PSGL-1 engagement by E-selectin signals through Src kinase Fgr and ITAM adapters DAP12 and FcRγ to induce slow leukocyte rolling

Binding of P-selectin to PSGL-1 induces Src kinase–dependent phosphorylation of downstream molecules (7). To test whether Src kinases are involved in E-selectin–mediated slow rolling, we investigated the rolling velocity of neutrophils from mice pretreated with the Src-family kinase inhibitor (PP2) or an inactive control (PP3) in an autoperfused flow chamber. In this system, neutrophils can be investigated in native whole blood, without any isolation procedure that might activate the cells and alter their ability to regulate rolling velocity. As previously shown (6), the rolling velocity of WT neutrophils rolling in autoperfused flow chambers coated with E-selectin plus ICAM-1 is significantly reduced compared with E-selectin alone (6). Pretreatment of mice with PP2 did not influence the rolling velocity of neutrophils on E-selectin, but the decrease of rolling velocity on E-selectin plus ICAM-1 was almost completely abolished after Src-family kinase blockade (Fig. 1 A).

To confirm our flow chamber data in vivo, we conducted intravital microscopy in mixed chimeric mice. hck^−/− lyn^−/− fgr^−/− triple knockout mice lack all three neutrophil Src-family kinases (33). Lethally irradiated WT mice received bone marrow cells from WT LysM-GFP⁺ mice (34) and GFP-negative hck^−/− lyn^−/− fgr^−/− mice mixed in a ratio of 1:1. Leukocyte rolling was analyzed in TNF-α–treated venules of the cremaster muscle after blocking P-selectin (to limit our observations to E-selectin) and Gα_i-signaling (to block chemokine signaling). In this system, gene-targeted leukocytes can be compared side-by-side with leukocytes from WT mice under the same hemodynamic conditions in the same venules. The measured mean blood flow velocity and the calculated wall shear rates in these venules were 3.2 ± 0.2 mm/s and 2,000 ± 200 s⁻¹, respectively. Under these conditions, mean rolling velocity (V_avg) of hck^−/− lyn^−/− fgr^−/− leukocytes in vivo was 7.3 ± 0.3 μm/s (Fig. 1 B), almost twice that of LysM-GFP⁺ control cells (V_avg = 4.4 ± 0.1 μm/s; P < 0.05). The high rolling velocity seen in hck^−/− lyn^−/− fgr^−/− leukocytes is equal to the rolling velocity of leukocytes seen in WT mice after antibody blockade of both β₂ integrins, Mac-1 and LFA-1 (6). These findings suggest that Src-family kinases participate in β₂ integrin activation and subsequent slow rolling after E-selectin engagement.

It is known that the Src-family kinases have a considerable redundancy and that elimination of more than one kinase is often necessary to impair downstream signaling and subsequent biological activity (19, 20). To test whether such a redundancy exists in the signaling pathways triggered by PSGL-1 engagement with E-selectin, we investigated the rolling velocity of WT, hck^−/−, lyn^−/−, fgr^−/−, and hck^−/− lyn^−/− fgr^−/− neutrophils using the autoperfused flow chamber described int the previous section. The rolling velocity of neutrophils from WT mice and gene-deficient mice on E-selectin alone was similar (Fig. 2 A). As expected, neutrophils from WT mice showed a reduction of the rolling velocity on E-selectin plus ICAM-1 (Fig. 2 A). However, fgr^−/− neutrophils failed to reduce their rolling velocity on E-selectin plus ICAM-1 compared with WT neutrophils (Fig. 2 A). This rolling velocity was not further increased in hck^−/− lyn^−/− fgr^−/− triple knockout mice, and hck^−/− or lyn^−/− mice showed no rolling defect.

