T Cell Receptor Signaling: Beyond Complex Complexes*

The adaptive phase of the immune response begins with engage- ment on CD4 (cid:1) helper T cells of the T cell antigen receptor (TCR) 1 by its ligand, a small foreign peptide bound to a cell surface protein of the class II major histocompatibility complex (peptide-MHC) expressed on an antigen-presenting cell. This engagement initiates a series of biochemical events that can differentially signal the naive T cell to: 1) enter into a pathway leading to generation of effector T cells with the onset of rapid proliferation and production of effector cytokines; 2) enter into a state of antigenic non-respon- siveness known as anergy; or 3) die by apoptosis. The type of response elicited depends on multiple factors including the affinity of the interaction, the duration of the interaction, and the presence or absence of various costimulatory signaling inputs such as those provided by the CD4 coreceptor and the CD28 costimulatory receptor. In this review we provide an overview of the signaling events that are associated with the first of these outcomes: T cell activation. To present an overview of sufficiently broad scope, the depth of discussion of each aspect of TCR signaling is by necessity limited, and the reader is referred to the reviews cited throughout this text for consideration of these events in more detail.

The adaptive phase of the immune response begins with engagement on CD4 ϩ helper T cells of the T cell antigen receptor (TCR) 1 by its ligand, a small foreign peptide bound to a cell surface protein of the class II major histocompatibility complex (peptide-MHC) expressed on an antigen-presenting cell. This engagement initiates a series of biochemical events that can differentially signal the naive T cell to: 1) enter into a pathway leading to generation of effector T cells with the onset of rapid proliferation and production of effector cytokines; 2) enter into a state of antigenic non-responsiveness known as anergy; or 3) die by apoptosis. The type of response elicited depends on multiple factors including the affinity of the interaction, the duration of the interaction, and the presence or absence of various costimulatory signaling inputs such as those provided by the CD4 coreceptor and the CD28 costimulatory receptor. In this review we provide an overview of the signaling events that are associated with the first of these outcomes: T cell activation. To present an overview of sufficiently broad scope, the depth of discussion of each aspect of TCR signaling is by necessity limited, and the reader is referred to the reviews cited throughout this text for consideration of these events in more detail.

TCR Structure
The TCR is composed of six different polypeptide chains. The specificity of ligand binding is dictated by the clonotypic TCR␣ and TCR␤ chains, which arise from a process of genetic rearrangement that results in millions of receptor variants. These chains form a heterodimer that binds directly to peptide-MHC. Communication of TCR␣␤ engagement by peptide-MHC to the intracellular signaling machinery occurs via the TCR-associated CD3 chains, which are arranged into three dimers: ␥⑀, ␦⑀, and (1). Each CD3 chain contains immunoreceptor tyrosine-based activation motifs (ITAMs); one each in ␥, ␦ and ⑀ and three in . The eponymous features of these motifs are a pair of tyrosine residues separated by 9 -11 amino acids. These tyrosines become rapidly phosphorylated by the Src-family kinase Lck following TCR stimulation; a required event for initiating TCR signaling (2,3).

