Fer and Fps/Fes Participate in a Lyn-dependent Pathway from FcϵRI to Platelet-Endothelial Cell Adhesion Molecule 1 to Limit Mast Cell Activation*

Mast cells express the high affinity IgE receptor FcϵRI, which upon aggregation by multivalent antigens elicits signals that cause rapid changes within the mast cell and in the surrounding tissue. We previously showed that FcϵRI aggregation caused a rapid increase in phosphorylation of both Fer and Fps/Fes kinases in bone marrow-derived mast cells. In this study, we report that FcϵRI aggregation leads to increased Fer/Fps kinase activities and that Fer phosphorylation downstream of FcϵRI is independent of Syk, Fyn, and Gab2 but requires Lyn. Activated Fer/Fps readily phosphorylate the C terminus of platelet-endothelial cell adhesion molecule 1 (Pecam-1) on immunoreceptor tyrosine-based inhibitory motifs (ITIMs) and a non-ITIM residue (Tyr700) in vitro and in transfected cells. Mast cells devoid of Fer/Fps kinase activities display a reduction in FcϵRI aggregation-induced tyrosine phosphorylation of Pecam-1, with no defects in recruitment of Shp1/Shp2 phosphatases observed. Lyn-deficient mast cells display a dramatic reduction in Pecam-1 phosphorylation at Tyr685 and a complete loss of Shp2 recruitment, suggesting a role as an initiator kinase for Pecam-1. Consistent with previous studies of Pecam-1-deficient mast cells, we observe an exaggerated degranulation response in mast cells lacking Fer/Fps kinases at low antigen dosages. Thus, Lyn and Fer/Fps kinases cooperate to phosphorylate Pecam-1 and activate Shp1/Shp2 phosphatases that function in part to limit mast cell activation.

are present in both the ␤ and ␥ subunits and serve as docking sites for the recruitment of signaling molecules (2). Fc⑀RI signaling is initiated by binding of IgE, which is sufficient for induction of survival pathways as well as cytokine production (3,4). Some highly cytokinergic IgEs can induce antigen-independent degranulation, survival, adhesion, and chemotaxis of mast cells (5). However, in most cases, aggregation of Fc⑀RI by multivalent antigens is required for a full mast cell response including degranulation, lipid mediator release, increased cell adhesion, and increased motility (6). The Src family protein-tyrosine kinase (PTK) Lyn is constitutively associated with Fc⑀RI (7), and upon antigen-mediated clustering of receptor chains, Lyn phosphorylates ITAMs on ␤and ␥-chains. The ␤-chain ITAMs are thought to recruit additional Lyn and Fyn kinases, the p85 subunit of phosphatidylinositol 3-kinase, SH2-containing inositol 5Ј-phosphatase, and Shp2 phosphatase (8). Phosphorylated ITAMs on the ␥-chains recruit the dual SH2 domain-containing PTK Syk (9). Syk activity is essential for signal transduction downstream of Fc⑀RI, because Syk-deficient mast cells fail to degranulate, synthesize leukotrienes, and secrete cytokines following antigen challenge (10). Although Fc⑀RI-induced tyrosine phosphorylation is greatly reduced in Syk-deficient mast cells, phosphorylation of the receptor ITAMs and Lyn are maintained (10). Lyn-deficient mast cells, despite severely reduced tyrosine phosphorylation and delayed calcium flux, are able to degranulate and secrete cytokines (11). In fact, Lyn-deficient mast cells release more of the granule constituent ␤-hexosaminidase than do wild type mast cells. Further studies have shown that multiple responses to Fc⑀RI aggregation are delayed in Lyn-deficient mast cells, including tyrosine phosphorylation of receptor subunits, calcium flux, and phosphatidylinositol 3,4,5-trisphosphate production but persist far longer than in wild type mast cells (12). Other notable characteristics of Lyn-deficient mast cells are that Fyn kinase is hyperactivate, whereas SH2-containing inositol 5-phosphatase is completely inactive (12). Thus, in addition to initiating signaling downstream of Fc⑀RI, Lyn is also involved in signal termination at least partly through activation of SHZ-containing inositol 5-phosphatase, which hydrolyzes phosphatidylinositol 3,4,5-trisphosphate and thereby reduces the plasma membrane localization of pleckstrin homology domain-containing proteins. The pleckstrin homology domain-containing adaptor protein Gab2 (Grb2-associated binding-2), is required for phosphatidylinositol 3-kinase activation (13) and, together with Fyn, contributes to activation of RhoA, microtubule formation, and delivery of granules to the plasma membrane (14).
