- Email: [email protected]
From TCR Engagement to T Cell Activation
Cell, Vol. 96, 1–4, January 8, 1999, Copyright 1999 by Cell Press
From TCR Engagement to T Cell Activation: A Kinetic View of T Cell Behavior Antonio Lanzavecchia,* Giandomenica Iezzi, and Antonella Viola Basel Institute for Immunology Grenzacherstrasse 487 CH-4005 Basel Switzerland
Sensitivity, specificity, and context discrimination are three key properties of T cell antigen recognition. T lymphocytes recognize antigen as peptides bound to major histocompatibility complex (MHC) molecules on the surface of antigen-presenting cells (APC). Since MHC molecules are highly promiscuous in peptide binding, T cell epitopes will be generally displayed on APC at low copy number, and consequently T cell recognition must be exquisitely sensitive. In addition, recognition must be highly specific, since T cells must discriminate few antigenic peptide–MHC complexes among a vast excess of irrelevant, yet highly homologous, complexes of the same MHC molecules carrying a variety of self-peptides. Finally, T cells must be able to interpret the context in which a given antigen is presented to mount an effector response to infectious antigens or ignore innocuous or self-antigens. In this minireview we will discuss recent evidence indicating that these three properties of T cells are intimately connected and are dependent on the kinetics of the T cell receptor (TCR)–ligand interaction and T cell stimulation. The Elusive Mechanism of TCR Stimulation While the biochemical events that follow TCR triggering are understood in increasing detail, the mechanism that initiates this cascade of events is still unresolved. In principle, the initial triggering event, induced by ligand binding, can be envisaged either as a conformational change of the TCR-CD3-z complex (hereafter referred to as TCR) or as the induction of some form of aggregation. The latter may involve the formation of either homologous clusters of two or more ligand-engaged TCRs or heterologous clusters of one ligand-engaged TCR with other signal transduction components. Structural studies have failed thus far to support the conformational change model. In addition, two recent experiments performed using soluble peptide–MHC complexes make this model highly unlikely. Both studies demonstrate that T cell triggering cannot be induced by monovalent ligation of TCR by soluble peptide–MHC complexes, but rather requires some form of oligomerization. When defined oligomers of MHC class II–peptide complexes were used to trigger T cells, it was found that at least three TCRs need to be brought together in order to induce a calcium response (Boniface et al., 1998). In another study however, T cell activation could be induced either by dimers of MHC class I–peptide complexes or by monomers that cross-link the TCR with the CD8 coreceptor–Lck kinase complex (Delon et al., * To whom correspondence should be addressed (e-mail: lanza [email protected]).
Minireview
1998). While these studies represent a clear step forward in defining the requirements for aggregation in initiating TCR signaling, they still leave several open questions. The first model does not explain how a few complexes (say, less than 10) per APC might rapidly trigger T cells, while the second does not account for CD8-independent responses. What is ultimately important is to understand the mechanism of TCR triggering within the constraints of the two cell membranes. The solution to this problem has so far been elusive due to the lack of appropriate experimental systems. These membrane constraints may favor partition of molecules at the T cell/APC contact region and influence the local concentrations of the reactants and the kinetics of receptor–ligand interaction (Shaw and Dustin, 1997; Monks et al., 1998). In this context, it is interesting to consider the possibility that monovalent TCR ligation may be sufficient to induce clustering with signaling molecules (Figure 1). Recent evidence indicates that many signaling components, such as Lck and LAT, which are key to initiating and propagating TCR signal transduction, are concentrated in cholesterol-enriched membrane microdomains, called rafts (Simons and Ikonen, 1997; Zhang et al., 1998). Two recent studies demonstrate that rafts are required for efficient T cell activation and that triggered TCRs are recruited to the rafts (Montixi et al., 1998; Xavier et al., 1998). We would like to suggest that monovalent ligation of TCR to the APC surface, by reducing its lateral mobility within the T cell membrane, might drive its recruitment into the kinase-rich environment of rafts, where
Figure 1. Initiation of TCR Triggering and Signal Transduction (a–c) Mechanisms that may promote TCR triggering. (a) Oligomerization of TCRs induced by ligand binding; (b) Cross-linking of TCR with the CD8-Lck complex induced by monovalent ligand. (c) Lateral translocation of a single TCR to a kinase-rich domain (raft) induced by actin-driven motility of the T cells and shear stress. The passive mobility of a single TCR complex is transiently decreased upon ligation to a peptide–MHC complex on APC. At the same time the rafts are laterally displaced by actin-driven motility, so that collisions between single TCRs and kinase-rich rafts are promoted. (d) Propagation of signal from the triggered TCR to raft-associated adaptors such as LAT. After TCR triggering is completed the peptide–MHC complex can dissociate and become available for another cycle of engagement and triggering.
