Two Signaling Mechanisms for Activation of αMβ2 Avidity in Polymorphonuclear Neutrophils*

Circulating polymorphonuclear neutrophils (PMN) are quiescent, nonadherent cells that rapidly activate at sites of inflammation, where they develop the capacity to perform a repertoire of functions that are essential for host defense. Induction of integrin-mediated adhesion, which requires an increase in integrin avidity, is critical for the development of these effector functions. Although a variety of stimuli can activate integrins in PMN, the signaling cascades involved are unclear. Phosphatidylinositol (PI) 3-kinase has been implicated in integrin activation in a variety of cells, including PMN. In this work, we have examined activation of the PMN integrin αMβ2, assessing both adhesion and generation of the epitope recognized by the activation-specific antibody CBRM1/5. We have found that PI 3-kinase has a role in activation of αMβ2 by immune complexes, but we have found no role for it in αMβ2 activation by ligands for trimeric G protein-coupled receptors, including formylmethionylleucylphenylalanine (fMLP), interleukin-8, and C5a. Cytochalasin D inhibition suggests a role for the actin cytoskeleton in immune complex activation of αMβ2, but cytochalasin has no effect on fMLP-induced activation. Similarly, immune complex activation of the Rac/Cdc42-dependent serine/threonine kinase Pak1 is blocked by PI 3-kinase inhibitors, but fMLP-induced activation is not. These results demonstrate that two signaling pathways exist in PMN for activation of αMβ2. One, induced by FcγR ligation, is PI 3-kinase-dependent and requires the actin cytoskeleton. The second, initiated by G protein-linked receptors, is PI 3-kinase-independent and cytochalasin-insensitive. Pak1 may be in a final common pathway leading to activation of αMβ2.

