Phosphatidylinositol 3-Kinase Is Required for CD28 But Not CD3 Regulation of the TEC Family Tyrosine Kinase EMT/ITK/TSK: Functional and Physical Interaction of EMT with Phosphatidylinositol 3-Kinase¹

Functional T cell activation requires at least two signals: one is delivered by the interaction of the Ag-specific TCR with peptide in the context of the MHC on APCs, and the second is delivered by costimulatory T cell surface molecules binding to their cognate receptors on APC (1, 2, 3). Engagement of the TCR without concurrent ligation of appropriate costimulatory molecules leads to T cell clonal anergy or programmed cell death through apoptosis (1, 2, 3). CD28, a T cell surface glycoprotein, is a potent costimulatory molecule for T cell activation (1, 2, 3, 4). Ligation of CD28 by its natural ligands, CD80 and CD86, on APC or by anti-CD28 Abs enhances T cell proliferation and cytokine production initiated via the TCR/CD3 complex (5, 6, 7, 8, 9). CD28 also plays a critical role in preventing the induction of T cell unresponsiveness and apoptosis (10, 11, 12, 13).

Ligation of CD28 induces an immediate and transient increase in tyrosine phosphorylation of specific intracellular substrates including CD28 itself (14, 15). Because CD28 does not possess intrinsic kinase activity (16), an intracellular nonreceptor tyrosine kinase(s) must be recruited to mediate this activity. Ligation of CD28 activates the SRC family tyrosine kinase members, LCK and FYN, and their kinase activities can be detected in CD28 immune complexes (17, 18). Following ligation of CD28, the EMT/ITK/TSK (EMT)³ TEC family protein tyrosine kinase is activated and recruited to CD28 in a LCK-dependent manner, placing LCK upstream of EMT in a tyrosine kinase cascade activated by CD28 (19, 20). Further, LCK can induce EMT phosphorylation following coexpression in Cos 7 cells (20), and LCK increases EMT kinase activity when coexpressed in bacculovirus-infected insect cells through phosphorylation of Tyr⁵¹¹ (21). The intracellular domain of CD28 contains four potential tyrosine phosphorylation sites (Y170, Y185, Y188, and Y197). At least in vitro, both EMT and LCK are capable of phosphorylating Tyr¹⁷⁰ in the YMNM phosphatidylinositol 3′ kinase (PI3K) and GRB2 consensus binding site in CD28, whereas EMT, but not LCK, appears competent to phosphorylate the distal three tyrosines of CD28 (22). Phosphorylation of Tyr¹⁷⁰ and recruitment of PI3K to CD28 is critical for optimal IL-2 production in some T cell lines but not others (23, 24, 25, 26, 27). In contrast, phosphorylation of the distal three tyrosines of CD28 and in particular Tyr¹⁸⁸ appears critical for optimal IL-2 production (27, 28, 29).

Knockout mice, lacking EMT, demonstrate decreased antiviral CTL immune responses and responses mediated by transgenic TCR, while maintaining normal B cell immune responses to viruses and mitogens (30, 31), indicating that EMT likely plays a positive role in some but not all TCR-mediated activation events. In contrast, EMT appears to negatively regulate at least some of the effects of CD28 costimulation, with EMT knockout mice being hyperresponsive to CD28 stimulation for cellular proliferation but not for IL-2 production (32). The mechanism(s) whereby EMT differentially regulates TCR and CD28 signaling events remains unclear, as ligation of either the TCR or CD28 induces similar LCK-dependent increases in EMT tyrosine phosphorylation and activation (20, 33). Indeed, concurrent ligation of CD3 and CD28 does not induce a significant alteration in the magnitude or kinetics of EMT activation (33). In contrast to the knockout mouse studies, transfection studies with EMT suggest that EMT plays a critical role in costimulation of the IL-2 promoter and probably also in IL-2 production (34).