Cell 2
the phosphorylation cascade can be triggered, resulting in complete phosphorylation of CD3 and z ITAMs by Lck, followed by docking and activation of ZAP-70, a cytoplasmic tyrosine kinase. This complex, which we refer to as a triggered TCR, can now propagate the signal by phosphorylating adaptors such as LAT that couple it to several downstream signaling pathways. In this model the actin-based T cell motility plays a key role in the initiation of TCR triggering (Valitutti et al., 1995a). Kinetic Threshold for TCR Triggering Peptide–MHC complexes do not need to bind with high affinity in order to trigger TCR. Strong agonists are typically characterized by a Kd of 1–90 mM and half-lives of z10 s (Davis et al., 1998 and references therein). However, small changes in peptide sequence, leading to slightly faster off rates, have been shown to greatly affect their capacity to trigger T cells. These altered peptide ligands may simply have a decreased potency (weak agonist), may trigger a partial response (partial agonists), or may even inhibit activation by TCR agonists (antagonists). The altered responses have been correlated to a distinct pattern of CD3/z phosphorylation and lack of ZAP-70 activation (Kersh and Allen, 1996). The kinetic proofreading model (McKeithan, 1995) provides a conceptual framework to analyze the relationship between kinetics of TCR–ligand interaction and T cell response. According to this model, the engagement of TCR by peptide–MHC does not immediately lead to TCR triggering, because a series of phosphorylation steps followed by recruitment and activation of ZAP-70 need to be performed, a process that requires time. If the duration of the interaction is sufficient, the phosphorylation and docking events will proceed until the fully active complex, which we refer to as a triggered TCR, is assembled and can transmit the signal to downstream pathways. In contrast, a premature dissociation before the process has been completed will lead to the formation of inactive intermediates that, by sequestering substrates, inhibit activation by agonists, resulting in TCR antagonism. This mechanism has been documented in the case of FceRI where antagonism is the result of competition for limiting amounts of kinase (Torigoe et al., 1998). An alternative possibility is that the intermediates generated might deliver an incomplete or even negative signal, thus explaining partial agonism and antagonism. The dependence of signaling on stepwise formation of multiple intermediates enhances the fidelity of kinetic discrimination by allowing for a greater number of proofreading steps (Rabinowitz et al., 1996). In this way, the kinetic proofreading model defines a minimum time threshold for the TCR/peptide–MHC interaction to be productive. An important aspect of kinetic proofreading is that the rate of the individual steps depends on the concentration of the reactants. Therefore, this rate may differ in different cell types and can be modulated by variations in the total level or local distribution of signaling molecules, thus potentially explaining developmentally regulated changes in responsiveness to TCR ligands. The kinetic proofreading model is supported by recent biochemical data showing a multistep phosphorylation of tyrosine residues on z chain ITAMs, dependent on the affinity of the ligand
(Kersh et al., 1998). This work shows that the TCR z phosphorylation state can translate the half-lives of the TCR-ligand interaction into discrete phospho-forms and possibly distinct functional outcomes, depending on the number or quality of adaptors and effectors recruited. T Cell–APC Interaction and TCR Serial Triggering Whereas the interaction between an individual TCR and its ligand occurs over a time frame of a few seconds, the interaction between a single T cell–APC pair has a time course of several hours. This long-lasting contact is necessary to ensure the sustained signaling that maintains gene transcription and promotes T cell cycle progression (Timmerman et al., 1996). This sustained signaling is the result of continuous engagements of TCRs by specific peptide–MHC complexes. In fact, in T cell–APC conjugates the ongoing signaling process can be rapidly terminated by treatments that prevent further TCR engagements, such as the addition of antibodies to the target MHC molecules, removal of antigen, as well as blocking of the T cell’s actin cytoskeleton (Valitutti et al., 1995a). The finding that the triggered TCRs are downregulated and degraded in an antigen dose- and time-dependent fashion provided a method to analyze the kinetics and extent of TCR triggering at the single cell level. Using this method it was estimated that a few (z100 per APC) agonistic peptide–MHC complexes engage and trigger, with time, a much larger number of TCRs (z20,000 per T cell) (Valitutti et al., 1995b). This serial triggering process implies that effective agonists must be able not only to engage and trigger, but also to dissociate from the triggered TCR to become available for a new interaction. The fast off-rate of this interaction is entirely consistent with this premise. Low-affinity TCR ligands such as peptide–MHC complexes and bacterial superantigens display logarithmic dose–response curves, with similar slopes but different potencies (Viola and Lanzavecchia, 1996) (Figure 2). In all cases, the capacity to induce TCR downregulation correlates with the capacity to activate T cells. Thus, strong agonists induce higher levels of downregulation than weak agonists, while antagonists fail to downregulate TCRs, but competitively inhibit TCR downregulation induced by agonists (Preckel et al., 1997; Viola et al., 1997a, 1997b). The logarithmic dose–response curve indicates that the efficiency of TCR triggering, defined as the ratio of displayed peptide–MHC complexes versus triggered TCRs, is highest when only a few complexes are displayed per APC. This is incompatible with TCR oligomerization as a necessary requirement for TCR triggering. In contrast to low-affinity ligands, high-affinity antibodies to CD3 are much less effective in TCR triggering at low copy number, since they display a fixed stoichiometry of z1:1 and a linear dose–response curve, indicative of a single cycle rather than serial triggering mode (Viola and Lanzavecchia, 1996). Furthermore, activation by high-affinity antibodies to CD3 is resistant to TCR antagonism (Viola et al., 1997b). In the context of serial triggering, TCR antagonism may be explained not only by sequestration or consumption of TCR substrates but also by the sterile occupancy of TCRs, that may limit the pool available for triggering by agonists. It makes sense that, in both cases, high-affinity interactions can override these inhibitory mechanisms.
