The Cytotoxicity Receptor CRACC (CS-1) Recruits EAT-2 and Activates the PI3K and Phospholipase Cγ Signaling Pathways in Human NK Cells¹

Natural killer cells express a broad repertoire of activating receptors that trigger multiple signaling pathways leading to cytotoxicity and secretion of IFN-γ (1, 2, 3, 4). Among these are a growing number of cell surface members of the Ig superfamily homologous to CD2, including CD2, CD48, CD58, signaling lymphocytic activation molecule (SLAM)³ (CD150), 2B4 (CD244), CD84, Ly-9 (CD229), NK-T-B-Ag (NTB-A), CD2-like receptor-activating cytotoxic cell (CRACC), and B lymphocyte activator macrophage expressed (BLAME) (5, 6, 7). These receptors mediate either homophilic adhesion or heterophilic interactions with other CD2 family members. The cytoplasmic domains of SLAM, 2B4, CD84, Ly-9, CRACC, and NTB-A contain unique amino acid motifs, TxYxxV/I, called immunoreceptor tyrosine-based switch motifs (ITSMs) (5, 6, 7). ITSMs associate with a Src homology 2 (SH2) domain-containing protein, SLAM-associated protein (SAP; also called SH2D1A or DSHP), predominantly expressed in T and NK cells (5, 6, 7, 8, 9, 10). SAP recruits and activates the Src-family kinase Fyn through a unique SH2-SH3 domains interaction (11, 12, 13, 14). Fyn induces phosphorylation of the cytoplasmic domain of ITSM-containing receptors, allowing sequential recruitment and activation of downstream signaling adaptors and effectors such as SHIP-1, Shc, Dok1/2, and Ras-GAP. Moreover, SAP can function as a blocker, inhibiting the recruitment of protein tyrosine phosphatases like Src homology-2 phosphotyrosine phosphatase (SHP-2) to the cytoplasmic domain of ITSM-containing receptors (8, 15, 16). The importance of these SAP-mediated signaling pathways in immune responses is underscored by the observation that mutations in the SAP gene lead to X-linked lymphoproliferative syndrome (XLP), an immunodeficiency associated with dysregulated proliferation of T and B lymphocytes during primary EBV infection (5, 6, 7). An SH2 domain-containing protein similar to SAP, Ewing’s sarcoma’s/FLI1-activated transcript 2 (EAT-2) (17), has been detected in human NK cells and T cells and mouse B cells and macrophages (18, 19, 20) and shown to interact with the cytoplasmic domain of CD2-family members in transfected cells (19, 20, 21). However, EAT-2 function is poorly defined in primary cells and in vivo.

NK cells express at least three CD2 family receptors, 2B4, NTB-A, and CRACC (18, 22, 23, 24, 25, 26). 2B4 binds CD48 (27, 28); CRACC and NTB-A mediate homophilic adhesion (29, 30, 31, 32). In humans, all three of these receptors trigger NK cell-mediated cytotoxicity. Although 2B4 and NTB-A recruit SAP (15, 25), CRACC does not, despite the presence of ITSMs in its cytoplasmic domain (18). Consistent with this, analysis of NK cells from XLP patients have shown that 2B4- and NTBA-mediated cytotoxicity is SAP-dependent (25, 33, 34, 35, 36), whereas CRACC activates cytotoxicity through an ERK-mediated pathway that is SAP-independent (18). Because of the close homology between SAP and EAT-2, it is possible that CRACC recruits EAT-2 through its cytoplasmic ITSMs and that EAT-2 effectively substitutes for SAP in mediating CRACC signaling. Although we failed to detect significant association of CRACC with EAT-2 in a preliminary experiment (18), we have more thoroughly investigated the possible association of these molecules in primary human cells using a newly generated EAT-2 antiserum. In this study, we demonstrate that CRACC does indeed associate with EAT-2 in human NK cells upon ligation with a specific Ab. We show that recruitment of EAT-2 mediates the phosphorylation of CRACC, possibly by recruiting a Src kinase, and we define the signaling mediators downstream of CRACC/EAT-2 implicated in triggering NK cell-mediated cytotoxicity.

