ICAM-1–activated Src and eNOS signaling increase endothelial cell surface PECAM-1 adhesivity and neutrophil transmigration

Tissue polymorphonuclear neutrophil (PMN) infiltration is required for effective host defense, but the recruitment and activation of PMNs, if unchecked, can also produce inflammation and injury.¹  For PMNs to cross microvessels, they must first adhere firmly to endothelial cells (ECs),²  which form a cup-like structure surrounding adherent leukocytes to direct the cell through the barrier.^3,4  Intercellular adhesion molecule-1 (ICAM-1)–mediated firm adhesion of leukocytes to ECs is a requisite step in PMN transmigration.²  Platelet/endothelial cell adhesion molecule 1 (PECAM-1) expressed in the EC plasma membrane is also critically involved in the mechanism of PMN transmigration.^5,6  Based on the key roles of both ICAM-1 and PECAM-1, a fundamental question is whether these adhesive proteins cooperate in the mechanism of PMN transmigration.

Activation of ICAM-1, as defined by ICAM-1 phosphorylation,^7,8  induces ICAM-1 clustering and its association with the actin cytoskeleton.^9,10  Multiple signaling pathways, such as Ca²⁺,¹¹  RhoA,^12,13  RhoG,⁴  MAPK (p38),¹⁴  and Src,^11,15-17  have been invoked to explain the mechanism of ICAM-1 activation and clustering. Recently, it was also shown that ICAM-1 ligation-dependent endothelial nitric oxide synthase (eNOS) activation via the AMPK pathway in rat brain endothelial cells signals lymphocyte transmigration.¹⁸  However, although Src and eNOS activation have been demonstrated to play key roles in leukocyte transmigration,¹⁶  their mechanism of action is still not known. We have shown⁸  Src-dependent ICAM-1 phosphorylation at Tyr⁵¹⁸. However, it is also possible that ICAM-1, on activation, can itself stimulate Src kinase activity placing Src both upstream and downstream of ICAM-1. Here we addressed the question whether the initial firm ICAM-1–mediated PMN adhesion can induce PECAM-1–dependent PMN transmigration and whether activation of Src and eNOS signaling downstream of ICAM-1 is an essential requirement of the PECAM-1–dependent inflammatory response.

There is precedence for invoking Src activation downstream of ICAM-1. The interaction of ICAM-1 with fibrinogen was shown to induce ICAM-1 phosphorylation on Tyr⁵¹⁸⁷  as well as monocyte transmigration.¹⁹  Tyrosine phosphatase SHP-2 interacting with Tyr⁵¹⁸ in the ICAM-1 C-terminal domain ITAM motif (I/V/L)XYXX(l/V)⁷  may regulate endothelial cell Src activity. An engineered SHP-2 mutant targeted to the plasma membrane was shown to stimulate Src activity,²⁰  suggesting a role of SHP-2 in mediating ICAM-1 dependent Src signaling. Src is thus a potentially important target that may be the key to understanding PMN transmigration. Src also induces PI3K-dependent Akt phosphorylation and eNOS activation by phosphorylating eNOS at Ser¹¹⁷.^7,21  Thus, ICAM-1 activation-dependent eNOS signaling downstream of ICAM-1 activation may also be involved in PMN transmigration.

Wild-type (WT) C57BL/6 and knockout strains (ICAM-1^−/− and eNOS^−/−) in the same background were purchased from The Jackson Laboratory. Mice weighing 20-30 g and 10-12 weeks of age were used. All animal procedures were approved by the University of Illinois Animal Care and Use Committee in accordance with Association for Assessment and Accreditation of Laboratory Animal Care guidelines.

Polyclonal ICAM-1, c-Src, and GAPDH antibodies and monoclonal PECAM-1 and SHP-2 antibodies were from Santa Cruz Biotechnology. Phospho-Tyr⁵¹⁸ ICAM-1 antibody was from Abcam. Active phospho-Src (mouse Y418 or human Y419), Src inactive site phospho-Tyr (mouse Y529 or human Y530), phospho-eNOS (S1177), phospho-Akt (T473), and total Akt were from Cell Signaling. Monoclonal anti-eNOS was from EMD Biosciences. Monoclonal anti–mouse ICAM-1 antibody (YN1/1.7.4), FITC-480–labeled YN1/1.7.4, and control rat IgG2b used in ICAM-1 crosslinking studies were from eBioscience. Phospho–PECAM-1 (Tyr⁶⁸⁶) polyclonal antibody was from Assay Biotechnology Company. Monoclonal M2 anti-Flag antibody and secondary anti–rat antibody used in crosslinking studies were from Sigma-Aldrich. PP2, wortmannin, and L-NAME were from Calbiochem. Carrageenan, protease inhibitor cocktail, and all other chemical reagents were from Sigma-Aldrich. AMP kinase inhibitor 5-iodotubercidin was from Biaffin GmbH & Co KG.

