Insufficient phosphorylation prevents fc gamma RIIB from recruiting the SH2 domain-containing protein-tyrosine phosphatase SHP-1.

Fc gamma RIIB are IgG receptors that inhibit immunoreceptor tyrosine-based activation motif (ITAM)-dependent cell activation. Inhibition depends on an immunoreceptor tyrosine-based inhibition motif (ITIM) that is phosphorylated upon Fc gamma RIIB coaggregation with ITAM-bearing receptors and recruits SH2 domain-containing phosphatases. Agarose bead-coated phosphorylated ITIM peptides (pITIMs) bind in vitro the single-SH2 inositol 5-phosphatases (SHIP1 and SHIP2) and the two-SH2 protein tyrosine phosphatases (SHP-1 and SHP-2). Phosphorylated Fc gamma RIIB, however, recruit selectively SHIP1/2 in vivo. We aimed here at explaining this discordance. We found that beads coated with low amounts of pITIM bound in vitro SHIP1, but not SHP-1, i.e. behaved as phosphorylated Fc gamma RIIB in vivo. The reason is that SHP-1 requires its two SH2 domains to bind on adjacent pITIMs. Consequently, the binding of SHP-1, but not of SHIP1, increased with pITIM density on beads. When trying to increase Fc gamma RIIB phosphorylation in B cells and mast cells, we found that concentrations of ligands optimal for Fc gamma RIIB phosphorylation failed to induce SHP-1 recruitment. SHP-1 was, however, recruited by Fc gamma RIIB when hyperphosphorylated following cell treatment with pervanadate. Our data suggest that Fc gamma RIIB phosphorylation may not be sufficient in vivo to enable the recruitment of SHP-1 but that (pathological?) conditions that would hyperphosphorylate Fc gamma RIIB might enable SHP-1 recruitment.


Fc␥RIIB are IgG receptors that inhibit immunoreceptor tyrosine-based activation motif (ITAM)-dependent cell activation. Inhibition depends on an immunoreceptor tyrosine-based inhibition motif (ITIM) that is phosphorylated upon Fc␥RIIB coaggregation with ITAMbearing receptors and recruits SH2 domain-containing phosphatases. Agarose bead-coated phosphorylated ITIM peptides (pITIMs) bind in vitro the single-SH2 inositol 5-phosphatases (SHIP1 and SHIP2) and the two-SH2 protein tyrosine phosphatases (SHP-1 and SHP-2).
Phosphorylated Fc␥RIIB, however, recruit selectively SHIP1/2 in vivo. We aimed here at explaining this discordance. We found that beads coated with low amounts of pITIM bound in vitro SHIP1, but not SHP-1, i.e. behaved as phosphorylated Fc␥RIIB in vivo. The reason is that SHP-1 requires its two SH2 domains to bind on adjacent pITIMs. Consequently, the binding of SHP-1, but not of SHIP1, increased with pITIM density on beads. When trying to increase Fc␥RIIB phosphorylation in B cells and mast cells, we found that concentrations of ligands optimal for Fc␥RIIB phosphorylation failed to induce SHP-1 recruitment. SHP-1 was, however, recruited by Fc␥RIIB when hyperphosphorylated following cell treatment with pervanadate. Our data suggest that Fc␥RIIB phosphorylation may not be sufficient in vivo to enable the recruitment of SHP-1 but that (pathological?) conditions that would hyperphosphorylate Fc␥RIIB might enable SHP-1 recruitment.
Fc␥RIIB are single-chain low-affinity receptors for the Fc portion of IgG antibodies that bind multivalent immune complexes. They exist as two (Fc␥RIIB1 and B2 in humans) or three (Fc␥RIIB1, B1Ј, and B2 in mice) alternatively spliced products of the FcgR2b gene (1). All murine and human Fc␥RIIB isoforms were shown to negatively regulate cell activation induced by all receptors bearing intracytoplasmic immunoreceptor tyrosine-based activation motifs (ITAMs) 1 (2).
Fc␥RIIB also negatively regulate cell proliferation induced by growth factor receptors with an intrinsic protein tyrosine kinase activity (3). Confirming these results, Fc␥RIIB-deficient mice were found: 1) to exhibit enhanced antibody responses (4); 2) to develop exaggerated IgE- (5) and IgG-dependent anaphylactic reactions (4); 3) to have an enhanced susceptibility to experimental murine models of IgG-dependent autoimmune diseases (6 -8); and 4) to exhibit enhanced antibody-dependent cell-mediated cytotoxic responses to the injection of therapeutic antibodies to tumor antigens (9).
To inhibit cell activation, Fc␥RIIB need to be coaggregated with ITAM-bearing receptors by immune complexes or by any extracellular ligand capable of interacting with the two receptors simultaneously (10 -12). Coaggregation indeed enables Fc␥RIIB to be tyrosyl-phosphorylated by Lyn (13), a Src family protein tyrosine kinase provided by ITAM-bearing receptors. Fc␥RIIB isoforms contain a variable number of tyrosine residues in their intracytoplasmic domain, one of which proved to be critical (2,14). This tyrosine stands within a 13-amino acid sequence that was found to be necessary (2,15) and sufficient (14) for inhibition. Related sequences subsequently found in a large number of transmembrane molecules with negative regulatory properties provided the molecular basis for the definition of an immunoreceptor tyrosine-based inhibition motif (ITIM) having the (I/V/L)xYxxL consensus sequence (16,17).
