriction (1). MHC genes are highly diverse in the population, making it extremely rare for pathogen-derived peptides to avoid binding with all MHC variants. By this means, MHC polymorphism helps to protect the population. Although MHC diversity is a clear advantage in terms of antigen presentation, this same diversity raises a problem during T cell development (2). A repertoire of T cells has to be selected whose receptors have the potential to recognize foreign antigenic peptides bound to the individual's own MHC molecules (3).

Customizing the T Cell Repertoire: Two Forms of Selection

After arriving as precursors from the bone marrow, small numbers of CD4CD8 (DN) thymocytes proliferate and differentiate into CD4+CD8+(DP) cells. Proliferation and differentiation of DN cells are driven by a distinct surface receptor, composed of a TCR β chain and an invariant pre-Tα chain (4). The majority (∼97%) of DP thymocytes die by neglect because they fail to recognize any of the available MHC molecules and consequently do not receive a survival signal. Remaining DP thymocytes (∼3%) expressing TCRs that bind self-MHC/self-peptide molecules with low affinity (5) are rescued through positive selection and differentiate into MHC-restricted, helper and cytotoxic T cells (Fig. 1). Ultimately, positive selection produces a customized T cell repertoire, consisting of mature T cells expressing receptors, which can identify the individual's MHC proteins. It is equally critical that mature T cells are not activated by self-peptide/MHC (pMHC) ligands, and negative selection eliminates thymocytes with high-affinity TCRs for self-MHC/self-peptide ligands by inducing apoptosis in those cells (Fig. 1). Consequently, the repertoire of mature T cells is considerably less likely to react against self-antigens and induce autoimmunity. To ensure that peripheral T cells remain tolerant to self-pMHC antigens, other mechanisms involving regulatory T cell subsets come into play (6).

Figure 1

The effect of ligand affinity on thymocyte and T cell fate. CD4+CD8+ (DP) cells constitute the majority of thymocytes. A DP cell, whose TCR fails to recognize an MHC molecule in the thymus, dies by neglect; i.e., it fails to receive a survival signal. Among the remaining DP thymocytes, those with low-affinity TCRs for pMHC ligands undergo positive selection (survival and differentiation). Thymocytes whose antigen receptors bind self–class I MHC/self-peptide complexes with low affinity differentiate into MHC class I–restricted, CD8+ cytotoxic T cells. Similarly, thymocytes whose antigen receptors bind self–class II MHC/self-peptide ligands differentiate into MHC class II–restricted, CD4+ helper T cells. In peripheral lymphoid organs, mature T cells utilize the interaction with low-affinity (self-MHC/self-peptide) ligands to ensure their survival, whereas mature T cells whose receptors engage a high-affinity ligand (self-MHC/foreign peptide) are activated and initiate an immune response.

Low-affinity, pMHC ligands are recognized by mature T cells, providing survival signals that maintain the peripheral T cell pool (7). In contrast, a T cell is activated when confronting a high-affinity pMHC ligand, generally consisting of a self-MHC and a foreign peptide (Fig. 1). The fate of individual thymocytes and T lymphocytes is largely determined by the affinity of their TCRs for the available pMHC ligands. Although much is known about positive and negative selection on a cellular level, a full understanding of how the TCR “reads” ligand affinity, initiates distinct intracellular signals, and induces the appropriate cellular response is still lacking.

The TCR: One Receptor, Two Kinds of Signals

The TCR is a rare example of a single receptor complex capable of delivering two signals that result in profoundly different cell fate decisions. The αβ TCR is composed of two polypeptides, α and β, whose variable regions are responsible for binding pMHC ligands. Constant regions of the α and β chains couple the antigen binding domains to the CD3 complex (Fig. 2), which contains three dimers (ɛγ, ɛδ, and ζζ) and is responsible for signal transduction (2). Although each of these chains expresses immuno-tyrosine activation motifs (ITAMs), which can be tyrosine phosphorylated by protein tyrosine kinases (PTKs) such as Lck, it is not yet clear whether each individual ITAM has a unique physiological role (8). However, a recent study indicates that the ITAM on CD3γ may influence the efficiency of positive selection (9). αβ TCRs themselves have several unique structural features; e.g., the α chain connecting peptide motif (α-CPM), that is present in the membrane-proximal domain of TCR α chains in all vertebrate species (10). The CD3δ chain is associated with the TCR α chain (11) and, along with the α-CPM, may have evolved to perform a specialized function for the αβ receptor (see below).

