Biochemical Interactions Integrating Itk with the T Cell Receptor-initiated Signaling Cascade*

Itk, a Tec family tyrosine kinase, acts downstream of Lck and phosphatidylinositol 3′-kinase to facilitate T cell receptor (TCR)-dependent calcium influxes and increases in extracellular-regulated kinase activity. Here we demonstrate interactions between Itk and crucial components of TCR-dependent signaling pathways. First, the inositide-binding pocket of the Itk pleckstrin homology domain directs the constitutive association of Itk with buoyant membranes that are the primary site of TCR activation and are enriched in both Lck and LAT. This association is required for the transphosphorylation of Itk. Second, the Itk proline-rich region binds to Grb2 and LAT. Third, the Itk Src homology (SH3) 3 and SH2 domains interact cooperatively with Syk-phosphorylated SLP-76. Notably, SLP-76 contains a predicted binding motif for the Itk SH2 domain and binds to full-length Itk in vitro. Finally, we show that kinase-inactive Itk can antagonize the SLP-76-dependent activation of NF-AT. The inhibition of NF-AT activation depends on the Itk pleckstrin homology domain, proline-rich region, and SH2 domain. Together, these observations suggest that multivalent interactions recruit Itk to LAT-nucleated signaling complexes and facilitate the activation of LAT-associated phospholipase Cγ1 by Itk.

The T cell receptor (TCR), 1 B cell receptor, and mast cell receptor (Fc⑀RI) are composed of multiple subunits, enabling the coupling of a variable antigen-recognition subunit to conserved signaling modules (1)(2)(3). Collectively, these receptors direct the antigen-specific responses of T cells, B cells, and mast cells. Although these antigen receptors do not contain intrinsic tyrosine kinases, the early events downstream of these antigen receptors are controlled by a relatively conserved series of tyrosine kinase-dependent events. In each case, Src family kinases initiate the signaling cascade by phosphorylating immunoreceptor tyrosine-based activation motifs present in the cytoplasmic tails of the signal-transducing receptor subunits. Complete immunoreceptor tyrosine-based activation motifs phosphorylation results in the recruitment of Syk/ZAP-70 family tyrosine kinases to the receptor. These kinases are subsequently activated by Src kinase-mediated transphosphorylation. The substrates of the Syk kinases include internal phosphorylation sites as well as transphosphorylation sites in LAT, SLP-76, and BLNK, a SLP-76-related adapter protein (1, 4 -7). These Syk kinase-dependent phosphorylations direct the recruitment of additional effector and adapter molecules to LAT, SLP-76, BLNK, and the Syk kinases themselves (7)(8)(9)(10)(11)(12)(13). In T cells LAT, which is recruited into lipid rafts by dual palmitoylation, nucleates a raft-associated complex containing multiple effectors, including Grb2 and Sos, Gads and SLP-76, Cbl, Vav, PLC␥1, and PI3K (5,14,15). The perturbation of this complex by the disruption of Lck, ZAP-70, LAT, or SLP-76 results in profound defects in the activation of calcium influxes and mitogen-activated protein kinases (MAPKs) downstream of the TCR (16 -20). Similar complexes may form in both B cells and mast cells.
The Tec family kinases (Tec, Itk, Btk, Bmx, and Txk (21,22)) also participate in signal transduction pathways downstream of antigen receptors (3,(23)(24)(25)(26)(27)(28)(29)(30)(31). In particular, Tec kinases have been observed to contribute to the PI3K-dependent phosphorylation and activation of PLC␥ isoforms, the concomitant induction of sustained calcium influxes (32), and the activation of MAPKs (33)(34)(35). However, the mechanisms that integrate the Tec kinases into antigen receptor-dependent signaling pathways are not yet well understood. Structurally, these kinases share the conserved arrangement of SH3, SH2, and kinase domains found in Src kinases, but are distinguished by the absence of amino-terminal acylations, the absence of the carboxyl-terminal regulatory tyrosine, and by the presence of a conserved amino-terminal region containing a pleckstrin homology (PH) domain, a Btk homology motif, and a proline-rich region (PRR). Although the functions of the protein-binding domains of Tec kinases remain unclear, the phosphatidylinositol 3,4,5-trisphosphate (PtdIns(3,4,5)P 3 ) binding pocket present in Tec kinase PH domains is clearly required for the cooperative activation of PLC␥ isoforms by PI3K and Tec kinases (32). The importance of the PH domain is underscored by the observation that its disruption can result in immunological disorders in both humans (XLA) and mice (Xid) (36). This lipid-protein interaction directs the recruitment of Tec kinases to the plasma membrane and facilitates their transphosphorylation and activation by Src kinases (23,24,(37)(38)(39)(40)(41)(42)(43).
Although protein interaction domains have been shown to recruit kinases to specific complexes containing physiological activators and substrates, the protein-protein interactions responsible for the facilitation of calcium influxes and the activation of MAPKs by Tec family kinases have not yet been identified. Given the similarity of the phenotypes observed upon the disruption of Src kinases, Syk kinases, adapter proteins (LAT, SLP-76, and BLNK), and Tec kinases, the Tec kinases are likely to interact with these proteins in order to facilitate the formation or activation of a competent signaling complex. However, the Tec kinase ligands reported to date have not supported this hypothesis. For example, it has been suggested that Tec kinase PRRs facilitate the transphosphorylation and activation of Tec kinases by recruiting Src kinases (44). Unfortunately, the contributions of Tec kinase PRRs to the physiological transphosphorylation of Tec kinases and to the induction of calcium influxes by antigen receptors remain untested. Furthermore, although ligands have been reported for the SH3 domains of Tec kinases, the SH3 domain does not appear to be relevant to the induction of calcium influxes (45). In contrast, ligands of the SH2 domains of Tec kinases remain unknown, yet the SH2 domain is clearly involved in the induction of calcium influxes and participates in Tec kinase-mediated transformation (39,45). Since Syk kinases act upstream of the Tec kinases but do not appear to phosphorylate the Tec kinases (46 -48), Syk kinases may facilitate the recruitment of Tec kinases to sites at which they encounter either activators (Src kinases) or substrates (such as PLC␥) by generating docking sites for Tec kinase SH2 domains. Since the substrates of the Syk kinases include LAT, SLP-76, BLNK, and the Syk kinases themselves, interactions mediated by Tec kinase SH2 domains may recruit the Tec kinases into complexes with physiologically relevant partners.
