The T Cell Receptor for Antigen: Signaling and Ligand Discrimination*

A defining characteristic of the vertebrate adaptive immune system is its specificity, which arises from clonally distributed receptors encoded by gene segments that undergo somatic rearrangement in B and T lymphocytes (1, 2). B lymphocyte receptors are membrane-bound immunoglobulin molecules that as a group react with a biochemically diverse array of ligands. In marked contrast, most receptors expressed by the major CD4 and CD8 subsets of T lymphocytes ( TCR) are narrowly focused on small protein fragments bound to cell surface molecules encoded by major histocompatibility complex (MHC) class I or class II genes (3). T cell responses to foreign organisms thus pose a major conceptual challenge (4). These lymphocytes develop effector activity when only a few hundred of the 10 MHC-encoded molecules on the plasma membrane of a cell are occupied by peptides derived from an infectious agent (3, 4). The remaining MHC molecules contain peptides from self-proteins that may differ from the foreign peptides by only a single amino acid (5, 6). Thus, a fundamental question is how TCR signaling can be both sensitive and specific for foreign determinants when there is a 10–10 greater concentration of structurally related self-ligands on the interacting cell membrane. Neither genetic nor developmental factors fully account for this capacity. TCR combining sites are generated anew in each individual by DNA rearrangement. The proteins from which self-peptides derive can be polymorphic and are mostly unlinked to TCR genes, and MHC molecules are highly diverse in structure and also unlinked to TCR loci. These are all features that prevent genetic matching of the TCR repertoire with the universe of ligands that it faces. Developing T cells do undergo a process termed negative selection, in which many clones with the most highly self-reactive TCR are eliminated by apoptosis (7). However, maturation of a precursor T cell requires TCR-dependent signals (7), yielding a repertoire with at least the capacity for self-recognition if not response. The basis for TCR antigen discrimination and sensitivity must thus lie in how receptor-ligand interactions are coupled to intracellular signaling pathways. We have an increasingly detailed knowledge of the biochemical cascade initiated by this recognition process (8, 9). Image analysis has revealed a complex spatiotemporal reorganization of lymphocyte proteins during responses to membrane-bound TCR ligands (10–13). The affinity and rate constants that characterize the interaction of soluble versions of TCRs and their ligands have been measured (14). Finally, differences in proximal TCR signaling and membrane protein reorganization have been documented following exposure of T cells to sets of related ligands (4, 11) and then correlated with TCRligand affinities. These observations have led to a “kinetic” model of ligand discrimination by the TCR that enjoys broad support (15). However this important concept does not explain how TCR engagement leads to initiation of signaling or how differences in TCR binding to related ligands produce distinct phosphorylation patterns and biological responses. This minireview will highlight what is known about these issues and also point out areas of controversy bearing on this fundamental aspect of T cell biology.

A defining characteristic of the vertebrate adaptive immune system is its specificity, which arises from clonally distributed receptors encoded by gene segments that undergo somatic rearrangement in B and T lymphocytes (1,2). 1 B lymphocyte receptors are membrane-bound immunoglobulin molecules that as a group react with a biochemically diverse array of ligands. In marked contrast, most receptors expressed by the major CD4 ϩ and CD8 ϩ subsets of T lymphocytes (␣␤ TCR) 2 are narrowly focused on small protein fragments bound to cell surface molecules encoded by major histocompatibility complex (MHC) class I or class II genes (3). T cell responses to foreign organisms thus pose a major conceptual challenge (4). These lymphocytes develop effector activity when only a few hundred of the Ͼ10 5 MHC-encoded molecules on the plasma membrane of a cell are occupied by peptides derived from an infectious agent (3,4). The remaining MHC molecules contain peptides from self-proteins that may differ from the foreign peptides by only a single amino acid (5,6). Thus, a fundamental question is how TCR signaling can be both sensitive and specific for foreign determinants when there is a 10 3 -10 4 greater concentration of structurally related self-ligands on the interacting cell membrane. Neither genetic nor developmental factors fully account for this capacity. TCR combining sites are generated anew in each individual by DNA rearrangement. The proteins from which self-peptides derive can be polymorphic and are mostly unlinked to TCR genes, and MHC molecules are highly diverse in structure and also unlinked to TCR loci. These are all features that prevent genetic matching of the TCR repertoire with the universe of ligands that it faces. Developing T cells do undergo a process termed negative selection, in which many clones with the most highly self-reactive TCR are eliminated by apoptosis (7). However, maturation of a precursor T cell requires TCR-dependent signals (7), yielding a repertoire with at least the capacity for self-recognition if not response.