Human NK cells were purified from peripheral blood and cultured in IL-2 as previously described (37) . CD4⁺, CD8⁺, CD4⁺ CD8⁺, and γ δ T cell clones were established from peripheral blood by limiting dilution and cultured in IL-2 as previously described (37) . Peripheral B cells and monocyte-derived dendritic cells (DC) were activated by incubation with CD40L-expressing cells and LPS, respectively. Cells lines used were: NK92 (human NK cell line that lacks the FcR CD16), P815 (murine mastocytoma), Daudi (human Burkitt lymphoma-derived B cell), 721.221 (human EBV-transformed B cell) and 293. The anti-CRACC mAbs 24 and 162 (18) and anti-2B4 mAb 2–69 (33) were generated in our laboratory. Antiserum to EAT-2 was generated by immunizing rabbits with the keyhole limpet hemocyanin-conjugated peptide DLPYYHGRLTKQDCETL. The anti-SAP Ab was a generous gift from Dr. S. G. Tangye (Centenary Institute of Cancer Medicine and Cell Biology, Newtown, New South Wales, Australia). Anti-2B4 mAb C1.7 (38) was purchased from the Immunotech laboratory, anti-phosphotyrosine 4G10 mAb was obtained from Upstate Biotechnology, and anti-Vav, anti-phospholipase Cγ1 (PLC γ 1), anti-PLCγ 2, and anti-SHIP-1 mAb were obtained from Santa Cruz Biotechnology. The pharmacological inhibitor of PLC γ U73122, the inactive analog U73343 and the src kinase inhibitor PP2 were purchased from Calbiochem.

Before stimulation with mAbs, NK92 cells (20 × 10⁶) were cultured overnight in the absence of IL-2 to reduce confounding signals due to IL-2-mediated activation. Cells were then incubated in ice for 15 min with a saturating dose of anti-CRACC mAb 24, anti-2B4 mAb C1.7, or anti-CD56. Cells were then washed and incubated at 37°C for the indicated period of time in the presence of goat anti-mouse IgG. When indicated, cells were incubated with the src kinase inhibitor PP2 (10 μ M) or solvent (DMSO) 20 min prior to Ab stimulation as well as during Ab stimulation at 37°C. For sodium pervanadate stimulation, cells were incubated with 200 μ M sodium pervanadate for 15 min at 37°C. After stimulation, cells were lysed with lysis buffer (1% v/v Triton X-100, 50 Mm Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1.5 mM MgCl₂, 10% glycerol, plus protease and phosphatase inhibitors) and immunoprecipitated with the indicated Abs. Precipitated proteins were fractionated by SDS-PAGE, transferred to nitrocellulose membranes and probed with the indicated mAb. To confirm that all substrates were adequately immunoprecipitated, immunoblots were reprobed with Abs directed against the various substrates. Because anti-CRACC Abs do not detect CRACC in immunoblotting, we ensured that each cell lysate used for CRACC immunoprecipitations contained equal amount of proteins with a control Ab (anti-EAT-2). To determine phosphorylation of AKT, cell lysates were immediately fractionated by SDS-PAGE and the active form of AKT was detected with phosphospecific Abs. Blots were reanalyzed with anti-AKT Abs to quantify proteins loaded in each lane. Anti-pAKT (Ser⁴⁷³) and anti-AKT Abs were obtained from New England Biolabs.

2B4 and SAP constructs have already been described (33) . To express EAT-2 and CRACC as N-terminal FLAG fusion proteins, EAT-2 and CRACC cDNA were subcloned in pCMV2-FLAG and pCMV3-FLAG (Sigma-Aldrich), respectively. 293 cells were transiently transfected with the above plasmids using Lipofectamine 2000 (Invitrogen Life Technologies) following the manufacturer’s instructions. When described, cells were pretreated with 10 μ M PP2 (Calbiochem) for 20 min at 37°C.

NK cell cytotoxicity was tested against [⁵¹Cr]-labeled P815 cells in the presence of 10 μg/ml of either mAb 162, mAb 2-69, or a control mouse IgG (Immunotech). In some experiments, NK92 cells were pretreated with the pharmacological inhibitor of PLCγ U73122 (3.3 μM) or the inactive analog U73343 (3.3 μM) for 5 min at 37°C. PP2 (2 μM) or DMSO were maintained throughout the cytotoxicity assays, as PP2 effects are rapidly reversible. At the concentration used, PP2 did not affect NK cell viability, as previously reported (39).

EAT-2 transcripts have been detected in human CD8 and CD4 T cells (19) and in mouse B cells and macrophages (20), while EAT-2 protein has been observed in human NK cells (18). To establish the expression pattern of the EAT-2 protein in human primary leukocytes, we analyzed a variety of cell lysates by immunoblotting with a rabbit anti-human EAT-2 serum. We found abundant EAT-2 protein in NK cells, either freshly isolated from blood or after culture in IL-2, and in the NK cell lines NK92 and YT (Fig. 1,A and data not shown). Moreover, we detected some EAT-2 protein in CD8⁺ αβT cells and γδ T cells before and after activation with an anti-CD3 Ab (Fig. 1,B). EAT-2 was absent in CD4⁺ T cells, CD4⁺CD8⁺ T cells (Fig. 1,B), B cells (Fig. 1,C), immature and LPS-activated DC (Fig. 1 D). Analysis of EAT-2 expression by RT-PCR confirmed that the EAT-2 transcript was more abundant in NK cells than CD8⁺ and γδ T cells (data not shown). We conclude that EAT-2 is preferentially expressed in human cytotoxic lymphocytes, particularly NK cells.

