Functional Association of Platelet Endothelial Cell Adhesion Molecule-1 and Phosphoinositide 3-Kinase in Human Neutrophils*

In this paper we show that the engagement of the platelet-endothelial cell adhesion molecule-1 (PECAM-1/CD31) up-regulates the adhesion of human neutrophils to the EA.hy926 endothelial cell line through a phosphoinositide 3-kinase (PI3K)-dependent pathway. Indeed, LY294002 and wortmannin prevented the effect of PECAM-1/CD31 cross-linking on cell adhesion, at concentrations known to inhibit PI3K without affecting other kinases. Both compounds blocked neutrophil binding to murine fibroblasts transfected with human ICAM-1, to purified ICAM-1 protein, or to fibronectin, suggesting that PECAM-1/CD31-mediated up-regulation of β2 and β1 integrin-mediated adhesion is PI3K-sensitive. We also provide evidence for the association of PECAM-1/CD31 to PI3K, because PI3K was detectable in neutrophil lysates after PECAM-1/CD31 cross-linking and immunoprecipitation. PECAM-1/CD31-dependent recruitment of PI3K was suggested by the finding that the serine/threonine kinase p70 S6 kinase (S6K), a signaling protein downstream of PI3K, is activated in neutrophils upon PECAM-1/CD31 cross-linking, based on the appearance of serine phosphorylation in S6K immunoprecipitates. In turn, S6K is not directly involved in the up-regulation of integrin function because rapamycin, which can inhibit S6K independent of PI3K, did not block PECAM-1/CD31-induced adhesion of neutrophils to β1 and β2 integrin substrates. In conclusion, PECAM-1/CD31 appears to be one of the molecules functionally coupled to PI3K, suggesting that this enzyme may represent a common pathway of integrin and adhesiveness regulation in leukocytes.

It is widely accepted that the process of leukocyte extravasation and tissue localization depends on several adhesion systems that are tightly regulated and connected one to each other (1)(2)(3)(4). In particular, adhesion of leukocytes to blood vessel wall, crawling across endothelial cells, and progression through the subendothelial matrix are regulated by changes in the avidity state of ␤2 and ␤1 integrins (3,4). Several molecules can modulate integrin function, up-or down-regulating their affinity to specific ligands (3,4), thus allowing leukocyte transendothelial migration. The platelet-endothelial cell adhesion molecule-1 (PECAM-1/CD31) 1 participates in this process both directly, as an adhesion molecule, and indirectly, as a modulator of integrin functions (5,6). Indeed, engagement of PECAM-1/CD31 by specific monoclonal antibodies (mAbs) or by homophilic interactions leads to up-regulation of ␤1 and ␤2 integrin functions, including cytoskeletal association and exposure of high affinity binding sites, in T lymphocytes, natural killer (NK) cells, platelets, and neutrophils (7)(8)(9)(10). These data point to a mechanism whereby oligomerization of PECAM-1/ CD31 on the cell surface results in the triggering of specific signals capable of modulating integrin affinity. However, the molecular basis of such signals is still unclear.
We have recently reported that, in NK cells, cross-linking of PECAM-1/CD31 leads to increased adhesion to extracellular matrix and endothelial cells, mostly mediated via ␤1 and ␤2 integrin through a cAMP-sensitive mechanism (11). In platelets, PECAM-1/CD31 becomes tyrosine phosphorylated during aggregation in an integrin-dependent process that results in the association of the SHP-2 phosphatase with tyrosine residues of the PECAM-1 cytoplasmic domain (12). In this paper, the relationship between PECAM-1 engagement and integrin activation in human neutrophils has been addressed. We provide evidence for a functional association between PECAM-1/ CD31 and phosphoinositide 3-kinase (PI3K). In addition, our data support the hypothesis that engagement of PECAM-1/ CD31 promotes downstream activation events, including increased ␤1and ␤2-mediated adhesion to specific substrates, which are regulated via a PI3K-dependent pathway.