The basis for TCR antigen discrimination and sensitivity must thus lie in how receptor-ligand interactions are coupled to intracellular signaling pathways. We have an increasingly detailed knowledge of the biochemical cascade initiated by this recognition process (8,9). Image analysis has revealed a complex spatiotemporal reorganization of lymphocyte proteins during responses to membrane-bound TCR ligands (10 -13). The affinity and rate constants that characterize the interaction of soluble versions of TCRs and their ligands have been measured (14). Finally, differences in proximal TCR signaling and membrane protein reorganization have been documented following exposure of T cells to sets of related ligands (4,11) and then correlated with TCRligand affinities. These observations have led to a "kinetic" model of ligand discrimination by the TCR that enjoys broad support (15). However this important concept does not explain how TCR engagement leads to initiation of signaling or how differences in TCR binding to related ligands produce distinct phosphorylation patterns and biological responses. This minireview will highlight what is known about these issues and also point out areas of controversy bearing on this fundamental aspect of T cell biology.

The TCR Complex and MHC Molecule Coreceptors
The ␣ and ␤ chains of the TCR are Type I integral membrane proteins with a membrane-distal immunoglobulin V-like domain and a membrane-proximal immunoglobulin C-like domain. The combining site is composed of germ line-encoded complementarity-determining regions (CDR) 1 and 2 of each chain and the two CDR3 regions that include germ line-and non-germ line-encoded residues. There is a relatively stereotypic docking position of the ␣␤ heterodimer on MHC proteins that orients the CDR3 region for contact with exposed peptide side chains (16), which determines TCR fine specificity. The receptor heterodimer is expressed in conjunction with a set of invariant integral membrane proteins termed the CD3 complex. The ␣ and ␤ cytoplasmic tails are short and devoid of recognizable signaling motifs, but the other subunits of the minimal octameric complex (␣, ␤, ␦, ␥, 2 CD3⑀, and 2 ) each contain one or more copies of an amino acid sequence (Y(2X)I/L(4 -6X)Y(2X)I/L) termed the immune tyrosine-based activation motif (ITAM) (17). ITAMs are the sites of Src family kinase tyrosine phosphorylation, a key early event in the TCR signaling cascade.
CD8 and CD4 are membrane glycoproteins expressed by T cells that bind MHC class I and MHC class II molecules, respectively, but lack the peptide specificity of the TCR. The interaction of CD4 and CD8 with their MHC ligands through structurally analogous sites (18 -20) facilitates T cell activation. These proteins are believed to form a ternary complex with the TCR and bound MHC ligand, hence the term "coreceptors" (21). Although some recent data on CD4-MHC class II interactions have raised questions about this model (22,23), the bulk of structural data and many biological experiments support the coreceptor view (24,25).