Mixed chimeric mice, generated with bone marrow from LysM-GFP⁺ mice and fgr^−/− mice, showed that the rolling velocity of fgr^−/− leukocytes was significantly increased (7.8 ± 0.2 μm/s, Fig. 2 B) compared with leukocytes from WT mice (4.5 ± 0.2 μm/s, Fig. 2 B). Collectively, these data show that Fgr, but not Hck or Lyn, is required for slow rolling triggered by PSGL-1 engagement.

To address whether ITAM-containing adaptor proteins are involved in E-selectin–dependent rolling, we investigated the rolling velocity of neutrophils from DAP12-deficient mice (Tyrobp^−/−), FcRγ-deficient mice (Fcrg^−/−), double mutant mice (Tyrobp^−/− Fcrg^−/−), and WT mice using the whole-blood autoperfused flow chamber. Elimination of either FcRγ or DAP12 alone slightly increased rolling velocity on E-selectin plus ICAM-1 compared with neutrophils from WT mice (Fig. 3 A). The rolling velocity on E-selectin plus ICAM-1 of neutrophils lacking both adaptor molecules was significantly increased compared with neutrophils from WT mice, and was similar to the rolling velocity on E-selectin alone (Fig. 3 A). These data suggest that the two ITAM-containing adaptor proteins, DAP12 and FcRγ, have a partially redundant function. When both are eliminated, E-selectin–dependent slow rolling is abolished.

To investigate the role of the ITAM-containing adaptor proteins DAP12 and FcRγ in E-selectin–dependent slow rolling in vivo, we generated mixed chimeric mice by injecting bone marrow cells from LysM-GFP⁺ mice and Tyrobp^−/− Fcrg^−/− mice into lethally irradiated WT mice and performed intravital microscopy. As before, we blocked P-selectin and Gα_i-coupled signaling to more clearly address the E-selectin pathway of neutrophil activation. In TNF-α–induced inflamed postcapillary venules, the rolling velocity of Tyrobp^−/− Fcrg^−/− leukocytes was 8.4 ± 0.4 μm/s (Fig. 3 B and Video 1, available), which is significantly higher than LysM-GFP⁺ control cells (4.3 ± 0.2 μm/s; P < 0.05). These findings show that β₂ integrin activation after E-selectin engagement is defective in Tyrobp^−/− Fcrg^−/− leukocytes.

Neutrophil recruitment into the peritoneal cavity after thioglycollate injection is promoted by E-selectin– and chemokine-dependent pathways (6, 35). To investigate the physiological importance of DAP12 and FcRγ in a model of acute inflammation, neutrophil recruitment in thioglycollate-induced peritonitis was investigated in WT, Fcrg^−/−, Tyrobp^−/−, and Tyrobp^−/− Fcrg^−/−mice with or without PTx treatment to block Gα_i-signaling. Consistent with previous data (6, 35), pretreatment with PTx reduced neutrophil recruitment to WT peritoneum 8 h after thioglycollate injection by ∼50% (Fig. 4). In the presence of intact GPCR signaling, Fcrg^−/−, Tyrobp^−/−, and Tyrobp^−/− Fcrg^−/− mice showed no major defect in neutrophil recruitment into the peritoneal cavity 8 h after thioglycollate injection. This is reminiscent of the phenotype of Syk^−/− neutrophils in bone marrow chimeras after induction of thioglycollate peritonitis (36). Blocking of Gα_i signaling by PTx reduced neutrophil recruitment in Fcrg^−/− and Tyrobp^−/− single knockout mice by ∼50%. However, treating Tyrobp^−/− Fcrg^−/− double knockout mice with PTx almost completely abolished neutrophil recruitment into the peritoneal cavity after thioglycollate injection.