The identification of ligands for the protein-interaction domains of Tec kinases may be complicated by the ability of these proteins to adopt a closed conformation. This closed conformation depends on intramolecular interactions between adjacent PRRs and SH3 domains and severely limits the participation of these domains in intermolecular interactions (49). To date, the functional consequences of this closed conformation have not been resolved. However, two lines of evidence suggest that the closed conformation limits the activity of Tec kinases. First, the deletion of the SH3 domain, which releases the Tec kinase PRR from intramolecular sequestration, potentiates both antigen receptor-dependent calcium influxes and Btk-dependent transformation (45,50). Second, Tec kinase autophosphorylation is strongly correlated with Tec kinase activation and is likely to disrupt the closed conformation, since the primary site of autophosphorylation is a tyrosine residue in the ligand-binding pocket of the SH3 domain (51)(52)(53). 2 Although the SH2 domain contributes to the stabilization of the closed conformation, the sequestration of the SH3 domain and PRR can be reversed by exposure to highly tyrosine-phosphorylated lysates (49), implying that intermolecular interactions driven by the SH2 domain may also antagonize the closed conformation of Tec kinases.
In this article we report the identification of ligands for the homology domains of Itk in both Jurkat T cells and in primary murine CD4 ϩ T cells. These observations suggest a model in which multivalent interactions overcome the closed conformation in order to recruit Itk to a LAT-nucleated signaling complex and facilitate the activation of LAT-associated PLC␥1 by Itk.
Expression Vectors-The native stop codon of Itk was deleted and replaced with a PCR-derived fragment encoding a short linker peptide and a carboxyl-terminal Myc epitope tag. These modifications extend Itk by 21 amino acids (AGAEFMEQKLISEEDLRKFDI). EF/Itk-myc expression vectors were generated by inserting the modified Itk cDNA into a derivative of pEF-BOS (54). EF/Itk-KR-myc includes a kinaseinactivating mutation (K390R (24)). Mutations inactivating the prolinerich region (P158A, P159A; PR*) and the binding pockets of the PH domain (R29C; Xid), SH3 domain (W208K; SH3*), and SH2 domain (R265A; SH2*) were introduced by either site-directed or PCR-directed mutagenesis. The EF/HA:SLP-76 expression vector was generated by subcloning the SLP-76 coding region into a transfer vector, and then into pEF-HA. This results in the insertion of a short linker-derived peptide (RSRTSGSPGRAA) between the triple-HA epitope tag present in pEF-HA (a kind gift of Anjana Rao) and the original start codon of SLP-76.
GST-Itk Fusion Proteins-GST fusion proteins were prepared as described previously (49,55). Briefly, inserts encoding the desired domains of Itk were generated by PCR and subcloned into the BamHI site of pGEX-2T. The TEC-, SH3-, and SH2-5Ј primers encode Itk products beginning at amino acids 97, 171, and 231, respectively. The TEC-, SH3-, and SH2-3Ј primers terminate at amino acids 174, 232, and 338, respectively. The soluble TEC fragment was prepared by thrombin cleavage of the TEC fusion protein as recommended by the manufacturer (Amersham Pharmacia Biotech).
Binding Properties of the Itk SH2 Domain-The Itk SH2 domain binding motif was obtained by the method of Songyang and Cantley (56), with minor modifications (57). In contrast to previous reports, the bound phosphopeptide mixture was eluted with 30% acetic acid rather than phenyl phosphate. The relative enrichments displayed in Fig. 5A were obtained normalizing the abundance (mol %) of amino acids in the bound mixture to their abundance in the starting library. Unlike previous reports of SH2 domain binding motifs, these data were not background corrected by subtracting the amino acid abundances observed after purification over GST prior to normalization. For comparison, the selection for proline at pYϩ3 is exceptionally strong, and is roughly equivalent to the selection of asparagine at pYϩ2 by the Grb2 SH2 domain.
Generation of Itk Overexpressing Stable Lines-Jurkat cells were transfected with 1 g of pBJneo and either 10 g of the empty pEF-HA expression vector, EF:Itk-KR-myc, EF:Itk-Xid/KR-myc, or EF:Itk-PR*/ KR-myc, as described (24). Stable cell lines were derived by selection in 1.4 mg/ml G418 (Life Technologies, Inc.). Lines expressing Itk-myc at approximately endogenous levels were expanded for use in subsequent experiments. These lines were determined to express normal levels of CD3⑀ by flow cytometry (FACScan, Beckton Dickinson).
Baculoviruses-The GST:Syk baculovirus was provided by Ian Mac-Neil (Ariad Pharmaceuticals, Cambridge, MA) and results in the expression of a GST fusion of full-length Syk. The Lck and Itk baculoviruses have been described (24). The HA:SLP-76 baculovirus was derived from the EF/HA:SLP-76 expression vector in the same manner.
Expansion of Murine CD4 ϩ Peripheral Lymphocytes-CD4 ϩ T cells were purified from murine spleens by removing red blood cells with red blood cell lysis buffer and killing CD8 ϩ T cells by incubation with anti-CD8 antibody (3.155) followed by complement (Cedarlane). Live lymphocytes were isolated and stimulated with 10 ng/ml PMA and 2 g/ml concanavalin A. Cells were rinsed into interleukin-2-containing media at 2, 4, and 6 days after the initiation of the cultures. By day 8 the TCR ϩ CD4 ϩ T cells had expanded approximately 100-fold and represented greater than 95% of the cells in culture, as assayed by FACS. These cells were rested 1 h at 37°C in the absence of interleukin-2 prior to stimulation. The CD4 ϩ T cells were stimulated by preincubating with biotinylated antibodies to both CD3⑀ and CD4 (10 g/ml each) at 4°C for 10 min and shifting to 37°C upon the addition of streptavidin (25 g/ml). After 2 min of stimulation, lysates were prepared as described.