The Immunological Synapse
The TCR and its ligands are membrane-bound proteins, and receptor engagement involves the apposition of two cell surfaces. It has recently become clear that the region of cell-cell contact shows partitioning of individual proteins into distinct zones (10, 11, 26 -28). This contact area is now termed the immunological synapse (28). Synapse formation begins with initial stabilization of cell-cell contact, presumably through adhesion between integrins on T cells and ICAM molecules on ligandbearing cells (11,27), and for specialized antigen-presenting cells, via non-integrin molecular recognition events (29). TCR engagement then leads to the development of a ring of the integrin LFA-1 surrounding a central region of (2-3-fold) increased TCR density (10,11). High resolution video microscopy studies of cells expressing -green fluorescent protein chimeras (12) suggest that engaged TCR complexes are initially distributed in small, irregularly sized clusters that over several minutes coalesce into a more definitive central zone (termed the c-SMAC for central zone of the supramolecular activation cluster (10)), with a corresponding accumulation of MHC ligands on the opposing cell.
Synapse formation requires participation of the actin cytoskeleton and signaling from the initial pool of engaged TCR (27), although some protein distribution patterns may arise directly from the physicochemical properties of molecules bound to ligands on an opposing cell membrane (30). The kinetics of synapse formation make clear that the central receptor accumulation of the mature synapse is not required for initiation of second messenger generation (4,28). But then what is the reason for the active formation of a topographically complex synapse? Two answers have been offered. First, the synapse controls the orientation of the microtubule organizing center of the T cell and secretory apparatus (31). Proteins exported to the cell surface or secreted by the lymphocyte thus achieve their highest concentration at the site of cell-cell contact (32). This facilitates non-TCR protein-protein interac-* This minireview will be reprinted in the 2001 Minireview Compendium, which will be available in December, 2001.
tions and diminishes the likelihood of exposing nearby cells to any exported mediators. Second, concentrating receptors and ligands in a small zone may augment TCR occupancy through rebinding effects and also enhance biochemical cross-talk among these clustered receptors, thus promoting sustained signaling (4,28,33). The latter is required for gene activation in the lymphocyte and for the reciprocal activation of the antigen-bearing cell (34). Serial engagement of TCRs by a single peptide-MHC ligand has also been suggested to contribute to prolonged signaling in T cells (35). However, it remains to be determined if ligands containing self-peptides, which are available on the presenting cell membrane but not counted in calculating the extent of serial engagement, contribute to effective TCR engagement in the presence of stimulatory foreign ligands.
Some investigators (36,37) have suggested that TCR occupancy is initiated by accessory molecules, especially CD2 and perhaps CD28, which help to form a zone of close membrane apposition as they bind their ligands on the presenting cell, thus "squeezing out" larger proteins such as CD45 or CD43. Nevertheless, TCR clustering and signaling precede the clearance of either of the latter from the zone of cell contact (27,38). 3 It is possible that membrane topology, and not the size of these proteins, is the more critical factor in synapse formation and signaling (30). 3 Also, in vivo many surface proteins of T cells have polarized distributions. 4 This suggests that measurements of the average density of proteins over relatively large surface areas defined by optical microscopy using cultured cells may not be nearly as relevant as knowing the patterns of protein distribution and membrane shape on a much finer scale on cells in vivo.

Biochemistry of Signaling and the Role of Rafts
A popular model proposes that the initiation and propagation of TCR signaling events occur within rafts (detergent-insoluble glycolipid domains and glycolipid-enriched membranes) (39,40), specialized membrane subdomains enriched in cholesterol and glycolipids (41,42). TCR engagement has been reported to promote inclusion of the receptor within these specialized regions, followed by phosphorylation of ITAMs via the raft-resident Src family kinases Lck or Fyn, creating sites bound by the tandem SH2 domains of the Syk family kinase ZAP-70 (43,44). Recruitment of the latter kinase leads to its phosphorylation and enzymatic activation (43). Active ZAP-70 then phosphorylates the integral membrane adapter protein LAT, a doubly acylated molecule associated with the detergent-insoluble (raft) membrane fraction before signaling begins (45). The cytoplasmic adapter SLP-76 binds to phosphorylated LAT, and together these proteins act as sites for the recruitment of additional adapters and key enzymes involved in further downstream signaling in T cells (9,45). Antibody-mediated aggregation of the TCR results in accumulation of the raft-associated ganglioside GM1 near the clustered receptors (46). This observation, the loss of LAT activity when mutated to prevent acylation (45), the isolation of phosphorylated signaling molecules in the Triton X-100-resistant fraction of TCR-engaged cells (39,40), and the failure to see early receptor phosphorylation events in cells whose TCRs are cross-linked after extraction of cholesterol (39, 47) constitute a seemingly persuasive set of results in favor of the raft hypothesis. However, experimental concerns and logical problems bedevil the interpretation of these results.