EXPERIMENTAL PROCEDURES
Cells and Antibodies-Peripheral blood neutrophils (PMN) were isolated from healthy donors (buffy coats, kindly provided by the Blood Transfusion Department of our institute) by dextran sedimentation and density gradient centrifugation, followed by hypotonic lysis of erythrocytes (10). The neutrophil pellet was harvested, washed in DPBS without calcium and magnesium (Biochrom, Berlin, Germany) and resuspended in ice-cold HEPES-Tyrode's buffer containing 129 mmol/liter NaCl, 9.9 mmol/liter NaHCO 3 , 2.8 mmol/liter KCl, 0.8 mmol/liter KH 2 PO 4 , 0.8 mmol/liter MgCl 2 -6H 2 0, 5.6 mmol/liter glucose monohydrate, 1 mmol/liter CaCl 2 , 10 mmol/liter HEPES, and 0.3 units/ml aprotinin, leupeptin, and antipain (Sigma). Cells were Ͼ98% neutrophils by Wright-Giemsa staining and Ͼ98% viable by exclusion of trypan blue. Cross-linking of the CD31 molecules expressed by PMN was obtained as described (11). Briefly, cells (10 7 /ml) were stained for 20 min at 4°C with saturating amounts (10 g/ml) of the anti-CD31 mAb M89D3, or mouse Ig (Sigma) as a control, washed and treated with 1 mg/ml of rabbit anti-mouse (RAM) immunoglobulins (Sigma) for 20 min at 4°C, followed by incubation at 37°C for 30 min. PMN were then washed and used in functional or biochemical assays. Purified and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. biotinylated anti-CD31 (11) or anti-CD11a (American Type Culture Collection) mAb and the F(abЈ) 2 of anti-CD31 mAb were obtained as described (11).
Adhesion Assay-The EA.hy926 cell line was kindly provided by C. Edgell (13). ICAM-1 transfectants were obtained using the murine NIH/3T3 fibroblast cell line as described (14). Adhesion of radiolabeled ( 51 Cr, as sodium chromate, 1 Ci/10 6 /cells, NEN Life Science Products) PMN to confluent EA.hy926 cells or ICAM-1 transfectants was performed as described (14), before or after cross-linking of CD31 molecules. In other experiments adhesion was assayed using plastic plates coated with soluble ICAM-1 (p-ICAM-1, 10 g/ml in DPBS, pH 9.3), purified by affinity chromatography from lysates of ICAM-1 transfectants (15) or fibronectin (FN, Sigma, 10 g/ml in DPBS, pH 9.3). In some samples, cells were pre-incubated for 30 min at room temperature with 100 nmol/liter wortmannin, or 50 mol/liter Ly294002 or 25 ng/ml rapamycin (all from Sigma) and washed twice, prior to the onset of the adhesion assay. After 30 min, nonadherent cells were washed, and adherent cells were lysed with 100 mmol/liter Tris buffer containing 1% Triton X-100. The radioactivity of the samples was measured in a ␥ counter (Packard, Sterling, VA). Cell adhesion was calculated as described (14). Results are expressed as the percentage of increase (or decrease) in adherent cells.