Minireview 3
Figure 3. A Kinetic Window Defines Strong TCR Agonists
Figure 2. Serial TCR Triggering by Low-Affinity Ligands For low-affinity ligands, such as peptide–MHC complexes, the relationship between the number of ligands displayed on an APC and the number of TCRs triggered and downregulated over time is logarithmic, indicating that the efficiency of the process is highest when only a few complexes are displayed on the APC. In contrast, in the case of high-affinity ligands the relationship is linear, indicating that they trigger TCRs on an z1:1 basis. TCR antagonists inhibit triggering by low-affinity but not by high-affinity ligands. Furthermore, dependency on CD4 coreceptor is a characteristic of weak agonists. The curves shown illustrate experimental observations. The dotted lines represent T cell activation thresholds in the absence or in the presence of costimulation (upper and lower lines, respectively).
The integration of the kinetic proofreading and TCR serial triggering models provides a kinetic definition for TCR agonists (Figure 3). Strong agonists will fall within a short kinetic window that represents a compromise between successful completion of all the steps required for TCR triggering and rapid dissociation of the ligand from the triggered TCR, to make it available for further cycles of ligation and triggering. In this context CD4 and CD8 coreceptors may act as kinetic tuners. By stabilizing the TCR/peptide–MHC interaction, they may allow weak agonists to engage TCRs long enough to reach the triggering threshold, while they may be dispensable in the case of optimal ligands. Accordingly, coreceptors may either increase the efficiency of TCR triggering by weak agonists or change a partial agonist into a full one (Viola et al., 1997a; Madrenas et al., 1997). Tunable Thresholds for T Cell Activation The number of TCRs triggered is an important parameter that determines the fate of antigenic stimulation. Irrespective of the nature and affinity of the ligand, T cells are activated to proliferate and produce cytokines when a threshold number of triggered TCRs is reached. However, this threshold is not absolute, since it depends on whether or not costimulation is present and is therefore characteristic for a particular T cell–APC interaction. In human T cell clones that express z30,000 TCRs, the threshold has been estimated to be z8000 TCRs in the absence and z1000 TCRs in the presence of CD28mediated costimulation (Viola and Lanzavecchia, 1996). Indeed, in extreme experimental conditions, when peptide–MHC complexes are offered on surfaces without costimulatory molecules, T cell activation is hardly
Hypothetical relationship between half-life of TCR-ligand interaction (black axes) and number of TCRs engaged (blue axes) or triggered (red axes) by a single ligand in a given time. Only ligands that engage TCRs within an optimal range of kinetics, that compromises between the stability necessary to perform all the steps required for triggering and the instability necessary for serial engagements, will be effective in the serial triggering process. The stippled vertical line defines the TCR triggering threshold.
achieved, even when most of the TCRs have been triggered. On the other hand, in the same experiments, CD28 costimulation allows activation at very low levels of TCR occupancy (Cai et al., 1997; Iezzi et al., 1998). The role of CD28 costimulation is not to increase the number of triggered TCRs, but rather to amplify the signal transmitted so that a cellular response can be achieved at lower numbers of triggered TCRs (Viola and Lanzavecchia, 1996). CD28 enhances early phosphorylation events including phosphorylation of the z chain (Tuosto and Acuto, 1998), and we hypothesize that CD28 may recruit rafts at the T cell–APC contact site. Thus, besides other specific functions, such as protection from apoptosis (Boise et al., 1995), CD28 may act as a quantitative amplifier of TCR signal transduction. In effector T cells that have encountered antigen, different responses may be achieved at different thresholds. Cytotoxicity, which requires local secretion of prestored mediators, can be rapidly induced by a single peptide–MHC complex and in the absence of measurable TCR downregulation (Sykulev et al., 1996; Valitutti et al., 1996). In contrast, cytokine production, which depends on gene transcription, is achieved at much higher thresholds. Given the definition of partial agonists as ligands that trigger only a subset of the effector responses, it is tempting to suggest that their property depends on how effectively they push signaling through the hierarchies of functional thresholds (Itoh and Germain, 1997). While effector T cells are rapidly committed to proliferate, naive T cells must be stimulated for several hours, which may reflect their need to increase size before entering the cell cycle (Iezzi et al., 1998). The required duration of signaling is dependent on the amount of antigen and the presence of costimulation. When high doses of antigens are presented by professional APC, naive T cells become committed after 6 hr; however, at lower antigen doses, or when costimulation is lacking, the duration of stimulation needs to be increased up to
Cell 4