Figure 2

Interactions of the TCR with its ligand and coreceptor. (A) A diagram of a class I MHC–restricted TCR with its associated CD3 complex, and CD8αβ coreceptor with the attached PTK, lck. (Right) The orientation of a class I MHC-restricted TCR's variable region (shown in pink) is superimposed on its MHC/peptide ligand (shown in yellow and white) (67). The class I MHCs coreceptor docking site (red circle) recruits CD8 to the α-chain side of the TCR. (B) A similar diagram of a class II MHC–restricted TCR with its associated CD4 coreceptor, which is also attached to lck. (Right) The orientation of this receptor's variable region (shown in light blue and green) is superimposed on its class II MHC/peptide ligand (shown in yellow, light green, and white) (68). Similar to the class I MHC, the class II MHC's coreceptor docking site (blue circle) recruits its coreceptor, CD4, to the α-chain side of the TCR.

How does a thymocyte discriminate between pMHC ligands of differing affinity? Because high-affinity ligands would tend to occupy more TCRs than would low-affinity ligands, a developing thymocyte could distinguish a low- from a high-affinity ligand by counting the number of receptors occupied over a given period of time (12). A prediction of the receptor occupancy (avidity) model is that low doses of a high-affinity peptide (occupying few TCRs) should lead to positive selection (13, 14). However, it has been difficult to consistently verify this prediction, raising the question of whether the occupancy/avidity explanation of thymic selection is in fact correct (15). An alternative view is that the individual receptor could distinguish ligand affinity by “reading” its occupancy time (16). As a result, the receptor could generate an ordered set of biochemical signals such that the earliest signals lead to positive selection and later signals result in negative selection, an idea known as kinetic proofreading (17). Because of their short occupancy time, low-affinity ligands would be limited to inducing only early TCR signals and thus lead to positive selection. High-affinity ligands, on the other hand, contact the TCR for a longer time and induce late TCR signals that initiate negative selection.

Affinities of positive- and negative-selecting ligands for their TCR have been examined by plasmon resonance. These experiments have shown that the TCR's affinity for a positively selecting pMHC complex is only one-third of its affinity for a ligand capable of negative selection (18), although the affinity of a negatively selecting ligand for its corresponding TCR may be increased at physiological temperatures (19). This difference arises largely because the positively selecting ligand has a faster off-rate. One limitation of plasmon resonance measurements is that they are carried out with soluble TCRs and pMHC ligands, in the absence of CD4 or CD8 coreceptors, which increase the affinity of the TCR. Another important issue is to define the ligand affinity where the cellular response makes a transition between positive and negative selection. In other words, what happens with a ligand of intermediate affinity? Would thymocytes responding to an intermediate-affinity ligand escape negative selection, and would these thymocytes be a source of mature T cells capable of initiating an autoimmune response?

TCR-Coreceptor Interactions

The coreceptors CD4 and CD8 are expressed on helper and cytotoxic T cells, respectively, and act to strengthen TCR engagement/pMHC by binding directly to MHC molecules (20, 21) (Fig. 2). From the perspective of a DP thymocyte, one means by which the thymocyte might discriminate whether its TCR has bound a class I or a class II pMHC ligand is by the coengagement of CD8 or CD4, respectively.

Although the mechanism by which DP thymocytes decide to differentiate toward the CD4 helper T cell or the CD8 cytotoxic T cell lineage is not well understood, one important parameter may be the strength of the signal, which in part is determined by the coreceptor (22). TCRs that have coengaged CD4 are thought to deliver a stronger signal, because the PTK, Lck, is bound to CD4 more tightly than it is to CD8 (23). Consistent with this idea is the observation that in the presence of a dominant-negative form of Lck, thymocytes expressing a class II MHC–restricted TCR develop into CD8 T cells. In contrast, DP thymocytes expressing a class I MHC–restricted TCR develop into CD4 T cells in the presence of a constitutively active form of this PTK (24). Thus, by altering the levels of Lck in the thymocyte, the lineage decision during positive selection can be modified (25).