RESULTS AND DISCUSSION
Ligation of PECAM-1/CD31 on Human Neutrophils Up-regulates Integrin-mediated Cell adhesion through a PI3K-dependent Pathway-Occupancy of PECAM-1/CD31 on the surface of leukocytes triggers a signal leading to integrin activation and engagement of their endothelial counter-receptors (7)(8)(9)(10)(11). To identify the intracellular mediator(s) of such a signal transduction, we analyzed the binding of PMN to the EA.hy926 cell line, upon PECAM-1/CD31 cross-linking, before and after PMN exposure to different kinase inhibitors. Fig. 1 shows that ligation of PECAM-1/CD31 on human neutrophils increases by 85% cell adhesion to this endothelial cell line, whereas cross-linking of unrelated molecules with mouse Ig followed by RAM or with RAM alone has no effect. Similar results were obtained using the F(abЈ) 2 of the anti-PECAM1/CD31 mAb (not shown). Interestingly, wortmannin, a fungal metabolite that inhibits PI3K (17), prevents the effect of PECAM-1/CD31 ligation, at variance with rapamycin, which does not affect PI3K (18) (Fig. 1). At the concentrations used (100 nmol/liter) wortmannin fully inhibits PI3K, although it has no significant effect on other kinases such as myosin light chain kinase (19) or phosphoinositide 4-kinase (20). Further evidence for the involvement of PI3K in the process of PECAM-1/CD31-induced PMN endothelial cell adhesion comes from the finding that LY294002, another compound that blocks this enzyme with a different mechanism as compared with wortmannin (21), can also prevent the effect of PECAM-1/CD31 engagement (Fig. 1). Because wortmannin is a

FIG. 2. Ligation of PECAM-1/CD31 up-regulates integrin-mediated neutrophil adhesion through a PI3K-dependent pathway.
Adhesion of radiolabeled PMN to confluent ICAM-1 transfectants (A), p-ICAM-1 (B), or FN (C) was evaluated before or after cross-linking of CD31 molecules, obtained as described in the legend to Fig. 1. In some samples, cells were pre-treated for 30 min at room temperature with 100 nmol/liter wortmannin, 50 mol/liter Ly294002, or 25 ng/ml rapamycin and washed twice prior to the onset of the adhesion assay. After 30 min, nonadherent cells were washed, and adherent cells were lysed. The radioactivity of the samples was measured in a ␥ counter (Packard, Sterling, VA). Results are expressed as the percentages of increase (or decrease) in adherent cells (means Ϯ S.D. from eight independent experiments). Basal PMN adhesion was 10%. Wart, wortmannin; LY, Ly294002.
FIG . 1. Ligation of PECAM-1/CD31 up-regulates neutrophil adhesion to endothelial cells involving PI3K. Adhesion of radiolabeled PMN to confluent EA.hy926 cells was evaluated before or after cross-linking of CD31 molecules obtained by treatment with 10 g/ml of the anti-CD31 mAb, followed by 1 mg/ml of RAM Ig and incubation at 37°C for 30 min. Unrelated mouse Ig were used as a control. In some samples, cells were pre-treated for 30 min at room temperature with 100 nmol/liter wortmannin, 50 mol/liter Ly294002, or 25 ng/ml rapamycin and washed twice, prior to the onset of the adhesion assay. After 30 min, nonadherent cells were washed, and adherent cells were lysed. The radioactivity of the samples was measured in a ␥ counter (Packard, Sterling, VA). Results are expressed as the percentages of increase (or decrease) in adherent cells (means Ϯ S.D. from eight independent experiments). Basal PMN adhesion, calculated as described (17), was 10%. Wart, wortmannin; LY, Ly294002. potent inhibitor of activation-dependent increases in integrin adhesiveness (17), we analyzed the involvement of ␤1 and ␤2 integrins in the process of PECAM-1/CD31-mediated up-regulation of PMN endothelial cell adhesion. To this aim, binding experiments were performed using murine fibroblasts (NIH/ 3T3) transfected with human ICAM-1, one major ligand for ␤2 integrins (3), or FN, an extracellular matrix protein that binds to most ␤1 integrins (4). As shown in Fig. 2A, ligation of PECAM-1/CD31 induced a significant increase in PMN adhesion to ICAM-1 transfectants, which was sensitive to wortmannin and LY249002 but not rapamycin treatment. Similar results were obtained when PMN adhesiveness was assayed using p-ICAM-1 (Fig. 2B) or FN (Fig. 2C). Wortmannin has been reported to block ␤1 integrin-mediated increases in T cell adhesiveness induced by CD2, CD28, or CD7 ligation (17,22,23); likewise, there is evidence for PI3K involvement in the modulation of integrin activation in B and NK cells (17). Thus, the function of several different integrin regulators can be blocked by pharmacological inhibition of PI3K activity. PE-CAM-1/CD31 appears to be one of the adhesion molecules functionally coupled to PI3K, suggesting that this enzyme may represent a common pathway of integrin and adhesiveness regulation in leukocytes.