Tyrosine Kinase Cascade, Phosphorylation of Linker
Proteins, and Assembly of Signalosome Given the primacy of ITAM phosphorylation by Lck in TCR signaling, an especially important question to answer is how is ITAM tyrosine phosphorylation maintained below the signaling threshold prior to TCR engagement. Current data suggest that multiple mechanisms act in concert to block spontaneous TCR signaling. At the first level, there is a physical sequestration of Lck away from the TCR by virtue of differential partitioning of Lck and the TCR into lipid rafts (4,5). Lipid rafts are heterogeneous lipid microdomains relatively enriched in sphingomyelin, glycosphingolipids, and cholesterol that spontaneously form in cell membranes as a consequence of the biophysical properties of the different lipids that comprise the membrane. Lck (by virtue of it being myristoylated and palmitoylated) constitutively partitions to the lipid rafts, whereas the unstimulated TCR is largely excluded from this fraction. The CD3 ITAMs are also maintained in a subcritical state of tyrosine phosphorylation by tyrosine phosphatases, which have ready access to the TCR prior to TCR stimulation but more limited access following TCR stimulation (6,7). In addition, prior to TCR engagement, Lck is maintained in an inactive state by the combined actions of Csk and PEP. Csk is a tyrosine kinase that phosphorylates the negative regulatory C-terminal tyrosine residue of Lck, and PEP is a hematopoietically restricted phosphatase that associates with Csk via the SH3 domain of Csk and dephosphorylates the activation loop tyrosine of Lck. Csk is co-localized to the lipid raft resident Lck via binding to tyrosine-phosphorylated Cbp/ PAG, which is constitutively associated with the lipid rafts (8,9).
Following stimulation, there is increased distribution of TCR to the lipid rafts and sequestration of negative regulatory tyrosine phosphatases away from the TCR (4,5). Concurrently Lck becomes activated via dephosphorylation of the regulatory C-terminal tyrosine in response to increased exposure to CD45, a transmembrane phosphatase that dephosphorylates the negative regulatory site, and decreased exposure to Csk as Cbp/PAG is transiently dephosphorylated (9). In addition, because a portion of Lck is constitutively associated with the CD4 coreceptor, the peptide-MHC-induced co-localization of TCR with CD4 results in an increased local concentration of Lck around the TCR. The ITAM sequences of the CD3 chains subsequently become fully tyrosine-phosphorylated. Once fully phosphorylated, these motifs serve as binding sites for ZAP-70, which binds via its tandem SH2 domains (10). ZAP-70 is activated following ITAM recruitment via Lck-mediated tyrosine phosphorylation of the activation loop tyrosine (Tyr-493) of ZAP-70. Activated ZAP-70 autophosphorylates at tyrosines 292, 315, and 319. These sites serve to recruit various positive and negative signaling effectors to the TCR complex (10,11). In addition to serving as a scaffold via self-phosphorylation, ZAP-70 also phosphorylates a restricted set of substrates following TCR stimulation. These include ␣-tubulin, Sam-68, Vav-1, VHR, Shc, Gab2, LAT, and SLP-76 (11). These latter two substrates in particular have been recognized to play a pivotal role in TCR signaling and are considered in more detail below.
When phosphorylated, both LAT and SLP-76 act as linker/ adapter proteins, which serve as nucleation points for the construction of higher order multimolecular signaling complexes (Fig. 1), often referred to as the signalosome (8,(12)(13)(14). Acting in concert, these linker/adapter proteins regulate the activation of PLC␥1 and the subsequent hydrolysis of phosphatidylinositol (PI)-4,5-P 2 to generate diacylglycerol (DAG) and inositol-1,4,5-P 3 (IP 3 ), second messengers in protein kinase C (PKC) and Ras activation (via DAG) and calcium mobilization (via IP 3 ). They also play an important role in the activation of another TCR-activated protein-tyrosine kinase, Itk (also known as Emt and Tsk). Itk serves as an integrator of signals arising from the signalosome and from phosphoinositide 3-kinase (PI3K; see below), because signals from both pathways are required for its activation (15,16). Itk has been implicated in 1) PLC␥1 activation, 2) regulation of TCR-stimulated actin cytoskeleton reorganization via its involvement in regulating the Vav-1/Cdc42/WASP pathway (see below), and 3) in so-called "inside-out" signaling, which results in activation of ␤ 1 integrin adhesion factors (15,16). The Lat and SLP-76 signaling complexes also support the activation of multiple other signaling proteins, including mitogen-activated protein kinases (p38, Jnk, and Erk) and small molecular weight G proteins (Ras, Rac, Rho, and Cdc42).