The ability of the CD8 coreceptor to bind to MHC molecules undergoes a developmentally regulated alteration that is coincident with positive selection (26, 27). The CD8 expressed on immature (DP) thymocytes contains O-glycans, which are poorly sialylated, whereas the CD8 on mature SP cytotoxic T cells contains an increased amount of sialic acid, especially on the CD8β chain. Poorly sialylated CD8 molecules on DP cells bind pMHC tetramers without the participation of the TCR; the more fully sialylated CD8 on mature (SP) thymocytes contributes to tetramer binding, but cannot bind these artificial ligands without TCR involvement. The avid binding of CD8 to class I MHC molecules may be important to the recognition of low-affinity pMHC ligands during positive selection of DP thymocytes. The addition of sialic acid to CD8 could explain why mature CD8+ T cells are unresponsive to low-affinity ligands expressed in the periphery.

Two mutations have been particularly informative regarding the ability of the TCR-CD3 complex to transmit distinct positive and negative selection signals. Thymocytes expressing TCRs, which lack CD3δ or a conserved motif in the α−chain constant region (α-CPM), fail to undergo positive selection, but remain competent to be negatively selected (28–31). The poor association of CD3δ with TCRs lacking the α-CPM indicates that CD3δ- and α-CPM–deficient receptors have a similar structural problem. The existence of such mutants implies that bifurcation between positive and negative selection signals originates on the cell surface. The CD8 coreceptor increases the ability of the TCR to bind pMHC complexes (32), and recent studies (33) suggest that CD8 is linked to the TCR by way of CD3δ and/or the α-CPM. TCRs lacking the α-CPM fail to profit from the presence of CD8 on the T cell surface. This phenomenon is most pronounced with low- affinity peptide ligands (34) and may explain the defect in positive selection seen in thymocytes expressing α-CPM mutant receptors. Because higher affinity ligands are less dependent on coreceptor assistance (35), negative selection would be maintained in the absence of the α-CPM (Fig. 3).

Figure 3

Ligand affinity and coreceptor recruitment. A class I MHC–restricted TCR (dark blue) can bind both high- and low-affinity pMHC ligands (yellow and white), which results in engagement of the coreceptor, CD8 (red). The class I MHC molecule recruits the NH2-terminal domain of CD8α using its coreceptor docking site. The TCR likely engages CD8 by way of the CD3δ chain and/or the α-CPM (light blue). In the presence of CD3δ and the α-CPM, CD8 enhances ligand binding and signal transduction from the TCR. In the absence of CD3δ or the α-CPM, CD8 can still participate in the recognition of a high-affinity pMHC ligand owing to the ligand's slow off-rate and long occupancy of the TCR. In contrast, CD3δ or α-CPM deficiency has a disproportionate effect on the response to low-affinity ligands. Because of its fast off-rate and short occupancy of the TCR, CD8 is not recruited, and signaling is only poorly initiated. This may explain the dependence of positive selection on CD3δ and the α-CPM.

Crystallographic studies of murine and human TCR-pMHC complexes have revealed a conserved orientation of the pMHC over the TCR (36), placing the coreceptor binding site of the MHC protein on the α-chain side of the TCR (Fig. 2). Linking CD8 to the TCR through the α-CPM/CD3δ places the coreceptor in an optimal position to engage its binding site on the MHC (Fig. 3). It is not yet known whether, like CD8, CD4 is linked to the TCR through the α-CPM/CD3δ, although a class II–restricted TCR lacking the α-CPM fails to generate positive selection signals (30).

Is the Distinction Between Selection Signals Regulated by the Kinetics of ERK Activation?

The family of MAPKs includes the extracellular signal–regulated kinase (BRK), p38, and the c-Jun NH2-terminal kinase (JNK); these enzymes play a crucial role in defining a cellular response (37). Studies of the MAPK cascades in thymocytes indicate that no single MAPK is uniquely activated by either a positively or a negatively selecting ligand (31). One distinction between positive and negative selection signaling may lie in the kinetics of ERK activation (31, 38). For example, positively selecting ligands induce a slow and sustained accumulation of ERK activity (Fig. 4A), whereas ligands that induce negative selection provoke a strong and transient burst of ERK activity (Fig. 4B). In comparison, both positively and negatively selecting ligands induce p38 and JNK with similar kinetics (31). Thus, during positive selection, maximal ERK activation occurs after p38 and JNK, whereas in response to a negatively selecting ligand, ERK is activated before p38 and JNK (Fig. 4).