30 hr. The fact that stimulation time can be shortened by increasing TCR occupancy or by providing costimulation demonstrates that what is important for the T cell is the cumulative (quantitative, not qualitative) amount of signal received. Common and Distinct Aspects of T and B Cell Antigen Recognition Can the information gained on TCR behavior serve as a model for the B cell receptor (BCR)? The overall structure and the signaling pathways emanating from the two receptors are very similar. In both cases receptor signaling is enhanced by costimulation, which in B cells is mediated by complement components bound to the antigen. TCR and BCR share the property of being downregulated and degraded upon triggering. In B cells this mechanism plays the essential role of delivering antigen to the processing compartment, while in T cells the role of TCR degradation remains a matter of speculation. By adjusting the surface TCR level, this mechanism may tune down and extinguish the signaling process, thus avoiding overstimulation. There are however important differences between T and B cells that relate to the mode of antigen recognition and to the different affinity of the receptor–ligand interaction. There is today no consensus as to the form of antigen that triggers BCRs. As for TCR, monovalent binding is not sufficient and some form of homo- or heteroclustering is required. The high affinity of the interaction will generally prevent a univalent ligand from engaging multiple BCRs, raising the problem of how B cells sustain signaling. Nonetheless, we speculate that the kinetics of BCR ligation to the antigenic determinants of a multivalent antigen (for instance a particle or a cell) may be like slowly pulling a zipper. In this way multivalent T cell–independent antigens may lead to sustained BCR triggering and commitment in B cells. Summary The three peculiar properties of T cell antigen recognition, specificity, sensitivity, and context discrimination, are dependent on the kinetics of TCR–ligand interaction and T cell stimulation. The exquisite capacity to discriminate between homologous ligands can be understood in terms of thresholds for TCR triggering, dependent on multistep phosphorylation and recruiting events, that allow a kinetic discrimination of the ligand. The sensitivity of T cells can be explained by the fact that a single agonist can engage many TCRs over time, thus amplifying and sustaining the signaling process. Finally, the capacity of the T cell to discriminate the context in which antigen is presented can be explained by the function of costimulatory molecules as amplifiers of TCR signaling. As proposed by Charles Janeway, the innate immune system controls the expression of costimulatory molecules on APC, thus providing a code for the degree of “danger” associated with incoming antigens that adjusts the “gain” of TCR signaling. Selected Reading Boise, L.H., Noel, P.J., and Thompson, C.B. (1995). Curr. Opin. Immunol. 7, 620–625. Boniface, J.J., Rabinowitz, J.D., Wulfing, C., Hampl, J., Reich, Z., Altman, J.D., Kantor, R.M., Beeson, C., McConnell, H.M., and Davis, M.M. (1998). Immunity 9, 459–466.
Cai, Z., Kishimoto, H., Brunmark, A., Jackson, M.R., Peterson, P.A., and Sprent, J. (1997). J. Exp. Med. 185, 641–651. Davis, M.M., Boniface, J.J., Reich, Z., Lyons, D., Hampl, J., Arden, B., and Chien, Y. (1998). Annu. Rev. Immunol. 16, 523–544. Delon, J., Gregoire, C., Malissen, B., Darche, S., Lemaitre, F., Kourilsky, P., Abastado, J.P., and Trautmann, A. (1998). Immunity 9, 467–473. Iezzi, G., Karjalainen, K., and Lanzavecchia, A. (1998). Immunity 8, 89–95. Itoh, Y., and Germain, R.N. (1997). J. Exp. Med. 186, 757–766. Kersh, E.N., Shaw, A.S., and Allen, P.M. (1998). Science 281, 572–575. Kersh, G.J., and Allen, P.M. (1996). Nature 380, 495–498. Madrenas, J., Chau, L.A., Smith, J., Bluestone, J.A., and Germain, R.N. (1997). J. Exp. Med. 185, 219–229. McKeithan, T.W. (1995). Proc. Natl. Acad. Sci. USA 92, 5042–5046. Monks, C.R., Freiberg, B.A., Kupfer, H., Sciaky, N., and Kupfer, A. (1998). Nature 395, 82–86. Montixi, C., Langlet, C., Bernard, A.M., Thimonier, J., Dubois, C., Wurbel, M.A., Chauvin, J.P., Pierres, M., and He, H.T. (1998). EMBO J. 17, 5334–5348. Preckel, T., Grimm, R., Marin, S., and Weltzien, H.U. (1997). J. Exp. Med. 185, 1803–1813. Rabinowitz, J.D., Beeson, C., Lyons, D.S., Davis, M.M., and McConnell, H.M. (1996). Proc. Natl. Acad. Sci. USA 93, 1401–1405. Shaw, A.S., and Dustin, M.L. (1997). Immunity 6, 361–369. Simons, K., and Ikonen, E. (1997). Nature 387, 569–572. Sykulev, Y., Joo, M., Vturina, I., Tsomides, T.J., and Eisen, H.N. (1996). Immunity 4, 565–571. Timmerman, L.A., Clipstone, N.A., Ho, S.N., Northrop, J.P., and Crabtree, G.R. (1996). Nature 383, 837–840. Torigoe, C., Inman, J.K., and Metzger, H. (1998). Science 281, 568–572. Tuosto, L., and Acuto, O. (1998). Eur. J. Immunol. 28, 2131–2142. Valitutti, S., Dessing, M., Aktories, K., Gallati, H., and Lanzavecchia, A. (1995a). J. Exp. Med. 181, 577–584. Valitutti, S., Muller, S., Cella, M., Padovan, E., and Lanzavecchia, A. (1995b). Nature 375, 148–151. Valitutti, S., Mueller, S., Dessing, M., and Lanzavecchia, A. (1996). J. Exp. Med. 183, 1917–1921. Viola, A., and Lanzavecchia, A. (1996). Science 273, 104–106. Viola, A., Salio, M., Tuosto, L., Linkert, S., Acuto, A., and Lanzavecchia, A. (1997a). J. Exp. Med. 186, 1775–1779. Viola, A., Linkert, S., and Lanzavecchia, A. (1997b). Eur. J. Immunol. 27, 3080–3083. Xavier, R., Brennan, T., Li, Q., McCormack, C., and Seed, B. (1998). Immunity 8, 723–732. Zhang, W., Trible, R.P., and Samelson, L.E. (1998). Immunity 9, 239–246.
