Src Phosphorylation of Endothelial Cell Surface ICAM-1 Mediates Neutrophil Adhesion and Contributes to the Mechanism of Lung Inflammation

Abstract

Objective

To determine whether TNFα-induced Src activation and ICAM-1 phosphorylation rapidly increases endothelial cell adhesivity and PMN sequestration independent of de novo ICAM-1 synthesis.

Methods and Results

TNFα exposure of mouse lungs for 5 min produced a 3-fold increase in ¹²⁵I-anti-ICAM-1 mAb binding and ¹¹¹In oxine-labeled PMN sequestration as well as Src activation, ICAM-1 Tyr⁵¹⁸ phosphorylation, and pTyr⁵¹⁸-ICAM-1 co-immunoprecipitation with actin. The response was absent in Nox2^−/− lungs or following Src inhibition. In COS-7 cells transfected with wild-type (WT), phospho-defective (Y518F), or phospho-mimicking (Y518D) mouse ICAM-1 cDNA constructs, TNFα increased the B_max of YN1/1.7.4 anti-ICAM-1 mAb binding to WT-ICAM-1 but not to Y518F-ICAM-1 indicating increased binding avidity secondary to ICAM-1 phosphorylation. This effect was mimicked by expression of the Y518D-ICAM-1 mutant. TNFα also increased the staining intensity and cell surface clustering of YN1/1.7.4 mAb-labeled WT-ICAM-1 that co-localized with F-actin which was not observed with Y518F-ICAM-1 but was recapitulated with Y518D-ICAM-1. Finally, overexpression of ICAM-1 in mouse lungs significantly increased LPS-induced transvascular albumin leakage and bronchoalveolar lavage PMN counts at 2 and 24 hrs after LPS inhalation compared to lungs expressing Y518F ICAM-1 mutant.

Conclusions

Src-dependent phosphorylation of endothelial cell ICAM-1 Tyr⁵¹⁸ induces PMN adhesion by promoting ICAM-1 clustering which we propose mediates rapid-phase lung vascular accumulation of PMNs during inflammation.

Introduction

The release of the cytokine TNFα (tumor necrosis factor alpha) in inflamed tissues stimulates β₂ integrin expression on polymorphonuclear leukocytes (PMNs) and the de novo synthesis of intercellular adhesion molecule-1 (ICAM-1) in endothelial cells (ECs).^1,2 Newly-synthesized ICAM-1 appears on the EC plasma membrane in significant amounts within 2 to 4 hours of cytokine exposure.^3–5 ICAM-1, a member of the immunoglobulin gene superfamily, is the counter-receptor for β₂ integrins (CD11/CD18) expressed on the PMN cell surface^6–10. Interactions between endothelial ICAM-1 and different β2-integrins, macrophage-1 antigen (CD11b/CD18; Mac-1) and leukocyte function-associated antigen-1 (CD11a/CD18; LFA-1) ^3–4,6–15 are required for firm adhesion of PMNs to the endothelium and PMN transmigration^3,6–8. This binding interaction represents an essential step in PMN extravasation into tissues at sites of infection^1,2,4,16.

Endothelial cells constitutively express cell surface ICAM-1 albeit at a low level^3,5,17,18. This basal ICAM-1 expression is thought to support the limited transmigration of PMNs and other leukocytes into tissue perhaps as an immune surveillance mechanism^3,4,7,8. However, ICAM-1 expression requiring de novo ICAM-protein synthesis increases markedly in response to pro-inflammatory mediators such as TNFα^3,5,17,18. We have shown that TNFα also induces a rapid phase of endothelial cell surface ICAM-1 adhesion within minutes of TNFα exposure¹¹. The purposes of the present study were to address the validity of this phenomenon in the intact circulation and to identify key signaling constituents mediating this post-translational ICAM-1 modification and its consequences in lung inflammation. We show herein that TNFα-activation of Src and subsequent phosphorylation of ICAM-1 on Tyr⁵¹⁸ in mouse endothelial cells induces association of ICAM-1 with cortical actin, causes cell surface clustering of ICAM-1, and mediates rapid-phase PMN adhesion to vascular intima. We also demonstrate the importance of the rapid induction of ICAM-1 activation as defined by its phosphorylation in the mechanism of lung inflammation induced by LPS.

Methods

An expanded Methods section is available in the Online Data Supplement at http://ATVB.ahajournals.org. Binding kinetics (K_d and B_max) of cell surface wild-type (WT) and mutant mouse ICAM-1 expressed in COS-7 cells were determined by cell-attached mAb binding ELISA. Confocal images (1024 × 1024 pixels) of mouse lung endothelial cells (MLECs) and HUVEC stimulated with TNFα (20 ng/ml) or vehicle for 20 min were acquired using a Zeiss LSM510 META confocal imaging system. To measure ICAM-1 mAb-specific binding in the isolated-perfused mouse lung, a dual tracer protocol was implemented. To measure PMN sequestration in mouse lungs, TNFα and ¹¹¹indium oxine-labeled PMNs were sequentially infused. ¹¹¹Indium oxine tracer washout was monitored at 5 min intervals and PMN binding was determined by calculating the ratio of recovered tissue counts over infused counts. To assess the role of ICAM-1 phosphorylation in lung PMN infiltration and protein-rich edema formation associated with LPS inhalation, C57BL6 mice injected with liposome/cDNA mixture were administered nebulized-LPS (10 mg dissolved in 10 ml water over 1 hr) and then mice were euthanatized under anesthesia 2 or 24 hrs later. PMN number and albumin concentration in bronchoalveolar lavage fluid were determined. Significance of differences was determined by one way analysis of variance (ANOVA) with P < 0.05 considered significant.