A key feature of LAT is that it is a transmembrane adapter * This minireview will be reprinted in the 2004 Minireview Compendium, which will be available in January, 2005.
protein that is constitutively targeted to the lipid rafts by virtue of palmitoylation of two juxtamembrane cysteine residues (8,13,14). It has no protein interaction domains other than the multiple tyrosine residues that are phosphorylated by ZAP-70 following TCR signaling. When phosphorylated, these tyrosines serve as binding sites for specific SH2 domain-containing proteins. There are a total of 8 tyrosine residues in LAT that are conserved between humans, mice, rats, and bovines (14). The C-terminal 5 tyrosine residues play critical roles in the ability of LAT to bind to PLC␥1 and the linker proteins Gads and Grb2. SLP-76 associates with LAT via Gads. SLP-76 can also directly bind to PLC␥1, indicating the existence of higher order interactions between these and possibly other signaling proteins in the signalosome. In addition to the proteins already mentioned, phosphorylated LAT also binds to PI3K, Grap, 3BP2, Shb, SOS, c-Cbl, Vav, and Itk, localizing these molecules in close proximity and a defined orientation to one another within the lipid raft domain of the plasma membrane. Jurkat T cells deficient for LAT expression exhibit severe signaling defects to Ca 2ϩ mobilization, mitogen-activated protein kinase activation, and NFAT activation (8,13,14).
Unlike LAT, SLP-76 is a cytosolic protein (12). The structure of SLP-76 includes an acidic N-terminal region, which includes three sites of tyrosine phosphorylation (Tyr-113, Tyr-128, and Tyr-145). When phosphorylated, these sites can bind the SH2 domains of Vav-1, Nck, and Itk. The acidic region is followed by an extended proline-rich region, which binds the SH3 domains of Gads, Itk, and PLC␥1. The C-terminal portion of SLP-76 is comprised of a single SH2 domain, which binds primarily to ADAP, which has been reported to play a key role in inside-out signaling to integrins (17). Jurkat T cells that lack SLP-76 are defective in their ability to activate: 1) PLC␥1 and consequently Ca 2ϩ mobilization and NFAT activation; 2) the Ras/Raf/Mek/Erk pathway; 3) NFB; and 4) inside-out signaling to integrins (12,17,18).

Cytoskeletal Reorganization and Formation of
Immunological Synapse A new layer of complexity in the process of TCR signaling has come to light in recent years in the form of the immunological synapse (IS), which has also been referred to as the supramolecular activation complex (SMAC). This is a dynamic yet highly ordered structure that forms at the site of T cell contact with an antigenpresenting cell (19). The mature IS is characterized by a central region (c-SMAC) that is enriched in clustered TCR and PKC, surrounded by a peripheral ring (pSMAC) of adhesion factors such as LFA-1 and a distal ring (dSMAC) containing proteins such as the tyrosine phosphatases CD148 and CD45. The IS, although not required for initiating TCR signaling, is required for sustained signaling, IL-2 production, and proliferation (19). It also has the ability to act as a positive or negative servo, either amplifying weak TCR signals or attenuating strong signals (20).
Clustering of lipid rafts at the contact site and formation of the IS is an active process that requires several upstream signaling events to occur. Most important of these are the signals that lead to reorganization of the actin cytoskeleton. The proximal catalyst for this is the recruitment and activation of the Arp2/3 complex, which catalyzes the formation of new nucleation sites for actin polymerization. A key upstream regulator of Arp2/3 is WASP, which is rapidly recruited to lipid rafts following TCR/CD28 costimulation (13,21). WASP is constitutively present at high stoichiometry in a complex with WIP and CrkL. WIP-bound WASP is refractory to activation. Upon TCR stimulation, this complex is recruited to the TCR (and consequently to the lipid rafts and the antigen-presenting cell contact site) via the binding of the CrkL SH2 domain to tyrosine-phosphorylated ZAP-70 (22,23). Co-localization of this complex with activated PKC at the lipid rafts results in the phosphorylation of WIP and disruption of the WIP-WASP association, thereby facilitating activation of WASP by Cdc42 (22). In this model, WIP serves both to keep WASP basally inactive and to facilitate WASP activation upon TCR stimulation. WASP can also be recruited to the lipid rafts via SH3 domain-mediated binding of Nck to a proline-rich region of WASP (24). In this case, the raft recruitment and activation of WASP are coordinated by SLP-76, which functions as a targeting scaffold, bringing WASP into close proximity with Cdc42 that has been activated by SLP-76-bound Vav-1 and also targeting WASP to the rafts via SLP-76-Gadsmediated binding to Lat. Itk also plays a critical role in this process, apparently at the level of supporting Vav-1 recruitment to the plasma membrane (25,26).

