Tyrosine Phosphorylation of the CD3-ε Subunit of the T Cell Antigen Receptor Mediates Enhanced Association with Phosphatidylinositol 3-Kinase in Jurkat T Cells*

T cell receptor signaling results both in T cell proliferation and apoptosis. A key enzyme at the intersection of these downstream pathways is phosphatidylinositol 3′-kinase (PI 3-kinase). In a previous report, we showed that the p85α subunit of the PI 3-kinase preferentially associated with the CD3-ζ membrane-proximal immunoreceptor tyrosine-based activation motif of the ζ chain (ζA-ITAM) (Exley, M., Varticovski, L., Peter, M., Sancho, J., and Terhorst, C. (1994) J. Biol. Chem.269, 15140–15146). Here, we demonstrate that tyrosine phosphorylation of CD3-ε can recruit the PI 3-kinase enzyme in a T cell activation-dependent manner. In vivostudies with Jurkat cells stably transfected with a CD8-CD3-ε chimera (termed CD8-ε) shows that ligation of endogenous CD3-ε or CD8-ε by specific antibodies induces tyrosine phosphorylation of CD3-ε or CD8-ε, respectively. Increased tyrosine phosphorylation correlates with increased binding of p85α PI 3-kinase and recruitment of PI 3-kinase enzymatic activity to CD3-ε or CD8-ε proteins. Mutagenesis studies in COS-7 cells, transiently transfected with CD8-ε, p85α, and Fyn cDNAs in various combinations, show that both Tyr170 and Tyr181 within the CD3-ε-ITAM are required for efficient binding of p85α PI 3-kinase. Thus, replacement of Tyr170 by Phe (Y170F), or Tyr181 by Phe (Y181F) significantly reduces binding of p85α PI 3-kinase, whereas it does not affect binding of Fyn. Further in vitroexperiments suggest that a direct binding of the tandem SH2 domains of p85α PI 3-kinase to the two phosphorylated tyrosines in a single CD3-ε-ITAM may occur. The data also support a model in which a single CD3 subunit can recruit distinct effector molecules by means of TCR-mediated differential ITAM phosphorylation.

The individual cytoplasmic tails of the CD3 complex are necessary and sufficient for transduction of activating stimuli. The signal transduction function of the CD3 complex has been mapped to small sequence motifs, originally defined by Reth (15). This motif, termed the immunoreceptor tyrosine-based activation motif (ITAM), contains consensus sequence around two characteristic tyrosine residues spaced by 10 or 11 amino acids (YXX(L/I)X 7-8 YXX(L/I)). There are three ITAMs within the chain, A, B, and C, and one ITAM in each CD3 chain, ⑀, ␦, and ␥. The presence of distinct motifs in individual cells offers the potential for regulation of activation pathways with different specificities and consequences (16). Motifs from different receptor subunits have partially distinct binding activity, consistent with their activation of distinct sets of effectors. The distinct selectivity of ITAMs for effector binding indicates that they may drive at least distinct biological responses (17).
The ITAMs contain structural information necessary for signal transduction, and they are rapidly tyrosine-phosphorylated on both tyrosines following receptor stimulation (7,8,18). The ability of phosphorylated CD3 cytoplasmic tails to transduce mitogenic signals is based on specific recruitment of signaling molecules. One candidate for signal transducing molecule is PI 3-kinase. This enzyme phosphorylates the D-3 position of the inositol ring of PI, PI-4-P, and PI-4,5-P 2 (19). PI 3-kinase is a heterodimer consisting of a regulatory subunit (p85) and a catalytic subunit (p110) (20,21). PI 3-kinase regulatory subunit p85 has an N-terminal Src homology 3 (SH3) region, two SH2 domains, and a domain with significant sequence homology to the product of the breakpoint cluster region gene (21)(22)(23). Although the downstream effectors of the phosphoinositide 3-phosphates generated by the enzyme are not well characterized, it has been shown that they activate protein kinase C, a non-calcium/diacylglycerol-regulated isoform of protein kinase C, and the RAC (related to the A and C kinases) serine threonine kinase (24). In addition, PI 3-kinase products have been implicated in regulation of multiple cellular events, such as vesicular trafficking, cytoskeletal rearrangements, and mitogenesis. T cell activation results in association of PI 3-kinase with the TCR and an increase in levels of 3Ј-phosphorylated polyphosphoinositides (1,25).
In this study, we report association of p85␣ PI 3-kinase with the CD3-⑀, which is dependent on phosphorylation of the two tyrosine residues of the CD3-⑀-ITAM. The data support a model of TCR-mediated activation in which the extent of CD3-⑀ phosphorylation may serve to recruit distinct SH2 domain-containing signaling proteins.

Materials
Cells-Jurkat human T cells (ATCC) were used. Cells were grown in RPMI 1640 (Life Technologies, Inc.) with 10% FCS (BioWhittaker), penicillin, streptomycin, and 2 mM glutamine. Jurkat T cells were stably transfected by electroporation with the cDNA coding the CD8-⑀ chimera, and clones were selected by growth in media with neomycin. Clones were screened for surface expression of the chimera by staining with anti-CD8 (OKT8) antibody and fluorescein-conjugated goat antimouse antibody and analyzed by flow cytometry. COS-7 monkey kidney epithelial cells were cultured in Dulbecco's modified Eagle's medium with 10% FCS, penicillin, streptomycin, glutamine, and sodium pyruvate minimal essential medium (Life Technologies, Inc.).