Results

TNFα Rapidly Increases Anti-ICAM-1 mAb Binding Activity and PMN Uptake in Mouse Lungs

We first determined the effects of TNFα on the binding activity of ICAM-1 to anti-ICAM-1 mAb in mouse lung vessels using the dual tracer method in which we used a mouse specific anti-ICAM-1 mAb and another control mAb (see online Methods section). TNFα infusion in WT lungs within 5 minutes resulted in a 2-fold increase in the binding of ¹²⁵I-labeled anti-ICAM-1 YN1/1.7.4 mAb (Fig. 1A). To address whether the binding was specific, we added a 10-fold excess unlabeled anti-ICAM-1 mAb to the perfusate just prior to addition of radio-labeled mAb. The excess mAb abolished TNFα-stimulated binding of tracer mAb (see Fig. 1A). Exposure of lung vessels to TNFα for up to 30 minutes showed that the increase in ¹²⁵I-labeled mAb binding was maximal within the first 10 minutes (Fig. 1A). We also observed an additional increase in anti-¹²⁵I-ICAM-1 mAb binding after a prolonged (4 hr) period of TNFα exposure (Fig. 1A) consistent with the increase in ICAM-1 protein expression in response to TNFα in mouse lungs seen at this later time.¹⁹

Figure 1. TNFα increases the binding of anti-¹²⁵I-ICAM-1 mAb and PMN accumulation within 5 min.

A: Isolated-perfused mouse lungs (see Methods) were challenged with TNFα (2,000 U/ml, for indicated times) and specific binding of labeled ICAM-1 mAb was determined using a dual tracer protocol (see Methods). B: WT PMNs were isolated from C57BL/6 mice and labeled with ¹¹¹Indium oxine. TNFα was provided to the lung perfusate for 5 minutes at a final concentration of 2,000 U/ml. The binding of labeled PMNs was determined after a brief vascular rinse to remove free cytokine. TNFα increased PMN sequestration 3-fold above basal level in lungs of WT but not ICAM-1^−/− mice. Plotted values are mean ± S.D. (n = 3 per group); * = P < 0.01 compared to basal PMN uptake. C: Western blot showing the absence of increase in lung ICAM-1 expression within the first 30 min after TNFα but a progressive increase thereafter. Result is representative of 3 similar experiments

We next addressed whether TNFα produces a similar rapid increase in PMN binding in situ in lung vessels. The vasculature of isolated-perfused mouse lungs was treated with saline or TNFα (2,000 U/ml) for 5 minutes, lungs were cleared of the TNFα for 1 minute by adding fresh perfusate, and 10^{6 111}In oxine-labeled mouse PMNs were added to the perfusate over 5 minutes. After a washout period to remove the unbound PMNs, we determined tissue counts as the fraction of infused counts. TNFα increased the fraction of infused PMNs retained in the lung 3-fold (from 0.15 ± 0.03 to 0.47 ± 0.11; p<0.05). The ICAM-1 dependence of TNFα-induced PMN binding to endothelial cell ICAM-1 in the lungs is supported by the finding that TNFα-induced increase in PMN binding was not seen in lungs of ICAM-1^−/− mice (Fig. 1B). As shown in Figure 1C, ICAM-1 expression did not increase for up to 30 min after TNFα treatment of mouse lung endothelial cells when binding of the anti-ICAM-1 mAb was maximal; however, predictably the expression began to increase slightly at 60 min but increased substantially after treatment with TNFα for 2 hr and 24 hr.

Rapid Phase of TNFα-Induced Increase in ICAM-1 Binding Activity and PMN Uptake in Lung Vessels Requires Nox2

TNFα activates NADPH oxidase in endothelial cells by inducing the phosphorylation of the cytosolic regulatory subunit p47^phox, mediating its subsequent translocation to the plasma membrane and association with other subunits to form an active NADPH oxidase (Nox2) complex.²⁰ To address whether Nox2 is required for the observed rapid increase in binding of anti-¹²⁵I-ICAM-1 mAb, we perfused isolated lungs from wild type (WT) and p47^phox−/− mice with TNFα for 10 minutes and measured mAb binding. TNFα treatment doubled the binding of labeled anti-ICAM-1 mAb in lungs of WT mice whereas no significant change was observed in lungs of p47^phox−/− mice (Fig. 2A).