HUVECs were used between passage 5 ad 7 and transiently transfected with mouse ICAM-1 cDNA by Lonza Nucleofector kit according to the manufacturer's instructions.⁸  Expression of mouse ICAM-1 in human endothelial cell monolayers was used to assess downstream signaling mechanisms evoked by ICAM-1 ligation per se compared with phospho-defective ICAM-1 mutant in the absence of cytokine stimulation known to activate similar signaling pathways.⁸  For coimmunoprecipitation assay, HEK cells cultured in DMEM with 10% FBS were used for transfection of ICAM-1.

ICAM-1 was crosslinked as described.^22,23  Briefly, HUVECs expressing mouse ICAM-1 were washed once with EBM-2 and incubated with 15 μg/mL rat anti–mouse ICAM-1 monoclonal antibody at room temperature for 20 minutes. Rat IgG2b was used as a control antibody. The cells were then washed twice and anti–rat IgG (50 μg/mL) was added to crosslink ICAM-1 for 20 minutes at 37°C.

After ICAM-1 crosslinking, HEK or HUVECs were placed on ice and lysed in buffer containing 50mM Tris-HCl, pH7.5, 150mM NaCl, 1% Triton X-100, 1mM PMSF, and Sigma protease inhibitor. After low-speed centrifugation (100g, 10 minutes, 4°C), the supernatants were collected for coimmunoprecipitation and Western blot. Protein concentration was determined with Bio-Rad D_c protein assay solution (Bio-Rad). For coimmunoprecipitation, SHP-2 antibody was incubated with the lysates overnight at 4°C and Protein A beads were added and incubated for 1 hour at 4°C. The precipitated proteins were eluted by sample buffer. SHP-2 and ICAM-1 in the precipitated proteins were analyzed with Western blot.

To determine PMN infiltration, myeloperoxidase (MPO) activity of lung lysates was measured and used as a marker of PMN accumulation. Four hours after intrapleural injection of 0.2 mL of 1% carrageenan or saline, mouse lungs were homogenized in 0.5% hexadecyl-trimethylammonium bromide dissolved in 10mM potassium phosphate buffer (pH 7). After centrifugation for 20 minutes at 20 000g at 4°C, aliquots of the supernatant were assessed for total protein concentration and MPO activity. A 10-μL aliquot of each sample (supernatant) was loaded per well of a 96-well plate, and then O-dianisidine dihydrochloride with 0.0005% hydrogen peroxide in phosphate buffer (190 μL per well) was added to the samples. Absorbance readings were measured at 460 nm for 3 minutes. MPO activity was expressed as the change in absorbance per minute per gram of tissue.

After ICAM-1 crosslinking or addition of activated PMNs to EC monolayers, NO produced by HUVECs in 12-well plates was measured using a porphyrinic NO electrode as described.²¹  Briefly, the electrode is created by coating carbon fibers with a metalloporphyrinic conductive polymer and subsequently sealed with Nafion. Each electrode is calibrated using a stock solution of NO-saturated water. NO diffusion into the Nafion membrane is oxidized to a nitrosyl ion, and the electron is transferred to the porphyrin of the conductive polymer, proceeding along the copper wire to a detector. The NO electrode is placed onto the surface of an EC monolayer, and 2 additional electrodes are added to the solution to generate a 650-mV potential. The system is coupled to a FAS1 femtostat and personal computer with electrochemical software (Gamry Instruments). Electrode current, which is proportional to NO concentration, is measured as a function of time. The cell culture medium temperature is maintained at 37°C.

PMNs were isolated from WT C57BL/6 mouse blood. After blood sedimentation, the PMNs were collected from the plasma layer by centrifugation. After removal of red blood cells by hypotonic shock, PMNs were further isolated with Ficoll-Pacque PLUS (GE Healthcare) gradient centrifugation. The yield of PMNs was approximately 3 × 10⁶ PMNs/mouse with a purity of > 90% and viability of > 95% as determined by Trypan blue exclusion.²³  Isolated PMNs were labeled with fluorescent LeukoTracker solution (Cell Biolabs) according to the manufacturer's protocol and used immediately in adhesion and transendothelial migration assays.