Peptide Synthesis Procedure-The peptides were synthesized by the solid phase procedure using Fmoc (N-(9-fluorenyl)methoxycarbonyl) amino acids (30, 31) on a Ranin Symphony peptide synthesizer. The initial residue was linked to a p-alkoxybenzyl alcohol resin providing a C-terminal carboxyl group. The peptides were deblocked and cleaved from the resin by a mixture of trifluoroacetic acid/thioanisole/ethanedithiol/anisol (9:0.5:0.3:0.2) to minimize side reactions. The individual peptides were precipitated from the trifluoroacetic acid cleavage mixture with diethylether and washed several times with cold ether to remove impurities generated by the cleavage reaction. Peptides were purified by high pressure liquid chromatography.
Preparation of Affinity Matrices-The synthetic peptides were chemically coupled via N-terminal Cys with maleimidobenzoyl-N-hydroxysulfosuccinimide ester (Pierce) to L-Lys-Sepharose (Sigma) according to the manufacturer's instructions and blocked with 1% bovine serum albumin. Coupling efficiency was monitored by determining the degree of binding of anti-peptide and anti-phosphotyrosine antibodies to the peptide affinity matrices as described (1).
Jurkat CD8-⑀ cells were serum starved by incubating for 4 h at 37°C in RPMI 1640 with 0.1% FCS. Cells were washed and resuspended at 10 7 cells/100 l in serum-free RPMI 1640 containing 20 mM HEPES, pH 7.4, and incubated at 37°C for 10 min before activation. Jurkat cells were then incubated with anti-receptor mAb (5 g/10 7 cells) at 37°C for 5 min. In indicated experiments, secondary antibody was used to further cross-link the receptors, e.g. goat anti-mouse IgG (whole molecule) or the F(abЈ) 2 fragments (Cappel). The concentration of the secondary antibody used for activation was 20 g/10 7 cells. In the latter case, cells were stimulated with the anti-receptor mAb for 2 min at 37°C followed by the addition of secondary antibody for an additional 3 min at 37°C. Lysis was carried out as described above. Cell lysates were clarified by centrifugation at 14,000 ϫ g for 15 min at 4°C. For immunoprecipitations with mAbs OKT3 and OKT8, 5 g/10 7 cells of Ab were bound to 20 l of protein A-Sepharose CL-4B (Pharmacia Biotech Inc.) for 1 h at 4°C after one preclearing with bovine serum albumin-Sepharose 1 h at 4°C. Washes before loading were carried out as described previously. COS-7 cells were lysed and subjected to immunoprecipitation with 5 g/10 7 cells of mAb OKT8 and 20 l of protein A-Sepharose CL-4B (Pharmacia) for 1 h at 4°C after one preclearing with normal mouse serum (Sigma) and protein A-Sepharose CL-4B for 1 h at 4°C.
DNA Transfections-COS-7 cells (10 ϫ 10 6 ) were transfected with 2 g of expression vectors, containing the appropriate cDNA insert, by the DEAE-Dextran method (32). Cells were harvested 72 h after transfection and lysed in a lysis buffer as described above. Aliquots of these lysates were subjected to immunoblotting with Abs to Fyn, p85␣ PI 3-kinase, and CD3-⑀ to confirm protein expression. In transfected cells, surface expression of the CD8-⑀ chimera was analyzed by flow cytometry with OKT8 antibody. Western Blotting-Proteins were separated on SDS-PAGE and transferred to PVDF Immobilon membrane (Millipore Corp.). Filters were probed with specific antibodies as described previously (8). For detection, anti-mouse IgG (HϩL) or anti-rabbit IgG (Fc) horseradish peroxidase conjugates (Promega) were used followed by chemiluminescence (ECL, Amersham Corp.). For anti-Tyr(P) Western blot, membranes were incubated with anti-phosphotyrosine directly coupled to horseradish peroxidase followed by chemiluminescence (ECL system, Amersham Corp.). For stripping and reprobing, membranes were incubated for 10 min in stripping buffer (150 mM NaCl, 10 mM Tris-HCl, pH 2.3) at room temperature. Membranes were then washed for 3 ϫ 10 min in TNA-Tween and reblocked.
PI 3-Kinase Assay-The PI 3-kinase assay was performed as described (33). Briefly, whole cell lysates were prepared as above and incubated with anti-phosphotyrosine antibody, with OKT3 or with OKT8 and protein A-Sepharose. PI 3-kinase activity immobilized on beads was measured using PI and phosphatidylinositol biphosphate substrates. Reaction products were extracted and subjected to thin layer chromatography. PI 3-kinase activity was quantified by scintillation counting of 32 P incorporated into phosphatidylinositol 3,4,5-phosphate. Results were expressed as net counts/min following background subtraction.

RESULTS
In Jurkat and Jurkat CD8-⑀ cells, stimulation of the TCR⅐CD3 or CD8-⑀ receptors induces the association of p85␣ PI 3-kinase with the phosphorylated forms of CD3-⑀ and CD8-⑀.