Figure 2. TNFα-induced anti-ICAM-1 mAb binding and PMN sequestration are abolished in Nox2-null mouse lungs.

A: Isolated-perfused lungs from wild-type (WT) or p47^phox−/− mice were challenged with TNFα (2,000 U/ml) for 10 minutes, and specific anti-¹²⁵I-ICAM-1 mAb binding was determined during the last 5-minutes of the treatment period. Unlabeled ICAM-1 mAb nearly abolished anti-¹²⁵I-ICAM-1 mAb binding indicating the specificity of binding to ICAM-1. TNFα had no effect on ¹²⁵I-ICAM-1 mAb binding to lungs from p47^phox−/− mice. Mean values ± S.D (n = 3–4) are plotted; * = P < 0.05 vs basal value. B: Isolated-perfused mouse lungs were exposed to TNFα for 5 min and then perfused with WT PMNs labeled with ¹¹¹1ndium oxine. TNFα increased PMN uptake 3-fold above basal level in WT mice whereas deletion of gp91^phox or p47^phox abrogated the effect of TNFα. In other experiments, Nox2 deficient lungs (gp91^{phox −/−}) received superoxide, generated in the perfusate for 10 minutes using the xanthine/xanthine oxidase (X/XO) system (see Methods), which reconstituted the 3-fold increase in PMN accumulation in lungs. Control values for WT lungs were replotted from data in Fig. 1B. Values shown are mean ± S.D. (n = 3–5); * = P < 0.01 vs basal value.

We also addressed the role of Nox2-derived oxidants in the rapid induction of PMN binding in lungs of p47^phox−/− mice or gp91^phox−/− mice (the latter carrying a deletion of the catalytic subunit of Nox2). The rapid stimulatory effect on PMN binding seen in WT lungs was absent in lungs of both knockout models (Fig. 2B). Nox2-deficient lungs, however, recovered their ability to recruit PMNs after superoxide was generated via perfusion of lungs with xanthine/xanthine oxidase (X/XO) mixture for 10 minutes (Fig. 2B) indicating that PMN uptake could be restored in Nox2-null lungs by chemically generating the oxidants. Superoxide generation by X/XO resulted in a 3-fold increase in the fraction of infused PMNs retained in the lung when compared with the unchallenged group (from 0.20 ± 0.02 to 0.67 ± 0.07; P<0.05). In control experiments, we observed that infusion with X/XO for 10 min at the concentrations used above (see Methods) did not injure the vessels of the lung preparation; i.e., superoxide generation did not alter ¹²⁵I-albumin pulmonary transvascular PS (permeability-surface area product) (24 ± 6 μl/min/g dry lung in control vs. 23 ± 3 μl/min/g dry lung in X/XO exposed lungs; n = 3 per group); thus, the oxidant-mediated PMN uptake was not the result of induction of lung vascular injury by oxidants produced by the X/XO reaction.

TNFα-Induced ICAM-1 Binding Activity and PMN Adhesion Require Phosphoinositide 3-Kinase-γ and PKCζ Signaling

As activation of PKCζ has been shown to induce Nox2 activation and ICAM-1 binding activity in endothelial cells,^11,21–23 we validated this observation in the present study and observed that TNFα induced a 7-fold increase in PKCζ activity in MLECs within 5 minutes (see Online Data Supplment, Fig. IA). Furthermore, TNFα–induced increase in ¹¹¹In oxine-PMN uptake in the mouse lung vasculature was blocked by pretreating mice (50 μg/mouse/day for 5 days, i.p.) with an inhibitory pseudosubstrate peptide specific for PKCζ (PKCζ-PS), whereas PKCθ-PS had no effect (see Online Fig. IB). As we previously showed in endothelial cells that phosphoinositide 3-kinase-γ (PI3-kinase-γ) activates PKCζ upstream of Nox2 activation,²⁴ we next addressed the role of PI3-kinaseγ in TNFα-mediated rapid PMN uptake in lungs. As shown in Online Fig. IB, we did not see any effect of TNFα in p110γ^−/− mice (deleted catalytic subunit of PI3-kinaseγ), indicating the requirement for PI3-kinase p110γ in mediating the rapid TNFα-induced PMN uptake in lung vessels.

TNFα Induction of Nox2-dependent Src Activation Mediates ICAM-1 Phosphorylation In Vivo