Structurally, EMT, which is expressed primarily in T cells, NK cells, and mast cells, possesses pleckstrin homology (PH), TEC homology, SRC homology 2 (SH2), and SRC homology 3 (SH3) domains, which have the capacity to bind to specific protein motifs. Yeast two-hybrid, phage display, in vitro fusion protein, and coimmunoprecipitation studies have identified a number of potential ligands and mediators of EMT function. The PH domain of EMT has been demonstrated to bind to G protein β/γ subunits, a number of different protein kinase C isoforms, and inositol lipids (35, 36, 37). Further, the PH domain targets EMT to the cell membrane (37). Membrane localization may be needed to regulate EMT function in response to members of the SRC family of kinases (37). The proline-rich region in the TEC homology domain can bind to the GRB2 and the SH3 domain of SRC family members including FYN, LYN, and HCK (38, 39). Furthermore, fusion proteins containing the EMT SH3 domain associate with CD28, SAM68, Wiskott-Aldrich syndrome protein (WASP), hnRNP-K, FYN, and CBL (40, 41). However, many of these associations have been difficult to detect in intact T cells (34, 38, 40, 41), and their role in regulating EMT function or action remains to be determined.

PI3K is involved in CD28 signaling, as indicated by the accumulation of its D-3 phosphoinositide products and recruitment of PI3K to CD28 in CD28-stimulated cells (42, 43, 44, 45, 46). These PI3K products act as second messengers, eventually leading to the activation of some forms of protein kinase C (47, 48), p70 ribosomal S-6 kinase, PDK1, and AKT (48, 49). PI3K consists of a p110 catalytic subunit and a p85 regulatory subunit, which contains a proline-rich region, a SH3 domain, and two SH2 domains (50). The proline-rich region of the p85 subunit of PI3K binds to the SH3 domains of LCK and FYN resulting in the activation of PI3K (51, 52, 53, 54). The SH2 domain of PI3K binds to the tyrosyl-phosphorylated YMNM motif (Y170 in mouse and Y173 in human CD28) in the cytoplasmic domain of CD28, recruiting PI3K to CD28 (23, 27, 55, 56).

Since both EMT and PI3K are inducibly recruited and activated by CD28 cross-linking, we evaluated the functional and physical interaction between these molecules. Further, as LCK is required for EMT activation and recruitment, we assessed the role of LCK in these processes. In Jurkat T cells, CD28-induced but not CD3-induced stimulation of EMT in vitro autokinase activity is sensitive to the PI3K inhibitors LY294002 and wortmannin (57, 58). Coexpression of constitutively activated LCK Y505F with EMT in Cos 7 cells results in increases in EMT in vitro autokinase activity, a process that is partially dependent on PI3K activity, as indicated by treatment of the cells with the PI3K inhibitors. CD28 ligation but not CD3 ligation induces a LCK-dependent physical association of EMT with PI3K. This association appears to be mediated by the SH2 domain of EMT binding to tyrosine phosphorylated PI3K or a PI3K-associated linker molecule. Thus, ligation of CD28, but not CD3, regulates EMT by a mechanism at least partially dependent on PI3K. Further, the recruitment of EMT to PI3K following CD28 ligation potentially places EMT in the context of a different set of targets and regulatory molecules as compared with CD3 ligation, which does not induce association of EMT with PI3K.

The anti-human CD28 mAb (9.3, IgG2a) was a gift of Dr. J. Ledbetter (Bristol-Myers Squibb Research Institute, Seattle, WA). The mAb against the ε-chain of the CD3 complex (UCHT1, IgG1) was purified from the supernatants of hybridoma cells provided by Dr. Peter Beverly (University College, London, U.K.). Rabbit anti-mouse IgG was purchased from Western Blotting (Toronto, Ontario, Canada). The rabbit polyclonal Ab against PI3K p85 subunit and the anti-phosphotyrosine mAb (4G10, IgG1) as well as the mAb against EMT (clone 2F12, used for immunoblotting) were purchased from Upstate Biotechnology (Lake Placid, NY). A goat anti-EMT Ab (used for immunoblotting) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The production and specificity of the rabbit anti-EMT serum used for immunoprecipitation has been described previously (59). In some experiments, the polyclonal anti-EMT Abs were further affinity purified using bacterially expressed glutathione S-transferase fusion proteins containing EMT, and the purified Ab was linked to biotin. Anti-TrpE mAb was purchased from Oncogene (Cambridge, MA). Protein A conjugated with horseradish peroxidase (HRP) and [γ-³²P]ATP were purchased from Amersham (Arlington Heights, IL). Goat anti-mouse IgG Ab conjugated with HRP was purchased from BioRad (Hercules, CA). Protein A conjugated Sepharose 4B was purchased from Pharmacia Biotech (Piscataway, NJ). TrpE fusion proteins containing the SH2 domain (amino acid 242 to 342) of EMT were derived from PCR products and verified by sequencing. LY294002 was purchased form Calbiochem (La Jolla, CA), and wortmannin was purchased from Sigma (St. Louis, MO).