Engagement of PSGL-1 can activate Src-family kinases (7) and the tyrosine kinase Syk (10). However, because Syk cannot directly interact with the cytoplasmatic tail of PSGL-1, ITAM-containing adaptor molecules are necessary for this interaction (10). During immunoreceptor and integrin signaling, Src-family kinases phosphorylate ITAM-containing adaptor proteins (21, 22, 32). This leads to activation of Syk, which in turn induces further downstream signaling. Consequently, we tested whether downstream signaling of PSGL-1 after E-selectin engagement required Fgr, DAP12, and FcRγ. Stimulation of WT neutrophils with E-selectin under shear stress conditions induced phosphorylation of the tyrosine in the catalytic domain of Fgr in a time-dependent manner (Fig. 5 A). E-selectin engagement also activated Hck (Fig. S1 A, available), but did not induce tyrosine phosphorylation of Lyn (Fig. S1B). As Tyrobp^−/− Fcrg^−/− mice showed an abolished E-selectin–mediated slow rolling, we speculated that these ITAM-containing adaptor molecules are phosphorylated after E-selectin stimulation. To investigate DAP12 phosphorylation, we used a glutathione S-transferase (GST) fusion protein with the tandem SH2 domains of Syk (GST-Syk-[SH2]2) to pull down tyrosine-phosphorylated proteins. Bound proteins were immunoblotted with a phosphospecific DAP12 antibody. Stimulation of neutrophils with E-selectin for 10 min led to an increase in DAP12 phosphorylation (Fig. 5 B). To investigate whether Fgr is involved in phosphorylation of the tyrosine residues in the ITAM domains, we stimulated WT and fgr^−/− neutrophils with E-selectin. Elimination of Fgr almost completely abolished DAP12 phosphorylation after stimulation with E-selectin (Fig. 5 C).

Phosphorylation of downstream effectors was abolished in neutrophils pretreated with an Src-family kinase inhibitor (7). To investigate whether only Fgr is responsible for signaling downstream of PSGL-1, we measured Syk phosphorylation in bone marrow–derived neutrophils from WT mice and fgr^−/− mice. Plating neutrophils on immobilized recombinant E-selectin under shear stress conditions for 10 min induced phosphorylation of Syk in neutrophils from WT, hck^−/−, and lyn^−/− (Fig. S1 C), but not fgr^−/−, neutrophils (Fig. 5 D). To directly demonstrate that DAP12 and FcRγ are also involved in this signaling pathway, we investigated Syk phosphorylation in unstimulated and stimulated neutrophils from WT mice and Tyrobp^−/− Fcrg^−/− mice. In contrast to stimulated WT neutrophils, no Syk phosphorylation was detectable in Tyrobp^−/− Fcrg^−/− neutrophils after stimulation (Fig. 5 E).

p38 MAPK is phosphorylated after E-selectin engagement and is involved in neutrophil slow rolling and adhesion (6, 8–10, 37). Pharmacological inhibition of p38 MAPK reduces slow rolling on E-selectin plus ICAM-1 through unknown mechanisms (6, 37). Consistent with this functional role, WT neutrophils incubated with immobilized E-selectin show significant induction of p38 MAPK phosphorylation that peaked at 10 min (Fig. 6, A and B). Tyrobp^−/− Fcrg^−/− neutrophils failed to phosphorylate p38 MAPK after stimulation with E-selectin (Fig. 6 C). In contrast, phosphorylation of p38 MAPK was normal in Fcrg^−/− or Tyrobp^−/− neutrophils (Fig. 6 C). These data suggest that ITAM domains provided either by DAP12 or FcRγ are sufficient for downstream signaling after E-selectin engagement.

Several prior studies have shown that engagement of selectin receptors on neutrophils can activate Src-family kinases (7, 38), induce Syk phosphorylation (10), integrin activation (6), slow rolling (6), and adhesion (9). However, the proximal signaling pathway between PSGL-1 and Syk was still unknown. In this study, we demonstrate that the Src-family kinase Fgr was indispensable for PSGL-1–dependent Syk activation and slow neutrophil rolling in vitro and in vivo. After E-selectin engagement, Fgr was activated and ITAM-containing adaptor proteins became phosphorylated and subsequently associated with Syk. Eliminating both DAP12 and FcRγ blocked this signaling pathway, and Gα_i-independent neutrophil recruitment into the peritoneal cavity was defective in Tyrobp^−/− Fcrg^−/− mice, confirming the physiological relevance of this signaling pathway.