Studies have demonstrated oxidants activate Src ^25,26 and that Src is capable of inducing the phosphorylation of ICAM-1 Tyr⁵¹⁸ (^27–32). To address whether TNFα-induced oxidant production and subsequent Src activation mediates phosphorylation of ICAM-1 Tyr⁵¹⁸, and thus induces the rapid increase in PMN binding in lungs, we administered TNFα (2,000 U/ml) to the isolated lung perfusate for 10 minutes, snap-froze the tissue in liquid nitrogen, and analyzed homogenates for phospho-ICAM-1 by immunoblotting using an antibody that recognizes phosphorylated form of mouse ICAM-1 Tyr⁵¹⁸. This brief TNFα exposure resulted in 2-fold increase in phospho-ICAM-1 (Fig. 3A and B). To test the requirement for Nox2-dependent oxidant generation in the response, we repeated the procedure in lungs from gp91^−/− mice. In these mice, the level of phospho-ICAM-1 in unstimulated conditions was the same as in WT mice. However, TNFα perfusion did not increase pICAM-1 Tyr⁵¹⁸ phosphorylation (Fig. 3A and B). In other experiments, we immunoprecipitated ICAM-1 from homogenates obtained from TNFα-perfused lungs and immunoblotted these with anti-phosphoserine and anti-phosphothreonine Abs. These studies showed that phosphorylation of ICAM-1 on either serine or threonine residues did not increase on exposure to TNFα (data not shown).

Figure 3. TNFα induces rapid Src activation and ICAM-1 phosphorylation in murine lungs.

Isolated-perfused lung preparations were treated with TNFα (2,000 U/ml) for 10 minutes and quick-frozen with liquid nitrogen. Western blots (A) show TNFα-induced c-Src activation as seen by increased phosphorylation of c-Src Tyr⁴¹⁸ and ICAM-1 on Tyr⁵¹⁸ which were abrogated in gp91^phox null lungs. Summarized data reveal that TNFα doubled c-Src activity (B) and ICAM-1 phosphorylation (C) in WT lungs but produced no increase in c-Src or ICAM-1 phosphorylation in gp91^phox−/− lungs. Values shown are mean ± S.D. (n=3); * = P < 0.05.

As the tyrosine kinase c-Src may have a role in ICAM-1 activation^27–33, we assessed the function of c-Src in mediating TNFα-induced ICAM-1 phosphorylation and increase in PMN sequestration in mouse lungs. Lungs were perfused with or without TNFα (10 minutes), snap-frozen, homogenized, and immunoblotted for active Src using anti-pY418-Src pAb as well as for pY518-ICAM-1 using anti-pY518-ICAM-1 pAb. We observed that TNFα rapidly induced both c-Src and ICAM-1 phosphorylation on Tyr⁵¹⁸ (Fig. 3A and B). To address the requirement for oxidant generation in signaling Src activation and thereby in mediating ICAM-1 phosphorylation, we performed the same experiment in lungs from gp91^−/− mice. Basal Src phosphorylation was not significantly different in gp91^−/− mouse lungs vs. WT lungs, and importantly, TNFα perfusion did not induce Src activation or ICAM-1 phosphorylation (Fig. 3A and C) indicating that Nox2 is required for TNFα-induced Src Tyr⁴¹⁸ phosphorylation (activation) and resultant ICAM-1 phosphorylation.

Src-mediated Phosphorylation of Endothelial Cell ICAM-1 Induces Rapid Lung PMN Sequestration through an ICAM-1/Actin-Cytoskeleton-dependent Interaction

We next determined whether Src-dependent phosphorylation of ICAM-1 Tyr⁵¹⁸ in its intracellular C-terminal tail induces association of ICAM-1 with the endothelial cell cytoskeleton, which may induce ICAM-1 clustering and thereby increase availability of ICAM-1 binding epitopes, i.e. increased avidity or “functional affinity”^14,15,28 of ICAM-1 for PMN CD11/CD18 β2 integrins. To assess this concept, we treated mouse lungs with TNFα in the presence and absence of Src inhibitor PP1. As shown in Figure 4A, TNFα treatment increased ICAM-1 Tyr⁵¹⁸ phosphorylation, and this response was blocked by PP1 pretreatment, indicating TNFα-induced Src activation is responsible for ICAM-1 Tyr⁵¹⁸ phosphorylation. To show whether Src phosphorylation of ICAM-1 Tyr⁵¹⁸ in vivo plays a role in rapid PMN sequestration induced by TNFα, isolated-perused mouse lungs were pretreated with Src inhibitor PP1 for 20 min prior to 5 min TNFα exposure, followed by perfusion of ¹¹¹In oxine-labeled PMNs for 10 min. As shown in Figure 4B, PP1 blocked the rapid TNFα-induced PMN sequestration in mouse lungs.

Figure 4. Src inhibition prevents TNFα-induced ICAM-1 phosphorylation and lung PMN sequestration (A, B) and TNFα-induced clustering of ICAM-1 and its co-localization with F-actin in MLECs (C, D).

A. Isolated-perused lungs were treated with TNFα (2,000 U/ml) for 10 minutes and quick-frozen in liquid nitrogen. Western blots show that TNFα-induced increase in ICAM-1 Tyr⁵¹⁸ phosphorylation was abolished in lungs pretreated with 10 μM PP1. Bar graph shows the average band intensities (mean ± SD; n = 3; * = P < 0.05 vs control). B: Isolated-perfused mouse lungs were pretreated for 10 min with 10 μM PP1, stimulated for 10 min with 2000 U/ml TNFα, and then perfused with ¹¹¹Indium oxine-labeled PMNs as described above. C: MLECs treated for 10, 20, or 120 min with TNFα in the absence and presence of Src inhibitor PP2 were lysed, immunoprecipitated with anti-ICAM-1 mAb, and immunoblotted for ICAM-1 and β-actin. D. MLECs cultured on 10 mm diameter #1 glass coverslips were treated with vehicle (media) or TNFα (100 ng/ml) for 20 min. Immunostaining of endogenous ICAM-1 (green) and labeling of polymerized actin (red) was detected by confocal microscopy (60x 1.25 NA objective; pinhole = 1 Airy unit). Scale bar = 20μm; images are representative of 10 separate fields analyzed.