It is first worth returning to the basis for CD4/8 coreceptor function. Both coreceptors bind the Src family kinase Lck through their cytoplasmic tails. Tyrosine kinase activity along with Fyn and Lck has also been detected in TCR immunoprecipitates (8). Based on these findings, one model for initiation of signaling involves recruitment of a CD4 or CD8 coreceptor to a MHC ligand-engaged TCR complex, bringing the Lck associated with the coreceptor into proximity with TCR-associated kinase (48) and resulting in transphosphorylation on activating tyrosines (Tyr-394 for Lck). This scenario raises several questions, however. What is the phosphorylation status of the tyrosine responsible for inactivation of Src family kinases (Tyr-505 in Lck) in the two pools prior to TCR engagement? If this site is phosphorylated, how does it become dephosphorylated to yield active kinase? What is the relationship between the pool of Lck that resides in rafts and that bound to the TCR complex?
One of the most abundant proteins on lymphocyte membranes is the phosphatase CD45, which removes the inhibitory tyrosine phosphates from Src family kinases such as Lck (49). Lymphocytes lacking CD45 do not show TCR-dependent signaling, presumably because the Src family kinase pools in such cells are uniformly inactivated via tyrosine phosphorylation mediated by Csk (50). CD45 is not believed to enter rafts, and available evidence indicates that the raft-associated pool of Lck is inactive and fully Tyr-505-phosphorylated (51). CD4-Lck and CD8-Lck have been reported to be partially associated with rafts (52), and hence, some fraction of the kinase associated with these coreceptors should be inactive as well. In contrast, the TCR-associated Lck pool outside rafts is free to undergo CD45 attack and partial activation.
We are then left to consider how full kinase activation occurs upon ligand recognition by the TCR. If a receptor had only a single associated kinase molecule, one possibility is the interaction of multiple occupied TCR assemblies with each other (Fig. 1A). If more than one kinase molecule is associated with a TCR-CD3 complex, reorganization of the subunits upon ligand binding could lead to transphosphorylation (Fig.  1B). A third possibility is for the TCR to associate with a kinase-linked coreceptor (Fig. 1C). Either of the first two cases should lead to transphosphorylation because the kinases associated with the receptor would be partially activated due to CD45 dephosphorylation. In the last case, however, for those CD4 and CD8 molecules residing in rafts, the bound Lck would be in a dormant state, making it difficult to see how initiation of signaling could be mediated by their recruitment (Fig. 1D). It thus seems that either TCRs outside rafts and lacking bound coreceptors initiate signaling or the relevant pool of coreceptors must be outside of rafts (Fig. 1, C and E). Either way, if current data on the inactivity of tyrosine kinases in rafts of resting lymphocytes are correct, the earliest phase of TCR signal transduction is unlikely to take place within cholesterol-defined lipid subdomains as often proposed.
Lymphocytes exposed for more than 30 min to cholesterol-depleting agents fail to signal upon subsequent TCR ligation, which is taken as evidence that receptor signaling normally occurs within cholesterol-dependent membrane subdomains. However there is an alternative explanation. The release of raft contents due to cholesterol extraction would expose previously sequestered kinases to the dephosphorylating effects of CD45. These partially active kinases could then associate with TCR complexes, mediating phosphorylation events that in turn activate negative regulatory processes normally involved in terminating receptor signaling. This could account for the inability of receptor ligation at this late stage to promote signaling. Support for this interpretation can be found in a recent study (47), which shows that cyclodextrin disruption of rafts leads to transient tyrosine phosphorylation of numerous

FIG. 1. Different possible modes of initial kinase activation upon TCR-ligand interaction.