A general property of ITIMs is to have an affinity for cytoplasmic SH2 domain-containing phosphatases when tyrosylphosphorylated (17). Phosphatases that are recruited to the membrane antagonize with activation signals transduced by ITAM-bearing receptors. ITIM-bearing receptors were found to recruit two classes of phosphatases which exert markedly different effects: protein-tyrosine phosphatases and inositol 5-phosphatases. Protein-tyrosine phosphatases are SHP-1 and SHP-2. They have two SH2 domains and their substrates are tyrosyl-phosphorylated proteins (18). SHP-1 is thought to dephosphorylate tyrosines in ITAMs (19), protein tyrosine kinases and/or adapter proteins such as SLP-76 (20) whose phosphorylation is critical for activation signals. SHP-1 thus stops the initial steps of transduction. The possible role of SHP-2 is not clear, as both positive and negative effects have been assigned to this phosphatase. Inositol 5-phosphatases are SHIP1 and SHIP2. They have a single SH2 domain and they remove 5Ј-phosphate groups in inositol phosphates and phosphatidylinositol phosphates that are 3Ј-phosphorylated (21). The preferred substrate of SHIP1 is phosphatidylinositol (3,4,5)-* This work was supported in part by the Institut National de la Santé et de la Recherche Médicale, the Institut Curie, and the Association pour la Recherche sur le Cancer. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. trisphosphate which enables the membrane translocation of the Bruton's tyrosine kinase via its pleckstrin homology domain (22). Bruton's tyrosine kinase is mandatory for phospholipase C␥ to be activated and to hydrolyze phosphatidylinositol (4,5)-bisphosphate into inositol (1,4,5)-trisphosphate, which induces a Ca 2ϩ response (23), and diacylglycerol, which activates protein kinase C. SHIP1 was recently shown to inhibit the Ras pathway by acting as an adapter molecule in B cells. When phosphorylated by Lyn, SHIP1 recruits Dok by its protein tyrosine-binding domain. Dok is phosphorylated and recruits RasGAP which inactivates Ras by exchanging GTP for GDP on the latter molecule (24). SHIP1 therefore arrests the propagation of intracellular signals leading to the Ca 2ϩ response and to the activation of the Ras pathway. The possible role of SHIP2 is not yet known.
The in vitro binding specificity of ITIMs was analyzed using phosphorylated synthetic peptides to precipitate phosphatases from cell lysates. Under these conditions, all known ITIMs, including the Fc␥RIIB ITIM, bound SHP-1 and SHP-2 (25). Remarkably, the Fc␥RIIB ITIM also bound SHIP1 (26) and SHIP2 (27). We previously identified two hydrophobic residues, at positions Y-2 and Yϩ2, that determine the binding of SHPs (28) and SHIPs (29), respectively. The in vivo recruitment of phosphatases by ITIM-bearing receptors was analyzed by coprecipitation, following their tyrosyl phosphorylation upon coaggregation with ITAM-bearing receptors. Tyrosyl-phosphorylated Fc␥RIIB were initially reported to recruit SHP-1 in B cells following their coaggregation with B cell receptors (BCR) (30,31). Fc␥RIIB, however, were found to recruit selectively SHIP1, when coaggregated with high-affinity IgE receptors (Fc⑀RI) in mast cells (26,32). SHIP1, but not SHP-1, was subsequently demonstrated to be necessary for Fc␥RIIB-dependent inhibition of cell activation in SHIP1-deficient DT40 B cells (33) and in SHP-1-deficient mast cells derived from motheaten mice (26,32). These results altogether generated some confusion, and whether Fc␥RIIB indeed recruit SHP-1 in vivo remains unclear. Depending on the answer, the following two issues may be addressed: 1) if they do, do they recruit SHP-1 exclusively in B cells or also in other cell types, and 2) if they do not, how can one reconcile the apparent discordance between the in vitro binding of phosphatases to ITIM peptides and the in vivo recruitment of phosphatases by Fc␥RIIB.
In the present work, we aimed at clarifying these questions by analyzing the conditions required for SHP-1 to bind to phosphorylated ITIM-coated beads in vitro and to be recruited by phosphorylated Fc␥RIIB in mast cells and in B cells. We failed to detect SHP-1 recruitment by Fc␥RIIB in vivo when coaggregated either with Fc⑀RI in mast cells, or with BCR in B cells. We found that the in vitro binding of SHP-1 required a higher level of Fc␥RIIB phosphorylation than SHIP1 binding. Indeed, the two SH2 domains of SHP-1 were required to bind phosphorylated ITIMs and, as a consequence, SHP-1 binding, but not SHIP1 binding, depended on the density of phosphorylated ITIMs. In vivo, SHP-1 recruitment also required a higher level of Fc␥RIIB phosphorylation than SHIP1 recruitment. The level of Fc␥RIIB phosphorylation that enabled the recruitment of SHP-1 was reached after treating cells with pervanadate, but not following coaggregation of Fc␥RIIB with BCR or Fc⑀RI, in B cells and mast cells, respectively.
In Vitro Phosphorylation of GST-ITIM and GST-ICIIB1Ј and in Vitro Binding of Phosphatases-Ten l of glutathione-agarose beads coated with GST-ITIM or GST-ICIIB1Ј were washed in kinase buffer containing 100 mM Tris-HCl, pH 7.4, 125 mM MgCl 2 , 2 mM EGTA, 0.25 mM Na 3 VO 4 , and 2 mM dithiothreitol, and incubated for the indicated periods at 30°C with 20 l of kinase buffer containing 2 units of the Src kinase Lyn (Chemicon) and 100 M ATP. Kinase reaction was stopped on ice; beads were immediately washed in lysis buffer, and incubated for 2 h at 4°C with lysates from 1 ϫ 10 7 RBL-2H3 cells.
Cell Stimulation and Immunoprecipitation-RBL transfectants, resuspended at 5 ϫ 10 6 /ml, were incubated or not for 1 h at 37°C with IgE anti-DNP (culture supernatant diluted 1/10) and with 3 g/ml 2.4G2 F(abЈ) 2 , washed, and resuspended at 1 ϫ 10 7 cells/ml. Cells were challenged or not for 5 min or for the indicated periods of time at 37°C with 10 g/ml TNP-MAR F(abЈ) 2 . Unless otherwise specified, IIA1.6 and K46 transfectants, resuspended at 5 ϫ 10 7 /ml, were stimulated at 37°C for 5 min with 0.3 M RAM IgG (45 g/ml) or RAM F(abЈ) 2 (30 g/ml). BMMCs were sensitized as RBL-2H3 cells with the indicated dilutions of IgE anti-DNP, and resuspended at 1 ϫ 10 7 /ml. BMMCs were stimulated at 37°C for 5 min with preformed immune complexes made with the indicated final concentrations of DNP 25 -BSA and mAb mouse IgG1 anti-DNP. K46 transfectants, resuspended at 5 ϫ 10 7 /ml, were stimulated with preformed immune complexes made with the indicated final concentrations of NP 9 -BSA-TNP 12 and of polyclonal rabbit IgG anti-DNP. Immune complexes were preformed at 37°C for 10 min immediately before stimulation.