To assess mechanisms of phospho-ICAM-1-mediated PMN sequestration, MLECs were grown to confluence, serum-deprived for 2 hr, treated for 20 minutes with vehicle (cell culture media) or Src inhibitor PP2, and stimulated for 0,10, 20, or 120 minutes with TNFα. As shown in Figure 4C, we observed that TNFα induced an increase in Src activation (autophosphorylation of Tyr⁴¹⁸) and ICAM-1 Tyr⁵¹⁸ phosphorylation, both of which were blocked by PP2. Furthermore, immunoprecipitation of ICAM-1 revealed increased association of ICAM-1 with β-actin, which peaked at 20 min and returned to basal at 2 hr (Fig. 4C). Pretreatment of cells with Src inhibitor PP2 prevented the increased association between ICAM-1 and actin (Fig. 4C), indicating that this interaction was the result of Src-dependent ICAM-1 phosphorylation.

Next we addressed whether TNFα stimulated the association of ICAM-1 specifically with polymerized F-actin. MLECs were treated with TNFα for 20 min, lightly fixed with 1% paraformaldehyde on ice to limit permeabilization, and immunostained with ICAM-1 mAb coupled with Alexa 488 goat-anti-mouse secondary Ab (green) and Alexa 546 phalloidin (red), a fluorescently-tagged small molecule which binds and labels F-actin stress fibers. This technique allowed visualization of membrane-associated ICAM-1 together with actin stress fibers. Figure 4D shows diffuse membrane-associated endogenous ICAM-1 staining and presence of some actin stress fibers in non-stimulated sub-confluent MLECs, and little localization of ICAM-1 along stress fibers. However, stimulation of MLECs for 20 min with 100 ng/ml TNFα induced ICAM-1 clustering which appeared along actin stress fibers.

Quantification of Anti-ICAM-1 mAb Binding Kinetics and Effects of ICAM-1 Tyr⁵¹⁸ Point Mutations on TNFα-Induced Endothelial Cell Surface ICAM-1 Clustering

To investigate quantitatively how ICAM-1 Tyr⁵¹⁸ phosphorylation mediates a rapid increase in PMN binding, COS-7 cells (Figure 5A–B) were transiently transfected with WT, Y518F-non-phosporylatable, and Y518D-phospho-micking mouse ICAM-1 cDNAs. We used COS-7 cells to express WT-, Y518F-, and Y518D-ICAM-1 to determine the role of ICAM-1 phosphorylation induced by TNFα on anti-ICAM-1 mAb YN1/1.7.4 (YN1) and KAT-1 binding kinetics. To exclude the possibility that differences in transfection efficiency contributed to the differences in YN-1 antibody binding, we also measured the binding activity of KAT-1 mAb. YN1 mAb binds to the LFA-1 eptitope^9,10,14–16 whereas eBioKAT-1 mAb binds to a non-overlapping D1 domain epitope (eBioscience.com) in the ICAM-1 extracellular domain. Figure 5A shows the effects of TNFα on YN1mAb concentration-dependent binding to WT-ICAM-1 (left panel) and Y518F-ICAM-1 expressed in COS-7, compared to mAb binding to the phospho-mimicking Y518D-ICAM-1 mutant in these cells (right panel). A summary of binding kinetics is shown in Figure 5B. TNFα induced a 23% increase in B_max of ICAM-1 YN1 mAb binding in COS-7 cells expressing WT-ICAM-1, and this effect was blocked in cells expressing the phospho-defective Y518F-ICAM-1 mutant. YN1 mAb binding to cells expressing Y518D-ICAM-1 was 31% greater (P<0.05) than in non-stimulated COS-7 cells expressing WT-ICAM-1. Thus, phosphorylation of ICAM-1 Tyr⁵¹⁸ was associated with increased ICAM-1 B_max for YN1 mAb, which was similarly increased with the phospho-mimicking Y518D ICAM-1 mutant. There was no effect of TNFα on YN1 mAb binding to phospho-defective Y518F ICAM-1 mutant. The K_d of mAb YN1 binding did not differ in any of the groups. Interestingly, there was no effect of TNFα on KAT-1 binding to WT-ICAM-1 and also KAT-1 binding to Y518F and Y518D-ICAM-1 mutants was similar (i.e., there was no difference in B_max or K_d)(Fig. 5B), indicating that WT and mutant ICAM-1 constructs were equally expressed. Thus, ICAM-1 intracellular domain (Tyr⁵¹⁸) phosphorylation increases its avidity for LFA-1 reflected by the specific increase in anti-ICAM-1 YN1 mAb binding.