To determine PMN adhesion, we used the CytoSelect Leukocyte-endothelium adhesion assay kit (Cell Biolabs). Briefly, confluent monolayers of transfected HUVECs were seeded onto gelatin-coated 24-well plates in 300 μL of EBM-2 medium supplemented with 2% FBS. Peripheral blood PMNs were labeled with fluorescent LeukoTracker for 60 minutes at 37°C and then added to the EC monolayer for 30 minutes. After washing with PBS, the remaining adherent PMNs were lysed and fluorescence measured with a fluorometer (FlexStation II; Molecular Devices) at excitation and emission wavelengths of 485 and 535 nm, respectively. To determine PMN transmigration, we used the CytoSelect Leukocyte Transmigration Assay kit (Cell Biolabs). HUVECs transfected with empty vector, WT mouse ICAM-1, and mutant mouse ICAM-1 cDNA were cultured in 24-well inserts (3 μm pore size) until confluent. After overnight 0.1% FBS incubation, LeukoTracker-labeled PMNs were added to ECs at a ratio of 10:1 and the inserts were placed in each well of a 24-well plate with culture medium containing PMN activator fMLP (1μM) for 3 hours. PMNs that transmigrated to the bottom chamber were collected, lysed, and the fluorescence level of supernatants determined with a fluorometer.

After ICAM-1 crosslinking of transfected HUVECs in 96-well plates for 1 hour, the cells were subjected to PECAM-1 mAb binding assay. Cells were washed with serum-free medium and incubated with the PECAM-1 mAb diluted in medium (10 μg/mL) at 4°C for 1 hour. Cells were then fixed with 4% paraformaldehyde for 15 minutes at 4°C followed by PBS washing and incubation with HRP-labeled secondary Ab. After incubation with substrate TMB (3,3′,5,5′-tetramethylbenzidine), the optical density at 450 nm was measured and recorded.

All data are expressed as the mean ± SD unless otherwise indicated. Statistical differences between groups were determined using ANOVA or Student t test depending on group size. P values less than .05 were considered significant.

We used the mouse model of carrageenan-induced lung inflammation²⁴  to address the role of ICAM-1–activated signaling in PMN extravasation. The initial phase of the carrageenan response involves secretion of TNF-α and IL-1β during the first 1 hour, followed by PMN infiltration over the next 2-6 hours.²⁴  Thus, C57BL/6 mice were injected with 1% carrageenan or saline (sham) into the pleural space; then, lungs were collected and analyzed after 4 hours. Carrageenan injection induced a 2-fold increase in ICAM-1 expression (Figure 1A). Using phospho-specific antibodies, we also observed increased phosphorylation of Src and eNOS in lung tissue of carrageen-challenged mice compared with the saline-injected control mice (Figure 1B).

Leukocyte firm adhesion requires several adhesion molecules, such as ICAM-1, ICAM-2, and VCAM-1.²⁵  Here we focused on the role of ICAM-1, which is essential for PMN firm adhesion to ECs²⁶  and thus also thought to be required for subsequent PMN transmigration. We examined whether carrageenan-induced phosphorylation of Src and eNOS was dependent on ICAM-1 expression. Western blotting showed that auto-phosphorylation of mouse Src on Tyr⁴¹⁸ and eNOS at Ser¹¹⁷⁷ was markedly reduced in ICAM-1^−/− lungs basally and after carrageenan challenge (Figure 2A).

To address in vivo relevance of eNOS activation downstream of ICAM-1, we assessed carrageenan-induced pleurisy in eNOS knockout mice (eNOS^−/−). WT or eNOS^−/− mice were injected with carrageenan and ICAM-1 expression, and MPO activity was measured after 4 hours. Interestingly, we observed a 6-fold increase in basal ICAM-1 expression in eNOS^−/− mice, which increased further on carrageenan challenge (Figure 2B), suggesting that eNOS, in addition to being essential for PMN sequestration and transmigration, may also negatively regulate ICAM-1 expression. PMN lung infiltration determined by MPO activity was increased 3-fold at 4 hours in WT control mice, whereas ICAM-1 knockout mice (ICAM-1^−/−) showed 70% less MPO activity in lungs 4 hours after injection of carrageenan compared with WT control (C57BL/6) mice (Figure 2C). The increase in lung MPO activity induced by carrageenan injection in eNOS^−/− mice was also 50% less than in WT mice (Figure 2D). Consistent with these data, we observed 50% fewer PMNs in bronchoalveolar lavage samples from carrageenan-treated ICAM-1^−/− and eNOS^−/− mice (supplemental Figure 3C, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Thus, regardless of the increase in ICAM-1 expression observed in eNOS^−/− mice, PMN transmigration was reduced, suggesting that NO production plays a critical role in the mechanism of PMN extravasation in mouse lungs.