Pervanadate was generated by mixing 1 ml of 20 mM Na 3 VO 4 with 330 l of 30% H 2 O 2 followed by a 5-min incubation at room temperature, yielding a solution of 6 mM pervanadate. Unless otherwise specified, cells were incubated with a final concentration of 100 M pervanadate for 15 min at 37°C.
Following stimulation, cells were lysed as described above, and cell lysates were used for immunoprecipitation. Protein G-Sepharose (Amersham Pharmacia Biotech) was used to precipitate 2.4G2-bound Fc␥RIIB in lysates from RBL transfectants and 2.4G2-coated Sepharose beads were used to precipitate Fc␥RIIB in lysates from BMMCs or IIA1.6 and K46 transfectants.
Western Blot Analysis-Adsorbents were washed in lysis buffer and boiled for 3 min in sample buffer. Eluted material was fractionated by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore, Bedford, MA). In most experiments, immunoprecipitates were fractionated and transferred onto two membranes. These were used to assess: 1) tyrosyl phosphorylation and Fc␥RIIB or GST on one membrane, and 2) SHIP1 and SHP-1 phosphatases on the other membrane. Membranes were saturated with either 5% BSA (Sigma) or 5% skimmed milk (Régilait, Saint-Martin-Belle-Roche, France) diluted in 10 mM Tris buffer, pH 7.4, containing 0.5% Tween 20 (Merck, Schuchardt, Germany). Membranes Western blotted with HRP-conjugated anti-phosphotyrosine antibodies were stripped and reblotted with anti-Fc␥RIIB or anti-GST antibodies followed by HRP-conjugated GAR or GAM. In all experiments, the same membrane was used for blotting with anti-SHIP1 and anti-SHP-1 antibodies after having been cut into two pieces. The upper part, containing molecules with a molecular mass higher than 100 kDa was hybridized with anti-SHIP1, and the lower part, containing molecules with a molecular mass lower than 100 kDa was hybridized with anti-SHP-1 antibodies. mAb anti-SHP-1 was used for mast cells analysis while polyclonal antibodies was used for B cells analysis. They were revealed with HRP-conjugated GAM or GAR, respectively. Labeled antibodies were detected using the Amersham ECL kit.

RESULTS
Tyrosyl-phosphorylated Fc␥RIIB Recruit SHIP1 but Not SHP-1 Both in Mast Cells and in B Cells-Because phosphatases were reported to be differentially recruited by Fc␥RIIB, when coaggregated with BCR in B cells or with Fc⑀RI in mast cells, we compared the ability of Fc␥RIIB1 to recruit SHP-1 and SHIP1 in a rat mast cell line, RBL-2H3, and in two Fc␥RIIBdeficient murine B cell lines, IIA1.6 cells and K46. The three cell lines were stably transfected with murine Fc␥RIIB1. Fc␥RIIB1 were coaggregated with Fc⑀RI, that are constitutively expressed in RBL cells, by challenging transfectants, previously sensitized with mouse IgE anti-DNP and coated with F(abЈ) 2 fragments of the rat anti-Fc␥RIIB mAb 2.4G2, using TNP-MAR F(abЈ) 2 . Fc␥RIIB1 were coaggregated with BCR, that are constitutively expressed in IIA1.6 cells and K46 cells, using RAM IgG. Fc␥RIIB1 were immunoprecipitated, their phosphorylation was assessed by Western blotting with anti-phosphotyrosine antibodies, and phosphatases co-precipitated with Fc␥RIIB1 were examined by Western blotting with anti-SHP-1 and anti-SHIP1 antibodies. As previously observed, Fc␥RIIB1 became tyrosyl phosphorylated following coaggregation with Fc⑀RI or with BCR, in mast cells and in B cells, respectively. SHIP1, but not SHP-1, co-precipitated with tyrosyl-phosphorylated Fc␥RIIB1 in RBL cells, but also in IIA1.6 cells and K46 cells (Fig. 1A). Because the recruitment of SHIP1 and SHP-1 may not have identical kinetics, we repeated the experiment at various time points in RBL transfectants. The induced phosphorylation of Fc␥RIIB1 was constant over the time range studied, and comparable amounts of SHIP1 co-precipitated with phosphorylated Fc␥RIIB1. No co-precipitation of SHP-1 was detected between 15 s and 15 min (Fig.  1B). SHIP1, but not SHP-1, was therefore detectably recruited by tyrosyl-phosphorylated Fc␥RIIB1, and we found no differential in vivo recruitment of phosphatases in a mast cell line and in two B cell lines.
In Vitro Binding of SHP-1 Requires a Higher Phosphorylation Level of Fc␥RIIB ITIM Than Binding of SHIP1-Such a selective in vivo recruitment of phosphatases by tyrosyl-phosphorylated Fc␥RIIB1 was in marked contrast with the ability of phosphorylated synthetic peptides corresponding to the Fc␥RIIB ITIM to bind in vitro not only to the two SHIPs, but also to the two SHPs, when bound to agarose beads. In an attempt to understand this discrepancy, we reasoned that, among other differences between the two types of experiments, the proportion of Fc␥RIIB1 whose ITIM was phosphorylated in vivo was unknown, whereas all ITIM peptides used to coat agarose beads for in vitro experiments were phosphorylated. To evaluate the possibility that variations in the intensity of ITIM phosphorylation might differentially affect the affinity for SHIPs and for SHPs, we coated beads with a constant amount of ITIM peptides, in which the proportion of phosphorylated peptides (pITIM) varied from 100 to 0%. These were incubated with cell lysate from RBL-2H3 cells, and the binding of SHIP1 and SHP-1 was examined by Western blotting. As previously observed, no phosphatase precipitated with beads coated with nonphosphorylated ITIM (0% pITIM) and both phosphatases precipitated with beads coated with 100% pITIM. SHIP1 precipitation did not detectably decrease with the proportion of pITIM until beads were coated with less than 12% pITIM, and a detectable amount of phosphatase remained precipitated by beads coated with 6% pITIM. By contrast, SHP-1 precipitation progressively decreased with the proportion of pITIM coated to beads and it was not detected when beads were coated with less than 25% pITIM ( Fig. 2A). Beads coated with ITIM peptides, a small proportion of which were phosphorylated, therefore bound in vitro SHIP1 but not SHP-1, i.e. displayed the same selectivity for SHIP1 as tyrosyl-phosphorylated Fc␥RIIB in vivo.