Figure 5. Anti-ICAM-1 mAb binding to COS-7 cells expressing WT and mutant ICAM-1 forms.

A–B: COS-7 cells were transfected with WT-, Y518F-, and Y518D-ICAM-1 cDNA and 48 hr later, stimulated with TNFα for 20 min and incubated with 0.1–30 μg/ml rat anti-mouse ICAM-1 mAbs YN1/1.7.4 or KAT-1 followed by fixation with 2% paraformaldehyde on ice. Primary mAb binding was assessed using alkaline phosphatase-labeled goat-anti mouse secondary Ab and spectrophotometrically measured at OD450 nm; ELISA data from 3 independent transfections and binding isotherms were analyzed by non-linear curve-fitting using GraphPad Prism. (A). Non-specific binding (OD450 = 0.27) measured in empty vector transfected cells was subtracted. (B) B_max values (mean ± SD; n=3) normalized relative to WT-ICAM-1 B_max in non-stimulated cells are shown along with the mean K_d values determined from each curve. Note equal KAT-1 mAb B_max values indicating equal expression levels of the constructs. * = P<0.05 vs WT-ICAM-1. C: Membrane staining of WT-, Y518F-, and Y518D-ICAM-1 expressed in HUVECs. Cells were transduced with ICAM-1 cDNA constructs and after 48 hr, stimulated with 100 ng/ml TNFα for 20 min, immunostained with anti-ICAM-1 mAb and goat-anti-mouse Alexa 488 secondary Ab (green), and co-stained with Alexa 546 phalloidin (red). Confocal images are representative of 3 separate studies.

Cellular localization of WT-ICAM-1 as well as Y518F and Y518D-ICAM-1 mutants is shown in Figure 5C and Online Data Supplement Figure II. HUVECs were transiently transfected with ICAM-1 cDNA constructs and after 48 hr, stimulated with 100 ng/ml TNFα for 20 min and immunostained with anti-ICAM-1 mAb plus goat-anti-mouse Alexa 488 (green) and Alexa 546 phalloidin (red). In unstimulated cells, WT-ICAM-1 was uniformly distributed in the plasma membrane whereas in TNFα-stimulated cells, ICAM-1 clustering was observed in association with phalloidin-labeled F-actin filaments. In contrast, TFNα-treated Y518F-ICAM-1 mutant expressing cells showed reduced ICAM-1 clustering and lack of association with actin. However, we observed robust clustering of phospho-mimicking Y518D-ICAM-1 mutant along actin filaments and thus ICAM-1 phosphorylation induces clustering and association with polymerized cortical actin filaments. Thus, membrane-associated ICAM-1 staining in non-stimulated WT-ICAM-1 expressing cells (see also, Online Fig. IIA) and TNFα-stimulated Y518F-ICAM-1 expressing cells (see also, Online Fig. IIC) was diffuse, while in TNFα-stimulated WT-ICAM-1 (see also, Online Fig. IIB) and non-stimulated Y518D-ICAM-1 (see also, Online Fig. IID) expressing cells, ICAM-1 staining intensity increased and was observed in clusters.

Expression of Phospho-defective ICAM-1 Mutant Prevents LPS-induced Lung Inflammatory Injury in Mice

Adult male C57BL6 mice (n=3/group) were transduced with pcDNA3.1 (empty vector), WT ICAM-1, or Y518F ICAM-1 cDNA constructs by retro-orbitol injection of liposome-cDNA complexes to assess responses to LPS-induced lung inflammatory injury. After 48 hr, mice were exposed to nebulized LPS (10 mg over 1 hr) and bronchoaveolar lavage (BAL) fluid was collected at time 0 and at 2 hr and 24 hr after LPS exposure. LPS induced a significant increase in ICAM-1 Tyr⁵¹⁸ phosphorylation at 2 and 24 hr as well as increased endogenous ICAM-1 expression (Fig. 6A). Western blotting of expressed ICAM-1 two days after cDNA transfection revealed equal levels of GFP-tagged WT and GFP-tagged Y518F mutant ICAM-1 expression (Fig. 6B, GFP blot). PMN counts (Fig. 6C) and albumin concentration (Fig. 6D) of BAL fluid were determined as described in Methods. As shown in Figure 6C, overexpression of WT mouse ICAM-1 in mouse lungs significantly increased PMN transmigration and accumulation in lungs compared to Y518F ICAM-1 or empty vector transfected lungs at 2 hr and 24 hr after LPS. Furthermore, overexpression of WT ICAM-1 significantly increased LPS-induced pulmonary transvascular vascular albumin leakage at 2 and 24 hr after LPS inhalation compared to lungs expressing Y518F phospho-defective ICAM-1 mutant and empty vector transfected lungs (Fig. 6D).