Fc⑀RI aggregation also leads to phosphorylation of plateletendothelial cell adhesion molecule (Pecam-1) on ITIMs that recruit Shp1 and Shp2 phosphatases (15). Both Pecam-1 and Shp1 have been shown to negatively regulate Fc⑀RI-triggered degranulation, although the mechanism has not been established (16,17). Pecam-1 knock-out mice display hypersensitivity to challenge with lipopolysaccharide (or endotoxin) (18 -20). This has been attributed to impaired signal transducer and activator of transcription 3 (STAT3) phosphorylation in endothelial cells and lymphocytes and elevated production of inflammatory cytokines in Pecam-deficient mice. Phosphorylation of a non-ITIM tyrosine (Tyr 701 ) in human Pecam-1 allows recruitment of STAT3 or STAT5 via their SH2 domains (19). While tethered to Pecam-1, STAT3 phosphorylation is likely regulated by a Pecam-1-associated kinase or Shp2 phosphatase (21).
We recently showed that Fps and Fer kinases are both activated within 1 min of Fc⑀RI aggregation (22). Although the mechanism by which Fc⑀RI aggregation leads to Fps and Fer activation is unknown, one or more of Lyn, Fyn, and Syk are likely required, because these are the earliest signaling molecules activated upon Fc⑀RI aggregation. Here, we report that Fc⑀RI aggregation leads to elevated Fer and Fps kinase activities. Phosphorylation of Fer and Fps downstream of Fc⑀RI is independent of Syk, Fyn, and Gab2 but requires Lyn kinase for rapid activation. Once activated, Fer and Fps can phosphorylate both ITIMs and a non-ITIM residue (Tyr 700 ) in Pecam-1. In mast cells devoid of Fer/Fps kinase activities, the overall phosphorylation of Pecam-1 is reduced, whereas recruitment of Shp1/Shp2 is unaffected. In contrast, Lyn-deficient mast cells display a significant defect in Pecam-1 phosphorylation and Shp2 recruitment. Similar to previous studies on Pecam-1-and Shp1-deficient mast cells, we observe a hyperdegranulation response in Fer/Fps-deficient mast cells at low antigen dosages. Thus, a Lyn/Fer/Fps pathway from Fc⑀RI to Pecam-1/Shp1 functions in limiting mast cell activation.
Pecam-1 Expression Plasmids-Bacterial expression constructs were generated using glutathione S-transferase (GST) fusions to the cytoplasmic tail of Pecam-1 by amplification of the desired sequences from either plasmid DNAs or from cDNA pools from BMMCs generated by reverse transcription-PCR. The full-length Pecam-1 C-tail (with and without ITIM mutations) was amplified from mouse Pecam-1 expression plasmids provided by Andre Veillette (Institut de Recherches Cliniques, Montreal, Canada) (28). The ⌬14/15 isoform was derived from an IMAGE clone (Open Biosystems), and the ⌬15 isoform was cloned from BMMC cDNA. GST fusion proteins were purified from transformed BL21 using glutathione-conjugated beads according to the manufacturer's instructions (Amersham Biosciences).
In Vitro Kinase Assays-The in vitro kinase assays using [␥-32 P]ATP (PerkinElmer Life Sciences) were performed on anti-DNP-IgE-sensitized BMMCs (WT, DR, and KR/DR) treated with or without DNP-HSA in kinase reaction buffer, as previously described (27). To assess the preferred sites of Pecam-1 tyrosine phosphorylation by Fer and Fps, we performed similar in vitro kinase assays, but with unlabeled ATP (1 M), and analyzed them by Western blotting with anti-pY.
Transfection of Pecam-1 and Fer Plasmids-COS-7 cells were grown on 60-mm plates and transfected with the indicated combinations of mouse Pecam-1 and Myc-tagged Fer (WT and KR) expression plasmids using Lipofectamine (Invitrogen). SCLs were prepared after 48 h of transfection and subjected to IPs, followed by SDS-PAGE and immunoblotting as indicated.