Another difference between the two types of experiments is that intact receptors are used for in vivo recruitment whereas 13-amino acid peptides are used for in vitro binding. One must therefore consider the possibility that flanking sequences might affect the recruitment of phosphatases by the phosphorylated ITIM. To examine this possibility, we constructed GST Immunoprecipitated material was fractionated by SDS-PAGE, transferred onto Immobilon, and Western blotted with anti-Fc␥RIIB, anti-Tyr(P), anti-SHIP1, and anti-SHP-1 antibodies. Whole cell lysates (WCL) corresponding to 5 ϫ 10 5 cells were used as positive controls for Western blotting with anti-phosphatase antibodies. fusion proteins containing either the whole intracytoplasmic domain of the Fc␥RIIB1Ј isoform (GST-ICIIB1Ј) or only the same ITIM-containing 13 amino acids as synthetic peptides (GST-ITIM). Purified proteins were in vitro phosphorylated by purified Lyn for various periods of time, incubated in RBL lysate, and their ability to precipitate SHIP1 and SHP-1 was assessed by Western blotting. When submitted to in vitro kinase assay with Lyn, GST alone failed to be phosphorylated and recruited no phosphatase (data not shown). GST-ICIIB1Ј and GST-ITIM were phosphorylated by Lyn, and phosphorylation increased as a function of time. No phosphatase was precipitated by nonphosphorylated proteins. Both GST-ITIM and GST-ICIIB1Ј precipitated SHIP1 and SHP-1 when heavily phosphorylated, but SHIP1 only when less phosphorylated (Fig. 2B). A tyrosyl-phosphorylated intact Fc␥RIIB intracytoplasmic domain could therefore bind not only SHIP1 but also SHP-1 in vitro. SHP-1 binding, however, required a higher degree of phosphorylation than SHIP1 binding. The degree of ITIM phosphorylation seems therefore to determine the in vitro binding for SHP-1 rather than the presence or absence of sequences flanking the phosphorylated Fc␥RIIB ITIM.
The Density of Phosphorylated ITIM Critically Determines the Cooperative Binding of the Two SHP-1 SH2 Domains-A major difference between SHIP1 and SHP-1 is that the phosphatidylinositol phosphatase contains one SH2 domain whereas the protein tyrosine phosphatase contains two tandem SH2 domains. We investigated the possible role of this difference in the differential binding of these phosphatases to pITIMcoated beads. Agarose beads were coated with increasing concentrations of Fc␥RIIB pITIM and incubated either with RBL cell lysates or with purified SH2 domain-containing GST fusion proteins. These were the SH2 domain of SHIP1 (GST-SH2 SHIP1), the two SH2 domains of SHP-1 (GST-SH2 (NϩC) SHP-1), the N-terminal SH2 domain of SHP-1 (GST-SH2 (N) SHP-1), or the C-terminal SH2 domain of SHP-1 (GST-SH2 (C) SHP-1). Phosphatases precipitated from cell lysates were identified by Western blotting with anti-SHIP1 and anti-SHP-1 antibodies. GST fusion proteins precipitated were identified by Western blotting with anti-GST antibodies. SHIP1 precipitation was detected with beads coated with as little as 0.2 nmol of pITIM and slowly increased with the amount of pITIM coated to beads. SHP-1 precipitation became detectable with beads coated with 0.4 nmol of pITIM and rapidly increased with the amount of pITIM. Parallel variations in binding were observed for GST-SH2 SHIP and GST-SH2 (NϩC) SHP-1. No binding of GST-SH2 (N) SHP-1 was detectable, whatever the amount of pITIM coated to beads, and a faint binding of GST-SH2 (C) SHP-1 was observed for beads coated with high amounts of pITIM only (Fig. 3A). The differential binding of phosphatases, in a cell lysate, can therefore be reproduced using GST-SH2 fusion proteins. The single SHIP1 SH2 domain bound readily to ITIM-coated beads, but not isolated SHP-1 SH2 domains. The two SHP-1 SH2 domains, however, bound with a similar pattern as SHP-1 in a cell lysate, indicating the requirement for a cooperative binding between the two SHP-1 SH2 domains.
When using increasing amounts of pITIM to coat beads, one increases not only the quantity of pITIM bound to beads, but also the density of pITIM on beads. To investigate whether pITIM density may determine the binding of the two SH2 domain-containing SHP-1, the same amounts of pITIM were used to coat different amounts of beads. These were incubated in RBL cell lysate, and phosphatase binding was then examined as above. For a given amount of pITIM, the binding of SHIP1 did not vary with the amount of beads. By contrast, for

FIG. 3. Cooperative binding of the two SH2 domains of SHP-1 and effect of the density of pITIMs on beads.