Full length wild-type EMT cDNA was cloned in pSG5. Constitutively activated LCK Y505F, a gift from Dr. Andre Veillette (McGill, Quebec, Canada), was cloned in pCDNA3.

The human leukemic T cell line Jurkat E6.1 and its LCK defective mutant J.CaM1.6 cells as well as Cos 7 cells were from American Type Culture Collection (Manassas, VA). HER cells are Rat1 fibroblasts engineered to overexpress the human epidermal growth factor receptor. Cells were cultured in RPMI 1640 (Life Technologies, Gaithersburg, MD) supplemented with 10% FBS (Sigma), 2 mM l-glutamine and gentamicin (10 μg/ml) (Life Technologies). Jurkat cells were serum starved for 18 h and preincubated with or without PI3K inhibitors LY294002 or wortmannin in the presence of 1% DMSO prior to stimulation as indicated. We have demonstrated that the incubation conditions and concentrations of LY294002 and wortmannin utilized were sufficient to inhibit anti-CD3 or anti-CD28 increases in phosphorylation of AKT (not shown), which provides a surrogate for inhibition of PI3K (49). Control cells were also incubated with equivalent amounts of DMSO (1%). CD28 or CD3 were cross-linked by incubating the cells in 1 ml volume with 5 μg of anti-CD28 or anti-CD3 mAb plus 10 μg of rabbit anti-mouse IgG. Cos 7 cells were transiently transfected with lipofectamine/DNA (Life Technologies) complex at 37°C for 2 h. Cells were cultured in the above medium for 24 h following transfection and then serum starved for another 24 h before harvesting.

Cells were lysed in 1 ml of 1% Nonidet P-40, 50 mM HEPES, pH 7.4, 150 mM NaCl, 50 μM ZnCl₂, 50 μM NaH₂PO₄, 50 μM NaF, 2 mM EDTA, 1 mM Na₃VO₄, 2 mM PMSF, and 10 μg/ml aprotinin. Postnuclear detergent lysate was precleared by protein A-Sepharose beads and then incubated with 5 μl of rabbit anti-EMT serum in the presence or absence of PI3K inhibitors for 90 min as indicated. The immune complexes were captured by protein A-Sepharose beads. Immunoprecipitates were washed with 0.5% Triton X-100, 0.5% Nonidet P-40, 10 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM Na₃VO₄, and 2 mM PMSF, and mixed in Laemmli sample buffer.

Proteins from total cell lysates or eluted from immunoprecipitates were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. The membranes were blocked with 3% BSA and then incubated for 2 h at room temperature with primary Ab: anti-phosphotyrosine mAb 4G10 (1 μg/ml), rabbit antiserum against p85 subunit of PI3K (1:1000 dilution), or anti-EMT mAb (1:1000 dilution). The membranes were washed in TBS-T (10 mM Tris/HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20). HRP-conjugated goat anti-mouse IgG (1:5000 dilution) and HRP-protein A (1:3000 dilution) were used as secondary reagents for a 1-h incubation at room temperature. After extensively washing, membranes were developed by enhanced chemiluminescence detection reagents (ECL; Amersham) and exposed to X-ray films.

EMT was immunoprecipitated from detergent cell lysate and washed with 1% Triton X-100, 50 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM Na₃VO₄, and 1 mM PMSF. The immune complex was incubated in 50 μl of kinase reaction buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 5 mM MnCl₂, 5 mM MgCl₂, 5 μM ATP, and 10 μCi [γ-³²P]ATP) at room temperature for 30 min. The reaction was stopped by washing the immune complex twice with the above wash buffer containing 1 mM EDTA. Proteins were eluted by Laemmli sample buffer, resolved by SDS-PAGE, and autoradiographed.

TrpE fusion protein containing the SH2 domain of EMT was coated to protein A-Sepharose beads by anti-TrpE mAb plus rabbit anti-mouse IgG. Cos 7 cells were starved overnight in serum-free medium and treated with 50 μM pervanadate for 30 min. Cells were harvested and lysed. Jurkat cells were serum starved, stimulated by CD28 cross-linking, and lysed as described above. Detergent cell lysates from Cos 7 or Jurkat cells were precleared with protein A-Sepharose and then incubated with the beads coated with fusion proteins at 4°C for 2 h. The beads were washed, and the proteins were eluted by Laemmli buffer, as described above, for immunoprecipitation. Proteins were resolved by SDS-PAGE, transferred to PVDF membranes, and immunoblotted by the Abs against p85 subunit of PI3K or against phosphotyrosine as described as above.