Activation of Transcription Factors
Changes in gene expression represent the culmination of the TCR signaling pathway and are required for the T cell to gain full proliferative competence and the ability to produce effector cytokines. Three transcription factors in particular have been found to play a key role in TCR-stimulated changes in gene expression; these are NFB, NFAT, and AP-1.
NFB-NFB is the collective name given to the dimeric transcription factors of the Rel family (27). Activation of NFB is primarily controlled via the nuclear cytoplasmic partitioning of NFB. In the absence of an activating signal, NFB is retained in the cytoplasm by tight binding to an inhibitory IB protein. Numerous stimuli including TNF␣, IL-1, and TCR/CD28 costimulation activate an IB kinase (IKK) complex containing two kinase subunits, IKK␣ and IKK␤, and a regulatory subunit, IKK␥ (also known as NEMO). The IKK complex phosphorylates IB and targets it for ubiquitination and proteolysis via the 26 S proteasome complex. The degradation of IB unmasks the nuclear localization sequence of NFB, allowing translocation to the nucleus, where NFB regulates the activity of its target genes (28).
Remarkably, TCR stimulation uses a completely different pathway for activating IKK than other stimuli (e.g. TNF␣ and IL-1), and many of the early signaling proteins that have been described above are required including Lck, ZAP-70, SLP-76, PLC␥1, and Vav-1 (29). A key step in NFB activation is the activation of PKC and its translocation to lipid rafts and the IS. PKC is a member of the "novel" class (DAG-responsive, Ca 2ϩ -independent) of PKCs, is selectively expressed primarily in T cells, and has the distinction of being the only PKC isozyme that is known to translocate to the immunological synapse (30). The mechanism of raft/IS recruitment remains undetermined but appears to involve CARMA-1 (see below). The target of PKC in activating IKK also remains unknown. It does not directly phosphorylate IKK and may act through calmodulin-dependent kinase II (29,30).
A critical upstream element in raft recruitment and activation of PKC is Vav-1-mediated activation of Rho family G proteins and reorganization of the actin cytoskeleton (30). Raft localization of PKC also requires Lck, PI3K, PDK1, SLP-76, PLC␥1, and CARMA-1 (29). Notably, the requirement for PLC␥1 is independent of its enzymatic activity, because neither pharmacological inhibition nor expression of a dominant-negative allele of PLC␥1 inhibits PKC raft recruitment (31,32). PLC␥1 is likely acting as a scaffolding protein in this pathway, possibly in a multimolecular complex with SLP-76 and Vav-1. PKC␣ but not PKC␤, both "conventional" PKC isotypes (responsive to DAG and Ca 2ϩ ), also plays an important but as yet undefined role upstream of PKC in TCR/ CD28-costimulated but not TNF␣-stimulated NFB activation (33).