Cellular Fractionation-This protocol was adapted from previously described methods for the isolation of buoyant membranes (58 -60). Both unstimulated and OKT3-stimulated Jurkat cells (10 8 ) were washed twice in ice-cold phosphate-buffered saline with 20 mM sodium fluoride and 1 mM sodium orthovanadate and resuspended in 1.6 ml of ice-cold homogenization buffer (20 mM Hepes, pH 7.4, 1 mM dithiothreitol, 2 mM MgCl 2 , 1 mM sodium orthovanadate, 12.5 mM KCl, 5 mM sodium fluoride, and 5 mM sodium pyrophosphate, supplemented with protease inhibitors phenylmethylsulfonyl fluoride, pepstatin A, aprotinin, and leupeptin (Sigma)). Cells were homogenized on ice using a Dounce homogenizer. Total homogenates were adjusted to gradient buffer conditions by the addition of homogenization buffer with concentrated salts. Gradient buffer contained 50 mM KCl, 20 mM sodium fluoride, and 20 mM sodium pyrophosphate, but was otherwise identical to homogenization buffer. Homogenates were adjusted to the desired density by the addition of 80% sucrose (w/v) in gradient buffer. The diluted homogenate was loaded into a SW41 ultracentrifuge tube and a gradient was created by successively overlaying 1-ml steps of the indicated sucrose solutions, also in gradient buffer. The step gradients were ultracentrifuged for 16 h at 39,000 rpm and at 4°C. Fractions were removed in 0.5-ml aliquots and the pellet was rinsed once in 0.5 ml of gradient buffer, then resuspended by sonication in 0.5 ml of gradient buffer. 20 l of each fraction was analyzed without concentration by Western blotting; thus each gel represents the total protein content from 20/500 (4%) of 10 8 cells. Optical density at 600 nm (Beckman DU-600 spectrophotometer) was used to monitor light scattering by vesicular structures (58). Protein content was determined by Bradford assay (Bio-Rad).
Surface Biotinylation-Jurkat cells were rinsed three times in phosphate-buffered saline supplemented with 1 mM MgCl 2 and 0.1 mM CaCl 2 , resuspended at 2 ϫ 10 7 cells/ml in rinse buffer, and incubated at 4°C with 1 mg/ml Sulfo-NHS-LC-biotin (Pierce) for 30 min. Excess labeling reagent was quenched by a 10-min incubation in rinse buffer supplemented with 50 mM glycine. Finally, the labeled cells were rinsed three times in rinse buffer prior to homogenization.
Modifications to Cellular Fractionation Procedures-Jurkat homogenates were prepared as descibed above. Prior to the addition of 80% sucrose various manipulations were performed. These homogenates were either untreated, cleared of nuclear material by a 10 min spin at 1000 ϫ g and 4°C, or adjusted to 0.5% Triton X-100. Membranes were also prepared from cleared homogenates by pelleting at 100,000 ϫ g and 4°C for 1 h. In the latter case, the pelleted material was rinsed clean and resuspended by sonication in an equivalent volume of homogenization buffer. The treated homogenates and purified membranes were adjusted to 80% sucrose and fractionated as described above. When included, Triton X-100 was present throughout the gradient at 0.5%.
Immunoprecipitations and Western Blotting-Itk-myc was immunoprecipitated from 2 ϫ 10 7 Jurkat cells using 2 g of the 9E.10 mAb prebound to 20 l of recombinant protein G-Sepharose (Life Technologies, Inc.). The in vitro reconstitution of the interaction between SLP-76 and Itk was performed by prebinding the anti-HA mAb 12CA5 to protein A-Sepharose and incubating aliquots of the 12CA5-Sepharose beads with saturating amounts of baculoviral HA:SLP-76. The resulting HA:SLP-76-Sepharose beads were rinsed extensively and incubated with fresh Sf9 lysates containing Itk for 3 h at 4°C. Recombinant GST-Grb2 (Santa Cruz Biotechnology) was added to the indicated in-cubations at 2 g/ml. Western blotting protocols have been described elsewhere (55).
Secreted Alkaline Phosphatase Assays for NF-AT Activation-Jurkat/ Tag cells were transiently transfected by electroporating 2 ϫ 10 7 cells in 0.4 ml of RPMI 1640, 25 mM HEPES, and 2 mM glutamine at 310 V and 960 microfarads with 10 g of pSX-NFAT/SEAP and a total of 40 g of pEF-BOS-derived vector. EF/HA:SLP-76 or pEF-HA accounted for 20 g of this total, and the EF/Itk-myc variants or pEF-HA accounted for the remaining 20 g of this total. After 24 h each transfection was split into 8 pools. These pools were transferred, in duplicate, to 24-well plates which were untreated, coated with OKT3 ascites, or contained 10 ng/ml PMA and 1 M ionomycin. The remaining two pools were used to confirm the uniform overexpression of SLP-76 and Itk. After 18 h of stimulation, the supernatants were assayed, in triplicate, for SEAP activity by the method of Spencer et al. (61). The results presented were normalized to a percentage of the response to PMA and ionomycin.

Itk Is Present in Surface-exposed Buoyant Vesicles Similar to
GEMs-The PH domain of Itk has been shown to be required for the PI3K-dependent recruitment of Itk to COS cell membranes (40), but the specific role of the lipid binding pocket of the PH domain has not been assessed. In order to determine the contribution of the lipid binding pocket of the Itk PH domain to the membrane recruitment of Itk in T lymphocytes, we used an assay for membrane association which is based on the buoyant density of membrane vesicles prepared by mechanical disruption in the absence of detergents. Our procedure is based on the method of Bourguignon et al. (58), and is similar to methods commonly used for the preparation of both caveolae and detergent-resistant structures variously referred to as lipid rafts, GEMs (glycosphingolipid-enriched membranes), detergent-insoluble GEMs, and detergent-insoluble membranes (62)(63)(64)(65).
This method generated at least three populations of buoyant vesicles (20 -30, 35, and 50% sucrose) and a pellet containing dense or insoluble material, as indicated in Fig. 1A. The buoyant density of the least dense population of vesicles observed by this method was similar to that of GEMs observed in T cells in previous studies, suggesting that these structures may be related (14,66). In addition, the distributions of fluorescent lipid analogues within these gradients were compatible with predictions based on the composition of GEMs. Whereas light vesicles were preferentially enriched in BODIPY-sphingomyelin and DiI, which contain two saturated C 18 acyl chains, FAST-DiI, which contains two unsaturated C 18 acyl chains, was preferentially enriched in the 50% sucrose fractions (Fig. 1B). Furthermore, these light vesicles contained many proteins characteristic of GEMs, including LAT (Fig. 1C), Lck (Fig. 2), Fyn (data not shown), and heterotrimeric G protein subunits (data not shown).