Figure 4

Kinetics of positive and negative selection signals. (A) During positive selection, LAT phosphorylation, Ca2+ mobilization, and ERK activation are slowly induced over a sustained period of time. This contrasts with the induction of p38 and JNK, which are activated with different kinetics. Low-affinity pMHC ligands, which occupy the TCR for a relatively short time, could induce the phosphorylation of Tyr175 and Tyr195 on LAT. This might recruit Gads, SLP-76, and PLC-γ1 and lead to the slow production of DAG and the slow activation of RasGRP, Ras, and ERK. (B) During negative selection, LAT phosphorylation, Ca2+mobilization, and ERK activation are rapidly and transiently induced. High-affinity ligands occupy the TCR for longer times and might induce the phosphorylation of Tyr175, Tyr195, and Tyr136. The phosphorylation of the Tyr136 residue might stabilize PLC-γ1 within the growing TCR signalosome, leading to the rapid activation of Ras and ERK. High-affinity ligands could also lead to the phosphorylation of Tyr113, Tyr132, and Tyr235, which recruit Grb-2 and the guanine nucleotide exchange factor, Sos. This could increase the speed of ERK activation. Diacylglycerol kinase (DGK) and RasGAP might terminate ERK activity. Interestingly, the kinetics of p38 and JNK activation during negative selection are the same as during positive selection.

The activation of the ERK cascade promotes contrary cellular responses (39), including cell proliferation, cycle arrest, differentiation, senescence, or apoptosis. In the pheochromocytoma cell line, PC12, stimulation through the epidermal growth factor receptor results in proliferation and is associated with a transient activation of ERK (40), whereas stimulation through the nerve growth factor receptor produces a sustained activation of ERK and results in differentiation (40). This model provides evidence that two distinct receptors, using separate ligands, can initiate ERK activity with differential kinetics to implement a cell fate decision. Even more astonishing, cell fate in DP thymocytes is determined by a single receptor, which can differentially modulate ERK activity, depending on the affinity of its ligand.

Dynamic changes in the composition of TCR signaling complexes (so called TCR-signalosomes) are induced by ligand binding to the receptor (41). What regulates changes in the composition of these complexes is not known, but an attractive candidate is the linker for activation of T cells (LAT), a protein thought to provide the principal scaffold of the TCR-signalosome. LAT has nine tyrosine residues, which can be differentially phosphorylated and serve as inducible binding domains for various proteins containing SH2 motifs (42–44). Potentially, ligand binding to the TCR might induce the sequential phosphorylation of these tyrosines, allowing an ordered recruitment of specific signaling components. In this model, the extent to which LAT is tyrosine phosphorylated would consequently determine the composition of the TCR-signalosome, exerting an influence on the outcome of thymic selection. Conceivably, positively selecting ligands may induce a delayed phosphorylation of LAT, involving Tyr175 and Tyr195, thus leading to the recruitment of Gads/SLP-76/ phospholipase C–γ1 (PLC-γ1) to the membrane (Fig. 4A). However, this four-component complex might be unstable if Gads were not tightly bound to LAT, resulting in a weak activation of PLC-γ1. This in turn would generate a low-amplitude but sustained mobilization of Ca2+. Weakly activated PLC-γ1 would also lead to a slow hydrolysis of phosphatidylinositol-4,5-bisphosphate (PIP2) and, consequently, a delayed accumulation of diacylglycerol (DAG), the metabolite that recruits the Ras activator, RasGRP, to the membrane (45). The slow but sustained production of DAG might determine the prolonged kinetics of RasGRP recruitment, and the activation of Ras and ERK (Fig. 4A). In mice deficient for RasGRP, thymocytes are prevented from undergoing positive selection but not negative selection, supporting a key role for this guanine nucleotide exchange factor (46, 47).