We have previously shown that in vitro PI 3-kinase preferentially associates with the membrane-proximal ITAM of the chain (A-ITAM). Maximal PI 3-kinase activity is recovered following TCR⅐CD3 activation and is dependent upon phosphorylation of both tyrosine residues of the A-ITAM (1). A chimera composed of the extracellular domain of CD8 and the cytoplasmic domain of CD3-⑀ (termed CD8-⑀) was constructed to determine in vivo whether the intracellular domain of CD3-⑀ chain was capable of associating with the p85 PI 3-kinase independently of the presence of the other TCR⅐CD3 subunits. To this end, the CD8-⑀ chimera was stably transfected in Jurkat cells expressing the wild-type TCR⅐CD3 complex (hereafter termed Jurkat CD8-⑀ cells or JCD8-⑀ cells) and stimulated by the addition of the anti-CD8 mAb OKT8 alone or further crosslinked with a secondary antibody against mouse immunoglobulins (G␣mIg). In these cells, the overall pattern of OKT8-induced substrate tyrosine phosphorylation was similar, although somewhat weaker, than that induced after TCR stimulation with the anti-CD3 mAb OKT3 (data not shown). Regarding tyrosine phosphorylation of the CD8-⑀ chimera, in unstimulated cells or cells stimulated with a goat anti-mouse IgG antibody alone, a basal level of tyrosine phosphorylation of CD8-⑀ was observed (Fig. 1A, lanes 5 and 6), which was higher than the basal level of tyrosine phosphorylation of endogenous CD3-⑀ (Fig. 1A, lanes 1 and 2). However, the level of CD8-⑀ tyrosine phosphorylation induced upon stimulation of the chimera with anti-CD8 antibodies was similar to that observed in endogenous CD3-⑀ upon stimulation with the anti-CD3 mAb OKT3 (Fig. 1A, compare lanes 7 and 8 with lanes 3 and 4). Despite that, more tyrosine-phosphorylated proteins seemed to co-immunoprecipitate with CD3-⑀ than with CD8-⑀ (Fig. 1A, compare lane 4 with lane 8). Therefore, evidence clearly demonstrates that both the CD8-⑀ chimera and endogenous CD3-⑀ result in tyrosine phosphorylation in response to receptor ligation.
To test in these cells whether an interaction between CD8-⑀ and p85 PI 3-kinase may occur in vivo, we reblotted the filter with an anti-p85␣ PI 3-kinase antibody. As shown in Fig. 1B, p85 PI 3-kinase was weakly detected in OKT8 immunoprecipitates from unstimulated or G␣mIg-stimulated Jurkat CD8-⑀ cells (Fig. 1B, lanes 5 and 6). This association was highly increased in OKT8 immunoprecipitates from OKT8-or OKT8 plus G␣mIg-stimulated cells (Fig. 1B, lanes 7 and 8). The basal level of association of p85 PI 3-kinase to CD8-⑀ in nonstimulated and secondary antibody-stimulated cells (Fig. 1B, lanes 5 and 6) could be due to the higher basal level of tyrosine phosphorylation of CD8-⑀ as compared with that of endogenous CD3-⑀ (Fig. 1A, compare lanes 5 and 6 with lanes 1 and 2). Thus, p85␣ PI 3-kinase was not detected in OKT3 immunoprecipitates from unstimulated Jurkat CD8-⑀ cells or from cells exposed to secondary antibody alone (Fig. 1B, lanes 1 and 2), whereas p85␣ was readily co-immunoprecipitated with endogenous CD3-⑀ when Jurkat CD8-⑀ cells were stimulated with OKT3 or OKT3 plus the secondary antibody p85␣ (Fig. 1B,  lanes 3 and 4). Altogether, these results suggest a tyrosine phosphorylation-dependent mechanism for p85 PI 3-kinase binding to the intracellular domain of CD3-⑀, in which none of the other CD3 chains seems to play a role.
In Jurkat CD8-⑀ Cells PI 3-Kinase Associates with CD3 and CD8-⑀ in Response to Receptor Engagement-To test whether PI 3-kinase enzymatic activity was detected in association with the cytoplasmic tail of CD3-⑀, we performed kinase reactions on OKT8 or OKT3 immunoprecipitates from Jurkat CD8-⑀ cells stimulated with anti-CD8 mAb OKT8 or with anti-CD3 mAb OKT3, respectively (Fig. 2). In agreement with the p85␣ immunoblot studies, an increase in PI 3-kinase activity associated with the chimera CD8-⑀ or with endogenous CD3-⑀ was detected upon engagement of these receptors with their specific antibodies (Fig. 2, first and third columns). This receptor-associated PI 3-kinase activity further increased by extensive crosslinking of OKT8 or OKT3 with G␣mIg (Fig. 2, second and fourth columns). Since for PI 3-kinase enzymatic activity the association of both p85␣ and p110 subunits is required, these results suggest that in JCD8-⑀ cells, CD8-⑀ or TCR⅐CD3 stimulation leads to the binding of a p85␣/p110 heterodimer to the tyrosine-phosphorylated cytoplasmic tail of CD3-⑀.