From TCR Engagement to T Cell Activation: A Kinetic View of T Cell Behavior Antonio Lanzavecchia,* Giandomenica Iezzi, and Antonella Viola Basel Institute for Immunology Grenzacherstrasse 487 CH-4005 Basel Switzerland
Sensitivity, specificity, and context discrimination are three key properties of T cell antigen recognition. T lymphocytes recognize antigen as peptides bound to major histocompatibility complex (MHC) molecules on the surface of antigen-presenting cells (APC). Since MHC molecules are highly promiscuous in peptide binding, T cell epitopes will be generally displayed on APC at low copy number, and consequently T cell recognition must be exquisitely sensitive. In addition, recognition must be highly specific, since T cells must discriminate few antigenic peptide–MHC complexes among a vast excess of irrelevant, yet highly homologous, complexes of the same MHC molecules carrying a variety of self-peptides. Finally, T cells must be able to interpret the context in which a given antigen is presented to mount an effector response to infectious antigens or ignore innocuous or self-antigens. In this minireview we will discuss recent evidence indicating that these three properties of T cells are intimately connected and are dependent on the kinetics of the T cell receptor (TCR)–ligand interaction and T cell stimulation. The Elusive Mechanism of TCR Stimulation While the biochemical events that follow TCR triggering are understood in increasing detail, the mechanism that initiates this cascade of events is still unresolved. In principle, the initial triggering event, induced by ligand binding, can be envisaged either as a conformational change of the TCR-CD3-z complex (hereafter referred to as TCR) or as the induction of some form of aggregation. The latter may involve the formation of either homologous clusters of two or more ligand-engaged TCRs or heterologous clusters of one ligand-engaged TCR with other signal transduction components. Structural studies have failed thus far to support the conformational change model. In addition, two recent experiments performed using soluble peptide–MHC complexes make this model highly unlikely. Both studies demonstrate that T cell triggering cannot be induced by monovalent ligation of TCR by soluble peptide–MHC complexes, but rather requires some form of oligomerization. When defined oligomers of MHC class II–peptide complexes were used to trigger T cells, it was found that at least three TCRs need to be brought together in order to induce a calcium response (Boniface et al., 1998). In another study however, T cell activation could be induced either by dimers of MHC class I–peptide complexes or by monomers that cross-link the TCR with the CD8 coreceptor–Lck kinase complex (Delon et al., * To whom correspondence should be addressed (e-mail: lanza [email protected]).
Minireview
1998). While these studies represent a clear step forward in defining the requirements for aggregation in initiating TCR signaling, they still leave several open questions. The first model does not explain how a few complexes (say, less than 10) per APC might rapidly trigger T cells, while the second does not account for CD8-independent responses. What is ultimately important is to understand the mechanism of TCR triggering within the constraints of the two cell membranes. The solution to this problem has so far been elusive due to the lack of appropriate experimental systems. These membrane constraints may favor partition of molecules at the T cell/APC contact region and influence the local concentrations of the reactants and the kinetics of receptor–ligand interaction (Shaw and Dustin, 1997; Monks et al., 1998). In this context, it is interesting to consider the possibility that monovalent TCR ligation may be sufficient to induce clustering with signaling molecules (Figure 1). Recent evidence indicates that many signaling components, such as Lck and LAT, which are key to initiating and propagating TCR signal transduction, are concentrated in cholesterol-enriched membrane microdomains, called rafts (Simons and Ikonen, 1997; Zhang et al., 1998). Two recent studies demonstrate that rafts are required for efficient T cell activation and that triggered TCRs are recruited to the rafts (Montixi et al., 1998; Xavier et al., 1998). We would like to suggest that monovalent ligation of TCR to the APC surface, by reducing its lateral mobility within the T cell membrane, might drive its recruitment into the kinase-rich environment of rafts, where
Figure 1. Initiation of TCR Triggering and Signal Transduction (a–c) Mechanisms that may promote TCR triggering. (a) Oligomerization of TCRs induced by ligand binding; (b) Cross-linking of TCR with the CD8-Lck complex induced by monovalent ligand. (c) Lateral translocation of a single TCR to a kinase-rich domain (raft) induced by actin-driven motility of the T cells and shear stress. The passive mobility of a single TCR complex is transiently decreased upon ligation to a peptide–MHC complex on APC. At the same time the rafts are laterally displaced by actin-driven motility, so that collisions between single TCRs and kinase-rich rafts are promoted. (d) Propagation of signal from the triggered TCR to raft-associated adaptors such as LAT. After TCR triggering is completed the peptide–MHC complex can dissociate and become available for another cycle of engagement and triggering.