As carrageenan-induced activation of Src and eNOS is attenuated in ICAM-1^−/− mice, we next examined the role of ICAM-1 and its phosphorylation in the mechanism of Src and eNOS activation and signaling. WT and Y518F phosphorylation-defective mouse ICAM-1 mutant were expressed in HUVECs. Mouse ICAM-1 cDNA transfection resulted in equivalent cell surface expression of WT- and Y518F–ICAM-1 as evidenced by Western blotting and whole-cell ELISA assays (Figure 3A,C). Flow cytometric analysis demonstrated that ∼ 50% of HUVECs expressed the WT and mutant ICAM-1 at the similar levels (supplemental Figure 1A). In addition, expressed mouse WT and mutant ICAM-1 levels were comparable with endogenous ICAM-1 expression in HUVECs treated with TNF-α for 2 to 4 hours (Figure 3B), suggesting that the level of expression of WT and mutant mouse ICAM-1 in HUVECs was similar to endogenous ICAM-1 expression during the early stages of inflammation.

On ICAM-1 crosslinking, WT–ICAM-1–expressing cells showed ICAM-1 phosphorylation at Tyr⁵¹⁸ as well as human Src auto-phosphorylation (pTyr⁴¹⁹), respectively (Figure 3C). However, cells expressing the phosphorylation-defective Y518F–ICAM-1 mutant showed neither ICAM-1 nor Src Tyr⁴¹⁹ phosphorylation after ICAM-1 crosslinking (Figure 3C). Thus, intracellular C-terminal ICAM-1 tyrosine Tyr⁵¹⁸ phosphorylation plays an important role in Src signal amplification downstream of ICAM-1 crosslinking and activation.

SHP-2 phosphatase binds to the ICAM-1 ITAM motif⁷  and hence may play a role in Src activation downstream of ICAM-1. ICAM-1 crosslinking-induced Src phosphorylation was analyzed by Western blotting to determine whether SHP-2 is required for Src activation. HUVECs were cotransfected with WT mouse ICAM-1 and WT- or phosphatase-defective Cys463Ser–SHP-2 mutant (C463S–SHP-2). After ICAM-1 crosslinking in ECs expressing WT–SHP-2, we observed increased Src Tyr⁴¹⁹ phosphorylation, whereas a much weaker response was observed in cells expressing C463S–SHP-2 (Figure 3D). Phosphorylation of human Src Tyr⁵³⁰, the inhibitory phospho-tyrosine in the Src C-terminal regulatory domain,^27,28  decreased in cells expressing WT–SHP-2 after addition of ICAM-1 crosslinking mAb but not in cells expressing C463S–SHP-2 (Figure 3D). Cells expressing the catalytically inactive C463S SHP-2 mutant also showed less Src Tyr⁴¹⁹ auto-phosphorylation and no change in inhibitory Src Tyr⁵³⁰ dephosphorylation compared with that in cells expressing WT–SHP-2 (Figure 3D). Both Tyr⁴¹⁹ and Tyr⁵³⁰ are important for accurately assessing Src activity by phospho-Src immunoblot analysis^27,28 ; thus, a ratio of active site and inhibitory site Src tyrosine phosphorylation (pTyr⁴¹⁹/pTyr⁵³⁰) was calculated. As shown in Figure 3D, the ratio increased on ICAM-1 crosslinking in cells transfected with WT–SHP-2 but did not change significantly in cells transfected with dominant-negative SHP-2 mutant (Figure 3D). We also observed that phosphorylation of ICAM-1 Tyr⁵¹⁸ induced by crosslinking increased in cells expressing WT–SHP-2 (Figure 3D). In contrast, there was no change of ICAM-1 Tyr⁵¹⁸ phosphorylation induced by the crosslinking protocol in cells expressing the C463S–SHP-2 mutant (Figure 3D), indicating less Src activity in these cells. Taken together, these data indicate SHP-2 activity is required for ICAM-1 crosslinking-induced Src activation and ICAM-1 phosphorylation, most likely via SHP-2 mediated dephosphorylation of Src inhibitory Tyr⁵³⁰.

To determine the mechanism by which SHP-2 mediates ICAM-1 crosslinking-induced Src activation, we examined the interaction between SHP-2 and ICAM-1 by coimmunoprecipitation in HEK cells transduced with WT mouse ICAM-1. ICAM-1 association with SHP-2 was reduced after ICAM-1 crosslinking (supplemental Figure 2A). Both Y518F and Y518D (phosphor-defective and phosphor-mimicking ICAM-1 mutants) showed reduced coimmunoprecipitation (supplemental Figure 2B) with SHP-2 compared with WT–ICAM-1. Taken together, these data suggest that the hydroxyl group of ICAM-1 Y518 may be involved in coordinating SHP-2 binding within the ITAM motif, and that crosslinking-induced activation of ICAM-1 leads to displacement and activation of SHP-2 and dephosphorylation of the Src inhibitory phospho-Tyr⁵³⁰ residue.