Strikingly, both Itk and CD3⑀ were constitutively enriched in the light vesicle fractions (Fig. 1C). The amount of Itk observed in the light vesicle fractions was somewhat variable, ranging from 10 to 25% (data not shown). Both Itk and CD3⑀ were also present in the insoluble pellet. The Itk and CD3⑀ present in the pellet resisted detergent extraction and could be cleared by low speed centrifugation, suggesting that they may be directly associated with the cytoskeleton (67) (Fig. 1C). Substantial amounts of Itk were also present in the 50% sucrose fractions. This pool of Itk could not be pelleted by centrifugation at 100,000 ϫ g in low sucrose, and is therefore freely soluble (data not shown).
Based on surface biotinylation, the light vesicles we detected were surface-exposed, suggesting that they are at least in part derived from the plasma membrane (Fig. 1C). Under the conditions used, most surface-exposed proteins were found in the insoluble pellet, and could be cleared from the lysates by a brief low speed centrifugation. This indicates that light vesicles are a specific subset of the total plasma membrane. The absence of both LAT and calnexin from the insoluble pellet proved that the presence of surface biotinylated proteins in the pellet was not a trivial result of incomplete cell disruption. Light vesicles were also distinct from the endoplasmic reticulum, as low speed centrifugation cleared the lysates of calnexin, an endoplasmic reticulum marker, without affecting the amounts of LAT, Itk, and CD3⑀ found in the buoyant fractions (Fig. 1C).
However, the light vesicles derived by this procedure were relatively sensitive to detergent extraction, and only proteins present in the fractions with the lowest buoyant density resisted extraction by 0.5% Triton X-100 (LAT, Fig. 1C (3)), DiI-C 18 (3), or BODIPY FL C 5sphingomyelin for 30 min at 37°C prior to homogenization. Relative fluorescence was determined using appropriate excitation and emission wavelengths. C, surface biotinylated Jurkat cells were fractionated in the presence (left panel) or absence (center and right panels) of 0.5% Triton X-100. In one case (right panels) insoluble material was cleared from the initial homogenate prior to fractionation by one 10-min spin at 1000 ϫ g and 4°C. Biotinylated proteins were detected by blotting with a streptavidin-horseradish peroxidase conjugate (only one representative strip is shown); all other proteins were detected by standard Western blotting.
teristics with lipid rafts, which have been proposed to act as sites of signal initiation (70 -72), we wished to determine if TCR-dependent signals were initiated in light vesicles. In order to address this question we fractionated homogenates of Jurkat cells stimulated for varying times and analyzed the distributions of signaling molecules within the resulting gradients (Fig.  2). TCR, Lck, and LAT were all present in the light vesicles, and the fraction of these proteins present in the light vesicles did not change significantly following TCR stimulation (data not shown). In contrast, ZAP-70, Grb-2, and CrkL were recruited to light vesicles after TCR stimulation. This recruitment was very rapid, and paralleled the tyrosine phosphorylation of TCR, which peaked at 30 s. The recruitment of these molecules was also transient, and declined within 2 min, despite the persistence of tyrosine-phosphorylated TCR. The transience of the recruitment of ZAP-70 to the light vesicles may explain our ability to detect an unusually high stoichiometry of ZAP-70 recruitment, as previous studies did not examine time points earlier than 2 min post-stimulation (5,66). Unexpectedly, we did not observe any changes in the distribution of Itk following stimulation.
A more extensive time course (data not shown) revealed that changes in the gel-mobility of light vesicle-associated Lck were apparent within 15 s of TCR ligation, and that the recruitment of ZAP-70 to the light vesicles could be observed within the same time frame. The recruitment of Grb2 to the light vesicles lagged slightly, and was maximal between 30 s and 2 min after TCR cross-linking. ERK activation, as assayed by Western blotting with phosphorylation site-specific antibodies, was not detected until 1 min after TCR cross-linking. Interestingly, peak tyrosine phosphorylation occurred in the cytosolic fractions between 2 and 5 min. These data suggested that TCRmediated responses are specifically initiated in the light vesicles; therefore, the association of Itk with the light vesicles may be critical for the activation of Itk downstream of the TCR.
The Inositide Binding Pocket of the Itk PH Domain Is Required for Itk Transphosphorylation-The PH domains of Tec kinases have been proposed to mediate membrane recruitment required for the transphosphorylation of Tec kinases by Src kinases. Additionally, the PRR present in Tec kinases has also been proposed to promote the transphosphorylation of Tec kinases by promoting a direct interaction with Src kinase SH3 domains (44). In order to compare the contributions of the lipid binding pocket in the Itk PH domain and the Itk PRR to the transphosphorylation of Itk we generated expression vectors encoding Myc-tagged kinase-dead Itk (Itk-KR-myc) with or without additional mutations in the lipid binding pocket of the PH domain or the PRR (Xid or PR*, Fig. 3A). These vectors were used to establish stable Jurkat cell lines expressing variants of Itk-myc KR at approximately the same level as wildtype Itk. These cell lines were stimulated with pervanadate, and the transphosphorylation of the epitope-tagged Itk was assessed by anti-phosphotyrosine immunoblotting. These experiments revealed that the inositide binding pocket of the PH domain, but not the PRR, was required for the transphosphorylation of Itk (Fig. 3B).
Fractionation experiments also demonstrated that the transphosphorylation of Itk correlated strictly with the light vesicle association of Itk (Fig. 3C). In addition, in vitro binding assays indicated that although the Itk PRR is capable of binding many SH3 domains, including the Fyn SH3 domain, it cannot bind to the Lck SH3 domain (Fig. 3D). Since the transphosphorylation of Itk in Jurkat cells depends on Lck, rather than Fyn (23,24), these data support a model in which productive interactions between Itk and Lck require their colocalization in lipid raft-like structures.