On the other hand, high-affinity ligands could induce a fast and strong phosphorylation of LAT, involving the majority of its tyrosines (Fig. 4B). As in positive selection, phosphorylation of Tyr175 and Tyr195 could recruit the Gads/SLP-76/ PLC-γ1 complex, but this complex might be stabilized by the additional phosphorylation of Tyr136 on LAT. A rapid and maximal phospholipase activity could ensue, inducing a fast and transient mobilization of intracellular Ca2+ (Fig. 4B). The importance of LAT's Tyr136 has been confirmed in gene-targeted mice, in which this tyrosine was mutated to phenylalanine (48, 49). T cells from these animals exhibit an autoimmune phenotype, consistent with a defect in T cell negative selection. Phosphorylation of additional tyrosine residues (Tyr113, Tyr132, and Tyr235) on LAT would recruit the adaptor, Grb-2, and bring about synergistic activation of Ras through recruitment of the guanine nucleotide exchange factor, Sos (Fig. 4B). In this regard, it is interesting that hemizygous expression of Grb-2 diminishes negative selection, but not positive selection (50), and that activation of p38 and JNK is decreased in Grb-2+/− mice as well (50). Targeted disruption of gads as well as deletion of the Gads binding site in SLP-76 impaired both positive and negative selection, revealing a central role for these proteins in transducing both types of thymic selection signals (51,52).

Several mechanisms may account for the transient kinetics of ERK activation during negative selection. A strong initial rise in ERK activity might be self-limited; indeed, ERK has been documented to phosphorylate Lck and decrease the activity of this PTK, leading to a decrease in receptor signaling (53). High-affinity agonist ligands also induce the down-regulation of TCRs from the cell surface, resulting in the termination of receptor-generated signals. TCR down-regulation is impaired in Cbl- and SLAP-deficient thymocytes (54, 55). The importance of these observations lies in the fact that the ubiquitin- ligase activity of the adaptor, Cbl, targets the TCR for degradation, whereas the SRC-like adaptor protein, SLAP, may stabilize the recruitment of Cbl to the TCR-signalosome (56). Interestingly, in Cbl deficient-thymocytes, ERK activity is enhanced, suggesting that Cbl may negatively regulate the ERK cascade. Finally, recent work has demonstrated the involvement of DAG kinases (DGKs) in TCR signaling (57, 58). DGKs modulate lipid-induced signal transduction by converting DAG to phosphatidic acid (PA), thereby impairing the activation of DAG-dependent signaling components. It has been shown that DGKs interfere with the Ras pathway by blocking DAG-induced RasGRP activation (57, 58). Furthermore, PA, the end product of DGK activity, triggers RasGAP, a negative regulator of Ras (59).

Duration of ERK Activity and the Cellular Response

High-affinity ligands activate the ERK pathway before they activate p38 and JNK (Fig. 4B) (31). The activation of ERK prior to the other MAPKs may commit the thymocyte to apoptosis and negative selection. On the other hand, a low- affinity ligand induces maximal ERK activity after the induction of p38 and JNK (Fig. 4A) (31). This could trigger a different set of transcription factors, which lead to thymocyte survival and differentiation. The transcription factor E2F-1 (60,61) is required for negative selection, whereas the deficiency of the transcription factors Schnurri-2 (Shn-2) or Id3 affects only positive selection (62, 63). Although Id3 is a target of the ERK cascade, its activation in relation to the kinetics of ERK has not yet been studied.

Another way in which a sustained activation of ERK could potentially lead to positive selection is by the phosphorylation of BAD through the ERK effector kinase, Rsk (64). Phosphorylation of BAD diminishes its apoptotic effect, leading to an increase in thymocyte survival, an essential element of positive selection (65). The strong and transient increase in ERK activity induced by a negative selection ligand might not allow for a sufficiently long phosphorylation of BAD to promote thymocyte survival. Finally, the Bcl-2 family member Bim is essential for negative selection because Bim−/− mice fail to delete thymocytes expressing high-affinity TCRs for self-pMHC ligands (66). How TCR engagement leads to the activation of this mediator of apoptosis is not yet understood. Future work should delineate the targets of the MAPK pathways to better appreciate the downstream signaling pathways regulating thymocyte differentiation and death.


To generate a population of T cells that can discriminate between self and nonself, thymocytes have a mechanism for responding to different ligands in very distinct ways. Several kinetic parameters play a role in determining the specificity of the TCR-generated signal, which include the off-rate of ligand binding, the rate of coreceptor recruitment, the speed of forming a TCR-signalosome, and the kinetics of MAPK activation. Thymocytes have apparently evolved distinct cellular responses to the timing of ERK activity, and the resulting capacity for self/nonself discrimination is critical to the generation of a functional immune system.

  • * To whom correspondence should be addressed. E-mail: guy.werlen{at} (G.W.), ed.palmer{at} (E.P.)


View Abstract

Stay Connected to Science

Navigate This Article