We next examined eNOS Ser¹¹⁷⁷ phosphorylation in ICAM-1–transfected HUVECs after ICAM-1 crosslinking to address the possible involvement of eNOS downstream of ICAM-1 activation in human ECs. ICAM-1 crosslinking induced both eNOS phosphorylation at Ser¹¹⁷⁷ and Akt phosphorylation at Ser⁴⁷³ in cells transduced with WT–ICAM-1, whereas no effect was observed in cells expressing the phospho-defective ICAM-1 mutant (Figure 4A). eNOS is known to be phosphorylated at Ser¹¹⁷⁷ by Akt²⁹ downstream of PI3K and Src,²¹  and also by AMP kinase (AMPK).^18,30  Thus, we next determined the effect of Src inhibitor PP2 (10μM), PI3K inhibitor wortmannin (100nM) on ICAM-1 induced Src, Akt, and/or eNOS phosphorylation. As shown in Figure 4B, phosphorylation of ICAM-1 and Src were blocked by Src inhibitor PP2, but not by PI3K inhibitor wortmannin. In contrast, phosphorylation of Akt Ser⁴⁷³ and eNOS Ser¹¹⁷⁷ was blocked by both PP2 and wortmannin (Figure 4B), indicating that Akt and eNOS activation lies downstream of ICAM-1–induced Src and PI3K activation.

Studies by Martinelli et al revealed that ICAM-1 monoclonal Ab alone activates eNOS via AMPK-dependent signaling in rat brain endothelial cells.¹⁸  Thus, we also examined whether primary antibody alone compared with crosslinking ICAM-1 (using both a primary Ab and secondary crosslinking Ab) induced eNOS phosphorylation and whether AMPK inhibition by 5-iodotubercidin³¹  blocks eNOS phosphorylation. Incubation with primary antibody alone had no effect on eNOS phosphorylation, and 5-indotubercidin reduced ICAM-1 induced eNOS phosphorylation (supplemental Figure 1B). Thus, the mechanism of mouse ICAM-1 induced eNOS activation appears to be dependent on PI3K and Akt, and to some extent by AMPK, whereas in rat brain endothelial cell eNOS activation occurs via AMPK-dependent signaling.¹⁸ 

We next measured ICAM-1–induced eNOS-derived NO production to assess the functional significance of eNOS phosphorylation via the Src/PI3K/Akt pathway. HUVECs transfected with WT- or the Y518F–ICAM-1 mutant were subjected to ICAM-1 crosslinking and real-time NO production was measured by porphyrinic electrode as described.²¹  We observed NO production in WT–ICAM-1–transfected cells but not in cells transfected with the phospho-defective ICAM-1 mutant (Figure 5A). We also determined whether activation of PMNs layered on ECs (ratio of 10 PMNs/1 EC) stimulated NO production in an ICAM-1–dependent manner. Mouse PMNs activated with 10μM fMLP were added to HUVEC cultures transfected with WT mouse ICAM-1 or phospho-defective mouse ICAM-1 mutant, and NO was measured. Activated PMNs added to ECs transfected with the phospho-defective ICAM-1 mutant failed to induce NO production, whereas WT–ICAM-1 produced a 2-fold increase in NO production via eNOS (supplemental Figure 2C). To determine the role of Src and PI3K signaling in mediating ICAM-1–mediated NO production, PP2 and wortmannin were added to HUVECs and NO production was measured 30 minutes after crosslinking. We observed that ICAM-1 activation-dependent NO production required the Src-PI3K-Akt signaling pathway (Figure 5B) consistent with the eNOS Ser¹¹⁷⁷ phospho-immunoblots (Figure 4B).

We next determined whether expression of Y518F, the phospho-defective ICAM-1 mutant, affects PMN adhesion or transmigration through transduced human EC monolayers. ICAM-1–dependent adhesion of mouse PMNs to EC monolayers expressing WT- or Y518F-mouse ICAM-1 was not different (supplemental Figure 2D). However, as shown in the transmigration assay, Y518F–ICAM-1 expression resulted in a 50% decrease in PMN migration relative to that observed in cells expressing WT–ICAM-1 (Figure 5C). WT–ICAM-1–dependent PMN transmigration was also blocked by PP2, wortmannin, and L-NAME (Figure 5D), suggesting that the Src-PI3K-Akt-eNOS pathway plays an important role in the mechanism of ICAM-1–dependent PMN transmigration.