A, binding of SHIP1 and SHP-1 SH2 domains to beads coated with variable amounts of pITIM. 12.5-ml agarose beads were coated with increasing amounts of pITIM. pITIM-coated beads were incubated in the same experiment either with RBL-2H3 cell lysate or with SH2 domain-containing GST fusion proteins. These were the SH2 domain of SHIP1 (GST-SH2 SHIP), the two SH2 domains of SHP-1 (GST-SH2 (NϩC) SHP-1), the N-terminal SH2 domain of SHP-1 (GST-SH2 (N) SHP-1), or the C-terminal SH2 domain of SHP-1 (GST-SH2 (C) SHP-1). Precipitated material was fractionated by SDS-PAGE and transferred onto Immobilon. Precipitates from cell lysates and whole cell lysate (WCL) were Western blotted with anti-SHIP1 and anti-SHP-1. Precipitates from fusion proteins were Western blotted with anti-GST antibodies. Aliquots of GST fusion proteins used for incubation were electrophoresed and Western blotted with anti-GST antibodies to control the relative amounts of fusion proteins. B, binding of SHIP1 and SHP-1 to beads coated with Fc␥RIIB pITIM at variable densities. 12.5 ml or 50 ml of agarose beads were incubated with the same amounts of pITIM. Beads were incubated with RBL-2H3 cell lysates corresponding to 1 ϫ 10 7 cells. Precipitated material and WCL were fractionated by SDS-PAGE, transferred onto Immobilon, and Western blotted with anti-SHIP1 and anti-SHP-1 antibodies. a given amount of pITIM, the binding of SHP-1 dramatically decreased when the amount of beads increased (Fig. 3B). SHP-1 binding, but not SHIP1 binding, therefore depends on the density of pITIM coated to beads.

Increasing Fc␥RIIB Phosphorylation by Increasing the Concentration of Immune Complexes Does Not Enable SHP-1 to Be
Recruited-Based on the above results, we wondered 1) whether the intensity of Fc␥RIIB phosphorylation would increase with the concentration of extracellular ligands used to coaggregate Fc␥RIIB with Fc⑀RI in mast cells, or with BCR in B cells, and 2) whether phosphorylation levels induced under these conditions might enable Fc␥RIIB to recruit SHP-1 in vivo.
We first tried to increase Fc␥RIIB phosphorylation by stimulating Fc␥RIIB1-transfected IIA1.6 cells with increasing concentrations of RAM IgG and, as negative controls, with the same molar concentrations of RAM F(abЈ) 2 . Fc␥RIIB1 were immunoprecipitated, their phosphorylation was assessed by Western blotting with anti-phosphotyrosine antibodies, and co-precipitated phosphatases were examined by Western blotting with anti-phosphatase antibodies. Fc␥RIIB1 phosphorylation induced by a RAM IgG concentration as high as 2.4 M (360 g/ml) was not higher than Fc␥RIIB1 phosphorylation induced by 0.3 M (45 g/ml) RAM IgG. Comparable amounts of SHIP1 co-precipitated with phosphorylated Fc␥RIIB1, whatever the concentration of RAM IgG, but no detectable SHP-1 (Fig. 4).
This negative result led us to use IgG immune complexes, which are the natural ligands of Fc␥RIIB, to coaggregate these receptors with ITAM-bearing receptors in mast cells and B cells. Fc⑀RI and Fc␥RIIB expressed constitutively in BMMCs (42) were coaggregated by challenging cells, sensitized with increasing concentrations of mouse anti-DNP IgE, with immune complexes made with increasing concentrations of DNP-BSA and of a monoclonal mouse IgG 1 anti-DNP antibody. Fc␥RIIB phosphorylation increased with the concentration of IgE used for sensitization and with the concentration of IgG in immune complexes. It also varied with the concentration of DNP-BSA in immune complexes, and optimal concentrations depended on the concentration of IgG antibodies (Fig. 5A). Likewise, Fc␥RIIB were coaggregated with BCR in Fc␥RIIB1transfected K46 B cells, which express an anti-NP BCR, by incubating cells with immune complexes made of increasing concentrations of polyclonal rabbit IgG anti-DNP and of NP-BSA-TNP. Fc␥RIIB1 phosphorylation varied with the concentration of IgG antibody and antigen: it peaked with higher concentrations of antigen as the antibody concentrations increased. Immune complexes that induced the highest Fc␥RIIB1 phosphorylation were made of 10 g/ml IgG and 1 g/ml NP-BSA-TNP (Fig. 5B).
The co-precipitation of phosphatases was next examined in BMMCs and in Fc␥RIIB1-transfected K46 cells stimulated with concentrations of antigen and antibodies in the range of those which induced maximal Fc␥RIIB phosphorylation. For comparison, K46 transfectants were also stimulated with RAM IgG or F(abЈ) 2 under the same conditions as in Fig. 1. In both cell types, Fc␥RIIB were markedly phosphorylated following stimulation with immune complexes. Fc␥RIIB phosphorylation was of comparable magnitude in K46 cells stimulated with immune complexes or with RAM IgG. SHIP1 co-precipitated with phosphorylated Fc␥RIIB both in mast cells and in B cells, but not SHP-1 (Fig. 5C). The above results altogether indicate that even when using concentrations of ligands that were optimal for Fc␥RIIB phosphorylation, SHIP1 but not SHP-1 co-precipitation was detectable, both in mast cells and in B cells.
Fc␥RIIB Phosphorylation following Pervanadate Treatment Enables SHP-1 Recruitment-A possible reason explaining the absence of detectable recruitment of SHP-1 by Fc␥RIIB phosphorylated following stimulation with high concentrations of extracellular ligands was that Fc␥RIIB phosphorylation levels reached under these conditions were not high enough. We therefore compared the effect of coaggregating Fc␥RIIB with Fc⑀RI, in RBL transfectants, or with BCR, in IIA1.6 and K46 transfectants, and of treating the same three cells with pervanadate.