EMT and PI3K are activated and inducibly recruited to CD28 following CD28 cross-linking (19, 20, 42, 43, 44, 45, 46). Further, both EMT and PI3K are activated following ligation of the TCR complex (33, 48, 61). This suggests that PI3K may play a role in CD28- or CD3-induced increases in EMT in vitro autokinase activity. Human Jurkat T leukemia cells were chosen for the exploration of the relationship between EMT and PI3K due to their well-characterized responses to ligation of CD3 and CD28 and the well-characterized effects of ligation of CD28 or CD3 on EMT activation and phosphorylation (19, 20, 33).

As indicated in Fig. 1, in human Jurkat T cells, cross-linking of CD28 or CD3 induces a similar increase in EMT immune complex kinase activity. Pretreatment of Jurkat cells with the relatively specific PI3K inhibitors LY294002 (IC₅₀, 1.4 μM) (57) and wortmannin (IC₅₀, 5 nM) (58) decreased CD28-induced EMT in vitro autokinase activity. Indeed, 10 μM of LY294002 or 100 nM of wortmannin induced a 70% inhibition of EMT in vitro autokinase activity (Fig. 1, A and C). This effect was selective for CD28-induced increases in EMT in vitro autokinase activity, as preincubation with LY294002 or wortmannin did not alter CD3-induced increases in EMT in vitro autokinase activity (Fig. 1, A and D). This latter observation also indicates that the effects of LY294002 or wortmannin are not due to nonspecific toxicity but rather are due to specific effects of these inhibitors on the pathway(s) leading to CD28-induced increases in EMT activity. Further, this demonstrates that EMT is regulated by PI3K-dependent (CD28) and PI3K-independent (CD3) mechanisms.

Ligation of both CD28 and CD3 results in a transient LCK-dependent increase in EMT tyrosine phosphorylation, which correlates with increases in EMT tyrosine kinase activity (20, 33). Strikingly, as compared with the effects on EMT autokinase activity (Fig. 1A), CD28-induced increases in EMT tyrosine phosphorylation were much more resistant to the effects of LY294002 and wortmannin (Fig. 1, B–D). Indeed, treatment with 10 μM of LY294002 or 100 nM of wortmannin virtually eliminated CD28-induced EMT in vitro autokinase activity without significantly altering CD28-induced EMT tyrosine phosphorylation. These data further suggest that phosphorylation of EMT, at least as induced by ligation of CD28, is not sufficient for EMT in vitro autokinase activity. The discordance between the effect of inhibition of PI3K kinase activity on CD28- and CD3-induced EMT in vitro autokinase activity (Fig. 1, A, C, and D) and EMT phosphorylation (Fig. 1, B–D) suggests that CD28- and CD3-mediated EMT phosphorylation could be on different tyrosines.

Functional LCK is required for both anti-CD28- and anti-CD3-induced activation of EMT, placing EMT downstream of LCK (20, 33). As indicated in Fig. 1, CD28-induced EMT activation is dependent on intact PI3K activity. To determine whether the effects of LCK on EMT activity also required intact PI3K activity, EMT was transiently transfected with or without constitutively activated LCK (LCK Y505F; LCK containing a tyrosine to phenylalanine mutation at the negative regulatory site in the carboxyl terminus of LCK) (60, 61), into Cos 7 cells. Cos 7 cells do not express detectable levels of either EMT or LCK, providing an excellent recipient for transient transfection studies. Coexpression of activated LCK and EMT resulted in an ∼10-fold increase in EMT in vitro autokinase activity as assessed by EMT autophosphorylation in in vitro kinase assays (Fig. 2, A and C). The kinase activity in EMT immunoprecipitates is probably not due to coimmunoprecipitation of LCK, as we and others have been unable to demonstrate LCK in anti-EMT immunoprecipitates from intact cells (19, 37, 38, 40), and LCK protein (Western blotting) or autophosphorylating activity (kinase assay) was not detected in the EMT immunoprecipitates from cotransfected Cos 7 cells, despite being readily detected in anti-LCK immunoprecipitates (not shown).