A, TCR complex homo-oligomerization; B, TCR complex internal rearrangement; C, TCR complex-coreceptor heterodimerization; D, recruitment of TCR complex into rafts containing coreceptorinactive Lck complexes; E, release of coreceptor-Lck from rafts for interaction with TCR complex. proteins including TCR, followed by refractoriness to TCR-cross-linking. The standard raft model provides no explanation for how their disruption would result in active modification of signaling proteins like . These data are consistent, however, with the idea that Src kinases must be outside of rafts to be activated by CD45 (51). Additional support for this possibility comes from a recent report showing selective LAT association with activated TCR, without association of other raft-resident molecules (53). From these findings it could be inferred that rafts are the sites of storage of inactive signaling components sequestered from the activating CD45 phosphatase and from receptor-associated substrates, in contrast to the view that they are sites of initial kinase activation and function.
Side chain order in cholesterol-enriched regions of the lipid bilayer plays an important role in the accumulation of specific molecules in these microdomains (42). Similar degrees of lipid side chain order can be attained around clustered transmembrane proteins. This suggests an alternative explanation for the results of detergent extraction and microscopic colocalization experiments. If raft content is released at the earliest stages of TCR signaling, these proteins would be likely to accumulate next to oligomerized membrane proteins, such as the ligand-engaged TCR or TCR-coreceptor pairs. When these integral membrane clusters are disrupted with detergent prior to gradient fractionation, their affinity for the released raft content will be lost. The signal transducing molecules being tracked could then re-partition into available cholesterol-rich (detergent-insoluble) regions of membrane. As a consequence, these molecules would appear to have been resident in rafts, which was not necessarily the true state of events before experimental manipulation of the cells.
Thus, whereas an increasing number of reports using a limited number of experimental tools offer results consistent with rafts as the site of early and possibly sustained TCR signaling, there are several problematic experimental as well as conceptual aspects to these studies. The preceding discussion points offer logically consistent alternative explanations for most of the available data.

Ligand-specific Control of Signaling
The details of synapse formation and the conventional signaling cascade do not by themselves explain dose-response relationships markedly disproportionate to the small measured differences in binding affinity of stimulatory and nonstimulatory ligands (14). Insight into how this occurs is beginning to emerge from studies of "altered ligands" (54 -56). These are peptide-MHC combinations in which only one or a few amino acids in either the peptide cargo or MHC molecule differ from those in the known agonist (stimulatory) complex for a particular TCR. Modified ligands of this type show pharmacological properties ranging from superagonist (more potent than the original stimulatory ligand) to partial agonist (able to evoke, at high ligand densities, some but not necessarily all effector responses induced by the agonist) to antagonist (incapable of stimulating responses but able to interfere with responses to simultaneously presented agonist) to null ligand (lacking any measurable effect on T cell function). Analysis of early TCR signaling events has revealed that partial agonists and even antagonists induce the preferential accumulation of a partially tyrosine-phosphorylated form of (called p21 in the monomer form) associated with ZAP-70 molecules that were neither phosphorylated nor kinase-active (4). Strongly stimulatory ligands, in contrast, produce both this p21 and an equal or greater amount of fully phosphorylated p23 associated with kinase-active ZAP-70. Because decreasing agonist concentrations result in a parallel diminution of the two signals rather than in a change to the low p23/p21 ratio seen with partial agonists or antagonists, these data suggested a qualitative effect of altered ligand structure on the signaling process.