Co-transfection of Fyn and CD8-⑀ Chimera Can Reconstitute CD8-⑀ Tyrosine Phosphorylation in a Nonhematopoietic Cell Line-Taken together, the above results are consistent with a  5). Cells were lysed with 1% Nonidet P-40 lysis buffer. Cell lysates (WL, lane 11) were precleared with bovine serum albumin-Sepharose (PC, lanes 9 and 10) and immunoprecipitated with OKT3 5 g/10 7 cells (lanes 1-4) or OKT8 5 g/10 7 cells (lanes 5-8). Proteins were analyzed by 12.5% SDS-PAGE, transferred to PVDF membranes, and probed with anti-phosphotyrosine RC20 (A, ␣Tyr(P)) anti-p85␣ PI 3-kinase pAb (B), or anti-CD3-⑀ pAb (C). Bands were visualized by the ECL system. mechanism for p85 PI3-kinase binding in which Lck (or Fyn) directly phosphorylates the CD8-⑀ chimera and subsequently p85␣ binds direct or indirectly to the chimera. However, Jurkat T cells contain other hematopoietic-specific tyrosine kinases (such as ZAP-70, etc.) and other signaling molecules, which confound the drawing of definitive conclusions from the above experiments. A previously described reconstitution system was used (28) to determine whether Fyn was capable of directly tyrosine phosphorylating the CD8-⑀ chimera. In this system, COS-7 cells were transiently transfected with mouse Fyn, or mouse p85␣ PI 3-kinase, or CD8-⑀ cDNAs alone or in various combinations. The CD8-⑀ chimera was isolated by immunoprecipitation with the anti-CD8 mAb OKT8 and analyzed by SDS-PAGE and Western blotting with anti-phosphotyrosine mAb (Fig. 3A). Since COS-7 cells lack all known hematopoietic specific tyrosine kinases, only very weak CD8-⑀ tyrosine phosphorylation occurred without exogenous kinase expression in COS-7 cells (Fig. 3A, lane 5). When the Fyn kinase was cotransfected with CD8-⑀ in these cells, a marked increase in the level of CD8-⑀ tyrosine phosphorylation occurred (Fig. 3A, lane  6). Moreover, co-expression of these two proteins resulted in the co-immunoprecipitation of a 59-kDa tyrosine phosphoprotein with CD8-⑀ that was identified as Fyn (Fig. 3, panel A and panel B, lane 6). In contrast, in cells transfected with Fyn cDNA alone, the OKT8 antibody did not co-immunoprecipitate any 59-kDa band (Fig. 3B, lane 3), suggesting that the interaction of phosphorylated CD8-⑀ with Fyn was specific. These results indicate that, without the ZAP-70 PTK, Fyn is able to directly tyrosine-phosphorylate the cytoplasmic tail of the CD8-⑀ chimera, resulting in a tight association of this kinase with the CD3-⑀-ITAM.
Fyn can efficiently bind to the CD8-⑀ chimera (Fig. 3B, lane  6), and Src homology 3 domain of Fyn can mediate binding to PI 3-kinase in T cells (34). Therefore, the increased association of p85␣ to the CD8-⑀ chimera observed in cells that were cotransfected with Fyn (Fig. 3C, lane 8) suggested that either Fyn was the linker between CD8-⑀ and p85 PI 3-kinase or that tyrosine phosphorylation of the chimera could provide a high affinity binding site to the SH2 domains of p85 PI 3-kinase. To distinguish between these two possibilities, two mutated CD8-⑀ chimeras were constructed. In one chimera, the first tyrosine of CD3-⑀-ITAM was replaced by phenylalanine (Y170F). In the second chimera the second tyrosine was replaced by phenylalanine (Y181F) (see "Experimental Procedures"). The data showed that in COS-7 cells co-transfected with Y170F or with Y181F and Fyn, the chimeras resulted in weak tyrosine phosphorylation as compared with wild-type CD8-⑀ (Fig. 3A, compare lanes 10 and 14 with lane 6). Despite that, Fyn could still be detected associated with Y170F and Y181F chimeras (Fig.  3B, lanes 10 and 14). These results suggest that phosphorylation of only a single tyrosine residue within the CD3-⑀-ITAM is sufficient for high affinity binding of Fyn to the cytoplasmic tail of CD3-⑀.
Moreover, in cells transfected with Y170F or Y181F and with p85␣ and Fyn cDNAs, the amount of p85␣ protein co-immunoprecipitated with either Y170F or Y181F chimeras did not increase as compared with that in the double transfectants (Fig. 3C, compare lanes 12 and 16 with lanes 11 and 15, respectively). Therefore, these results suggest that Fyn is not the linker that couples p85 PI 3-kinase to the CD3-⑀-ITAM. Moreover, the data suggest that there are two types of p85⅐CD8-⑀ complexes depending upon the phosphorylation state of the two tyrosines within the CD3-⑀-ITAM.