The PSGL-1–Fgr–DAP12–FcRγ pathway has striking similarities to the immunoreceptor and integrin outside-in (21) signaling pathways. However, there are also significant differences. Elimination of only one Src-family kinase does not affect β₂ integrin–dependent neutrophil functions (39) or the inflammatory response in an endotoxic shock model (40). At least two Src-family kinases must be eliminated to reduce β₂ integrin–dependent neutrophil adhesion and superoxide production (39). However, such a redundancy does not exist in the E-selectin–mediated signaling pathway. Elimination of Fgr abolishes downstream signaling, including Syk phosphorylation and slow neutrophil rolling. This study is one of the few examples demonstrating a specific and nonredundant function for a Src-family kinase. In most other signaling pathways studied, one Src-family kinase may be dominant (such as Src in β₃ integrin signaling in osteoclasts or Lyn in inhibitory signaling in B lymphocytes), or there may be no apparent kinase selectivity (such as in integrin signaling). The specific role of Fgr in neutrophil PSGL-1 signaling is unique and provides a potential explanation for the evolution of this individual member of the Src kinase family.

Another difference between PSGL-1 and immunoreceptor or integrin signaling is the strength of the signal. E-selectin engagement of PSGL-1 induces downstream signaling and leads to partial activation of β₂ integrins on the leukocyte surface, resulting in slow rolling on ICAM-1. In the autoperfused flow chamber, this activation is insufficient to cause arrest. This suggests that signaling through PSGL-1 is weak compared with other immunoreceptor pathways such as TREM1, which can induce firm adhesion, and even activation, of effector functions like cytokine production and degranulation (29).

A study with isolated human neutrophils showed that P-selectin engagement of PSGL-1 induced Lyn and Hck activation (38). However, this study may be complicated by the ligands used. Piccardoni et al. stimulated neutrophils with either fixed platelets or a P-selectin–IgG chimera containing an Fc portion that does not rule out the possibility that the signal was induced by outside-in integrin signaling or engagement of Fc receptors. In our in vitro assay, we can exclude signaling through engagement of Fc receptors, because neutrophils from Fcrg^−/− mice, which cannot signal through Fc receptors, show the same phosphorylation pattern of p38 as WT mice. Together with the present study, these data suggest that engagement of E- or P-selectin may activate different Src-family kinases.

Although Fgr is activated after E-selectin engagement of PSGL-1, the mechanism of this interaction remains unknown. It is tempting to speculate that phosphatases like CD45 (41) and/or kinases like Csk (42) may be involved in this process. Csk phosphorylates tyrosine 527 near the C-terminus of Fgr, which blocks Fgr activation. CD45 is a plausible candidate phosphatase that may dephosphorylate Y527 and activate Fgr (43, 44). Indeed, elimination of phosphatases is associated with a proinflammatory phenotype (42). Further detailed studies are needed to determine how Fgr is activated after E-selectin engagement of PSGL-1.

The tyrosine kinase Syk cannot directly interact with the cytoplasmatic tail of PSGL-1 (10). In HL-60 cells, the ITAM-containing adaptor proteins ezrin and moesin can act as linkers between the two molecules (10). In this study, we demonstrate that the ITAM-containing adaptor proteins DAP12 and FcRγ provide alternative linkers, and eliminating both adaptor proteins abolishes E-selectin–mediated downstream signaling and slow rolling. This suggests that the ITAM domains of ezrin and moesin cannot substitute for DAP12 and FcRγ. One likely explanation for this is that ITAM-containing adaptor molecules may be compartmentalized in the neutrophil, with one set (DAP12 and FcRγ) promoting slow rolling and another (ezrin and moesin) involved in directing PSGL-1 to the uropod of migrating cells (10). Compartmentalization of multimolecular signaling complexes is known to be important for linking different signals at the appropriate time and location to specific signaling cascades in the cell (45). Both FcRγ and DAP12 associate with many cell surface receptors in various hematopoietic cell types (22). However, PSGL-1, unlike TREM-1, myeloid DAP12-associating lectin-1, and signal-regulatory protein β, does not contain a basic amino acid in its transmembrane domain. Therefore, the interaction between DAP12 or FcRγ and PSGL-1 may be indirect. Other transmembrane proteins may be involved in linking PSGL-1 to DAP12 and FcRγ. The finding that PSGL-1 resides in lipid rafts (46) may bring it into close enough proximity to allow productive interactions with DAP12 and FcRγ.