As indicated in Fig. 2, LY294002 induced a marked decrease in the ability of activated LCK to increase EMT in vitro autokinase activity. In contrast, there was no effect of either LY294002 or wortmannin on LCK immune complex autokinase activity. When corrected for EMT protein levels (Fig. 2 C), preincubation with LY294002 induced an ∼50% decrease in LCK-induced increases in the specific activity of EMT. It is important to note, however, that even following a 4-h pretreatment with 10 μM of LY294002 (which is sufficient to completely inhibit PI3K; data not presented) (57, 58), activated LCK induced a fourfold increase in EMT in vitro autokinase activity (Fig. 2C). This indicates, once again, that EMT is regulated by PI3K-dependent and -independent mechanisms and that LCK appears to be able to regulate EMT by both PI3K-dependent and -independent mechanisms.

Interestingly, as in CD28- or CD3-induced increases in EMT tyrosine phosphorylation in Jurkat cells, coexpression of activated LCK with EMT in Cos 7 cells induced an increase in tyrosine phosphorylation of EMT. This increase in EMT tyrosine phosphorylation was resistant to a 4-h treatment with 10 μM of LY294002 (Fig. 2, B and D), suggesting that LCK-induced tyrosine phosphorylation of EMT is not sufficient for optimal EMT autokinase activity.

Based on the observation that both PI3K and EMT associate with CD28 following the ligation of CD28 (19, 20, 42, 43, 44, 45, 46) and on the involvement of PI3K in CD28- but not CD3-induced EMT activation (Fig. 1), we evaluated whether cross-linking of CD28 or CD3 would induce a physical association of EMT with PI3K. CD28-induced association of EMT with PI3K was assessed by immunoprecipitation of EMT followed by immunoblotting for the p85 subunit of PI3K. To prevent precipitation of CD28 or CD28-associated PI3K by the anti-CD28 Abs used to activate CD28, the EMT Ab was biotin-conjugated and precipitated with avidin-Sepharose beads (beads alone were used as the control, and anti-CD28 Abs were added to the control cells postlysis to ensure that equal amounts of anti-CD28 Abs were present in each immunoprecipitate). Following CD28 cross-linking, the amount of EMT associated with PI3K underwent a time-dependent increase, peaking at 30 min (Fig. 3). The recruitment of PI3K to CD28 was more rapid, peaking at 5 min following cross-linking (Fig. 3). Furthermore, the time course of recruitment of EMT to PI3K was markedly different from that of recruitment of PI3K to CD28 (Fig. 3A). The amount of PI3K associated with EMT represents a small portion of the PI3K present in the cell (<1%; not shown). However, as indicated in Fig. 3A, the amount of PI3K associated with EMT is ∼50% of the amount of PI3K associated with CD28.

Strikingly, CD3 cross-linking or PHA stimulation did not induce a detectable increase in the association of EMT with PI3K (Fig. 3B). Thus, similar to the dependence of CD28-induced EMT activation on functional PI3K, CD28 but not CD3 cross-linking induces a physical association of EMT with PI3K.

LCK is required for optimal CD28-induced activation and phosphorylation of EMT (20). Cross-linking of CD28 in J.CaM1.6 cells, which lack functional LCK (62), failed to induce the association of EMT with PI3K (Fig. 4), suggesting that tyrosine phosphorylation initiated by LCK is involved in the association of EMT with PI3K. Strikingly, cross-linking of CD28 in J.CaM1.6 cells induced detectable association of PI3K with CD28 (cf Figs. 3 and 4). This indicates that CD28 is at least partially functional in J.CaM1.6 cells and that functional LCK is not required for the recruitment of PI3K to CD28. Thus, LCK is required for CD28-induced increases in EMT phosphorylation, increases in EMT in vitro autokinase activity, and induced association of EMT with PI3K. Furthermore, the mechanisms regulating recruitment of PI3K to CD28, probably phosphorylation of Y170 (23, 27, 55, 56), must be different from those regulating the recruitment of EMT to PI3K and to the activation of EMT by CD28.