p85␣ Subunit of PI 3-Kinase Binds in Vitro to Tyrosinephosphorylated CD3-⑀-ITAM-To examine whether segments representing the CD3-⑀-ITAM specifically bind p85␣ PI 3-kinase, we constructed chemically synthesized peptides representing individual ITAMs. The peptides were coupled via an N-terminal L-cysteine to L-lysine-Sepharose beads. The peptides used (see Table I) represented the unphosphorylated CD3-⑀-ITAM (CT26), the phosphorylated CD3-⑀-ITAM on both tyrosines (CT26P), and an unphosphorylated peptide representing the entire cytoplasmic tail of CD3-⑀ (CD3). Similar amounts of each synthetic peptide bound to lysine-Sepharose beads were used to precipitate proteins from Jurkat whole lysates. In these experiments, the efficiency of CD3-⑀-ITAM binding was judged by the efficiency of the ITAM peptides to recruit the proteins from Jurkat T cell lysates. Fig. 4A shows an anti-phosphotyrosine Western blot of the proteins bound to the peptides. Multiple tyrosine-phosphorylated proteins bound both phosphorylated and unphosphorylated peptides, and only a phosphoprotein at 72-kDa bound exclusively to the phosphorylated CT26P peptide.
We next tested whether p85␣ PI 3-kinase interacted with CD3-⑀-ITAM peptides by Western blot analysis with an anti-p85␣ pAb. A preferential binding of p85␣ PI 3-kinase subunit to doubly phosphorylated peptide (CT26P) was observed as compared with that of unphosphorylated CT26 or CD3 peptides (Fig. 4B, lane 5 versus lanes 4 and 3). These results suggested that association of the CD3-⑀-ITAM with p85 PI 3-kinase is dependent on the phosphorylation of both tyrosine residues.
Requirements for p85␣ PI 3-Kinase Binding to CD3-⑀-ITAM Peptides-That the in vitro binding of p85␣ PI 3-kinase to CD3-⑀ requires tyrosine phosphorylation of CD3-⑀-ITAM was further confirmed by testing the ability of phosphotyrosine to block p85␣ PI 3-kinase with tyrosine-phosphorylated CT26P peptide. The addition of soluble phosphotyrosine to lysates from Jurkat cells just before mixing them with CT26P-Sepharose beads caused a dose-dependent inhibition of the association of p85␣ PI 3-kinase with CT26P (Fig. 5A, lanes 3-6).
We also tested the ability of other compounds to compete the association of p85 PI 3-kinase with CT26P peptide (Fig. 5B). The addition of 80 mM phosphotyrosine or 50 mM phenyl phosphate to the binding reaction completely blocked the association of p85 PI 3-kinase with the immobilized CT26P peptide (Fig. 5B, lanes 3 and 6), whereas similar amounts of phosphothreonine or phosphoserine had only a partial effect (Fig. 5B,  lanes 4 and 5, respectively). Therefore, the results indicate that binding of p85␣ PI 3-kinase to CD3-⑀ depends mostly on the phosphorylation of the tyrosines of the CD3-⑀-ITAM.
Once bound, the p85␣ PI 3-kinase⅐CD3-⑀-ITAM complex appeared to be very stable and fulfilled similar requirements for binding of p85␣ PI 3-kinase to phosphorylated platelet-derived growth factor-␤ receptor (35). Thus, when the immobilized CT26P peptides with bound proteins were washed in nonionic Nonidet P-40 detergent and then incubated for 15 min in solutions containing 2 M NaCl or 1 M urea, no dissociation of p85 PI 3-kinase was detected (Fig. 5B, lanes 7 and 9, respectively). More strongly dissociating conditions such as the presence of ionic detergents (SDS-containing radioimmune precipitation buffer) partially disrupted the p85␣ PI-3K binding to CT26P peptide (Fig. 5B, lane 8). Taken together, these experiments indicate the high affinity and specific association of p85␣ PI 3-kinase with phosphorylated CD3-⑀-ITAM.
In further experiments, we compared the efficiency of p85␣ PI 3-kinase binding to doubly phosphorylated CD3-⑀-ITAM with the previously described high affinity binding of ZAP-70 to CD3-ITAM (36). A negative control in these experiments was the comparison with the low affinity binding of Grb-2 to CD3-⑀-ITAM (37). As shown in Fig. 5, C-E, p85 PI 3-kinase and ZAP-70 did bind to different concentrations of the doubly phosphorylated CD3-⑀-ITAM peptide (CT26P), whereas Grb-2 only bound at the highest concentration of CT26P peptide tested, suggesting the existence of a hierarchy of the SH2-containing proteins binding to doubly phosphorylated CD3-⑀-ITAM: p85 Ն ZAP-70 Ͼ Ͼ Grb-2.