Based on the expectation that less potent ligands would have a lower affinity for (in particular, a faster off-rate from) a given TCR, McKeithan (15) proposed the idea of "kinetic proofreading" in the context of TCR recognition (15). This model argues that different downstream signaling events occur in a time-delayed fashion after initiation of receptor occupancy, with amplification at each step of the cascade. Loss of receptor occupancy not only eliminates input into the system but also has a disproportionate effect on distal parts of the signaling pathway due to the nonlinear amplification process. This would both attenuate overall signaling and change the balance of signal-transducing mediators. Support for this model has come from measurements of the binding kinetics of soluble TCR with soluble peptide-MHC ligands (14). Most, though not all, of these data show an inverse correlation between either K d or k off and ligand potency (and when measured, the quality of signaling as assessed by and ZAP-70 phosphorylation patterns). This hypothesis provides a simple explanation for the altered ratio of phosphorylation as well as the recruitment of ZAP-70 without activation seen with low affinity partial agonist and antagonists, based on the notion that full phosphorylation of and activation of recruited ZAP-70 take longer than partial phosphorylation and ZAP-70 binding itself (57).
What are the actual molecular events that translate differences in duration of TCR occupancy into distinct signaling events? Given that even antagonists can promote phosphorylation, TCR occupancy by weakly binding ligands must still be adequate to activate Src kinases. What inactivates these kinases so quickly once TCR occupancy ends that ITAM phosphorylation ceases as the simple kinetic model proposes? Perhaps the removal of the activating tyrosine phosphate on Lck/Fyn could be mediated by CD45 in the vicinity of the previously occupied TCR because coalescence of TCRs into the core of the synapse away from this phosphatase does not occur with antagonist ligands (11). However, altered patterns of phosphorylation are seen with antagonists at times well before even agonist ligands promote formation of microscopically visible TCR clusters; that is, when CD45 is still intermixed with engaged TCR (58,59). This makes it difficult to ascribe differential ITAM phosphorylation to CD45 effects. Csk is another molecule that inactivates Src family kinases in T cells. Csk rephosphorylation of the inactivating tyrosine requires the presence of both this kinase and the target Src kinase in the same membrane compartment. At present, evidence argues that antagonist-engaged TCRs are not raft-associated, whereas Csk is localized to these microdomains unless its docking protein is dephosphorylated (60,61). Unless antagonist specifically promotes this dephosphorylation, Csk cannot account for the required rapid loss of kinase function upon TCR-ligand dissociation.
Published findings can be combined with emerging data to produce a more complex but nevertheless coherent model. TCR engagement leads to the tyrosine phosphorylation of the cytosolic phosphatase SHP-1 (62). We have found that this promotes binding of the modified SHP-1 to TCR-associated Lck, whether or not the TCR in question has been engaged ("signal spreading") (63). 5 This binding is followed by activation of the phosphatase and then the inactivation of the associated Lck and ZAP-70 (63)(64)(65). 5 Unexpectedly, the pace of this SHP-1 binding to the TCR is inversely related to the potency of the TCR ligand used, fast for antagonists and slow for agonists, even though the latter would be expected to be better activators of Lck, enhancing SHP-1 phosphorylation and its recruitment to the TCR. The unexpected delay in phosphatase binding to the TCR after the onset of agonist recognition has been traced to a novel role of ERK, whose serine phosphorylation of Lck prevents binding of tyrosine-phosphorylated SHP-1 to the Lck SH2 domain, allowing prolonged TCR signaling. 5 Thus, two feedback loops operate in opposition once TCR engagement begins, with SHP-1 acting to terminate TCR signaling and ERK acting to preserve receptors from such desensitization. Agonists, but not antagonists, effect the rapid up-regulation of ERK activity. Perhaps by modification of Lck in adjacent unengaged receptors, there is "ERKmediated positive signal spreading," creating a pool of TCRs whose subsequent stimulation by agonist is protected against SHP-1 negative feedback. Mathematical modeling of TCR-dependent T cell activation is consistent with the concept that by combining such negative and positive feedback loops with signal spreading (63), both high sensitivity and good discrimination by the receptor can be achieved (66). Finally, these new data suggest that the partially phosphorylated -nonphosphorylated ZAP-70 pattern seen with these ligands, rather than representing a failure of processive ITAM phosphorylation, might arise from complete phosphorylation that is induced by such ligands then rapidly lost due to the action of the SHP-1.