Interactions of the Itk Homology Domains-Although the Itk PH domain clearly has a dominant role in the transphosphorylation of Itk by Src kinases, the remaining Itk homology domains may facilitate the colocalization of Itk with activating receptors or with substrate-containing complexes. In order to identify binding proteins for the Itk PRR, SH3 domain, and SH2 domain we performed binding assays using GST-Itk fusion proteins (Fig. 4A). We used lysates of CD4 ϩ murine T cell lines stimulated with antibodies to both CD3⑀ and CD4 for these binding assays, which allowed us to avoid artifacts of cellular transformation/immortalization and to take advantage of the augmentation of TCR-dependent signals by the CD4 coreceptor.
The proteins bound by the Itk-domain fusion proteins included 150, 120, 95, 76, 70, and 36-kDa tyrosine phosphoproteins (Fig. 4B). These bands corresponded to PLC␥1, Cbl, Vav, SLP-76, ZAP-70, and LAT, as indicated. PLC␥1 was observed to bind the Itk SH3 domain by Western blotting, and comigrated with the 150-kDa tyrosine phosphoprotein (data not shown). Cbl, Vav, SLP-76, and ZAP-70 were identified by specific re-immunoprecipiation from denatured binding assays (data not shown). Finally, LAT, which bound both PRR-and SH2 domain-containing fusion proteins, was identified by reprecipitation with the Grb2 SH2 domain fusion protein (data not shown). Similar results were obtained using Jurkat T cells (data not shown). It should be noted that the Itk PRR is unlikely to interact directly with LAT, which lacks an SH3 domain. Instead, this interaction is likely to be bridged by Grb2, which binds to the Itk PRR via its amino-terminal SH3 domain ( Fig. 3D and Ref. 49) and to LAT via its SH2 domain (5). Gads, a Grb2-related adaptor protein might also function in this manner (15). Since PLC␥1, Cbl, Vav, SLP-76, and Grb2 all co-immunoprecipitate with phosphorylated LAT (5), the binding of these signaling molecules by distinct Itk protein domains implies that Itk can interact with the LAT-nucleated signaling complex by more than one mechanism.
Identification of a Binding Motif for the Itk SH2 Domain-Since SLP-76 was the most prominent TCR-induced tyrosine

FIG. 2. Signal initiation occurs in the light vesicles.
Jurkat cells were preincubated for 15 min on ice in the presence or absence of OKT3 (3 g/ml). Cells were stimulated at 37°C for the indicated times by the addition of rabbit anti-mouse IgG (15 g/ml). Homogenates and gradients were generated as described. The resulting fractions were analyzed by Western blotting. Phospho-TCR was detected with the antiphosphotyrosine mAb 4G. 10. phosphoprotein bound to the Itk SH2 domain in both murine T cells and Jurkat T cells (Fig. 4B and data not shown), we suspected that phosphorylated SLP-76 could bind directly to the Itk SH2 domain. This hypothesis was supported by the observation that the Itk SH2 domain could precipitate SLP-76 from denatured lysates (data not shown). In order to facilitate the identification of Itk SH2 domain-binding sites in SLP-76 we determined a consensus binding motif for the Itk SH2 domain using oriented peptide library screening. A summary of the enrichment values for the most strongly selected residues at each position is shown in Fig. 5A. The Itk SH2 domain displays a strong preference for alanine and glutamate at the first degenerate position (pYϩ1), moderate preferences at the second degenerate position (pYϩ2), and an exceptionally strong preference for proline at the third degenerate position (pYϩ3). Notably, the SH2 domains of Vav and Nck share a strong preference for proline at pYϩ3 (56,73). This is striking since Vav and Nck have been shown to interact with the predicted Itk SH2 domain-binding sites in SLP-76 (9, 74 -76). Finally, the binding motif of the Itk SH2 domain predicts binding sites in BLNK and CD19, as well as Syk, ZAP-70, and SLP-76 (Fig. 5B). Thus, direct biochemical interactions mediated by the SH2 domains of Tec kinases may link the Tec kinases to mediators of antigen-induced calcium influxes in multiple systems (17,(77)(78)(79).
Characterization of the Interaction between Itk and SLP-76 -The predicted Itk SH2 domain-binding sites in SLP-76 (Fig. 5B) are known to be ZAP-70/Syk substrates (6). To confirm that these sites in SLP-76 were capable of binding the Itk SH2 domain, we tested the ability of the Itk SH2 domain to precipitate HA-tagged SLP-76 which had been expressed alone in Sf9 cells, or co-expressed with either Lck or GST:Syk (Fig.  5C). As expected, the binding of SLP-76 to the Itk SH2 domain was induced by the co-expression of SLP-76 with Syk, but not Lck. Further confirmation that the Syk-phosphorylation sites in SLP-76 are capable of binding to the Itk SH2 domain was obtained by using SLP-76-derived phosphopeptides in competition assays (Fig. 5D). For comparison, a phosphopeptide based on a predicted Grb2 SH2 domain-binding site (pYENV) in SHP-1 was included in these assays. The pYEPP peptide corresponding to tyrosine 145 in SLP-76 was the most effective (pY145) at inhibiting the interaction between the Itk SH2 domain and pervanadate-phosphorylated SLP-76; the pYENV peptide and the pYESP peptide corresponding to tyrosine 113 in SLP-76 (pY113) were marginally less effective. A second pYESP-containing peptide corresponding to tyrosine 128 in SLP-76 (pY128) was observed to inhibit binding of total tyrosine phosphoproteins to the Itk SH2 domain at 1 mM (data not shown), but bound the Itk SH2 domain with substantially lower affinity than the other peptides. As expected, the remaining peptides, pYITR (pY426) and pYLLG (pY483), did not detectably inhibit the binding of SLP-76 to the Itk SH2 domain. The inability of the pY113 and pY145 peptides to compete with phospho-LAT for binding to the Grb2 SH2 domain confirmed the specificity of these peptides for the Itk SH2 domain (Fig.  5E). These data suggest that Itk is recruited to SLP-76 following the phosphorylation of SLP-76 by ZAP-70. The Syk kinasedependent recruitment of Tec kinases to signaling complexes, such as the LAT⅐SLP-76 complex, could explain the apparent placement of Syk kinases upstream of Tec kinases.