A comparison of p85␣, ZAP-70, Shc, Lck, Cbl, and Grb-2 Binding to Unphosphorylated and Doubly Phosphorylated I Amino acid sequence of the peptides synthesized All peptides were chemically synthesized with an additional N-terminal Cys for coupling to the Lys-Sepharose matrix. Peptides CT26 and CT26P were chemically synthesized with an additional C-terminal sequence, Arg-Arg-Ala-Ser-Val (RRASV), for quantitative measure of the coupling efficiency of the peptides. The Ser residue of the RRASV sequence can be phosphorylated by the heart muscle kinase and [␥- 32  CD3-⑀-ITAM-Specific high affinity interactions between individual SH2 domains of p85␣ PI 3-kinase and phosphorylated tyrosine residues (Y PO4 ) have been shown to require methionine as the third amino acid residue C-terminal of the Y PO4 (Y PO4 -XXM), whereas SH2 domains of ZAP-70, Shc, and Fyn require the consensus binding sequence Y PO4 -XX(L/I) present in CD3-⑀-ITAM (38,39). SH3 domains of Fyn could bind to two proline-rich motifs within the p85 PI 3-kinase (34,40), and the SH3 domain of p85 PI 3-kinase could bind to the N-terminal proline-rich region of Shc (41) or Cbl (42). Therefore, any of these molecules would be a candidate for an intermediate between phosphorylated CD3-⑀-ITAM and p85␣ PI 3-kinase. It is also important to note that phosphorylation of both tyrosine residues within the CD3-⑀-ITAM could create a tighter binding site for proteins containing SH2 domains in tandem (such as ZAP-70 or p85␣ PI 3-kinase) than for single SH2-containing proteins (such as Fyn, Lck, Shc, etc.). Therefore, direct binding of p85␣ PI 3-kinase to CD3-⑀-ITAM cannot be ruled out.
Western blotting analysis was performed to determine whether various SH2-containing proteins such as p85␣ PI 3kinase, ZAP-70, Shc, Lck, Cbl, and Grb-2 could bind selectively to doubly phosphorylated CD3-⑀-ITAM peptide as compared with nonphosphorylated CD3-⑀-ITAM peptide. As an additional control, a nonphosphorylated peptide corresponding to the entire cytoplasmic tail of CD3-⑀ was used. Fig. 6 shows that the interactions of the SH2-containing proteins with the CD3-⑀-ITAM were quite selective, depending upon the phosphorylation state of the peptide used. Thus, proteins with two SH2 domains in tandem such as p85␣ PI3-kinase and ZAP-70 bound exclusively to doubly phosphorylated CD3-⑀-ITAM peptide (Fig. 6, A and B, respectively), whereas proteins with a single SH2 domain bound both phosphorylated and nonphosphorylated peptides although with different efficiencies (Fig. 6). Thus, Lck and to a lesser extent Shc showed a preferential binding to phosphorylated CD3-⑀-ITAM peptide as compared with the nonphosphorylated peptides (Fig. 6, D and C, respectively). In contrast, Grb-2 that bound weakly to both CD3-⑀-ITAM peptides (Fig. 6F, lanes 4 and 5) seemed to strongly interact with the CD3 peptide corresponding to the cytoplasmic tail of CD3-⑀ (Fig. 6F, compare lane 3 with lanes 4 and 5 and with lane 3 of the other panels). Last, Cbl, which does not contain SH2 domains, bound all peptides tested (Fig. 6E). DISCUSSION Tyrosine phosphorylation of ITAMs within the CD3-⑀ and TCRchains is crucial for the recruitment of protein-tyrosine kinases and effector molecules into the TCR⅐CD3 complex. In the present study, the ability of p85␣ PI 3-kinase to bind the CD3-⑀-ITAM was analyzed. In vivo studies show that the p85␣ PI 3-kinase regulatory subunit associates with the CD3-⑀ intracellular region in Jurkat cells and Jurkat cells stably transfected with the cDNA encoding a CD8-⑀ chimera (Fig. 1B, lanes  1 and 5). Such interaction occurs despite a different basal level of CD3-⑀ or CD8-⑀ tyrosine phosphorylation (Fig. 1A, lanes 1  and 5). However, CD3-⑀ or CD8-⑀ ligation with OKT3 or OKT8 increases the amount of p85␣ PI 3-kinase associated with CD3-⑀ or CD8-⑀. Likewise, in COS-7 cells, co-transfection of the cDNA encoding the CD8-⑀ chimera with the cDNA of mouse p85␣ PI 3-kinase results in stable association of p85␣ PI 3kinase with CD8-⑀, although the chimera is not tyrosine-phosphorylated. However, in these cells, triple transfection of cDNAs coding CD8-⑀, p85␣ PI 3-kinase, and Fyn results in additional binding of p85␣ PI 3-kinase to CD8-⑀ and increased CD8-⑀ tyrosine phosphorylation. Overall, these results seem to indicate that although p85␣ PI 3-kinase may bind to the intracellular domain of CD3-⑀ independently of its tyrosine phos- phorylation state, it preferentially binds to tyrosine-phosphorylated CD3-⑀. What is more important, we find a dramatic receptor-induced association of PI 3-kinase with the intracellular domain of CD3-⑀.
Although it is unclear how TCR ligation leads to initiation of tyrosine phosphorylation, Fyn and Lck appear to be directly implicated in the tyrosine phosphorylation of the CD3 chains (4,6,28). In COS-7 cells, the data show that Fyn induces strong tyrosine phosphorylation of wild-type CD8-⑀ chimera and weak phosphorylation of Y170F or Y181F chimeras in which one of the tyrosines within CD3-⑀-ITAM is substituted by phenylalanine (Fig. 3A). These results were surprising, since it has been previously shown that single point mutations along the CD3-⑀-ITAM do not affect the ability of Fyn to phosphorylate the cytoplasmic tail of CD3-⑀ expressed as a chimera with the vesicular stomatitis virus cytoplasmic glycoprotein (43). A possible explanation for these apparently contradictory results is that we have done our experiments in vivo, whereas Timson-Gauen et al. (43) used an in vitro assay to detect CD3-⑀-associated Fyn kinase activity. In the latter case, it is likely that little CD3-⑀-associated tyrosine phosphatase activity was found, and Fyn could easily tyrosine-phosphorylate CD3-⑀ and their mutants. In contrast, in intact COS-7 cells where tyrosine phosphatases are fully active, single tyrosine-phosphorylated CD3-⑀-ITAMs are likely to be more easily dephosphorylated than the doubly tyrosine-phosphorylated ones.