PECAM-1/CD31 Associates with PI3K in Human Neutrophils-Because PI3K is a central signaling molecule for integrin regulators (17), we asked whether PECAM-1/CD31 is as-sociated to PI3K in PMN. To address this question, coprecipitation experiments were performed using microwell plates coated with the anti-PECAM-1/CD31 mAb and PMN, which were lysed after cross-linking. As shown in Fig. 3, PI3K could be selectively detected in the immunoprecipitates obtained with the anti-PECAM-1/CD31 mAb after cross-linking of the molecule (Fig. 3A for enzyme-linked immunosorbent assay detection; Fig. 3B, upper panel, for Western blot; Fig. 3B, lower panel, shows the same blot hybridized with the anti-PECAM-1/CD31 mAb). Conversely, CD11b does not associate PI3K (Fig. 3A).
To address the functional significance of the PECAM-1/ CD31-associated PI3K, we tested indirectly the activity of this enzyme in PMN lysates after PECAM-1/CD31 cross-linking. To this aim, we analyzed the activation of S6K (17,23,24). Autophosphorylation of S6K is a hallmark of activation of this enzyme, which may regulate a wide array of cellular processes (17,(22)(23)(24). Interestingly, activation of S6K was observed in PMN upon PECAM-1/CD31 cross-linking, as documented by the appearance of serine phosphorylation, detectable by Western blot with an anti-Ser(P) antibody, in S6K immunoprecipitates ( Fig. 4 shows one representative experiment of three). These results, together with the data on wortmannin-dependent PECAM-1/CD31-induced adhesion of PMN to integrin substrates (Fig. 2), suggest that PECAM-1/CD31-associated PI3K is functionally active in PMN. Nevertheless, it is unlikely that S6K plays a direct role in the regulation of integrin adhesiveness, because it has been described to regulate mainly cell cycle progression (17). Indeed, we found that rapamycin, which can inhibit S6K independent of PI3K (22), is not effective in blocking PECAM-1/CD31-induced PMN adhesion to integrin substrates (Fig. 2) nor in preventing S6K activation triggered via PECAM-1/CD31 cross-linking (not shown). A number of downstream targets for PI3K have been identified, including protein kinase C isoenzymes, the structurally related kinases Akt and phospholypase C␥ and the FK506-binding rapamycin-associated protein (17,23). The precise sequence of events downstream PI3K and the resulting cellular behavior need to be elucidated. It is tempting to speculate that differential ligation of PECAM-1/CD31 on neutrophils by distinct endothelial-specific ligands (either homo-or heterophilic) leads to activation of selected kinase cascades, resulting in a finely tuned modulation of integrin-mediated leukocyte binding to endothelial cells.

FIG. 3. Association of PECAM-1/CD31 with PI3K in human neutrophils.
A, immunoprecipitation of CD31 or CD11b was performed from PMN in microwell plates coated with the anti-CD31 or the anti-CD11b mAb. After PMN lysis (0.5% CHAPS), plates were stained with biotinylated anti-CD31 or anti-CD11b mAbs or with the anti-PI3K mAbs followed by biotinylated RAM. After incubation with HRP-conjugated streptavidin, plates were developed by luminol reaction and the optical density determined at ϭ 405 nm (means Ϯ S.D. from three independent experiments). B, Western blot of lysates from PMN recovered at different time points from CD31 cross-linking and immunoprecipitation. After SDS-PAGE (10%) and electrotransfer, nitrocellulose filters were hybridized with the anti-p85 subunit of PI3K antiserum (upper panel) followed by HRP-conjugated GAR Ig. The same blot was re-hybridized with the anti-CD31 mAb (lower panel) followed by HRPconjugated goat anti-mouse Ig. The immunoreactive bands were revealed by luminol reaction. One representative experiment of three is shown.