Since the Itk SH3 domain also appeared to bind SLP-76 directly (Fig. 5C) we also mapped Itk SH3 domain-interacting sites in SLP-76 by peptide competition (Fig. 5F). Candidate Itk SH3 domain-binding sites within SLP-76 were identified by similarity to predicted Itk SH3 domain binding motifs (KXXP-PXP and PPXPXXR (55)). For comparison, a high affinity SH3 domain-binding peptide was also included in these assays (GW-YSKPPPPIP (55)). The SLP-76 peptide 196 -208 competed with  Thus, we assessed the relative contributions of these two domains of Itk to the binding of SLP-76 using peptide competitors. For these experiments we used the high affinity Itk SH3 domain ligand described above (GWYSKPPPPIP) and the SLP-76-derived pY145 phosphopeptide to compete with pervanadate-phosphorylated SLP-76 for binding to the Itk SH32 and TEC2 fusion proteins. In each case, the maximal inhibition of SLP-76 binding was only observed when both peptides were present (Fig. 6A). We also tested the relative contributions of the Itk PRR, SH3 domain, and SH2 domain to the engagement of SLP-76 using a panel of point mutant Itk TEC2 fusion proteins. These experiments confirmed that the Itk SH3 and SH2 domains each contribute to SLP-76 binding, even in the context of the conformationally closed TEC2 fusion protein (Fig. 6A)). Since the contribution of the Itk SH3 domain to the interaction with SLP-76 must come at the expense of its involvement in the sequestration of the Itk PRR, the engagement of SLP-76 must induce a conformational change in Itk.
We expected that the occupation of the Itk SH3 domain by SLP-76 would promote Grb2-mediated interactions between the Itk PRR and LAT, stabilizing Itk in a LAT-nucleated, SLP-76-containing complex. To mimic the effects of a bidentate interaction between Itk and SLP-76 we limited the availability of the Itk SH3 domain present in the TEC2 fusion protein. As expected, the inactivation or occupation of the Itk SH3 domain promoted the interaction of LAT with the TEC2 fusion protein (Fig. 6B). Vav, a member of the LAT⅐SLP-76 complex, also bound more effectively to the SH3 domain-mutant TEC2 fusion protein. The contribution of the Itk PRR to the interaction between the TEC2 fusion protein and LAT was confirmed by inactivating the Itk PRR (Fig. 6B). In contrast, Cbl, a direct ligand of the Itk SH3 domain, depended on the Itk SH3 domain for its interaction with the TEC2 fusion protein (Fig. 6B). Thus, conformational changes induced in Itk as a result of its bidentate interaction with SLP-76 could facilitate Grb2-mediated interactions between Itk and the LAT-nucleated signaling complex. Furthermore, the autophosphorylation of the Itk SH3 domain could have a similar effect on the ability of Itk to interact with LAT through the Itk PRR.
In Vitro Reconstitution of the Interaction between Itk and SLP-76 -In order to demonstrate that an intramolecular interaction similar to that observed in the Itk TEC2 fusion protein also occurs in full-length Itk, we performed a series of in vitro binding assays. First, we compared the ability of fulllength Itk to bind SLP-76 to the ability of a GST-Itk chimera lacking the amino terminus of Itk (see Fig. 4A, GST-32KR) to bind SLP-76. As we expected, the GST fusion protein, but not full-length Itk, bound to Sepharose beads baited with unphosphorylated SLP-76, implying that the Itk SH3 domain is sequestered in the context of full-length Itk (data not shown). Second, by baiting beads with differentially phosphorylated SLP-76 we established that Syk-phosphorylated SLP-76 is capable of binding full-length Itk, despite the sequestration of the Itk SH3 domain (Fig. 7). Finally, the involvement of the Itk SH3 domain in the interaction between full-length Itk and Syk-phosphorylated SLP-76 was suggested by the observation that the addition of soluble GST-Grb2 to these binding assays promoted the interaction between Itk and SLP-76 (Fig. 7). These experiments are consistent with a model in which Grb2 stabilizes bidentate interactions between Syk-phosphorylated SLP-76 and full-length Itk by occupying the Itk PRR and limiting its ability to compete with SLP-76 for access to the Itk SH3 domain. Based on the biochemical evidence presented above, this interaction could occur in the context of a lipid raft-associated LAT-nucleated signaling complex.
Itk Antagonizes the Effect of SLP-76 on NF-AT Transcription-Since biochemical studies in Itk-deficient T cells revealed that Itk is required for the activation of MAPKs and calcium influxes downstream of the TCR, we expected that Itk would contribute to the TCR-dependent activation of the nuclear factor of activated T cells (NF-AT) (35,80). This would be consistent with a role for Itk downstream of LAT and SLP-76, which are required for the activation of NF-AT, the activation of MAPKs, and the activation of calcium influxes (18,20). Because SLP-76 promotes transcription from NF-AT response elements, we wished to test the ability of kinase-inactive Itk to inhibit NF-AT-dependent transcription in the presence or absence of SLP-76 (81). We found that kinase-inactive Itk sup- B, murine CD4 ϩ T cells were stimulated by co-crosslinking CD3 and CD4 for 2 min. Lysates were purifed over the indicated panel of GST fusion proteins. Bound phosphoproteins were detected by Western blotting with the anti-phosphotyrosine mAb 4G. 10. Similar experiments were performed in CD3-and pervanadate-stimulated Jurkat T cells with similar results. The identities of these proteins were variously determined by co-migration, by direct Western blotting, immunodepletion, and specific re-immunoprecipitation from denatured binding assays followed by anti-phosphotyrosine Western blotting. Migration positions of GST fusion proteins are indicated by asterisks (*).
pressed the ability of SLP-76 to enhance both basal and TCRinduced transcription from NF-AT response elements, and thus acted as a dominant-negative inhibitor of this signaling pathway (Fig. 8A). Kinase-inactive Itk also suppressed the TCRinduced activation of NF-AT in the absence of exogenous SLP-76. These results suggested that SLP-76 and Itk may interact in vivo.