BMMC Degranulation Assay-Degranulation assays using annexin V-reactivity of cells that have undergone fusion with exocytic vesicles were performed as described (29,30). Briefly, WT and KR/DR BMMCs were sensitized with anti-TNP-IgE (1 g/ml; BD Biosciences) for 18 h and washed once in warm Tyrode's buffer, and 5 ϫ 10 5 cells were stimulated in the absence and presence of DNP-HSA (0.1, 1, 10, and 100 ng/ml) for 15 min at 37°C. BMMCs were washed once prior to staining with propidium iodide (2 g/ml) and FITC-conjugated annexin V (5 l/sample). The percentage of annexin V-positive/propidium iodide-negative cells and mean fluorescence intensity (of annexin V staining) were analyzed on an EPICS Altra HSS flow cytometer (Beckman Coulter).

Increased Activities of Fer and Fps Kinases upon Aggregation of Fc⑀RI on
Mast Cells-We previously reported that Fc⑀RI aggregation induces phosphorylation of Fer and Fps kinases in mast cells (22). The fact that kinase-defective mutants of Fps and Fer also become phosphorylated upon Fc⑀RI aggregation (24) suggests that Fer and Fps are substrates for an upstream kinase. To address potential effects of Fc⑀RI aggregation-induced phosphorylation on Fer and Fps kinase activities, we cultured BMMCs from WT, fer DR/DR (DR), and fps KR/KR fer DR/DR (KR/DR) mice, sensitized with anti-DNP-IgE, and treated them with or without antigen (DNP-HSA). Soluble cell lysates were subjected to IP with anti-Fps/Fer antisera (which recognizes both proteins (27)), and in vitro kinase assays were per- formed (Fig. 1A). The lysates from either untreated or DNP-HSAtreated BMMCs are indicated at the top. Recombinant, purified GST, or a GST fusion to the cytoplasmic tail of Pecam-1 (⌬14/15 isoform, containing only one ITIM residue, Tyr 662 ) were added as potential substrates. In WT cells, IPs recovered both Fer and Fps kinases (lanes 1-3, lower panel). Fer and Fps displayed increased autophosphorylation and substrate phosphorylation when recovered from cells treated with antigen (lanes 1-3, top panel). To address Fps activity directly, we performed a similar assay on DR BMMCs (which lack Fer kinase activity and express very low levels of Fer DR because of protein instability (23)). Both autophosphorylation of Fps and substrate phosphorylation were significantly elevated upon antigen treatment (compare lanes 5 and 6, with lane 4 in top panel). To address whether substrate phosphorylation was indeed carried out by Fer and Fps kinases, we performed a parallel experiment with KR/DR BMMCs (which lack both Fer and Fps kinase activities (24)). Despite recovering similar amounts of Fer and Fps (lanes 7-9, bottom panel), we detected no signals for autophosphorylation or substrate phosphorylation. Coomassie staining of our substrate preparations indicated that the proteins were relatively pure and ran at their expected molecular masses (Fig.  1B).
Fc⑀RI-induced Phosphorylation of Fer Is Independent of Syk-Because Syk kinase plays a central role in signaling pathways emanating from the Fc⑀RI (9), we wished to address its potential involvement in Fer kinase activation. Using rat basophilic leukemia cells (RBL-2H3), a Syk-deficient subline (B2), and B2 cells that were stably transfected with a Syk expression plasmid (B2-Sykϩ), we sensitized the cells with anti-DNP-IgE and stimulated them with antigen (DNP-HSA) for various times (Fig. 2). Antigen-induced changes in tyrosine phosphorylation were noted in all cell lines (top panel), but the B2 cells were defective in phosphorylation of many proteins, including a prominent 70-kilodalton protein that co-migrated with Syk (data not shown). We tested Fer phosphorylation in each cell line and found that Fer phosphorylation downstream of Fc⑀RI aggregation does not require Syk (compare lanes 1-3 with 4 -6). Although the basal amount of Fer phosphorylation was higher in B2-Sykϩ cells, a similar increase in Fer phosphorylation was observed upon antigen treatment (compare lanes 7-9 with lanes 1-3). Similar results were obtained using Fps/Fer antisera; however, we have not been able to convincingly resolve endogenous Fer and Fps proteins in RBL-2H3 cells. These experiments, together with picetannol treatment of BMMCs (data not shown), suggest that Fer activation downstream of Fc⑀RI is Syk-independent.