Another pathway that affects PKC activation is PI3K. PI3K phosphorylates the D3 position on the inositol ring of PI-4,5-P 2 to generate PI-3,4,5-P 3 . This lipid serves as a selective binding site in the plasma membrane for certain pleckstrin homology domaincontaining proteins. The serine/threonine kinase Akt is one of these proteins (as is Itk). Akt is activated in response to PI3K activation, and Akt and PKC physically and functionally interact to synergistically activate NFB (34,35). PI-3,4,5-P 3 may also affect PKC via activation of Vav-1 although activation of Vav-1 by PI-3,4,5-P 3 is controversial (36,37). More recently, Vav-1 has been implicated as an upstream regulator of PI3K, so Vav-1 may function in a positive feedback loop for PI3K activation. Interestingly, the placement of Vav-1 upstream of PI3K may help to explain why Vav-1 is required for PKC recruitment to lipid rafts in T cells from Vav Ϫ/Ϫ mice but is dispensable for recruitment in Vav-negative Jurkat T cells (37,38). A notable difference between these two model sys-tems is that Jurkat T cells have constitutively high levels of PI-3,4,5-P 3 , because they fail to express the phosphatases that catabolize PI-3,4,5-P 3 (39,40). Thus Jurkat T cells have substantial resting levels of PI-3,4,5-P 3 in the absence of Vav-1-mediated activation of PI3K.
Most recently a series of genetic studies have established critical roles for a number of novel proteins in TCR/CD28-costimulated NFB activation. These include: 1) CARMA-1 (also known as CARD11 and Bimp3), which is a lymphocyte-specific scaffold molecule containing a caspase-recruitment domain (CARD) and a membrane-associated guanylate kinase-like (MAGUK) domain; 2) Bcl10, which is a CARD-containing serine/threonine kinase; and 3) MALT1/paracaspase, which is composed of an N-terminal death domain followed by two immunoglobulin-like domains and a Cterminal caspase-like domain (41)(42)(43). CARMA-1 is constitutively associated with lipid rafts. Upon TCR stimulation, CARMA-1 acts as a scaffold/adaptor protein recruiting Bcl10, PKC-, and IKK to the lipid rafts/IS and placing them in juxtaposition to one another and to the TCR (44 -47).
Bcl10, in turn, acts by targeting IKK␥ for lysine 63-linked ubiquitination on lysine 399 (48). Bcl10-induced IKK␥ ubiquitination and subsequent NFB activation requires the Bcl10-associated proteins, MALT1/paracaspase, and the ubiquitin-conjugating enzyme UBC13. Unlike lysine 48-linked polyubiquitination, which generally targets proteins for degradation in the proteasome (as is the case with IB), lysine 63-linked ubiquitination can often regulate protein function or intermolecular interactions (49). Notably K399R IKK␥ was defective in reconstituting NFB activation in an IKK␥-deficient Jurkat T cell line (48). The precise mechanism by which IKK␥ ubiquitination activates NFB activation is unknown but presumably involves changes in the structure of IKK that lead to increased IKK activity toward IB. Bcl10 is also found in complex with IKK␥, IKK␣, and IKK␤ following stimulation and may play a role in recruiting or stabilizing association of the IKK complex with the lipid rafts of the immunological synapse via the association of Bcl10 with CARMA-1.
That there is yet further complexity to the signaling pathway leading to TCR/CD28-costimulated NFB activation is suggested by a number or recent studies, including the observation that the cytosolic adaptor protein Shc is required for nuclear transport of c-Rel (50). In addition, another CARD-containing serine/threonine kinase, Rip2 (also known as RICK, CARDIAK, CCK, and Ripk2) has also been implicated in TCR/CD28-costimulated NFB activation (51)(52)(53). There are also additional kinases that clearly play a role in NFB activation. These include calmodulin-dependent kinase II, MEKK1, MEKK2, MLK3, COT/Tpl-2, and NIK (29).