Unexpectedly, kinase-active Itk also suppressed the ability of SLP-76 to enhance transcription from NF-AT response elements (Fig. 8A). This revealed that Itk catalytic activity is not limiting for the activation of NF-AT in Jurkat cells, and furthermore, that the inhibition of NF-AT mediated by kinaseinactive Itk does not depend on the exclusion of catalytically active Itk from its normal sites of action. Instead, we postulate that this transcriptional suppression arises from competition between Itk and other effector molecules for access to a limiting number of functional docking sites in the LAT⅐SLP-76 complex. In order to identify the domains contributing to the suppres-sion of SLP-76-mediated transcription, we assessed the ability of domain-inactivated variants of Itk to suppress SLP-76-dependent NF-AT activation. These experiments revealed that the binding properties of the Itk PH domain, PRR, and SH2 domain were all required for the suppression of SLP-76-dependent NF-AT activation by Itk (Fig. 8B). DISCUSSION The PH domains of Tec kinases bind PtdIns(3,4,5)P 3 and have been shown to mediate the PI3K-dependent recruitment of Tec kinases to the plasma membrane (see Fig. 9). In this article we confirm that the inositide binding pocket of the Itk PH domain is required for the membrane localization of Itk. Furthermore, we demonstrate that the lipid binding pocket of the Itk PH domain directs the association of Itk with a specific subset of the plasma membrane that is similar in composition to lipid rafts. This raft-related subset of the plasma membrane is enriched in Itk, as well as TCR subunits, Lck, and LAT, and FIG. 5. Predicted ligands for the Itk SH2 and SH3 domains. A, a degenerate phosphopeptide library containing a fixed phosphotyrosine residue was applied to a GST-Itk SH2 domain affinity matrix. The eluted peptides were microsequenced and relative enrichments were obtained as described under "Experimental Procedures." pYϩ1, pYϩ2, and pYϩ3 indicate the first, second, and third degenerate position carboxyl-terminal to the fixed phosphotyrosine residue. Relative enrichments are presented for strongly selected amino acids. There was no preferred selection observed in positions NH 2 -terminal to the phosphotyrosine resdiue. B, potential Tec kinase SH2 domain ligands were scanned by eye for potential Itk SH2 domain binding sites. The positions of the tyrosine residues in the full-length protein are indicated at the right in parentheses. All the sites presented are derived from human sequences and are cytoplasmic. All of these sites, with the exception of those in BLNK, are known to be phosphorylated in vivo. The PIDs for these proteins are as follows: SLP-76 (g1082893), BLNK (g3406749), CD19 (g1705708), CD28 (g338445), ZAP-70 (g1177044), and SYK (g627588). C, Sf9 cells were infected with the indicated baculoviruses. Lysates were affinity purified over Itk SH3 and SH2 domain fusion proteins; bound SLP-76 was detected by Western blotting for the HA-epitope tag. D, Jurkat cells were treated with pervanadate prior to lysis. Lysates were preincubated for 30 min at 4°C with the phosphopeptides, as indicated. The GST-Itk SH2 domain was added to this mixture and incubated for 90 min at 4°C. Unbound proteins were removed by rapid washes at 4°C, and SLP-76 was detected by Western blotting. E, Jurkat cells were stimulated by CD3 cross-linking prior to lysis. Peptide competition experiments were performed as in D, and the 36-kDa Grb2 SH2 domain-binding tyrosine phosphoprotein (LAT) was detected by anti-phosphotyrosine Western blotting. F, lysates of unstimulated Jurkat cells were preincubated with the indicated peptides and binding assays were performed using the Itk SH3 domain. Bound SLP-76 was detected by Western blotting. The GWYSKPPPPIP peptide used as a control for the inhibition of SLP-76 binding was initially generated based on the selection of Itk SH3 domain ligands from phage display libraries.
is involved in signaling events immediately proximal to the TCR.
The enrichment of Itk in these raft-related structures could contribute to the function of Itk in a number of ways. First, the colocalization of Itk with Lck could promote the transphosphorylation of Itk by Lck. Second, the colocalization of Itk with LAT could stabilize interactions between these proteins. Third, since phosphatidylinositol phosphates have been observed to be enriched in rafts, the localization of Itk in these raft-related structures could favor the specific phosphorylation and activation of PLC␥1 molecules poised to hydrolyze their substrates (70,82). As expected, the inactivation of the lipid binding pocket of the Itk PH domain prevented the association of Itk with these raft-like structures and inhibited the transphosphorylation of Itk.
The association of Itk with Jurkat-derived raft-like structures was observed at all times, regardless of the activation state of the Jurkat cells. Given that Tec kinases do not typically interact constitutively with the plasma membrane, and that the lipid binding properties of the Tec kinase PH domains are quite similar, we propose that the constitutive association of Itk with plasma membrane-derived raft-like structures is a unique feature of Jurkat T cells (37). This idea is supported by our inability to detect Itk in raft-like structures derived from resting murine thymocytes (data not shown). Since Jurkat cells do not express PTEN, a phosphatidylinositol 3Ј-phosphatase, the basal association of Itk with raft-like structures in Jurkat T cells may result from constitutively elevated levels of PtdIns(3,4,5)P 3 in these cells. 3 Notably, the constitutive association of Itk with Lck-contain-ing raft-like structures did not result in constitutive Itk phosphorylation, demonstrating that the colocalization of Itk with Lck may be a prerequisite for Itk transphosphorylation, but is not sufficient for Itk transphosphorylation. Activation-induced changes in the conformation of Lck and Itk may regulate the transphosphorylation of Itk by Lck (83,84). These changes could result in the activation of Lck and could increase the accessibility of the Src kinase transphosphorylation site in the Itk kinase domain. In support of this hypothesis, we have shown that multivalent interactions between Itk, SLP-76 and Grb2 can modify the conformation of Itk, as indicated by the increased accessibility of the Itk PRR and SH3 domain. We expect that an interaction between the unphosphorylated Itk SH3 domain and SLP-76 will contribute to the generation of conformational changes in Itk upon its initial recruitment to the LAT⅐SLP-76 complex. As outlined above, we expect this to promote the transphosphorylation of Itk by Lck (see Fig. 9). Although Yablonski et al. (20) have shown that the TCR-induced phosphorylation of Itk can proceed in the absence of SLP-76, the phosphorylation of Itk at time points proximal to receptor cross-linking was not examined. It is possible that a more extensive kinetic analysis will reveal an influence of SLP-76 on the rate of Itk phosphorylation. Alternate biochemical interactions could contribute to the observed phosphorylation of Itk in the absence of SLP-76. For example, ZAP-70, which we have predicted will interact with the Itk SH2 domain, could provide a docking site capable of promoting the activation 3 L. Kane and A. Weiss, personal communication. of Itk by Lck. In support of this hypothesis, the phosphorylation of Itk has recently been observed to depend on the expression of ZAP-70 (48).