Fc⑀RI-induced Phosphorylation of Fer Is Independent of Fyn Kinase and the Gab2 Adaptor-Recently, Fyn kinase and the adaptor protein Gab2 were shown to regulate Fc⑀RI aggregation-induced degranulation (31). Studies in RBL-2H3 cells and  BMMCs have shown potential involvement of Gab2 in Fc⑀RI aggregation-induced p38 MAPK activation (13,32). Because we have observed defects in p38 MAPK activation in Fer-deficient BMMCs (22), we wanted to address whether Fer may signal in the Fyn/Gab2 pathway. We generated BMMCs from wild type (fyn ϩ/ϩ ) and fyn knock-out (fyn Ϫ/Ϫ ) mice, sensitized them with anti-DNP-IgE, and treated them with antigen for various times (Fig. 3A). Analysis of Fer phosphorylation revealed that peak phosphorylation of Fer was similar between genotypes (compare lanes 1-4 with lanes 5-8). A similar experiment using wild type (gab2 ϩ/ϩ ) and gab2 Ϫ/Ϫ BMMCs, revealed no involvement of Gab2 for Fc⑀RI aggregation-induced phosphorylation of Fer (Fig. 3B).
Rapid Fc⑀RI-induced Phosphorylation of Fer Requires Lyn Kinase-To assess whether the upstream kinase acting on Fer is potentially Fc⑀RI-associated Lyn kinase, we generated BMMCs from wild type (lyn ϩ/ϩ ) and lyn Ϫ/Ϫ mice, sensitized them with anti-DNP-IgE, and treated them with antigen for various times (Fig. 4). SCLs were prepared, and immunoblotting confirmed the absence of both Lyn isoforms (p51/p53) in lyn Ϫ/Ϫ cells. Characterization of Fer phosphorylation revealed that the rapid phosphorylation of Fer, which was maximal at 1 min in wild type cells, was lost in lyn Ϫ/Ϫ cells (compare lanes 1-4 with lanes 5-8). This difference was not due to the amounts of Fer recovered, because we have recovered more Fer in lyn Ϫ/Ϫ samples (bottom panel). The phosphorylation of Fer increased gradually in Lyn-deficient cells and did not reach the levels of phosphorylation that occurred within 1 min in lyn ϩ/ϩ cells. Comparable results were obtained for two independent cultures of lyn ϩ/ϩ and lyn Ϫ/Ϫ BMMCs and suggest that Lyn is required for rapid phosphorylation of Fer but that another kinase, or perhaps Fer itself, can partially compensate for Lyn in mediating Fer activation.
Fer and Fps Phosphorylate ITIM and Non-ITIM Sites in Pecam-1-Several putative substrates of Fer have been identified in endothelial cells, including Pecam-1, Shp2, and Gab1 (33). The cytoplasmic tail of Pecam-1 contains several sites of phosphorylation, including ITIM tyrosines (Tyr 662 and Tyr 685 in mouse Pecam-1) that bind Shp1 and Shp2 phosphatases. Phosphorylation of Tyr 700 in mouse Pecam-1 (Tyr 701 in human) has been shown to promote recruitment of STAT3 or STAT5 (18). To address the sites in Pecam-1 preferred by Fer or Fps kinases isolated from activated BMMCs, we performed parallel in vitro kinase assays for Fer and Fps with a series of Pecam-1 cytoplasmic tail substrates (expressed as GST fusion proteins; Fig. 5). Kinase assays were performed with unlabeled   JULY 28, 2006 • VOLUME 281 • NUMBER 30 ATP and analyzed by immunoblotting with anti-pY to avoid potential spurious signals caused by serine/threonine phosphorylation. Fer was found to phosphorylate the full-length Pecam-1 C-tail, but not GST (top panel, lanes 1 and 3). A control reaction from lysates of DR BMMCs shows that the activity being measured was solely that of Fer (lane 2). Mutation of either ITIM (Y662F or Y685F) resulted in a 60% decrease in substrate phosphorylation but not autophosphorylation of Fer (compare lanes 3-5, upper panel). Using a substrate derived from the exon 15-deleted isoform of Pecam-1 (⌬15), we observed a significant reduction in phosphorylation by Fer ( compare lanes 3 and 6). A parallel experiment to examine preferred Fps phosphorylation sites in Pecam-1 using DR BMMCs (to avoid recovery of active Fer kinase in our IPs) was carried out. The results showed that Fps has higher activity toward Tyr 685 than Tyr 662 (60% versus 30% reduction because of mutation; third panel, lanes [3][4][5] and showed a significant difference in phosphorylation of the ⌬15 isoform (70% reduction; lane 6). Exon 15 contains one tyrosine (Tyr 700 ) that has been implicated in the recruitment of STAT3/STAT5 (18).