Figure 6. Expression of phospho-defective ICAM-1 mutant in mouse lungs prevents LPS-induced PMN recruitment and vascular injury.

pcDNA3.1 (empty vector), WT-ICAM-1, or Y518F-ICAM-1 cDNA were reconstituted in cationic liposomes and injected i.v. in mice to induce lung ICAM-1 expression as described²⁷. After 48 hr, mice were exposed to nebulized LPS (10 mg over 1 hr) and bronchoaveolar lavage (BAL) fluid was collected at 2 hr or 24 hr after initiation of LPS exposure. A. LPS exposure of control mouse lungs for 1 hr induced an increase ICAM-1 Tyr⁵¹⁸ phosphorylation at 2 and 24 hrs as well as increased ICAM-1 expression. B. Western blot analysis shows equal expression of GFP-tagged WT-ICAM-1 and Y518F-ICAM-1 48 hr after i.v. delivery of liposome/cDNA complexes. PMN infiltration (C) and transvascular albumin leakage (D) measured in BAL fluid increased at 2 and 24 hr after LPS challenge in mice transduced with the empty vector or WT-ICAM-1, but was reduced in mice expressing the phospho-defective ICAM-1 mutant. * = P<0.05 vs. empty vector and Y518F ICAM-1 transfected mice; data shown are mean ± S.D.; n=3/group.

Discussion

In the present study, we addressed the molecular mechanism and functional significance of rapid-phase TNFα-induced phosphorylation of Tyr⁵¹⁸ in the C-terminal intracellular tail of ICAM-1 in mouse lungs and cultured mouse lung endothelial cells (MLECs). Our results demonstrate that TNFα increases anti-ICAM-1 mAb binding to ICAM-1 and recruitment of PMNs in lungs that is dependent on Nox2 and its upstream activators, PI3 kinase-γ and PKCζ, and Src activation. PMNs were not recruited into lungs of ICAM-1^−/− mice after TNFα challenge indicating the requirement of Src-dependent phosphorylation of ICAM-1 in mediating the response. To assess changes in ICAM-1 adhesivity in vessels, we measured TNFα-induced changes in binding activity of ICAM-1 using a specific ¹²⁵I-labeled anti-ICAM-1 mAb (YN1/1.7.4), which we corrected for non-specific binding by co-administering a non-binding ¹³¹I-labeled control mAb as described^12,13. The rapid increase in anti-ICAM-1 mAb binding occurred within 5 minutes of TNFα treatment, and was thus independent of de novo ICAM-1 synthesis which requires at least 2 hr. We detected a further increase in mAb binding in mouse lungs that resulted from de novo ICAM-1 synthesis only after 2 hr of TNFα exposure, consistent with previous observations^{5,24,17,18,34}. Thus, TNFα mediates rapid recruitment and sequestration of PMNs through the phosphorylation-dependent modification of endothelial cell surface ICAM-1.

As TNFα has been shown to stimulate the generation of reactive oxygen species (ROS) in endothelial cells,^{17,21,22,24,25,35} we investigated whether NADPH oxidase, specifically Nox2, plays a role in signaling the rapid increase in ICAM-1 binding and recruitment of PMNs. We found that p47^phox and gp91^phox subunits, essential protein constituents of Nox2 NADPH oxidase, were required for the rapid-phase TNFα-induced increase in ICAM-1 binding activity and lung PMN sequestration. To address the requirement for ROS, we demonstrated that brief exposure of lung vessels to subtoxic levels of superoxide, generated by addition of xanthine and xanthine oxidase, stimulated PMN recruitment in gp91^phox−/− mouse lung preparations.

Because PI3 kinaseγ and PKCζ are sequential upstream activators of Nox2 in endothelial cells²⁴, we predicted that TNFα should activate both kinases. Using histone H1 as a substrate to measure PKCζ activity,¹¹ we observed that TNFα induced PKCζ activation in MLECs within 5 minutes. We also established the physiological relevance of increased PKCζ activity using a specific PKCζ blocking peptide, which was shown to prevent TNFα-induced PMN sequestration in murine lungs. Our results fit with the role of PKCζ activation of NADPH oxidase assembly through serine-phosphorylation of cytoplasmic p47^phox and its subsequent translocation to the membrane to induce NADPH oxidase complex formation^17,21,22,24.

Because PKCζ-dependent activation of NADPH oxidase was required for TNFα-induced increase in PMN binding to endothelial cells, we surmised that ROS generated downstream of PKCζ might activate tyrosine kinases such as c-Src ^25,26. We observed that TNFα caused a rapid phosphorylation of ICAM-1 on Tyr⁵¹⁸. We did not detect an increase in the phosphorylation of ICAM-1 on serine or threonine residues, making it unlikely that PKCζ directly phosphorylated ICAM-1 in MLECs stimulated with TNFα; thus, PKCζ most likely acts upstream of a tyrosine kinase such as Src. Studies have shown that Src is important in the ICAM-1-dependent PMN transmigration response^19,27–33. Data presented here support such a mechanism since TNFα rapidly activated c-Src kinase in lungs of WT but not Nox2-deficient mice. Based on our findings, we propose that TNFα increases PI3-kinaseγ activity which stimulates PKCζ-dependent phosphorylation of p47^phox, thereby promoting Nox2 assembly and ROS production. Thus, in this scheme TNFα-induces the activation of c-Src by oxidants which in turn leads to Src-mediated phosphorylation of ICAM-1 Tyr⁵¹⁸.