Transphosphorylated Tec kinases rapidly autophosphorylate a tyrosine residue which forms one of the binding pockets of the SH3 domain (53). This phosphorylation should inactivate the SH3 domain of Itk, inducing the dissociation of Itk from its SH3 domain ligands and stabilizing interactions mediated by the Itk PRR (49). In contrast, autophosphorylation should not influence the binding properties of the Itk PH domain or SH2 domain. Consequently, we expect activated, autophosphorylated Itk to engage substrate-containing complexes through PH domain-, PRR-, and SH2 domain-mediated interactions. Since the PH domain directs interactions with LAT-enriched light vesicles, the PRR directs interactions with Grb2 and LAT, and the SH2 domain binds SLP-76, it is likely that activated Itk interacts with the raft-associated, LAT-nucleated signaling complex containing SLP-76 (see Fig. 9). Recently Itk was observed to co-immunoprecipitate with LAT following TCR stimulation, supporting this hypothesis (48). Thus, in SLP-76-deficient Jurkat cells, a failure to recruit Itk to LAT may prevent the phosphorylation and activation of the raft-associated, LATbound pool of PLC␥1, resulting in the defective activation of calcium fluxes and MAPKs downstream of the TCR (Fig. 9). Heavy dark lines indicate protein-protein interactions, thin dark lines indicate flow through the cycle of activation, and grey lines indicate phosphorylation events. Itk is maintained in an inactive conformation by an intramolecular interaction which involves its PRR, SH3 domain, and SH2 domain and sequesters the PRR and SH3 domain (bottom). TCR cross-linking promotes the recruitment of Itk to PtdIns(3,4,5)P 3 -containing lipid rafts proximal to the receptor and enriched in Lck and LAT. This colocalization promotes multivalent interactions between Itk and the LAT signaling complex and therefore disrupts the intramolecular interaction between the Itk PRR and SH3 domain. The resulting conformational change enables the transphosphorylation of Itk by Lck (top left). Subsequently, activated Itk autophosphorylates the binding pocket of its SH3 domain, inactivating the SH3 domain and increasing the accessiblity of the Itk PRR. The PRR then cooperates with the PH domain and SH2 domain to stabilize Itk in LAT-nucleated complexes containing PLC␥1, an Itk substrate. This stabilization depends on the association of the Itk PH domain with PtdIns(3,4,5)P 3 -containing lipid rafts enriched in LAT, the association of the Itk PRR with Grb2 bound to LAT, and the association of the Itk SH2 domain with SLP-76 bound to LAT via GADS (top right). Because Itk engages the LAT complex through sites which may also be involved in the recruitment of Vav and Nck to the LAT complex (PtdIns(3,4,5)P 3 and Syk-phosphorylation sites in SLP-76), the overexpression of Itk may exclude either Vav or Nck from their sites of action. Finally, the dephosphorylation of Itk and PtdIns(3,4,5)P 3 allows the termination of signaling.
In this article we also report that the overexpression of Itk suppresses transcription from NF-AT response elements. We propose that this effect depends on the competitive exclusion of effectors from their docking sites in the raft-associated LAT⅐SLP-76 complex rather than the normal function of Itk downstream of SLP-76. In support of this proposal, the suppression of transcription is independent of Itk kinase activity. Furthermore, the suppression of transcription requires precisely the domains which we expect to direct interactions between activated Itk and the raft-associated LAT⅐SLP-76 complex: the PH domain, PRR, and SH2 domain.
The effectors antagonized by Itk have not been identified, but may include Nck and Vav. Nck, a SH2 and SH3 domaincontaining adaptor protein, engages SLP-76 via its SH2 domain and has been shown to collaborate with SLP-76 to promote the activation of NF-AT downstream of the TCR (85). Nck is capable of recruiting both PAK and WASP to SLP-76, and has been shown to promote actin polymerization downstream of SLP-76 (9,75). Thus, the Itk SH2 domain, which can bind to Nck SH2 domain docking sites in SLP-76, could compete directly with Nck for access to these sites, resulting in the inhibition of NF-AT activation. The Itk PH domain and PRR could enable the Itk SH2 domain to compete more effectively for access to SLP-76 by stabilizing Itk in the LAT⅐SLP-76 complex. Vav, an exchange factor for the Rac and Cdc42 GTPases, can also bind SLP-76 through its SH2 domain (8). However, Vav can cooperate with SLP-76 to activate NF-AT downstream of the TCR in the absence of this interaction (74,76). For this reason, it is unlikely that the overexpression of Itk antagonizes the effects of SLP-76 by directly interfering with an SH2 domain-mediated interaction between SLP-76 and Vav. Instead, the PRR-and SH2 domain-directed stabilization of the interaction between Itk and the raft-associated LAT⅐SLP-76 complex may enable the Itk PH domain to compete with the Vav PH domain for access to PtdIns(3,4,5)P 3 , an activator of Vav (86). Thus, the competitive exclusion of either Nck or Vav from the LAT⅐SLP-76 complex could explain the suppression of transcription from NF-AT response elements by Itk.
The data presented in this article are consistent with a model in which Tec kinases are regulated by a reversible intramolecular interaction that sequesters the homology domains, limiting the intermolecular interactions of Tec kinases. In this model Tec kinases are recruited to their sites of activation by mulitvalent interactions, autophosphorylated within their SH3 domain, and targeted to downstream effectors by their PH domain, PRR, and SH2 domain. Since many predicted ligands of the Tec kinase SH2 domains are substrates of Syk kinases, this model explains the placement of the Tec kinases downstream of the Syk kinases as a consequence of the Syk-dependent recruitment of Tec kinases to multimolecular complexes associated with Tec kinase activators, such as Lck, and containing Tec kinase substrates, such as PLC␥1. In T cells this complex appears to be nucleated by the ZAP-70 substrates LAT and SLP-76. A related complex, nucleated by CD19 and/or BLNK, may mediate similar functions in B cells.