How do differences in the quality of TCR binding lead to variation in the pace of ERK activation? A possible answer comes from data relating coreceptor function to signaling phenotype (67,68). Most coreceptors are physically separate from the pool of surface TCR. Thus, some time must pass between ligand binding and the association of the engaged TCR with a coreceptor. If the rate of translational movement of coreceptors in the plasma membrane is sufficiently slow, or if effective binding of the coreceptor to an engaged TCR is a rare event and requires many attempts to be successful, formation of ternary complexes will occur only with slow off-rate (agonist) and not fast off-rate (antagonist) ligands. When ligand binding is sufficiently long lived to allow coreceptor recruitment, interaction of this Lck-associated molecule with the TCR complex could contribute to phosphorylation of the recruited ZAP-70 and/or permit activated ZAP-70 in the TCR complex to phosphorylate coreceptor-associated LAT (52). LAT phosphorylation is key to secondary adapter recruitment and communication with the Ras pathway leading to ERK activation.
On the other hand, if the ligand dissociates almost immediately from the TCR, coreceptor binding to form a stable ternary complex would not occur and rapidly activate the ERK pathway. This leads to the eventual phosphorylation of SHP-1 and the binding of this modified phosphatase to the Lck molecules of engaged as well as nearby unoccupied TCR. Upon phosphorylation of additional TCR-associated proteins by the activated Lck, engagement of the SH2 domains of the bound phosphatase takes place. This results in SHP-1 activation and removal of tyrosine phosphates from proteins in the TCR complex, including ZAP-70, CD3, and Lck. The result is a "dead" receptor with a partial phosphorylation pattern.

Concluding Remarks
Many of the major players in TCR-mediated intracellular signaling have been identified, the protein-protein interactions involving these molecules have been defined, and the post-translational modifications that contribute to these associations have been characterized. Nevertheless, the most fundamental property of T lymphocytes, namely their capacity for exquisite ligand discrimination, has not been accounted for in detailed molecular terms. In large part this is because, until recently, antibody-mediated cross-linking of receptors has been the primary tool used to dissect TCR signaling pathways. Antibody binding to TCR-CD3 complexes does not have the kinetic characteristics of physiologic MHC ligand-receptor interaction nor does it support the recruitment of the key CD4 or CD8 coreceptors.
Very recently, many laboratories have moved toward the use of "real" TCR ligands for signaling studies and to the application of new microscopic tools to the analysis of the spatiotemporal features of TCR-dependent lymphocyte activation. Yet even these approaches have not defined either the earliest steps or the very late events in the cascade leading to gene activation in T cells. We have some understanding of what happens between tens of seconds to tens of minutes, but not much before or after that. It seems likely that the quality of TCR-ligand interaction is read out by individual receptors or small clusters of these TCRs well before available methodology is capable of detecting or measuring these events, and persistent signaling over hours is necessary for gene activation even though little proximal TCR-associated tyrosine phosphorylation can be seen at these late time periods. Our understanding of protein movement within the plasma membrane and the role of lipid domains is also still too limited to warrant confidence in the current view of these matters.
Because of the unique nature of the relationship between TCR and ligand, understanding the link between TCR recognition and cellular activation is a major goal in the field of signal transduction. As the data on the competing SHP-1 and ERK pathways are beginning to demonstrate, an integrated "systems analysis" that goes beyond just a careful reporting of the players and their connections will be required to move to the next level in our understanding of a recognition-response process that is one of the defining features of the adaptive immune system.