To understand how eNOS activation regulates PMN transmigration, we examined the role of the ICAM-1/Src/eNOS pathway in the regulation of cell surface expression and phosphorylation of PECAM-1 Tyr⁶⁸⁶. It was previously reported that PECAM-1, which is essential for leukocyte transmigration,^6,32  can be phosphorylated on 2 tyrosine residues within the intracellular C-terminal domain.^33,34  In transfected HUVECs, crosslinking of mouse WT–ICAM-1, but not that of the phospho-defective Y518F– ICAM-1 mutant, induced the phosphorylation of endogenous human PECAM-1 Tyr⁶⁸⁶ (Figure 6A-B). Furthermore, ICAM-1 crosslinking-induced PECAM-1 phosphorylation was blocked by PP2 and L-NAME, suggesting that ICAM-1 activation-dependent phosphorylation of PECAM-1 is the consequence of Src and eNOS (Figure 6B). Moreover, in eNOS^−/− mouse lungs, basal and carrageenan-stimulated PECAM-1 Tyr⁶⁸⁶ phosphorylation was significantly reduced compared with that observed in WT mouse lungs (Figure 6C), suggesting that activation of PECAM-1 downstream of ICAM-1 and eNOS is an essential component of the inflammatory response in vivo.

Homophilic interaction between leukocyte PECAM-1 and endothelial cell PECAM-1 is thought to mediate leukocyte transmigration.⁶  Thus, we next examined the effect of ICAM-1 activation on PECAM-1 binding activity in cultured endothelial monolayers. Mouse ICAM-1 expressing HUVECs were treated with ICAM-1 crosslinking mAb and assayed for cell surface PECAM-1 binding activity by a cell-attached ELISA assay.⁸ Figure 6D shows that ICAM-1 activation significantly increased PECAM-1 mAb binding activity. Moreover, HUVECs expressing the phosphorylation-defective mouse ICAM-1 mutant (Tyr⁵¹⁸Phe) showed significantly reduced PECAM-1 binding activity compared with WT–ICAM-1. Interestingly, ICAM-1 activation-dependent increase in PECAM-1 binding activity was not blocked by L-NAME (data not shown). Thus, NO appears to regulate PECAM-1 phosphorylation and PMN transmigration but does not appear to play a role in the mechanism of increased PECAM-1 mAb binding activity.

In this report, we show an important functional relationship between ICAM-1 activation and PECAM-1 binding activity and subsequent PMN transendothelial migration during inflammation (ie, ICAM-1 activation-dependent Src and NO signaling in human ECs increases PECAM-1–dependent PMN transmigration). ICAM-1 was demonstrated to function through the phosphorylation of its Tyr⁵¹⁸ residue that leads to activation of tyrosine phosphatase SHP-2, which, via dephosphorylation of Src inhibitory Tyr⁵³⁰, increased Src activity. Src activation by this mechanism induced PECAM-1 Tyr⁶⁸⁶ phosphorylation and also stimulated PI3K, Akt, and eNOS activation. The Src-Akt-eNOS signaling initiated by activation of mouse ICAM-1 expressed in human endothelial cells, which we used to assess the direct effects of ICAM-1 mediated signaling by avoiding Src activation induced by cytokines, such as TNF-α,⁸  was demonstrated to be essential for PECAM-1–dependent PMN transmigration. We also showed in the mouse model of carrageenan-induced lung inflammation the pathophysiologic relevance of ICAM-1–induced PECAM-1 activation in mediating inflammation. In addition to ICAM-1, ICAM-2 and VCAM-1 are also involved in leukocyte adhesion to endothelial cells.²⁵  For example, in T-cell transmigration, ICAM-1 and VCAM-1 mediate T-cell arrest, whereas ICAM-1 and ICAM-2 mediate T-cell crawling to permissive sites for diapedesis.³⁵  Therefore, our studies do not exclude the possibility that ICAM-2 and VCAM-1 also participate in ICAM-1 activation-dependent PMN transmigration.

Pleural space injection of carrageenan increased ICAM-1 expression and phosphorylation, as well as activation of Src and eNOS, leading to increased infiltration of PMNs in the lung. In this model, L-NAME treatment or eNOS deletion reduced lung PMN infiltration in support of cultured EC data showing that ICAM-1–activated NO production via enhanced eNOS Ser¹¹⁷⁷ phosphorylation downstream of Akt plays a key role in the mechanism of PMN transmigration in vivo. WT–ICAM-1 and phospho-defective Y518F mutant ICAM-1 expressed in ECs both supported PMN adhesion, whereas Y518F–ICAM-1 failed to induce PMN transmigration, suggesting that ICAM-1 phosphorylation-dependent signaling is required for PMN transmigration. Inhibition of the Src-Akt-eNOS pathway also blocked PMN transmigration without reducing PMN adhesion to ECs. In eNOS^−/− mice in which we observed reduced PMN infiltration (total lung MPO activity) and transmigration (bronchoalveolar lavage PMN cell counts), both basal and carrageenan-induced ICAM-1 expression was greater compared with WT mice, consistent with the negative regulatory role of NO on NF-κB signaling³⁶  and ICAM-1 gene transcription.³⁷  We also observed reduced PMN lung infiltration in iNOS^−/− mice (supplemental Figure 3A), as reported previously by Cuzzocrea et al.²⁴  In addition, eNOS inhibitor L-NAME was able to further reduce PMN infiltration in iNOS^−/− mice (supplemental Figure 3B), suggesting that both eNOS- and iNOS-derived NO production play important roles in the inflammatory response induced by carrageenan.