Several CD2-family receptors have been shown to bind EAT-2 in transfected cell lines (19, 20, 21). To determine whether CRACC recruits EAT-2 in human NK cells, we took advantage of a well-established NK cell line NK92. We ligated CRACC on NK92 using a specific Ab and a cross-linker and analyzed CRACC immunoprecipitates from resting and CRACC-activated cells with anti-phosphotyrosine and anti-EAT-2 Abs. We observed that upon ligation, CRACC is phosphorylated and associates with EAT-2 (Fig. 2,A). Analysis of CRACC immunoprecipitates with an anti-SAP Ab confirmed our previous finding that CRACC does not recruit SAP in either resting or stimulated cells (18). We also investigated association of CRACC with EAT-2 in NK92 cells stimulated with sodium pervanadate, which inhibits tyrosine phosphatases and thereby induces strong tyrosine phosphorylation of CRACC. In CRACC immunoprecipitates from pervanadate-treated cells, CRACC was associated with EAT-2 but not with SAP (Fig. 2 B), consistent with our results in CRACC-stimulated NK cells. We conclude that, upon activation, CRACC selectively recruits EAT-2 in human NK cells.

Because the receptor 2B4 has been shown to bind EAT-2 in transfected cells (20), we determined whether this also occurs in human NK cells. In unstimulated NK92 cells, 2B4 recruited EAT-2 and exhibited a modest tyrosine phosphorylation; after Ab-mediated ligation, 2B4 association with EAT-2 was reduced, while 2B4 effectively recruited SAP, as previously reported (Ref.15 ; Fig. 2,A). Consistent with this result, in pervanadate-treated cells, 2B4 association with SAP increased, while association with EAT-2 decreased (Fig. 2 B). These data suggest that 2B4 can recruit both EAT-2 and SAP and that tyrosine phosphorylation of cytoplasmic ITSMs of 2B4 augments their affinity for SAP, while reducing the affinity for EAT-2.

EAT-2 has been previously shown to induce phosphorylation of CD84, SLAM, Ly-9, and 2B4 in transfected cells (19, 20, 21). To address whether EAT-2 mediates phosphorylation of CRACC, we transiently expressed CRACC in 293 cells together with either EAT-2 or SAP, immunoprecipitated CRACC and analyzed its tyrosine phosphorylation by immunoblotting. CRACC was evidently phosphorylated in the presence of EAT-2 (Fig. 3,A). 2B4 was also phosphorylated in the presence of either EAT-2 or SAP (Fig. 3,B), consistent with previous reports (15, 20). To investigate whether EAT-2 induces tyrosine phosphorylation of CRACC and 2B4 by recruiting src kinases, we tested a pharmacological inhibitor of src tyrosine kinases (PP2) on EAT-2-mediated phosphorylation of CRACC and 2B4 in transfected cells. PP2 reduced CRACC and 2B4 phosphorylation (Fig. 3), suggesting that EAT-2 may induce phosphorylation by recruiting a src kinase. PP2 also reduced SAP-induced phosphorylation of 2B4, consistent with the known association of SAP with the src kinase Fyn (11, 12, 13, 14).

To investigate whether EAT-2 induces CRACC phosphorylation by recruiting a src kinase in human NK cells, we investigated the effect of PP2 on CRACC phosphorylation and CRACC-mediated cytotoxicity in NK92 cells. PP2 reduced phosphorylation of CRACC induced by Ab-mediated cross-linking (Fig. 4,A). In control experiments, PP2 also reduced phosphorylation of 2B4 (Fig. 4,B), which has been shown to be mediated by SAP through recruitment of Fyn (40). Thus, results suggests that EAT-2-induced phosphorylation of CRACC may involve a src kinase. In support of this hypothesis, we also observed that PP2 potently inhibited CRACC- and 2B4-mediated redirected lysis of P815, whereas the PP2 solvent DMSO did not (Fig. 4 C). It has been reported that EAT-2 induces tyrosine phosphorylation of associated receptors by blocking their ability to recruit protein tyrosine phosphatases, particularly SHP-2 (20). However, we detected no association of CRACC with SHP-1, SHP-2, or SHIP-1 in resting or Ab-stimulated NK92 cells (data not shown). Thus, although in transfected cells EAT-2 may induce CRACC phosphorylation by interfering with CRACC-SHP-2 association, this is unlikely to be a crucial mechanism in human NK cells.