The in vitro studies carried out with synthetic peptides al-lowed us to further investigate the nature of the p85␣ PI 3-kinase/CD3-⑀ interaction. Thus, as it occurred in the in vivo experiments, p85 PI 3-kinase from Jurkat cell lysates binds efficiently to doubly phosphorylated CD3-⑀-ITAM peptides but weakly to nonphosphorylated CD3-⑀-ITAM (Fig. 4). p85␣ PI 3-kinase binding to tyrosine-phosphorylated CD3-⑀-ITAM is specifically abrogated by an excess of free phosphotyrosine or phenyl phosphate, whereas free phosphoserine or phosphothreonine has little effect on binding (Fig. 5). Furthermore, p85/CD3-⑀-ITAM interaction is resistant to the presence of high salt concentration, SDS, or urea in the medium, which are the requirements for high affinity binding of p85␣ PI 3-kinase to the platelet-derived growth factor-␤ receptor (35). These data confirm and extend previous results reported by Cambier and Johnson (17) and show that p85␣ binds weakly to nonphosphorylated and tightly to doubly phosphorylated CD3-⑀-ITAM-containing peptides. Thus, both p85␣ PI 3-kinase and ZAP-70 could bind to nanomolar levels of the phosphorylated CD3-⑀-ITAM, whereas Grb-2 could only be detected with a 100-fold higher concentration of the ITAM (Fig. 5, C-E). It is also noteworthy that a high percentage of the total cellular pool of p85␣ PI 3-kinase and ZAP-70 could bind the CD3-⑀-ITAM, whereas Shc, Lck, Cbl, and Grb-2 did so in a smaller proportion. The present data do not agree with previously published data from Osman et al. (37) in T lymphoblasts. In that study, only ZAP-70 could bind to CD3-⑀-ITAM, whereas Fyn, Shc, Grb-2, and p85␣ PI 3-kinase could not bind to CD3-⑀-ITAM peptides, even at a concentration 10-fold higher than that used in our own experiments. The differences with our data were particularly marked for p85␣ binding to phosphorylated CD3-⑀-ITAM peptide (CT26P), and they do not seem to be related to the RRASV sequence introduced at the C terminus of our CT26P peptide. Thus, a doubly phosphorylated CD3-⑀-ITAM peptide devoid of the Cys and RRASV sequences recruits similar amounts of p85␣ PI 3-kinase as CT26P (data not shown). Therefore, it is likely that the differences between the results of these studies may reflect differences in the relative levels of kinases and adaptors in different populations of T cells.
Hypothetically, direct association of CD3-⑀-ITAM with the tandem SH2 domains of p85␣ PI 3-kinase could occur. However, it is well established that individual SH2 domains of p85␣ bind to single phosphotyrosine moieties in the consensus Y PO4 -MXM/Y PO4 XXM (38), which is not present in the CD3-⑀-ITAM. It has been observed, for the binding of p85 to the plateletderived growth factor receptor, that high affinity binding is only detected when both SH2 domains of p85 are able to bind to two phosphotyrosine residues on the platelet-derived growth factor receptor that are 11 residues apart (44). In this sense, it is interesting that binding of p85 SH2 domains to the phosphorylated YXXM motif activates the enzyme specifically only when both p85 SH2 domains are involved (45). A similar spacing (10 amino acids) between YEPI and YSGL binding sites occurs in the CD3-⑀-ITAM, and the Ile or Leu at the ϩ3position carboxyl terminus of the Tyr residues are the most hydrophobic amino acid residues after Met. Because of the stronger selection for Met at the ϩ3-position, single p85 SH2 domains are expected to bind much more tightly at phosphorylated YXXM sites, yet the spacing between the tyrosine residues in the YEPI and YSGL sites of CD3-⑀ might provide a relative high affinity binding site for the two SH2 domains of p85 PI 3-kinase. We find that mutating Tyr 170 or Tyr 181 within CD3-⑀-ITAM reduces binding of p85␣ PI 3-kinase (Fig. 3C). One possible model to explain these results would be that mutations of a single tyrosine of CD3-⑀-ITAM would leave only the potential for weak interaction of the remaining phosphotyrosine with a single p85␣ SH2 domain, whereas if both phos- photyrosines are available, interactions with both SH2 domains of p85␣ would be possible, and the overall affinity of interaction would be increased significantly, allowing detection of binding. This model implicitly acknowledges that doubly phosphorylated CD3-⑀-ITAMs exist and that both SH2 domains of a single p85␣ molecule participate. If this were not so, the effect of the mutations of Tyr 170 and Tyr 181 on the quantity of p85␣ PI 3-kinase bound would be simply additive.