Since SH2 domains bind to tyrosine-phosphorylated protein domains and the association of EMT with PI3K is most likely dependent on tyrosine phosphorylation, we assessed whether the SH2 domains of EMT or PI3K were sufficient for formation of a PI3K:EMT complex. EMT does not contain the YmxM consensus binding motif for PI3K SH2 domains (50, 59). We therefore first assessed whether the SH2 domain of EMT was capable of mediating the association of EMT with PI3K. Cos 7 cells, which do not express CD28 or EMT, were used as the PI3K donor. As noted in Fig. 5A, a trpE-SH2 fusion protein containing the SH2 domain of EMT was able to bind PI3K or a PI3K-associated protein from pervanadate-treated but not resting Cos 7 cells, as assessed by Western blotting with anti-p85 Abs. Further, as indicated in Fig. 5A, the PI3K that associated with the EMT SH2-trpE fusion protein was tyrosine phosphorylated, as demonstrated by Western blotting with anti-phosphotyrosine Abs. Indeed, the degree of association of PI3K with EMT correlated with the degree of tyrosine phosphorylation of PI3K. Once again, this suggests that EMT can bind PI3K or a PI3K-associated protein in a tyrosine phosphorylation-dependent manner.

We also assessed whether the SH2 domain of EMT could bind to PI3K derived from resting or CD28-activated Jurkat cells. The interpretation of this approach is complicated by the presence of CD28 in Jurkat T cells, which could act as an intermediary in the formation of a ternary complex. Nevertheless, as indicated in Fig. 5B, the SH2 domain of EMT was able to form a complex with PI3K from lysates of CD28-treated Jurkat T cells as compared with resting Jurkat T cells.

As expected, since EMT does not contain a PI3K SH2 consensus-binding motif and tyrosine phosphorylation of the PI3K donor cell appears to be required for the association of the SH2 domain of EMT with PI3K, we were unable to detect an association of EMT from resting or CD28-stimulated T cells with GST fusion proteins containing the two SH2 domains of the p85 PI3K subunit, as assessed by Far Western or protein overlay blotting (data not presented). The SH2 domains of PI3K readily bound to a number of proteins from epidermal growth factor- or pervanadate-treated HER cells, demonstrating the validity of the approach.

Taken together, the data suggest that the SH2 domain of EMT can associate with tyrosine-phosphorylated PI3K. However, the possibility of indirect association between EMT and PI3K through other adapter molecule(s) cannot be excluded. CD28 is not an obligatory linker molecule for the formation of this complex. Whether CD28 plays a role in the formation of a ternary complex in T cells remains to be determined.

Ligation of both CD28 and CD3 induces tyrosine phosphorylation and activation of EMT (19, 33). We demonstrate herein that CD28-induced stimulation of EMT tyrosine autokinase activity is due, in part, to activation of PI3K. Strikingly, anti-CD3-induced activation of EMT autokinase activity does not require functional PI3K, indicating that EMT enzyme activity is regulated by PI3K-dependent and -independent mechanisms. Further, LCK-induced increases in EMT in vitro autokinase activity also demonstrate P13K-dependent and -independent components (Fig. 2). The PI3K-independent mechanism of EMT in vitro autokinase activation probably involves tyrosine phosphorylation of EMT mediated by LCK, as stimulation of EMT kinase activity is dependent on the presence of LCK (21, 33). Indeed, Berg and colleagues have recently demonstrated that LCK induces tyrosine phosphorylation of EMT on Tyr⁵¹¹, a process that is required for the optimal activation of EMT (21). Even following overexpression of EMT and LCK in bacculovirus infected insect cells, Berg and colleagues were unable to demonstrate significant phosphorylation of other tyrosines in EMT, indicating that Tyr⁵¹¹ is likely to be the major target of LCK-mediated phosphorylation (21). Phosphorylation of a serine, threonine, or tyrosine residue in the activation loop of kinases appears to be a common mechanism for regulation of kinase activity (63).

The PI3K-dependent mechanism of EMT activation is detectable following immunoprecipitation of EMT, suggesting that PI3K leads to a posttranslational alteration in EMT or an associated molecule, or alternatively, that the products of PI3K remain bound to EMT under the immunoprecipitation conditions utilized. August et al. (37) have demonstrated that the PH domain of EMT, similar to several other PH domains (49), has the ability to bind products of PI3K. This may target EMT to the cell membrane, a location potentially required for its phosphorylation by SRC family kinases (37), or alternatively, may directly regulate EMT in vitro autokinase activity as it does other PH-containing kinases such as AKT (49).