To identify downstream mediators that become activated in response to ligation of CRACC, we cross-linked CRACC on NK92 cells and analyzed several downstream effectors by immunoprecipitation and anti-phosphotyrosine immunoblotting. Ligation of CRACC induced phosphorylation of PLCγ1 and PLCγ2 (Fig. 5, A and B). Moreover, we detected a significant phosphorylation of Akt, which is an indicator of PI3K activation (Fig. 5,C), and the E3 ubiquitin ligase c-Cbl (Fig. 5,D). Ligation of CRACC induced modest tyrosine phosphorylation of the guanine nucleotide exchange factors Vav and the 5′ inositol phosphatase SHIP-1 (Fig. 5, E and F). No phosphorylation of Shc and LAT was observed (data not shown). In comparison with CRACC-mediated activation, ligation of 2B4 was as effective in phosphorylating PLCγ1, PLCγ2, Akt, and c-Cbl and more effective in activating Vav and SHIP-1. Although two studies have shown that LAT is constitutively associated with 2B4 (41, 42), we detected no phosphorylation of LAT (data not shown). Moreover, we observed no 2B4-mediated phosphorylation of Shc. Collectively, these results suggest that PLCγ1, PLCγ2, PI3K, and c-Cbl are the major substrates involved in downstream signaling events initiated by CRACC. Moreover, our data indicate that, despite differential recruitment of EAT-2 and SAP, CRACC and 2B4 activate similar signaling adaptor and effector molecules.

To substantiate the implication that activation of PLCγ1 and PLCγ2 is critical for CRACC- and 2B4-mediated cytotoxicity, we determined whether this function could be inhibited by the pharmacological PLCγ inhibitor U73122. NK92 cells effectively killed the FcR bearing P815 cell in the presence of anti-CRACC or anti-2B4 Abs (Fig. 6). U73122 potently inhibited CRACC- and 2B4-mediated redirected lysis of P815, whereas the pharmacologically inactive analog U73343 did not (Fig. 6), confirming a role for PLCγ in CRACC- and 2B4-mediated cytotoxicity.

Our results provide the first demonstration that CRACC recruits EAT-2 in NK cells. We previously showed that CRACC is a unique CD2-family receptor in that it triggers cytotoxicity independently of SAP, despite the presence of ITSMs in its cytoplasmic domain (18). As EAT-2 is closely related to SAP, we hypothesized that CRACC may trigger cytotoxicity through EAT-2. Although we were not able to detect significant association of CRACC with EAT-2 in a preliminary experiment (18), using a newly generated anti-EAT-2 antiserum we now provide conclusive evidence that CRACC recruits EAT-2 after Ab-mediated cross-linking in NK cells. We also show that EAT-2 protein is preferentially expressed in primary NK cells, CD8⁺ T cells, and γδ T cells, whereas we detect no EAT-2 protein in CD4⁺ T cells, CD4⁺CD8⁺ T cells, B cells, and DC. Yet, CRACC is expressed in CD4⁺ T cells, B cells and DC as well as NK cells and CD8⁺ T cells, suggesting that CRACC signaling differs in distinct cell types. Moreover, as EAT-2 transcripts have been detected in murine B cells and macrophages, our data provide evidence for a significant difference in the cellular distribution of human and mouse EAT-2.

We show that 2B4 also recruits EAT-2 in NK cells, consistent with a previous report showing that 2B4 and other CD2-family receptors associate with EAT-2 in transfected cells (19, 20, 21). Because 2B4 associates with SAP, this observation raises the question of whether SAP and EAT-2 compete for binding to 2B4 or associate independently of one another (19) and whether SAP and EAT-2 play redundant roles. We observed that EAT-2 preferentially binds 2B4 in nonactivated cells, whereas SAP binds better after cell activation. Thus, it is possible that EAT-2 and SAP compete for one or more cytoplasmic ITSMs of 2B4 and that ITSM phosphorylation following activation favors recruitment of SAP over EAT-2. Interestingly, in SAP-deficient NK cells from XLP patients, 2B4 is no longer capable of triggering cytotoxicity (25, 33, 34, 35, 36), despite the presence of EAT-2. Thus, SAP and EAT-2 appear to play nonredundant roles in 2B4 signaling and function.