Fc␥RIIB1Ј-expressing RBL cells were sensitized with mouse IgE anti-DNP and incubated with 2.4G2 F(abЈ) 2 , treated or not treated with pervanadate and challenged or not with TNP-MAR-F(abЈ) 2 . Likewise, Fc␥RIIB1-expressing IIA1.6 and K46 cells were treated or not with pervanadate and challenged with RAM F(abЈ) 2 or IgG. Fc␥RIIB phosphorylation and phosphatase recruitment were assessed as in previous experiments. As expected, Fc␥RIIB phosphorylation was induced by coaggregating Fc␥RIIB1Ј with Fc⑀RI in RBL transfectants, and by coaggregating Fc␥RIIB1 with BCR in IIA1.6 and K46 transfectants. In all three cells, pervanadate treatment alone induced a much higher level of Fc␥RIIB phosphorylation that did not further increase by coaggregating Fc␥RIIB with Fc⑀RI or with BCR. SHIP1 co-precipitated with Fc␥RIIB phosphorylated following their coaggregation with Fc⑀RI or with BCR. SHIP1 co-precipitated also with Fc␥RIIB phosphorylated following treatment of cells with pervanadate (in higher amounts than following coaggregation of Fc␥RIIB with Fc⑀RI, in RBL cells, or with BCR, in K46 cells). SHP-1 did not co-precipitate with Fc␥RIIB phosphorylated following their coaggregation with FIG. 4. Absence of co-precipitation of SHP-1 with Fc␥RIIB1 in IIA1.6 transfectants stimulated with increasing concentrations of RAM IgG. 6.5 ϫ 10 7 IIA1.6-Fc␥RIIB1 cells were stimulated with increasing concentrations of RAM F(abЈ)2 or IgG. Cells were lysed and Fc␥RIIB were immunoprecipitated. Immunoprecipitated material and whole cell lysate (WCL) were fractionated by SDS-PAGE, transferred onto Immobilon, and Western blotted with anti-Fc␥RIIB, anti-Tyr(P), anti-SHIP1, and anti-SHP-1 antibodies.
Fc⑀RI or BCR. SHP-1, however, co-precipitated with Fc␥RIIB phosphorylated following pervanadate treatment in all three cells (Fig. 6A). Treating Fc␥RIIB1-expressing RBL transfectants with decreasing concentrations of pervanadate induced a dose-dependent tyrosyl phosphorylation of Fc␥RIIB1. Interestingly, as Fc␥RIIB1 phosphorylation decreased, the co-precipitation of SHP-1 was lost before that of SHIP1 (Fig. 6B). Treating cells with pervanadate, but not coaggregating Fc␥RIIB with ITAM-bearing receptors, could therefore induce a phosphorylation of Fc␥RIIB that was high enough to enable the recruitment of SHP-1. DISCUSSION We show here that murine Fc␥RIIB recruit the inositol 5-phosphatase SHIP1, but not the protein-tyrosine phosphatase SHP-1 in vivo, although the Fc␥RIIB ITIM has an affinity for both phosphatases in vitro, because the binding of SHP-1 requires a higher degree of Fc␥RIIB phosphorylation than the binding of SHIP1. The same phosphorylation-dependent preference for SHIP1 was observed 1) in vitro using beads coated with suboptimal concentrations of pITIM, 2) in vitro using GST-ICIIB1Ј phosphorylated by Lyn for short periods of time, 3) in vivo using phosphorylated Fc␥RIIB precipitated from cells treated with low concentrations of pervanadate, and 4) in vivo, when Fc␥RIIB was phosphorylated following coaggregation with BCR or Fc⑀RI, in B cells and in mast cells, respectively. Our results suggest that, depending on their level of phosphorylation, Fc␥RIIB could potentially use the two phosphatases, with different consequences.
Evidence that, when tyrosyl phosphorylated, the Fc␥RIIB ITIM has an affinity for SH2 domain-containing phosphatases was first provided in 1995 by D'Ambrosio et al. (30) who demonstrated that phosphorylated synthetic peptides containing the Fc␥RIIB ITIM precipitated several molecular species from [ 35 S]methionine-labeled cell lysates, one of which was identified as SHP-1. Other molecules precipitated by these peptides were subsequently shown to be SHP-2 (28) and SHIP1 (26). Similar experiments confirmed D'Ambrosio's results (28,32). SHIP2, a second SH2 domain-containing inositol phosphatase BMMCs sensitized with mouse IgE anti-DNP supernatant (diluted 1/10) were stimulated or not with immune complexes made with 5 g/ml DNP-BSA and 50 g/ml mouse IgG 1 anti-DNP (final concentrations). K46-Fc␥RIIB1 cells were stimulated for 5 min with 10 g/ml rabbit IgG anti-DNP or with immune complexes made with 1 g/ml NP-BSA-TNP and 10 g/ml rabbit IgG anti-DNP (final concentrations). For comparison, K46 transfectants were also stimulated with 0.3 M RAM IgG or F(abЈ) 2 . BMMCs and K46 transfectants were lysed after stimulation, and Fc␥RIIB were immunoprecipitated. Immunoprecipitated materials and whole cell lysate from K46 transfectants (WCL) were fractionated by SDS-PAGE, transferred onto Immobilon, and Western blotted with anti-Fc␥RIIB, anti-Tyr(P) (A, B, and C), anti-SHIP1 and anti-SHP-1 antibodies (C).
was recently found to bind also to phosphorylated Fc␥RIIB ITIM (27,29). It follows that phosphorylated Fc␥RIIB ITIM peptides can bind all four known SH2 domain-containing phosphatases in vitro.
In the same 1995 paper, D'Ambrosio et al. (30) reported that SHP-1 co-precipitated with Fc␥RIIB bearing an intact ITIM, following coaggregation with BCR in A20 and in IIA1.6 B cells reconstituted with Fc␥RIIB, and that Fc␥RIIB-dependent inhibition of B cell proliferation was impaired in B cells from SHP-1-deficient motheaten mice. In 1996, Ono et al. (26) reported that SHIP1 co-precipitated with Fc␥RIIB following coaggregation with Fc⑀RI in BMMCs or with BCR in A20 cells, and that Fc␥RIIB-dependent inhibition of IgE-induced serotonin release was unaffected in BMMCs derived from motheaten mice. Fong et al. (32) reported that SHIP1, but not SHP-1 or SHP-2, co-precipitated with Fc␥RIIB following coaggregation with Fc⑀RI in BMMCs. In 1997, Ono et al. (33) showed that Fc␥RIIB-dependent inhibition of Ca 2ϩ responses and of NF-AT activity was abolished in SHIP1-deficient, but not in SHP-1deficient, DT40 chicken B cells, and that SHIP1, but not SHP-1, was detectably co-precipitated with Fc␥RIIB following coaggregation with BCR in A20 cells. In 1998, however, Sato et al. (31) observed the co-precipitation of both SHIP1 and SHP-1 with Fc␥RIIB in A20 cells expressing an anti-TNP BCR following coaggregation with intact anti-idiotypic antibodies. Contrasting with the consensus that Fc␥RIIB recruit SHIP1 both in B cells and in mast cells, their ability to recruit SHP-1 in vivo therefore remains controversial.