Lyn and Fer/Fes Kinases Phosphorylate Pecam-1 in Mast Cells
To determine whether Fer contributes to Pecam-1 phosphorylation on ITIM and non-ITIM residues in cells, we performed co-transfections of COS-7 cells with Myc-tagged Fer (WT and kinase-dead (KR)), and various full-length Pecam-1 expression plasmids (Fig. 6). SCLs were prepared, and Western blotting showed comparable expression of Myc-Fer WT and Myc-Fer KR between samples (top panel, alternating lanes), and high level expression of Myc-Fer compared with endogenous Fer in all of the samples except those lacking Pecam-1 (lanes 1 and 2). Pecam-1 phosphorylation by Fer was reduced upon mutation of Tyr 662 (30% reduction, middle panel, lanes 3 and 5), which is consistent with a previous study (33). A single mutation at Tyr 700 did not significantly reduce the overall phosphorylation (lanes 3 and 7). However, a double mutation at Y662F and Y700F resulted in a 50% reduction in Pecam-1 phosphorylation compared with mutation of the Tyr 662 site alone (lanes 5 and 9). These studies suggest that Fer and Fps kinases can phosphorylate Pecam-1 ITIMs and a non-ITIM residue in vitro and in vivo.
Fc⑀RI-induced Tyrosine Phosphorylation of Pecam-1 Is Reduced in the Absence of Fps and Fer-Because Fc⑀RI aggregation in mast cells causes increased phosphorylation of Pecam-1 and subsequent recruitment of Shp2, we wanted to address whether Fer and/or Fps kinases play a role in signaling from Fc⑀RI to Pecam-1. We generated BMMCs from WT, fer DR/DR (DR), and fps KR/KR / fer DR/DR (KR/DR) mice, sensitized them with anti-DNP-IgE, and stimulated the cells with antigen for various times. Using a phospho-specific antibody raised against a peptide encompassing . IPs with rat anti-Pecam-1 antibody (Pecam-1 NT ) were fractionated by SDS-PAGE, and duplicate blots were IB with anti-pY and anti-Pecam-1 CT . The control blot was stripped and reprobed with Shp2 antisera, and the position of Shp2 is indicated with an arrow on the left. The ratio of phosphorylated Pecam-1 at Tyr 685 relative to total Pecam-1 in SCLs, was determined by densitometry. The ratio of total Pecam-1 phosphorylation relative to Pecam-1 recovery in IPs with Pecam-1 NT was determined by densitometry. B, WT and KR/DR BMMCs were sensitized with anti-DNP-IgE and starved of IL-3 for 18 h, stimulated with DNP-HSA (100 ng/ml) for 0, 2, 10, or 20 min, and lysed, and IPs using goat anti-Pecam-1 CT were fractionated by SDS-PAGE in duplicate and immunoblotted with anti-pY or anti-Pecam-1 CT . The ratio of total Pecam-1 phosphorylation relative to Pecam-1 recovery in IPs with Pecam-1 CT was determined by densitometry. Tyr 686 of human Pecam-1 (which cross-reacts with Tyr 685 of mouse Pecam-1; Tyr(P) 685 ), we observed no differences in Pecam-1 phosphorylation between genotypes (Fig. 7A). However, we did observe reduced overall tyrosine phosphorylation of Pecam-1 in KR/DR BMMCs following IgE/DNP treatment ( Fig.  7A; Ϸ30 -40% reduction). Despite the reduced phosphorylation, there is no defect in Shp2 recruitment to Pecam-1 in KR/DR BMMCs following Fc⑀RI aggregation (Fig. 7A, arrow indicates position of Shp2). Similar results were obtained for Shp1 recruitment (data not shown). Using an antibody raised against the extreme C terminus for immunoprecipitation that should preferentially recover the full-length Pecam-1 isoform harboring both ITIMs and Tyr 700 , we observed a more profound reduction in Pecam-1 tyrosine phosphorylation in KR/DR BMMCs compared with WT ( Fig. 7B; 60 -70% reduction). Thus, Fer and Fps play redundant roles in promoting Pecam-1 phosphorylation at positions other than Tyr 685 . Given our in vitro kinase assay results, Fer and Fps may phosphorylate Pecam-1 at positions Tyr 662 and/or Tyr 700 .