Our data suggest that DAP12 and FcRγ are partially redundant in slow rolling after PSGL-1 engagement by E-selectin. However, they seem to be fully redundant for p38 MAPK phosphorylation, because the phosphorylation level is indistinguishable in Tyrobp^−/− or Fcrg^−/− neutrophils, but completely absent in double knockouts. This raises the possibility that more than one signaling pathway may be involved in LFA-1 activation after PSGL-1 binding to E-selectin. The p38 MAPK-dependent arm of the pathway may require only one ITAM domain-containing species, DAP12 or FcRγ, whereas another arm of the pathway may require both, resulting in partial loss of slow rolling function in Tyrobp^−/− or Fcrg^−/− mice.

The reduction of neutrophil recruitment in Tyrobp^−/− Fcrg^−/− mice after blocking GPCR signaling is quantitatively similar to the reduction seen in PTx-treated syk^−/− chimeric mice (6). This suggests that eliminating DAP12 and FcRγ is as effective in blocking neutrophil recruitment as eliminating Syk. Like syk^−/− chimeric mice (6) or E-selectin–deficient (Sele^−/−) mice (35), Tyrobp^−/− Fcrg^−/− mice show defective neutrophil recruitment to the inflamed peritoneum when chemokine receptor signaling is blocked. These data demonstrate the physiological relevance of the E-selectin–mediated PSGL-1 pathway in vivo. The new mechanistic insights we provide here suggest that Syk, DAP12, and FcRγ may be therapeutic targets for antiinflammatory therapy. The specific role of Fgr suggests that specific Fgr inhibitors that do not block Lyn or Hck could be devised that may specifically block E-selectin–dependent neutrophil activation, perhaps without compromising host defense.

8–12-wk-old C57BL/6 mice (The Jackson Laboratory), DAP12 (Tyrobp^−/−) (47), FcRγ (Fcrg^−/−) (31), Tyrobp^−/− Fcrg^−/− (32), Fgr (fgr^−/−) (48), Lyn (lyn^−/−) (49), Hck (hck^−/−) (48), Hck/Lyn/Fgr (hck^−/− lyn^−/− fgr^−/−) (33), and Syk (syk^+/−) (6) mice were housed in a specific pathogen–free facility. The Animal Care and Use Committees of the University of Virginia, the La Jolla Institute for Allergy and Immunology, and the University of California at San Francisco approved all animal experiments. Chimeric mice were generated by performing bone marrow transplantation, as previously described (50). In brief, bone marrow cells isolated from Lys-M-GFP⁺ (34) and gene-deficient mice were mixed 1:1, and 5 × 10⁶ unfractionated cells were injected intravenously into lethally irradiated mice. Experiments were performed 6–8 wk after bone marrow transplantation.

2 h before cremaster muscle exteriorization, mice received 4 μg PTx i.v. (Sigma-Aldrich) and 500 ng TNF-α intrascrotally (R&D Systems). Mice were anesthetized with an i.p. injection of 125 mg/kg ketamine hydrochloride (Sanofi), 0.025 mg/kg atropine sulfate (Fujisawa), and 12.5 mg/kg xylazine (TranquiVed; Phoenix Scientific) and placed on a heating pad. The cremaster muscle was prepared as previously described (51). Postcapillary venules with a diameter between 20 and 40 μm were recorded using an intravital microscope (Axioskop, SW 40/0.75 objective; Carl Zeiss, Inc.) through a digital camera (sensicam qe; Cooke Corporation). Blood flow centerline velocity was measured using a dual-photodiode sensor system (CircuSoft Instrumentation).