That SHP-1 was found by two groups to co-precipitate with Fc␥RIIB in B cells, but not in mast cells, suggested the possibility that some discrepancies might be accounted for by a cell type-specific differential in vivo recruitment. To address this issue, we examined the co-precipitation of SHP-1 with recombinant Fc␥RIIB1 stably expressed by transfecting the same cDNA into the rat mast cells RBL-2H3, and into the two Fc␥RIIB-deficient mouse lymphoma B cells IIA1.6 and K46. The coaggregation of Fc␥RIIB1 with Fc⑀RI or with BCR in-duced a comparable tyrosyl phosphorylation of Fc␥RIIB1 and the co-precipitation of SHIP1, but not of SHP-1, in all three cells. The same result, observed at 5 min in the three cell lines, was also observed between 15 s and 15 min in RBL cells. Failure to detect SHP-1 co-precipitation cannot be accounted for an insufficient sensitivity of Western blotting because traces of SHP-1 could be seen on overexposed films, but in equal amounts in unstimulated and in stimulated cells (data not shown). Due to experimental conditions inherent to the co-precipitation technique, however, we cannot exclude that SHP-1, possibly recruited by Fc␥RIIB in vivo was lost. Whatever the explanation, we observed no difference between the three cells examined in which Fc␥RIIB preferentially, if not exclusively, recruited SHIP1. There is therefore a discrepancy between the ability of SHP-1 and SHIP1 to bind in vitro to Fc␥RIIB pITIM peptides and to co-precipitate with phosphorylated Fc␥RIIB in vivo.
Due to the different experimental conditions used for the two assay systems many differences can possibly explain this discrepancy. One difference could bear on the level of ITIM phosphorylation. Indeed, all ITIMs are phosphorylated on beads used for in vitro binding assay whereas an unknown proportion of Fc␥RIIB are phosphorylated following coaggregation with ITAM-bearing receptors. To explore the possible role of quantitative differences in ITIM phosphorylation, we studied the binding of SHIP1 and SHP-1 to beads coated with Fc␥RIIB ITIMs phosphorylated in varying proportions and incubated in a cell lysate. We found that SHP-1 binding decreased more sharply with the proportion of pITIM than SHIP1 binding, so that beads coated with 12% pITIM bound selectively SHIP1. Comparable results were obtained when incubating beads coated with increasing concentrations of pITIM with GST fusion proteins containing the SH2 domain of SHIP1 or the two SH2 domains of SHP-1. The use of SH2 domains permitted comparison between the binding of the two molecules using the same anti-GST antibodies for blotting. It also excluded that phosphatase binding was mediated by unknown intermediates FIG. 6. Co-precipitation of SHP-1 with Fc␥RIIB phosphorylated following pervanadate treatment. A, co-precipitation of phosphatases with Fc␥RIIB1 in mast cell and B cell transfectants. 6.5 ϫ 10 7 Fc␥RIIB1Ј-expressing RBL cells were sensitized with mouse IgE anti-DNP and incubated with 2.4G2 F(abЈ) 2 before they were treated or not with 100 M pervanadate for 15 min. During the last 5 min of pervanadate treatment, cells were stimulated (coaggregation ϩ) or not (coaggregation Ϫ) with TNP-MAR F(abЈ) 2 . 6.5 ϫ 10 7 IIA1.6-Fc␥RIIB1 or K46-Fc␥RIIB1 transfectants were treated or not with pervanadate and stimulated (coaggregation ϩ) or not (coaggregation Ϫ) with RAM IgG, under the same conditions as RBL cells. B, dose-dependent co-precipitation of phosphatases with Fc␥RIIB1 in mast cell transfectants treated with pervanadate. 9.5 ϫ 10 7 Fc␥RIIB1-expressing RBL cells were treated or not with the indicated concentrations of pervanadate for 15 min. Following stimulation, cells were lysed, and Fc␥RIIB were immunoprecipitated. Immunoprecipitated material from RBL, IIA1.6, and K46 transfectants were fractionated by SDS-PAGE, transferred onto Immobilon, and Western blotted with anti-Fc␥RIIB, anti-Tyr(P), anti-SHIP1, and anti-SHP-1 antibodies. present in cell lysates. Supporting these results, the Fc␥RIIB pITIM was reported to have a higher affinity for the SH2 domain of SHIP1 than for the two SH2 domains of SHP-1, when measured by Biacore analysis (43). Results obtained in these two sets of experiments are reminiscent of the selective in vivo recruitment of SHIP1 by Fc␥RIIB.
Based on data previously reported by others (43,44), the binding of SHP-1 is likely to involve the two tandem SH2 domains of this phosphatase. Compared with GST fusion proteins containing the two SH2 domains of SHP-1, no GST fusion proteins containing the N-terminal domain of SHP-1 and minute amounts of GST fusion proteins containing the C-terminal SH2 domain of SHP-1 bound to pITIM-coated beads, whatever the concentration of peptides on beads. This suggests that, since there is one tyrosine only in pITIM, GST fusion proteins containing the two SHP-1 SH2 domains bound to adjacent pITIMs on the same bead. The same holds for the binding of SHP-1 when incubating pITIM-coated beads with a cell lysate. If so, variations in the affinity of SHP-1 to beads coated with increasing amounts of pITIM might depend on the density of peptides coated to beads. To examine this possibility, we used a constant amount of pITIM to coat variable numbers of beads that were incubated in a cell lysate, and we compared the ability of these beads to bind SHIP1 and SHP-1. The binding of SHIP1 was proportional to the amount of pITIM on beads, and did not vary with the pITIM density. By contrast, the binding of SHP-1 depended not only on the amount of pITIM but also, critically, on the density of pITIM bound to beads. The in vitro binding of SHP-1 therefore requires a cooperative binding of its two SH2 domains to two adjacent pITIMs in trans, and this feature explains that pITIMs need to be closer to each other for enabling the binding of SHP-1 than for enabling the binding of SHIP1.