Antibody-induced ICAM-1 crosslinking activates Src¹⁵ ; however, the mechanism of crosslinking induced Src activation is unclear. In our cotransfection experiments using WT or phosphatase-defective SHP-2 mutant, we showed that phosphatase activity is required for ICAM-1 crosslinking-induced Src activation. Furthermore, we observed decreased Src Tyr⁵³⁰ phosphorylation after ICAM-1 crosslinking, which was abolished in cells expressing the phosphatase-dead SHP-2 mutant. These findings demonstrate that SHP-2 increases Src activity by de-phosphorylating the Src inhibitory Tyr⁵³⁰ residue and thereby dis-inhibiting tyrosine kinase activity. This function of SHP-2 probably becomes apparent after ICAM ligation (or crosslinking) by displacement from the C-terminal ICAM-1 ITAM motif.⁷  Displacement and activation of SHP-2 then dephosphorylates and activates Src C-terminal Tyr⁵³⁰.¹⁵  Coimmunoprecipitation results suggest that basal SHP-2 association with ICAM-1 is reduced after ICAM-1 crosslinking, supporting the hypothesis that SHP-2 release from ICAM-1 mediates the activation of Src. ICAM-1 was previously demonstrated to stimulate SHP-2–mediated dephosphorylation of human Src Tyr⁵³⁰ and induce Src activation.³⁸  The present study identifies a novel feed-forward Src signaling mechanism leading to PMN transmigration induced by phospho–ICAM-1–dependent SHP-2–mediated dephosphorylation of Src-negative regulatory Tyr⁵³⁰ and subsequent Src Tyr⁴¹⁹ auto-phosphorylation. This signaling culminates in Src-dependent PECAM-1 Tyr⁶⁸⁶ phosphorylation and enhanced PECAM-1–dependent transmigration of PMN.

PI3K is also a key downstream effector of Src.³⁹  Here we demonstrate that activation of eNOS induced by ICAM-1 crosslinking is mediated via Src-PI3K-Akt signaling. In these studies, ICAM-1 crosslinking induced Akt phosphorylation, eNOS phosphorylation, and NO production. Both Src and PI3K inhibition prevented eNOS and Akt phosphorylation and NO production, showing that eNOS activation occurs downstream of Src and PI3K. In these studies, compared with the study by Martinelli et al,¹⁸  primary antibody alone did not induce eNOS phosphorylation or NO production. Similar to that reported previously,¹⁸  eNOS phosphorylation was blocked in part by AMPK inhibition although we also observed eNOS phosphorylation to be blocked fully by PI3K inhibition, suggesting that eNOS activation downstream of ICAM-1 occurs through both PI3K/Akt- and AMPK-dependent pathways. Thus, the mechanism of eNOS activation varies slightly between rat brain endothelial cells and HUVECs expressing mouse ICAM-1. Nevertheless, both studies suggest that ICAM-1–mediated eNOS activation may be a common mechanism regulating leukocyte transmigration in the lung and brain.

The finding that ICAM-1–activated NO signaling contributes to the mechanism of PMN transmigration and lung PMN sequestration is seemingly contrary to the reported inhibitory role of NO on PMN migration,⁴⁰  which has been attributed to its ability to dampen PMN superoxide production.⁴¹  As migration of PMNs was also shown to be dependent on NO concentration,⁴²  it is possible that the amount of NO generated on ICAM-1 activation is sufficient to induce PMN migration but not to inhibit PMN NADPH oxidase. Importantly, in our studies, ICAM-1–dependent generation of NO did not inhibit PMN adhesion to endothelial cells. Rather, NO production increased PMN transmigration, consistent with previous observations.^43,44  The finding that ICAM-1 knockout mice exhibited reduced production of NO^45,46  also fits with our observation that ICAM-1 signaling plays a key role in regulating NO production by stimulating the phosphorylation of eNOS.