Some other pieces of evidence suggest that phosphorylation of both tyrosine residues within the ITAM would create a tighter binding site for proteins containing SH2 domains in tandem (as ZAP-70 or p85␣ PI 3-kinase) than for single SH2containing proteins (such as Fyn, Lck, Shc, etc.). In this sense, we have shown that either in vivo or in vitro p85␣ PI3-kinase binds preferentially to doubly phosphorylated CD3-⑀-ITAM (Figs. 1, 3C, and 4B), whereas Fyn could bind both doubly and single phosphorylated CD3-⑀-ITAMs (Fig. 3B). Moreover, the in vitro data show quantitative and qualitative differences in the abilities of different SH2-containing proteins to bind unphosphorylated or doubly phosphorylated CD3-⑀-ITAM peptides (Fig. 6). These differences are particularly marked between proteins having two SH2 domains (ZAP-70 and p85␣) and those with a single SH2 domain (Lck, Shc, Grb-2). This is even more evident upon titrating down the concentration of doubly phosphorylated CD3-⑀-ITAM peptide used in the binding studies (Fig. 5). This again may reflect the requirement for a duplicated SH2 domain in order for p85␣ to bind with high affinity to doubly phosphorylated CD3-⑀-ITAM.
Alternatively, the association of p85␣ PI 3-kinase with the CD3-⑀-ITAM may occur indirectly via interaction with a tyrosine kinase of the Src or Syk family or with the SH2-containing adaptor molecule Shc. Thus, the multiple ITAMs in T and B cell receptors are well suited to bind multiple SH2-containing proteins. It was suggested by Songyang et al. (46) that the Y PO4 -XX(L/I) sites with Ile at the ϩ3 position are able to bind with high affinity to either Shc or Src family members (Lck, Fyn, Lyn, etc.), whereas sites with Leu at the ϩ3-position are better suited to interact with Syk or ZAP-70 SH2 domains. In this scenario, it is predicted that several SH2-containing proteins might compete for the same binding site. Thus, binding of some SH2-containing molecules such as ZAP-70 is restricted to double phosphorylated ITAMs (47)(48)(49), and molecules containing a single SH2 domain may compete advantageously with ZAP-70 (or p85␣ PI 3-kinase) for ITAM binding when only one tyrosine within the ITAM is phosphorylated (50).
Several authors have shown that Fyn uses its own intrinsic SH3 domain to bind to p85 PI 3-kinase (34,51), and similar data have been obtained with the SH3 domain of Lck (52). Potential binding motifs within p85 PI 3-kinase including PPT-PKPRPPRPLP (amino acids 84 -96) or PAPALPPKPPKP (amino acids 303-314) could be detected associated to Fyn without prestimulation of the TCR⅐CD3 complex, so it would be predicted that Fyn may recruit associated PI 3-kinase independently of tyrosine kinase activity and phosphorylation. The data in COS-7 cells triple-transfected with CD8-⑀ (or Y170F or Y181F mutants), Fyn, and p85␣ PI 3-kinase cDNAs suggest that Fyn is not the adaptor molecule bringing p85␣ PI 3-kinase in the proximity of CD8-⑀. In contrast, the data on Lck binding to nonphosphorylated or doubly phosphorylated CD3-⑀-ITAM parallels the binding of p85␣ PI 3-kinase to these peptides although with somewhat lower efficiency. Taken together, these studies would suggest that there is only a marginal contribution of the SH3 domain of Fyn to the p85 PI 3-kinase binding to CD3-⑀-ITAM. Alternatively, since the nature of the interaction Fyn (or Lck) and CD3-⑀-ITAM is different, depending upon the phosphorylation state of the motif (18, 28, 43, 50), the binding of Fyn (or Lck) to tyrosine-phosphorylated CD3-⑀-ITAM could change Fyn conformation, rendering its SH3 domain more available to interact with p85␣ PI 3-kinase.
Our results raise the question of the biological function of PI 3-kinase binding to the activated TCR⅐CD3 complex. In a previous study we showed that p85␣ PI 3-kinase preferentially associated with the chain membrane-proximal A-ITAM (1). In this study, we show an alternative binding site for p85␣, the CD3-⑀-ITAM. One consequence of the binding of p85␣ to phosphorylated Aand CD3-⑀-ITAMs may be to bring p85␣ to the receptor so that it can be phosphorylated by an associated tyrosine kinase (e.g. ZAP-70). Another interesting possibility is that the binding of both SH2 domains of p85␣ to these phosphorylated ITAMs could induce a conformational change in the p85␣/p110 structure, leading to PI 3-kinase activation. Such mechanism has been shown to be involved in the activation of PI 3-kinase, when its tandem SH2 domains are engaged on phosphorylated peptides (53,54). The presence of two A-ITAMs on eachdimer and two CD3-⑀-ITAMs on the CD3-⑀: CD3-␥ and CD3-⑀:CD3-␦ pairs provides at least four binding sites for p85␣ in a single TCR⅐CD3 complex, in the case of no steric hindrance. Docking of several molecules of p85␣ PI 3kinase onto a single receptor could perhaps facilitate the activation of this important signaling molecule by adjacent tyrosine kinases or contact with other effector molecules.