Cell 2
the phosphorylation cascade can be triggered, resulting in complete phosphorylation of CD3 and z ITAMs by Lck, followed by docking and activation of ZAP-70, a cytoplasmic tyrosine kinase. This complex, which we refer to as a triggered TCR, can now propagate the signal by phosphorylating adaptors such as LAT that couple it to several downstream signaling pathways. In this model the actin-based T cell motility plays a key role in the initiation of TCR triggering (Valitutti et al., 1995a). Kinetic Threshold for TCR Triggering Peptide–MHC complexes do not need to bind with high affinity in order to trigger TCR. Strong agonists are typically characterized by a Kd of 1–90 mM and half-lives of z10 s (Davis et al., 1998 and references therein). However, small changes in peptide sequence, leading to slightly faster off rates, have been shown to greatly affect their capacity to trigger T cells. These altered peptide ligands may simply have a decreased potency (weak agonist), may trigger a partial response (partial agonists), or may even inhibit activation by TCR agonists (antagonists). The altered responses have been correlated to a distinct pattern of CD3/z phosphorylation and lack of ZAP-70 activation (Kersh and Allen, 1996). The kinetic proofreading model (McKeithan, 1995) provides a conceptual framework to analyze the relationship between kinetics of TCR–ligand interaction and T cell response. According to this model, the engagement of TCR by peptide–MHC does not immediately lead to TCR triggering, because a series of phosphorylation steps followed by recruitment and activation of ZAP-70 need to be performed, a process that requires time. If the duration of the interaction is sufficient, the phosphorylation and docking events will proceed until the fully active complex, which we refer to as a triggered TCR, is assembled and can transmit the signal to downstream pathways. In contrast, a premature dissociation before the process has been completed will lead to the formation of inactive intermediates that, by sequestering substrates, inhibit activation by agonists, resulting in TCR antagonism. This mechanism has been documented in the case of FceRI where antagonism is the result of competition for limiting amounts of kinase (Torigoe et al., 1998). An alternative possibility is that the intermediates generated might deliver an incomplete or even negative signal, thus explaining partial agonism and antagonism. The dependence of signaling on stepwise formation of multiple intermediates enhances the fidelity of kinetic discrimination by allowing for a greater number of proofreading steps (Rabinowitz et al., 1996). In this way, the kinetic proofreading model defines a minimum time threshold for the TCR/peptide–MHC interaction to be productive. An important aspect of kinetic proofreading is that the rate of the individual steps depends on the concentration of the reactants. Therefore, this rate may differ in different cell types and can be modulated by variations in the total level or local distribution of signaling molecules, thus potentially explaining developmentally regulated changes in responsiveness to TCR ligands. The kinetic proofreading model is supported by recent biochemical data showing a multistep phosphorylation of tyrosine residues on z chain ITAMs, dependent on the affinity of the ligand
(Kersh et al., 1998). This work shows that the TCR z phosphorylation state can translate the half-lives of the TCR-ligand interaction into discrete phospho-forms and possibly distinct functional outcomes, depending on the number or quality of adaptors and effectors recruited. T Cell–APC Interaction and TCR Serial Triggering Whereas the interaction between an individual TCR and its ligand occurs over a time frame of a few seconds, the interaction between a single T cell–APC pair has a time course of several hours. This long-lasting contact is necessary to ensure the sustained signaling that maintains gene transcription and promotes T cell cycle progression (Timmerman et al., 1996). This sustained signaling is the result of continuous engagements of TCRs by specific peptide–MHC complexes. In fact, in T cell–APC conjugates the ongoing signaling process can be rapidly terminated by treatments that prevent further TCR engagements, such as the addition of antibodies to the target MHC molecules, removal of antigen, as well as blocking of the T cell’s actin cytoskeleton (Valitutti et al., 1995a). The finding that the triggered TCRs are downregulated and degraded in an antigen dose- and time-dependent fashion provided a method to analyze the kinetics and extent of TCR triggering at the single cell level. Using this method it was estimated that a few (z100 per APC) agonistic peptide–MHC complexes engage and trigger, with time, a much larger number of TCRs (z20,000 per T cell) (Valitutti et al., 1995b). This serial triggering process implies that effective agonists must be able not only to engage and trigger, but also to dissociate from the triggered TCR to become available for a new interaction. The fast off-rate of this interaction is entirely consistent with this premise. Low-affinity TCR ligands such as peptide–MHC complexes and bacterial superantigens display logarithmic dose–response curves, with similar slopes but different potencies (Viola and Lanzavecchia, 1996) (Figure 2). In all cases, the capacity to induce TCR downregulation correlates with the capacity to activate T cells. Thus, strong agonists induce higher levels of downregulation than weak agonists, while antagonists fail to downregulate TCRs, but competitively inhibit TCR downregulation induced by agonists (Preckel et al., 1997; Viola et al., 1997a, 1997b). The logarithmic dose–response curve indicates that the efficiency of TCR triggering, defined as the ratio of displayed peptide–MHC complexes versus triggered TCRs, is highest when only a few complexes are displayed per APC. This is incompatible with TCR oligomerization as a necessary requirement for TCR triggering. In contrast to low-affinity ligands, high-affinity antibodies to CD3 are much less effective in TCR triggering at low copy number, since they display a fixed stoichiometry of z1:1 and a linear dose–response curve, indicative of a single cycle rather than serial triggering mode (Viola and Lanzavecchia, 1996). Furthermore, activation by high-affinity antibodies to CD3 is resistant to TCR antagonism (Viola et al., 1997b). In the context of serial triggering, TCR antagonism may be explained not only by sequestration or consumption of TCR substrates but also by the sterile occupancy of TCRs, that may limit the pool available for triggering by agonists. It makes sense that, in both cases, high-affinity interactions can override these inhibitory mechanisms.