August et al. have demonstrated that coexpression of SRC and EMT in Cos 7 cells results in activation of EMT, a process that is likely to be dependent, in part, on PI3K, as it was inhibited by LY294002 and wortmannin, albeit at very high concentrations (37). The relevance of this observation to activation of EMT in T lymphocytes is uncertain, as this same group had previously demonstrated that similar concentrations of LY294002 and wortmannin as used in Cos 7 cells augmented rather than inhibited CD28-induced activation of EMT in Jurkat T cells (22). The studies presented herein serve to clarify the role of PI3K in EMT activation by demonstrating that LCK activates EMT, at least partly, in a PI3K-dependent manner. Further, in Jurkat T cells, inhibition of PI3K activity by LY294002 or wortmannin at concentrations likely to be relatively specific for PI3K (56, 57) induced a significant decrease in CD28-mediated activation of EMT. The reason for the discrepancy between the results reported herein and those of Dupont and colleagues (22) is currently unknown. However, the results observed herein cannot simply be due to the toxicity of the inhibitors, as they failed to alter anti-CD3-induced activation of EMT.

Consistent with the observation of a role for PI3K in the activation of EMT by ligation of CD28 and not ligation of CD3, EMT associates with PI3K following ligation of CD28 but not CD3. The SH2 domain of EMT is competent to bind to PI3K in a tyrosine phosphorylation-dependent manner. Whether the SH2 domain of EMT binds to tyrosine-phosphorylated PI3K p85 or p110 directly or to a phosphorylated PI3K-associated protein has not yet been determined. PI3K has been demonstrated to bind to a number of proteins including CBL, VAV, and FYN among others (19, 51, 52, 53, 54, 64), any of which may mediate binding to EMT. Indeed, EMT has been suggested in surrogate model systems to bind to VAV, CBL, and FYN (19, 21, 38, 39, 40). Further, although EMT can interact with PI3K in the absence of CD28, both EMT and PI3K are recruited to CD28 following cross-linking of CD28, suggesting that CD28 may provide a framework for the recruitment of both PI3K and CD28.

The association of EMT with PI3K may be important in placing EMT in the context of upstream or downstream regulatory molecules or targets. If this is the case, the failure of ligation of CD3 to induce a PI3K:EMT complex is compatible with EMT playing differential roles downstream of CD28 and CD3. Genetic “knockout” experiments, which suggest that EMT plays a positive role in some TCR-mediated signaling events (30, 31) and a negative role in some CD28-mediated signaling events (32), support this contention.

As with the activation and phosphorylation of EMT (19, 20, 21, 33), LCK is required for the recruitment of EMT to PI3K. Surprisingly, although LCK is required for the recruitment of EMT to PI3K (Fig. 4) and of EMT to CD28 (20) as well as for the activation of EMT (20, 33), LCK is not required for recruitment of PI3K to CD28 (Fig. 4). Thus, phosphorylation of Y170, which recruits PI3K to CD28, does not require functional LCK or activation of EMT; this despite the observation that, at least in vitro, both EMT and LCK are capable of phosphorylating Y170 (22).

Hirano and colleagues have demonstrated that the SH2 domain of TEC can bind to the p85 PI3K subunit as well to a 55-kDa PI3K-associated molecule in a process dependent on the kinase activity of TEC (65) in yeast two-hybrid systems and in surrogate model systems. Further, the association of TEC with PI3K, similar to the association of EMT with PI3K, is increased by activation of cells by ligands that increase intracellular tyrosine phosphorylation, in this case, IL-3 and IL-6 (65). This finding suggests that TEC family kinases, at least TEC and EMT, and potentially other members of the family such as BTK, BMX, TXK, and DSRC28 (66), interact with PI3K. Determining whether PI3K plays a role in the regulation of TEC family members other than EMT will require additional investigation.

In summary, EMT activity is regulated through at least two different mechanisms, one of which requires PI3K. PI3K activity is required for activation of EMT following CD28 ligation but not CD3 ligation. Consistent with the requirement of PI3K activity for CD28- but not CD3-induced activation of EMT, a small but significant portion of cellular EMT associates with PI3K following CD28 cross-linking but not following CD3 stimulation. This association is most likely mediated by the binding of the SH2 domain of EMT to tyrosine-phosphorylated PI3K or a PI3K-associated protein. CD28-induced association of EMT with PI3K requires functional expression of LCK as does CD28-induced EMT activation and phosphorylation. Further, LCK-induced EMT activation is dependent, at least in part, on PI3K activity. Thus ligation of CD28 and CD3 regulates EMT through different mechanisms and recruits EMT to different signaling complexes, perhaps contributing to differences in the role of EMT downstream of CD28 and the TCR.