Cell surface β2 integrins on PMNs bind endothelial expressed ICAM-1^6–10. The effect of TNFα in inducing Tyr⁵¹⁸ phosphorylation-dependent ICAM-1 binding may be due to increased ICAM-1 affinity or ICAM-1 avidity^14,15. Both alterations in ICAM-1 can mediate increased PMN binding and sequestration of PMNs in lungs. To address whether pTyr⁵¹⁸-ICAM-1 could account for the enhanced adhesivity of ICAM-1, we first investigated whether TNFα induces ICAM-1 clustering through association with the actin cytoskeleton. Src inhibition in endothelial cells blocked TNFα-induced ICAM-1 tyrosine phosphorylation and ICAM-1 association with actin. Furthermore, inhibition of Src blocked PMN sequestration in mouse lungs induced by 10 minute exposure to TNFα indicating ICAM-1 phosphorylation by Src mediates the rapid recruitment of PMNs to lungs.

How ICAM-1 phosphorylation modification increases binding to CD11/CD18 integrins on PMNs is not clear. To identify whether ICAM-1 binding affinity or avidity were modified by the phosphorylation of ICAM-1 at Tyr⁵¹⁸_, we assessed ICAM-1 binding kinetics and membrane localization. For this we generated a non-phosphorylatable (Y518F) ICAM-1 mutant mouse cDNA as well as a phospho-mimicking (Y518D) mutant and studied binding kinetics of non-overlapping rat anti-mouse ICAM-1 extracellular D1 domain mAbs YN1/1.7.4^6–8 and eBioKAT-1 (eBioscience.com) as well as ICAM-1 localization in the plasma membrane. The data presented show that ICAM-1 Tyr⁵¹⁸ phosphorylation leads to ICAM-1 clustering, increased binding of YN1/1.7.4 mAb, and a 3-fold increase in PMN adhesion. Since YN1/1.7.4 recognizes the neutrophil LFA-1 integrin binding site on ICAM-1⁹, we surmise that TNFα-induced ICAM-1 phosphorylation at Tyr⁵¹⁸ by Src makes the extracellular ICAM-1 LFA-1-interacting epitope more available for binding to PMNs via β2 integrin. Furthermore, we did not observe an increase in KAT-1 mAb binding (B_max or K_d) to WT-ICAM-1 following treatment of cells with TNFα or to phospho-mimicking Y518D-ICAM-1 mutant relative to phospho-defective ICAM-1 mutant Y518F. Thus, consistent with the hypothesis that ICAM-1 phosphorylation increases its avidity, only YN1/1.7.4 mAb binding increased. Therefore, we conclude that TNFα-induced rapid PMN sequestration in lung microvessels is due to enhanced LFA-1 integrin/ICAM-1 interaction rather than increased expression level of ICAM-1.

We also assessed the pathophysiological significance of ICAM-1 phosphorylation in LPS-induced lung PMN infiltration and vascular hyperpermeability. For these studies, control mice were injected i.v. with liposome-ICAM-1 cDNA complexes and after 48 hr when the proteins are expressed³⁶, the mice were challenged with LPS. Inhalation of nebulized LPS induced a significant increase in ICAM-1 Tyr⁵¹⁸ phosphorylation, lung lavage PMN counts, and albumin concentration by 2 hr which further increased at 24 hr. LPS-induced increase in expression of ICAM-1 at 24 hr significantly increased PMN infiltrates in lungs and this was associated with increased albumin levels in lavage fluid indicative of inflammation and vascular hyper-permeability, the hallmarks of acute lung injury. Interestingly, in mice transduced with the phospho-defective Y518F ICAM-1 mutant, LPS-induced increase in lavage PMN counts and albumin accumulation in lungs did not increase over values observed in mice transduced with the empty vector. Thus, ICAM-1 phosphorylation at Tyr⁵¹⁸ plays a pivotal role in vascular inflammation and lung vascular hyper-permeability induced by endotoxin.

In summary, we showed that TNFα rapidly increased both ICAM-1 binding avidity and PMN sequestration in lung microvessels and cultured endothelial cells. These events were independent of de novo ICAM-1 synthesis but required generation of ROS through Nox2, Src activation, and Src-dependent ICAM-1 phosphorylation at Tyr⁵¹⁸. The phosphorylation of ICAM-1 induced ICAM-1 clustering in endothelial cells and its increased avidity for PMNs resulting in PMN adhesion and recruitment in tissue. We posit that this rapid post-translational modification dependent mechanism of ICAM-1 binding to PMNs may play an important role in inflammatory diseases such as acute lung injury.