Although in our study, EAT-2 promotes CRACC and 2B4 phosphorylation, recently, Roncagalli et al. (43) showed that EAT-2 is a negative regulator of mouse NK cell function and phosphorylation. Moreover, they demonstrated that this inhibitory function depends on two C-terminal tyrosines of EAT-2. However, because human EAT-2 lacks one of these tyrosines, it may be devoid of inhibitory functions.

In this report, we found that EAT-2 induces tyrosine phosphorylation of CRACC. Two mechanisms may be involved in this phenomenon. EAT-2 may recruit a src kinase, as SAP does (11, 12, 13, 14). Supporting this mechanism, the inhibitor of src kinases PP2 reduced EAT-2-induced phosphorylation of CRACC or 2B4 in transfected cells. Moreover, PP2 inhibited Ab-mediated CRACC phosphorylation and CRACC-induced cytotoxicity in human NK cells. Although we were unable to detect association of Fyn or other src kinases with CRACC (data not shown), it is possible that the binding affinity of EAT-2 for src kinases is too low to be detected in immunoprecipitation experiments. Surface plasmon resonance studies are required to conclusively assess the binding of soluble EAT-2 SH2 domain to soluble SH3 domains of various src kinases. Another mechanism for EAT-2-induced phosphorylation may rely on the ability of EAT-2 to block the recruitment of protein tyrosine phosphatases. It was previously shown that several CD2-family receptors associate with SHP-2 in transfected cells, and that this association is reduced by EAT-2 (20). However, we were unable to detect association of CRACC with SHP-2 or other phosphatases in resting NK cells as well as after Ab-mediated stimulation, when CRACC is phosphorylated and recruits more EAT-2. Thus, EAT-2-mediated blockade of protein tyrosine phosphatases recruitment does not appear to be a crucial mechanism in inducing phosphorylation of CRACC in human NK cells.

We provide biochemical demonstration that CRACC activates PLCγ1, PLCγ2, and PI3K. These intracellular signaling molecules are likely to be major intermediates in CRACC-induced NK cell activation and cytotoxicity. Critical involvement of PLCγ1 and PLCγ2 in CRACC-mediated cytotoxicity is corroborated by blocking of CRACC-mediated cytotoxicity in the presence of a pharmacological inhibitor of PLCγ. Moreover, activation of PLCγ1 and PLCγ2 is most likely responsible for the intracellular Ca²⁺ flux mediated by CRACC in a transfected NK cell line (44). PI3K may trigger CRACC-mediated cytotoxicity by generating phosphatidylinositol 3,4,5 trisphosphate (PIP3), which is essential for membrane recruitment and activation of PLCγ1 and PLCγ2. Moreover, it has been shown that PI3K can activate the NK cytolytic machinery by inducing sequential activation of the GTP-binding-protein Rac1, the cytoplasmic kinases Pak1 and MEK, and ERK1/2 (45). Our previous demonstration that pharmacological inhibitors of ERK block CRACC-mediated cytotoxicity suggests that this pathway may be involved (18). Ligation of CRACC also induced phosphorylation of c-Cbl and, to a minor extent, of Vav and SHIP-1. Whether these effectors are critical in regulating CRACC-mediated cytotoxicity remains to be determined.

Interestingly, despite differential recruitment of EAT-2 and SAP, CRACC and 2B4 activated a remarkably similar spectrum of downstream effectors triggering cytotoxicity. We found that ligation of 2B4 induces activation of PLCγ1 and PLCγ2. Although phosphorylation of PLCγ1 was not reported in a previous study (40), these experiments were performed in the NK cell line YT, which differs significantly from the NK92 cell line used in our study in that it does not require IL-2 for expansion. 2B4-mediated phosphorylation of PI3K, Vav1, c-Cbl, and SHIP-1 has been corroborated in previous studies (40, 46). Although 2B4 has been reported to constitutively bind the adaptor protein LAT (41, 42), like Chen et al. (40), we were unable to detect phosphorylation of LAT. These discrepancies in detecting 2B4-LAT associations and/or LAT phosphorylation may be due to the use of different detergents or cell lines.

In conclusion, we have shown that the natural cytotoxicity receptor CRACC is unique among CD2-family receptors in that it recruits the adaptor EAT-2 but not SAP in human NK cells after activation. EAT-2 induces phosphorylation of CRACC, possibly by recruiting a src-family kinase to the cytoplasmic domain of CRACC. Upon ligation and phosphorylation, CRACC activates downstream effectors of cytotoxicity, including PLCγ and PI3K. These results outline a new signaling pathway leading to NK cell-mediated cytotoxicity.

We thank Susan Gilfillan for reading the manuscript and Marina Cella for helpful advice throughout the experiments.

The authors have no financial conflict of interest.