Minireview 3
Figure 3. A Kinetic Window Defines Strong TCR Agonists
Figure 2. Serial TCR Triggering by Low-Affinity Ligands For low-affinity ligands, such as peptide–MHC complexes, the relationship between the number of ligands displayed on an APC and the number of TCRs triggered and downregulated over time is logarithmic, indicating that the efficiency of the process is highest when only a few complexes are displayed on the APC. In contrast, in the case of high-affinity ligands the relationship is linear, indicating that they trigger TCRs on an z1:1 basis. TCR antagonists inhibit triggering by low-affinity but not by high-affinity ligands. Furthermore, dependency on CD4 coreceptor is a characteristic of weak agonists. The curves shown illustrate experimental observations. The dotted lines represent T cell activation thresholds in the absence or in the presence of costimulation (upper and lower lines, respectively).
The integration of the kinetic proofreading and TCR serial triggering models provides a kinetic definition for TCR agonists (Figure 3). Strong agonists will fall within a short kinetic window that represents a compromise between successful completion of all the steps required for TCR triggering and rapid dissociation of the ligand from the triggered TCR, to make it available for further cycles of ligation and triggering. In this context CD4 and CD8 coreceptors may act as kinetic tuners. By stabilizing the TCR/peptide–MHC interaction, they may allow weak agonists to engage TCRs long enough to reach the triggering threshold, while they may be dispensable in the case of optimal ligands. Accordingly, coreceptors may either increase the efficiency of TCR triggering by weak agonists or change a partial agonist into a full one (Viola et al., 1997a; Madrenas et al., 1997). Tunable Thresholds for T Cell Activation The number of TCRs triggered is an important parameter that determines the fate of antigenic stimulation. Irrespective of the nature and affinity of the ligand, T cells are activated to proliferate and produce cytokines when a threshold number of triggered TCRs is reached. However, this threshold is not absolute, since it depends on whether or not costimulation is present and is therefore characteristic for a particular T cell–APC interaction. In human T cell clones that express z30,000 TCRs, the threshold has been estimated to be z8000 TCRs in the absence and z1000 TCRs in the presence of CD28mediated costimulation (Viola and Lanzavecchia, 1996). Indeed, in extreme experimental conditions, when peptide–MHC complexes are offered on surfaces without costimulatory molecules, T cell activation is hardly
Hypothetical relationship between half-life of TCR-ligand interaction (black axes) and number of TCRs engaged (blue axes) or triggered (red axes) by a single ligand in a given time. Only ligands that engage TCRs within an optimal range of kinetics, that compromises between the stability necessary to perform all the steps required for triggering and the instability necessary for serial engagements, will be effective in the serial triggering process. The stippled vertical line defines the TCR triggering threshold.
achieved, even when most of the TCRs have been triggered. On the other hand, in the same experiments, CD28 costimulation allows activation at very low levels of TCR occupancy (Cai et al., 1997; Iezzi et al., 1998). The role of CD28 costimulation is not to increase the number of triggered TCRs, but rather to amplify the signal transmitted so that a cellular response can be achieved at lower numbers of triggered TCRs (Viola and Lanzavecchia, 1996). CD28 enhances early phosphorylation events including phosphorylation of the z chain (Tuosto and Acuto, 1998), and we hypothesize that CD28 may recruit rafts at the T cell–APC contact site. Thus, besides other specific functions, such as protection from apoptosis (Boise et al., 1995), CD28 may act as a quantitative amplifier of TCR signal transduction. In effector T cells that have encountered antigen, different responses may be achieved at different thresholds. Cytotoxicity, which requires local secretion of prestored mediators, can be rapidly induced by a single peptide–MHC complex and in the absence of measurable TCR downregulation (Sykulev et al., 1996; Valitutti et al., 1996). In contrast, cytokine production, which depends on gene transcription, is achieved at much higher thresholds. Given the definition of partial agonists as ligands that trigger only a subset of the effector responses, it is tempting to suggest that their property depends on how effectively they push signaling through the hierarchies of functional thresholds (Itoh and Germain, 1997). While effector T cells are rapidly committed to proliferate, naive T cells must be stimulated for several hours, which may reflect their need to increase size before entering the cell cycle (Iezzi et al., 1998). The required duration of signaling is dependent on the amount of antigen and the presence of costimulation. When high doses of antigens are presented by professional APC, naive T cells become committed after 6 hr; however, at lower antigen doses, or when costimulation is lacking, the duration of stimulation needs to be increased up to
Cell 4