Another difference that might explain the discrepancy between the in vitro binding and the in vivo recruitment of SHP-1 is that isolated ITIMs are used in vitro whereas whole receptors are used in vivo. One cannot exclude that the recruitment of SHP-1 might be hampered by non-ITIM sequences or by molecules that could possibly bind to these sequences. Supporting this possibility, the N-terminal KIR2DL3 ITIM that could recruit SHP-2 in vivo, when kept in its original context, failed to recruit this phosphatase, when transposed in the intracytoplasmic domain of Fc␥RIIB1 (29). To answer this question, we compared the ability of GST fusion proteins containing the Fc␥RIIB ITIM only or the whole intracytoplasmic domain of Fc␥RIIB1Ј to bind SHIP1 and SHP-1, when incubated with a cell lysate, following their phosphorylation with Lyn for various periods of time. No difference was observed between the two fusion proteins and, like the isolated ITIM, the intracytoplasmic domain of Fc␥RIIB could bind SHP-1 when high enough phosphorylated.
Based on the latter result, we searched for experimental conditions that would induce a Fc␥RIIB phosphorylation sufficient to enable them to recruit SHP-1 in vivo. To this aim, we used several extracellular ligands including IgG immune complexes that are the physiological ligands of Fc␥RIIB, at various concentrations, in mast cells and in B cells. We found that indeed, the phosphorylation of Fc␥RIIB varied with the concentrations of antigen and antibody in immune complexes but that ligands which induced a maximal phosphorylation of Fc␥RIIB readily induced the recruitment of SHIP1 but failed to induce a detectable recruitment of SHP-1. Based on our results of in vitro binding with pITIM-coated beads, this suggests that a small proportion (less than 12%?) of Fc␥RIIB become tyrosyl phosphorylated in vivo upon coaggregation with ITAM-bearing receptors by physiological ligands. If so, we wondered whether Fc␥RIIB phosphorylation would reach a level high enough to enable the recruitment of SHP-1 following treatment of cells with pervanadate.
In both mast cells and B cells, pervanadate treatment indeed induced a higher degree of Fc␥RIIB phosphorylation than coaggregation with Fc⑀RI or BCR, respectively, and under these conditions, not only SHIP1 but also SHP-1 co-precipitated with phosphorylated Fc␥RIIB. This observation indicates that, in resting cells, Fc␥RIIB are tyrosyl phosphorylated but that protein-tyrosine phosphatases maintain this phosphorylation below the detection level. This implies that Fc␥RIIB are the substrates of both protein-tyrosine kinases and phosphatases and that, under resting conditions, the effect of phosphatases is dominant over that of kinases. Fc␥RIIB phosphorylation observed following their coaggregation with ITAM-bearing receptors results from the additional effect of a Src kinase, brought by activating receptors (13), leading to a displacement of the balance so that the effect of kinases becomes dominant over that of phosphatases. It should be emphasized that the higher intensity of Fc␥RIIB phosphorylation induced by pervanadate, compared with phosphorylation induced by coaggregation, may be due to the phosphorylation of a higher number of receptors and/or to the phosphorylation of a higher number of tyrosine residues in each receptor. The recruitment of SHP-1 by Fc␥RIIB phosphorylated after pervanadate treatment may indeed simply be explained by a quantitatively different phosphorylation, resulting in an increased density of phosphorylated ITIMs that might permit the binding in trans of the two SHP-1 SH2 domains. Supporting this interpretation, the recruitment of SHP-1 was lost before that of SHIP1 when pervanadate-induced Fc␥RIIB phosphorylation decreased following treatment of cells with decreasing concentrations of pervanadate. Alternatively, the recruitment of SHP-1 after pervanadate treatment may be explained by a qualitatively different phosphorylation of Fc␥RIIB, enabling SHP-1 to be recruited through the binding in cis of its two SH2 domains to two tyrosines borne by the same receptor. The intracytoplasmic domain of Fc␥RIIB1 contains four tyrosine residues. Supporting this possibility, the recruitment of SHP-1 by KIR2DL3 was found to require the conservation of its two ITIMs (39) and all ITIM-bearing receptors that were shown to recruit SHPs in vivo bear more than one ITIM. Finally, SHP-1 may co-precipitate with Fc␥RIIB phosphorylated following pervanadate treatment because, under these conditions, the enzymatic activity of the phosphatase is inhibited. Several ITIM-bearing receptors were shown both to recruit and to be the substrates of SHP-1 (45) or SHP-2 (46,47). If recruited by phosphorylated Fc␥RIIB under physiological conditions, SHP-1 might thus decrease Fc␥RIIB phosphorylation, thereby giving an advantage for the recruitment of SHIP over that of SHP-1.
The latter interpretation of the effect of pervanadate may explain our failure to co-precipitate SHP-1 with Fc␥RIIB phosphorylated upon coaggregation with ITAM bearing receptors. This hypothesis would also endow Fc␥RIIB with additional regulatory properties. These receptors could indeed transiently recruit SHP-1 which could dephosphorylate not only ITIMs, but also ITAMs and other signaling molecules whose phosphorylation is critical for positive signaling. This would have important consequences. By displacing the balance between kinases and phosphatases recruited to the receptor complex in favor of the latter, it would increase the signaling threshold and/or dampen activation signals. Fc␥RIIB might thus use two different mechanisms, i.e. SHIP-mediated and SHP-mediated, to adjust negative regulation to the intensity of extracellular signals. By decreasing positive signals, SHP-1 would in turn decrease ITIM phosphorylation bringing Fc␥RIIB back to con-ditions under which they recruit SHIPs. It remains to be determined whether Fc␥RIIB could be phosphorylated enough to recruit SHP-1 under physiological or pathological situations such as diseases associated with exaggerated antibody responses or immune complexes.