Autoperfused flow chambers were used to investigate the rolling velocity of neutrophils, as previously described (6, 37, 52). Rectangular glass capillaries (20 × 200 μm) were coated either with 30 μg/ml E-selectin (R&D Systems) alone or in combination with 15 μg/ml ICAM-1 (R&D Systems) for 2 h and blocked with 10% casein (Thermo Fisher Scientific). Each capillary was connected to a PE 10 catheter (Becton Dickinson) inserted into a mouse carotid artery. Wall shear stress (37) was controlled by hydrostatic pressure applied to the downstream side of the flow chamber. One representative field of view was recorded for 1 min using an SW40/0.75 objective and a charge-coupled device camera (VE-1000CD; Dage-MTI).

In some experiments, mice were pretreated with the specific Src-family kinase inhibitor PP2 (10 μg/kg, i.v., 30 min before the experiments, Calbiochem) or the inactive control (PP3).

For biochemical assays, 15-mm Petri dishes were coated with 3 μg/ml E-selectin for 2 h and blocked with casein. Bone marrow neutrophils (53) were suspended in PBS (containing 1 mM each CaCl₂ and MgCl₂) and incubated at 65 rpm for 10 min at 37°C. Neutrophils were lysed with RIPA buffer (32). Lysates were boiled with sample buffer or incubated with protein A–coated magnetic beads (Miltenyi Biotec) and rabbit anti-Syk (N-19; Santa Cruz Biotechnology, Inc.) antibody for 30 min on ice and separated using a magnet (Miltenyi Biotec). Beads were washed four times and bound proteins were eluted by adding boiling sample buffer. To test DAP12 phosphorylation, GST-Syk-(SH2)2 was coupled to glutathione–Sepharose beads (GE Healthcare), as previously described (32).

Cell lysates and immunoprecipitates were run on 10% SDS-PAGE and immunoblotted using antibodies against phosphotyrosine (4G10; Millipore), p38 MAP kinase (3D7), and phospho-p38 MAP kinase (both from Cell Signaling Technology), developed using GE Healthcare's ECL system, and analyzed by ImageJ (National Institutes of Health).

Peritonitis was induced by injecting sterile 4% thioglycollate i.p. (Sigma-Aldrich) (6). Some mice received 4 μg PTx i.v. 2 h before thioglycollate injection. After 8 h, mice were killed, the peritoneal cavity was rinsed with 10 ml PBS (containing 2 mM EDTA), and the number of leukocytes was counted. Neutrophils were detected by flow cytometry (FACSCalibur; BD Biosciences) based on expression of CD45 (clone 30-F11), 7/4 (clone 7/4; both from BD Biosciences), and GR-1 (clone RB6-8C5).

Statistical analysis was performed with SPSS (version 14.0, Chicago, IL) and included one-way analysis of variance, Student-Newman-Keuls test, and Student's t test where appropriate. All data are presented as mean ± SEM. P < 0.05 was considered significant.

Fig. S1 shows Src and Syk phosphorylation by E-selectin engagement in hck^−/− and lyn^−/− mice. Video 1 shows intravital microscopy of an inflamed cremaster muscle venule of a mixed chimeric mouse treated with PTx and a monoclonal blocking P-selectin antibody (GFP⁺, WT leukocytes; GFP⁻, Tyrob^−/− Fcrg^−/− leukocytes). The online version of this article is available.

This study was supported by grants from Deutsche Forschungsgemeinschaft (AZ 428/2-1 and 428/3-1 to A. Zarbock) and from the National Institutes of Health (NIH HL73361 to K. Ley and AI065495 to C.A. Lowell).

The authors have no conflicting financial interests.