30 hr. The fact that stimulation time can be shortened by increasing TCR occupancy or by providing costimulation demonstrates that what is important for the T cell is the cumulative (quantitative, not qualitative) amount of signal received. Common and Distinct Aspects of T and B Cell Antigen Recognition Can the information gained on TCR behavior serve as a model for the B cell receptor (BCR)? The overall structure and the signaling pathways emanating from the two receptors are very similar. In both cases receptor signaling is enhanced by costimulation, which in B cells is mediated by complement components bound to the antigen. TCR and BCR share the property of being downregulated and degraded upon triggering. In B cells this mechanism plays the essential role of delivering antigen to the processing compartment, while in T cells the role of TCR degradation remains a matter of speculation. By adjusting the surface TCR level, this mechanism may tune down and extinguish the signaling process, thus avoiding overstimulation. There are however important differences between T and B cells that relate to the mode of antigen recognition and to the different affinity of the receptor–ligand interaction. There is today no consensus as to the form of antigen that triggers BCRs. As for TCR, monovalent binding is not sufficient and some form of homo- or heteroclustering is required. The high affinity of the interaction will generally prevent a univalent ligand from engaging multiple BCRs, raising the problem of how B cells sustain signaling. Nonetheless, we speculate that the kinetics of BCR ligation to the antigenic determinants of a multivalent antigen (for instance a particle or a cell) may be like slowly pulling a zipper. In this way multivalent T cell–independent antigens may lead to sustained BCR triggering and commitment in B cells. Summary The three peculiar properties of T cell antigen recognition, specificity, sensitivity, and context discrimination, are dependent on the kinetics of TCR–ligand interaction and T cell stimulation. The exquisite capacity to discriminate between homologous ligands can be understood in terms of thresholds for TCR triggering, dependent on multistep phosphorylation and recruiting events, that allow a kinetic discrimination of the ligand. The sensitivity of T cells can be explained by the fact that a single agonist can engage many TCRs over time, thus amplifying and sustaining the signaling process. Finally, the capacity of the T cell to discriminate the context in which antigen is presented can be explained by the function of costimulatory molecules as amplifiers of TCR signaling. As proposed by Charles Janeway, the innate immune system controls the expression of costimulatory molecules on APC, thus providing a code for the degree of “danger” associated with incoming antigens that adjusts the “gain” of TCR signaling. Selected Reading Boise, L.H., Noel, P.J., and Thompson, C.B. (1995). Curr. Opin. Immunol. 7, 620–625. Boniface, J.J., Rabinowitz, J.D., Wulfing, C., Hampl, J., Reich, Z., Altman, J.D., Kantor, R.M., Beeson, C., McConnell, H.M., and Davis, M.M. (1998). Immunity 9, 459–466.
Cai, Z., Kishimoto, H., Brunmark, A., Jackson, M.R., Peterson, P.A., and Sprent, J. (1997). J. Exp. Med. 185, 641–651. Davis, M.M., Boniface, J.J., Reich, Z., Lyons, D., Hampl, J., Arden, B., and Chien, Y. (1998). Annu. Rev. Immunol. 16, 523–544. Delon, J., Gregoire, C., Malissen, B., Darche, S., Lemaitre, F., Kourilsky, P., Abastado, J.P., and Trautmann, A. (1998). Immunity 9, 467–473. Iezzi, G., Karjalainen, K., and Lanzavecchia, A. (1998). Immunity 8, 89–95. Itoh, Y., and Germain, R.N. (1997). J. Exp. Med. 186, 757–766. Kersh, E.N., Shaw, A.S., and Allen, P.M. (1998). Science 281, 572–575. Kersh, G.J., and Allen, P.M. (1996). Nature 380, 495–498. Madrenas, J., Chau, L.A., Smith, J., Bluestone, J.A., and Germain, R.N. (1997). J. Exp. Med. 185, 219–229. McKeithan, T.W. (1995). Proc. Natl. Acad. Sci. USA 92, 5042–5046. Monks, C.R., Freiberg, B.A., Kupfer, H., Sciaky, N., and Kupfer, A. (1998). Nature 395, 82–86. Montixi, C., Langlet, C., Bernard, A.M., Thimonier, J., Dubois, C., Wurbel, M.A., Chauvin, J.P., Pierres, M., and He, H.T. (1998). EMBO J. 17, 5334–5348. Preckel, T., Grimm, R., Marin, S., and Weltzien, H.U. (1997). J. Exp. Med. 185, 1803–1813. Rabinowitz, J.D., Beeson, C., Lyons, D.S., Davis, M.M., and McConnell, H.M. (1996). Proc. Natl. Acad. Sci. USA 93, 1401–1405. Shaw, A.S., and Dustin, M.L. (1997). Immunity 6, 361–369. Simons, K., and Ikonen, E. (1997). Nature 387, 569–572. Sykulev, Y., Joo, M., Vturina, I., Tsomides, T.J., and Eisen, H.N. (1996). Immunity 4, 565–571. Timmerman, L.A., Clipstone, N.A., Ho, S.N., Northrop, J.P., and Crabtree, G.R. (1996). Nature 383, 837–840. Torigoe, C., Inman, J.K., and Metzger, H. (1998). Science 281, 568–572. Tuosto, L., and Acuto, O. (1998). Eur. J. Immunol. 28, 2131–2142. Valitutti, S., Dessing, M., Aktories, K., Gallati, H., and Lanzavecchia, A. (1995a). J. Exp. Med. 181, 577–584. Valitutti, S., Muller, S., Cella, M., Padovan, E., and Lanzavecchia, A. (1995b). Nature 375, 148–151. Valitutti, S., Mueller, S., Dessing, M., and Lanzavecchia, A. (1996). J. Exp. Med. 183, 1917–1921. Viola, A., and Lanzavecchia, A. (1996). Science 273, 104–106. Viola, A., Salio, M., Tuosto, L., Linkert, S., Acuto, A., and Lanzavecchia, A. (1997a). J. Exp. Med. 186, 1775–1779. Viola, A., Linkert, S., and Lanzavecchia, A. (1997b). Eur. J. Immunol. 27, 3080–3083. Xavier, R., Brennan, T., Li, Q., McCormack, C., and Seed, B. (1998). Immunity 8, 723–732. Zhang, W., Trible, R.P., and Samelson, L.E. (1998). Immunity 9, 239–246.