NFAT-In comparison with NFB, the pathway leading to activation of NFAT is much simpler (54,55). The rate-limiting step in NFAT activation is the removal of key phosphate groups from the N terminus of the NFAT protein. Phosphorylation of these residues masks the nuclear localization sequence on NFAT, and when the phosphates are removed NFAT can translocate to the nucleus and regulate the expression of various genes. The dephosphorylation of NFAT is specifically carried out by the Ca 2ϩ -calmodulin-regulated phosphatase, calcineurin. Consequently, TCR-stimulated activation of PLC␥1, with the subsequent production of IP 3 and increase in intracellular Ca 2ϩ , is a critical component of NFAT activation.
Vav-1 is also recognized as playing a key role in NFAT activation (30,56). The importance of Vav-1 in TCR-stimulated NFAT activity is manifest at multiple steps. In mouse T cells, loss of Vav-1 is associated with greatly impaired Ca 2ϩ flux as a consequence of impaired activation of Itk, Tec, and PLC␥1. These effects are at least in part because of defective PI3K activation in these cells. Keeping in mind the previously discussed defect in PI-3,4,5-P 3 metabolism that is characteristic of Jurkat T cells, the important effector role that PI3K plays in Vav-1 signaling may explain why the Vav-1-negative Jurkat T cell line fails to recapitulate the phenotype of the Vav-1 Ϫ/Ϫ mouse T cells (37,38). However, despite high basal PI-3,4,5-P 3 , Vav-1-negative Jurkat T cells still show defective NFAT activation, which has been attributed to defective opening of calcium release-activated channels (57) and defective Jnk activation (38).
The NFAT activation/deactivation cycle is completed by the rephosphorylation of NFAT, which causes NFAT to repartition into the cytosol. This reaction can be carried out by several different serine/threonine kinases including CK1, CK2, Jnk, Erk, and p38; however, it is the GSK3 serine/threonine kinase that appears to play the dominant role in inactivating NFAT in T cells (54,55). The prominent role of GSK3 in NFAT deactivation also provides an additional pathway by which PI3K can lead to increased or sustained NFAT activation. GSK3 is basally highly activated and is inactivated when phosphorylated upon an N-terminal serine residue. Akt is particularly effective at phosphorylating and inactivating GSK3. Thus, PI3K activation induces Akt activation, phosphorylation, and inactivation of GSK3 and a slower rate of NFAT inactivation.
AP-1-Like NFB activation, the activation of AP-1 requires PKC activation, and PKC Ϫ/Ϫ mice fail to activate AP-1 in response to TCR stimulation (30,58). The AP-1 transcription factor is composed of dimers of c-Jun and c-Fos family proteins and can be activated both by phosphorylation of c-Jun by Jnk and by upregulation of c-Fos and c-Jun expression (59). Even though transient overexpression of PKC in T cell lines suggests an important role for PKC in Jnk activation, genetic ablation of PKC does not impair Jnk activation (30,58). The observation that genetic ablation of Jnk1 and Jnk2 interferes with NFAT activation but does not affect AP-1 activation is also consistent with PKC acting through an as yet undefined Jnk-independent pathway, possibly involving increased levels of c-Fos and/or c-Jun (58). AP-1 is also activated by PKC-independent pathways including the Ras/Raf/Mek/Erk pathway, which signals for increased expression of c-Fos. An important aspect of AP-1 function is its ability to form complexes with the NFAT and NFB transcription factors. It is particularly notable that the proximal NFAT binding sites of the IL-2 promoter cooperatively bind both NFAT and AP-1. Likewise the CD28RE site of the IL-2 promoter is a cooperative binding site for NFB and AP-1 (30,54,55,59).

Concluding Remarks
Over the past 15 years tremendous strides have been made toward understanding the molecular signaling events initiated by TCR engagement (11). A great many (maybe even most) of the molecular players have now been identified, and we have begun to comprehend a few of the lines and some of the ways in which the players interact with one another. However, we are still far from understanding the play in its entirety, but one thing is clear: it is not a simple matter to activate a resting T cell. Given how dangerous an inappropriately activated T cell has the potential to be, this is probably a good thing.