Lyn Is an Initiator Kinase for Pecam-1 and Is Essential for Shp2 Recruitment-Because Lyn is also a candidate kinase for Pecam-1 (33) and is required for rapid activation of Fer kinase (Fig. 4), we addressed the phosphorylation status of Pecam-1 in Lyn-deficient mast cells. Wild type (lyn ϩ/ϩ ) and lyn Ϫ/Ϫ BMMCs were starved and sensitized with anti-DNP-IgE prior to exposure to antigen for the times indicated (Fig. 8). Immu-noblots of SCLs with Tyr(P) 685 Pecam-1 revealed a profound defect in Pecam-1 phosphorylation at this site in Lyn-deficient BMMCs. Likewise, when we analyzed the overall levels of tyrosine phosphorylation of Pecam-1 following immunoprecipitation, we observed a 90% reduction in peak phosphorylation in Lyn-deficient BMMCs. This also resulted in a complete block in the recruitment of Shp2 to Pecam-1 following Fc⑀RI aggregation in lyn Ϫ/Ϫ BMMCs. We analyzed Pecam-1 Tyr 685 phosphorylation in Btk-deficient BMMCs, because Btk is known to act downstream of Lyn and observed no defects in phosphorylation (data not shown). Thus, Lyn is likely an initiator kinase for Pecam-1 phosphorylation at the ITIM Tyr 685 position, which is required for Shp2 recruitment. Additional kinases (e.g. Fer/Fps) may contribute to Tyr 662 and/or Tyr 700 phosphorylation.
Increased Sensitivity of Mast Cells Deficient for Fer/Fps Kinases to Degranulation at Low Dose Antigen Exposure-Recent studies have implicated Pecam-1 and Shp1 as negative regulators of mast cell degranulation (16,17). Because Fer/Fps kinase contribute to Pecam-1 phosphorylation in a Lyn-dependent pathway, we wished to assess the degranulation response of mast cells devoid of Fer/Fps kinases. Initial experiments using ␤-hexosaminidase release assays and high dose antigen (100 ng/ml) revealed no differences in degranulation between WT and KR/DR BMMCs (data not shown). However, using a flow cytometric assay measuring annexin V staining of exocytic vesicles that have fused with the plasma membrane (29, 30), we observed a significant increase in basal degranulation in IgE-sensitized KR/DR BMMCs compared with WT (Fig. 9A, base; p Ͻ 0.01). At increasing antigen dosages there was no difference in the percentage of cells that underwent degranulation. However, the mean fluorescent intensity of annexin V staining was significantly higher at low doses of antigen in KR/DR BMMCs compared with WT (Fig. 9B, 0.1 and 1 ng/ml; p Ͻ 0.05). Taken together, these results suggest that Fer/Fps kinases function to limit the extent of granule mobilization in IgEsensitized mast cells. At higher intensity stimulation, this function of Fer/Fps kinases may be compensated for by other kinases.