An unanswered question in the field has been whether ICAM-1 activation regulates the function of PECAM-1 as an adhesive EC protein required for PMN transmigration. We observed that ICAM-1 activation increases the phosphorylation of PECAM-1 Tyr⁶⁸⁶ via the Src-Akt-eNOS pathway because PECAM-1 phosphorylation at this residue was blocked by L-NAME and was abolished in eNOS^−/− mouse lungs. The antibody used for detecting PECAM-1 phosphorylation only recognizes phosphorylated Tyr⁶⁸⁶ (this residue is equivalent to Tyr⁷¹³ in full-length signal peptide-containing PECAM-1). It is currently not known whether Src activation by the mechanism described can also regulate the phosphorylation of PECAM-1 Tyr⁶⁶³, which was previously demonstrated to be important for leukocyte transmigration.^5,47,48  Phosphorylation of the Tyr⁶⁶³ residue was suggested to be essential for PMN transmigration by regulating the location and trafficking of PECAM-1 in endothelial cells.⁵ 

The mechanism by which eNOS activation and NO production increase PECAM-1 phosphorylation and homotypic PECAM-1 adhesion is not clear. Our data suggest that eNOS activity is required for PECAM-1 phosphorylation but not for increased cell surface PECAM-1 expression/binding activity. Interestingly, eNOS and PECAM-1 were previously reported to form a complex⁴⁹ ; thus, it is tempting to speculate that localized activation of eNOS and resultant NO production near the lateral membrane may regulate PMN transmigration via PECAM-1–mediated binding of PMNs.

Leukocyte PECAM-1 homophilic adhesion to endothelial cell PECAM-1 is required for leukocyte transmigration.⁶  An increase in PECAM-1 binding activity downstream of ICAM-1 suggests that firm adhesion of PMN via ICAM-1 promotes PMN transmigration in association with PECAM-1 located in the lateral membrane. The increase of available PECAM-1 mAb binding sites may be the result of an increase in cell surface PECAM-1 expression or increased adhesivity of available PECAM-1. Phosphorylation of PECAM-1 may lead to increased trafficking of PECAM-1 to the lateral membrane from stored pools in the “lateral membrane recycling compartment”⁶  or perhaps to PECAM-1 clustering-induced increase in binding avidity similar to that recently described for ICAM-1.⁸  As the increase in PECAM-1 binding was much less in HUVECs expressing phosphorylation-defective ICAM-1 mutant, ICAM-1 phosphorylation and downstream Src and eNOS signaling appear to play an important role in the mechanism of increased PECAM-1 binding activity and transendothelial migration of PMNs. However, although L-NAME did not block the increase in PECAM-1 binding activity, it did significantly reduce PECAM-1 phosphorylation and PMN transmigration, and there was significantly less PMN infiltration in eNOS^−/− mouse lungs after carrageenan-induced pleurisy. Thus, Src appears to regulate PECAM-1–mediated expression/adhesivity, whereas NO functions independent of PECAM-1–mediated adhesion to facilitate PMN transmigration. Taken together, these data indicate that ICAM-1 activation of Src induces eNOS-dependent regulation of PECAM-1 phosphorylation as well as PMN transmigration, and eNOS-independent regulation of PECAM-1 binding activity.

Interestingly, PECAM-1 was also shown to counteract ICAM-1–mediated signaling.⁵⁰  Thus, signaling via ICAM-1 and PECAM-1 activation may be bidirectional and complementary, thereby dynamically regulating leukocyte transmigration via feed-forward signaling and PECAM-1 adhesivity and feedback inhibition of ICAM-1 expression and signaling. During an inflammatory response, engagement of ICAM-1 signals Src and eNOS activation, which together promote PECAM-1–dependent PMN transmigration (Figure 7). Src phosphorylation of PECAM-1 and NO production were associated with increased PECAM-1 binding activity and PMN transmigration, whereas NO also limited ICAM-1 expression and thereby PMN adhesion, which may prevent excessive neutrophilic inflammation. These results raise the prospect of manipulating inappropriate PMN trafficking by interfering with ICAM-1–induced Src and eNOS signaling, thereby mitigating tissue inflammation dependent on PECAM-1–mediated transmigration.

The authors thank Maricela Castellon and David Visintine for excellent technical assistance, mouse husbandry, and development of the carrageenan pleurisy and lung inflammation mouse model.

This work was supported by the National Institutes of Health National Heart, Lung, and Blood Institute (grant P01 HL60678, R.D.M., R.A.S., and A.B.M.; grant R01 HL71626, R.D.M.; grant R01 HL046849 and R37 HL064774, W.A.M.).

National Institutes of Health

Contribution: G.L. designed research, performed experiments, analyzed data, made the figures, and wrote the paper; A.T.P. analyzed data, made the figures, and wrote the paper; Z.C. and V.M.B. performed experiments and analyzed data; S.M.V. and R.A.S. designed research and analyzed data; and W.A.M., A.B.M., and R.D.M. designed research, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Richard D. Minshall, Department of Pharmacology (m/c 868), University of Illinois, College of Medicine at Chicago, 835 S Wolcott Ave, Chicago, IL 60612; e-mail: rminsh@uic.edu.