DISCUSSION
Mast cell activation via clustering of Fc⑀RI elicits rapid changes in protein phosphorylation and localization, cytoskeletal organization, calcium mobilization, degranulation, and gene expression (1). Our research is aimed at trying to identify the roles played by two related PTKs, called Fer and Fps (34). We recently showed that both Fer and Fps are rapidly phosphorylated upon aggregation of Fc⑀RI (22) and that this also occurs in BMMCs expressing only kinase-dead Fer or Fps (24). This suggests that Fer and Fps are themselves substrates for another PTK(s) that is activated by this receptor. In this study, we provide evidence for Lyn involvement in early activation of Fer kinase. However, Fer phosphorylation does occur in Lyn-deficient mast cells but at a much slower rate and overall level of phosphorylation (Fig. 4). Because Fyn has been reported to act upstream of Fer in cells responding to changes in cell volume (35), it is possible that Fyn is compensating for loss of Lyn, albeit less efficiently. Future studies with BMMCs from Lyn/Fyn compound knock-out mice should address this issue. It is also possible FIGURE 8. Pecam-1 phosphorylation is greatly reduced in Lyn-deficient mast cells following Fc⑀RI aggregation. Wild type (lyn ϩ/ϩ ) and lyn Ϫ/Ϫ BMMCs were sensitized with anti-DNP-IgE and starved of IL-3 for 18 h and stimulated with DNP-HSA for 0, 2, 10, or 20 min, and SCLs were prepared. SCLs were IB with either a phospho-specific antibody to Tyr(P) 685 position of Pecam-1 or with a goat anti-mouse Pecam-1 (Pecam-1 CT ). IPs were also performed using anti-Pecam-1 NT , followed by IBs of duplicate gels with either anti-pY or anti-Pecam-1 CT . The control blot was stripped and reprobed with Shp2 antisera. The ratio of phosphorylated Pecam-1 at Tyr 685 relative to total Pecam-1 in SCLs, was determined by densitometry. The ratio of total Pecam-1 phosphorylation relative to Pecam-1 recovery in IPs with Pecam-1 NT was determined by densitometry.
that Lyn does not phosphorylate Fer kinase directly but that a downstream kinase participates.
In this study, we show that Fc⑀RI aggregation-induced Pecam-1 phosphorylation is partially dependent on Fer and Fps kinases (Fig. 7) and greatly dependent on Lyn (Fig. 8). Src family kinases were previously shown to efficiently phosphorylate Tyr 685 but not Tyr 662 ITIMs of mouse Pecam-1 (28). We propose that Lyn phosphorylates Tyr 685 of Pecam-1 to promote recruitment of Shp1/Shp2 phosphatases. Lyn is also required for Fer activation (Fig. 4), and subsequent recruitment of Fer/ Fps kinases may participate in Shp1/Shp2 activation by phosphorylation of Tyr 662 . Unfortunately, phospho-specific antibodies to this site are not available currently to address this model directly. Interestingly, Fer/Fps-deficient mast cells dis-play some hyperdegranulation phenotypes associated with Pecam-1-or Shp1-deficient mast cells (Fig. 9B and Refs. 16 and 17). Future studies will attempt to identify the sites of Pecam-1 phosphorylation in mast cells and their involvement in downstream signaling. In endothelial cells, Pecam-1 phosphorylation elicits Shp2 recruitment and regulation of cell adhesion and migration (21). It will be interesting to determine whether Pecam-1/Shp2 plays similar roles in mast cells.
Although mice lacking Pecam-1, Fps, and Fer kinases are viable (23,36,37), they all share phenotypes associated with hypersensitivity to lipopolysaccharide-induced inflammation (19,20,37,38). A recent study also showed that Pecam-1-deficient endothelial cells and lymphocytes are partially defective in STAT3 activation in response to lipopolysaccharide treatment. This is thought to involve recruitment of STAT3 to a non-ITIM tyrosine (Tyr 700 ) of Pecam-1 (19). We show in this study that Fer and Fps kinases can phosphorylate Tyr 700 of Pecam-1. It is worth noting that the peptide sequence surrounding Tyr 700 of Pecam-1 is similar to the C-terminal phosphorylation site of STAT3 (Tyr 705 ) and that previous studies have implicated Fer and Fps as STAT3 kinases in some cell types and conditions (39,40). Although STAT3 phosphorylation downstream of Fc⑀RI has not been reported, it was shown to occur downstream of the T cell receptor (which also signals via Src family kinases and ITAMs) (41). Activated mast cells also produce the STAT3 target gene vascular endothelial growth factor (42,43), a known angiogenic factor and chemotactic factor for mast cells (44). Interestingly, a high proportion of lung adenocarinomas contain infiltrating mast cells that express vascular endothelial growth factor, and this correlates with poor prognosis (45). Therefore, defining the pathway that controls vascular endothelial growth factor production by mast cells, could provide therapeutic targets to limit tumor-associated angiogenesis.