Phagocytes are essential cells in host defense of metazoan organisms because they prevent the systemic spread of invading pathogens. Phagocytic cells such as monocytes and polymorphonuclear leukocytes (PMN) 1 circulate throughout tissues to be able to initiate a rapid response to injury and infection. At sites of inflammation and infection, these cells perform many functions, including ingestion and killing of invading organisms, generation of inflammatory mediators, and initiation of an immune response. The acquisition of these effector functions required for successful host defense is called phagocyte activation. Adhesion is required to develop the full effector phenotype in phagocytes and, indeed, in other leukocytes as well (reviewed in Refs. 1 and 2). We have used human PMN as a model cell to study how adhesion regulates this phenotypic change and the critical role of leukocyte integrins in this process. PMN express ␤ 1 , ␤ 2 , and ␤ 3 integrins, but integrins other than the ␤ 2 family (also known as LeuCAM or CD18 integrins) are present in low number. In particular, the CD18 integrin ␣ M ␤ 2 plays a central role in PMN activation at sites of inflammation (3)(4)(5)(6)(7)(8)(9). PMN integrins including ␣ M ␤ 2 bind poorly to their ligands unless the cells are exposed to inflammatory stimuli such as chemokines, bacterial products, cytokines, complement fragments, or immune complexes (10 -15). These stimuli cause an increase in integrin avidity through a process called "insideout" signaling. The molecular pathways of inside-out signaling are uncertain, but increases in receptor affinity (16 -19), receptor clustering (20), cytoskeletal reorganization (19,21,22), and association with guanine nucleotide exchange factors (23) may all be involved in the enhancement of ␤ 2 integrin avidity.
Phosphatidylinositol-3 kinase (PI 3-kinase) has been implicated in the inside-out signaling for integrin activation (24). PI 3-kinase phosphorylates phosphatidylinositols (PI) at the D3 position, producing PI 3-phosphate, PI (3,4)-bisphosphate, and PI (3,4,5)-trisphosphate (PIP 3 ) (25). Five isoforms of mammalian PI 3-kinase have been discovered which appear to be products of distinct genes but have overlapping patterns of expression (26 -29). All known isoforms of mammalian PI 3-kinase share sensitivity to the pharmacologic agent wortmannin (26,30,31), which inhibits PI 3-kinase activity by binding to the lipid-binding domain of the catalytic subunit (32,33). Whereas wortmannin specifically inhibits PI 3-kinase activity at concentrations in the low nanomolar range (26,30,31), higher concentrations inhibit PI 4-kinase and myosin light chain kinase activity (34,35). A second agent called LY294002 inhibits PI 3-kinase activity of the p85/p110 isoforms by binding the ATP-binding site of p110, but it has no effect on PI 4-kinase activity at doses up to 100 M (36). These pharmacologic agents have been extremely useful in delineating cellular activities in which PI 3-kinase has a role, including regulation of adhesion. Wortmannin has been shown to inhibit ␤ 1 integrin-mediated adhesion to fibronectin of stem cell factortreated mast cells (37) and CD-2 transfected HL-60 cells (38); thrombin-and Fc␥RII-induced, ␤ 3 integrin-mediated aggregation of platelets (39 -41); and ␤ 2 integrin-dependent homotypic adhesion of IL-2-treated lymphocytes (42). Use of PDGF and CD28 receptor mutants that no longer bind PI 3-kinase has provided strong evidence that PI 3-kinase is important for regulating PDGF-and CD28-induced adhesion in mast cells and HL-60 cells, respectively (43,44). Furthermore, expression of dominant negative mutants of the p85 subunit of PI 3-kinase blocks CD7-induced activation of ␤ 1 integrins in human T cells (45). Although these data strongly suggest that PI 3-kinase activity is an important early event in inside-out signaling regulating integrin-mediated adhesion, the mechanism by which PI 3-kinase regulates adhesion is not clear nor is the generality of the requirement for PI 3-kinase in integrin activation. PDGF receptors, for example, can activate ␤ 1 integrins by PI 3-kinase-dependent and -independent mechanisms (44).
Whether PI 3-kinase has a role in ␣ M ␤ 2 activation in PMN is not known. Fc␥R-induced phagocytosis (46), fMLP-induced respiratory burst activity (32,33,47), and PDGF-induced chemotaxis (48) in PMN are inhibited by wortmannin. Since each of these events depends on activated integrins, these data suggest the hypothesis that PI 3-kinase is a component of the inside-out signaling pathway regulating integrin activation in PMN. However, PMN contain an intracellular pool of ␣ M ␤ 2 which is rapidly expressed at the plasma membrane upon activation (15,49). While this intracellular pool is not required for PMN binding to endothelia or for aggregation (50 -52), it is necessary for optimal adhesion (53). The role of PI 3-kinase in regulating the expression of this intracellular pool at the plasma membrane is unknown.
We tested the importance of PI 3-kinase in regulating integrin activation in PMN in two well characterized experimental systems for activating ␤ 2 integrin-dependent adhesion. Fc␥R ligation activates ␣ M ␤ 2 through initiation of a tyrosine kinase cascade, whereas fMLP requires a heterotrimeric G-protein to initiate signaling in PMN. Our results suggest that two pathways exist for activating ␤ 2 integrin-dependent adhesion in PMN. The Fc␥R-initiated pathway is dependent on PI 3-kinase activity and is inhibited by cytochalasin D, whereas the fMLPinduced increase in ␣ M ␤ 2 avidity is independent of PI 3-kinase and unaffected by cytochalasin D. Fc␥R-mediated enhancement of ␣ M ␤ 2 expression is inhibited by wortmannin, but increased expression is not required for adhesion. Importantly, both pathways activate Pak1, a recently described serine/threonine kinase implicated in membrane ruffling and focal adhesion formation (54). Fc␥R-induced activation of Pak1 is PI 3-kinase-dependent, whereas fMLP-induced activation of Pak1 is independent of PI 3-kinase, potentially placing Pak1 in a common pathway leading to activation of ␣ M ␤ 2 avidity. These data demonstrate that there is more than one molecular pathway for inside-out signaling and suggest that the effects of tyrosine kinase cascades, and G protein-dependent signaling on integrin function may be mediated by distinct mechanisms that converge on a common pathway involving Pak1.  (58), and B6H12 (anti-IAP, CD47) (59) were purified, and F(abЈ) 2 was prepared as described (60). CBRM1/5 (61) concentrated tissue culture supernatant and anti-Pak1 polyclonal antiserum (62) were prepared as described.

Reagents-Cytochalasin
Preparation of PMN Suspensions-Human PMN were isolated from whole blood exactly as described (63) except hypotonic lysis was not performed. PMN were greater than 98% viable as indicated by the exclusion of trypan blue dye. Cells were suspended in HBSS (Hanks' buffered salts solution with 20 mM Hepes and 8.9 mM sodium bicarbonate) with 1.0 mM Mg 2ϩ and 1 mM Ca 2ϩ (HBSS 2ϩ ) or HBSS with 0.5 mM Mn 2ϩ for adhesion assays and flow cytometry.
Adhesion Assay-Purified human PMN (1 ϫ 10 7 /ml) were incubated with 2 g/ml calcein in HBSS for 30 min at RT. The cells were washed once and resuspended in HBSS 2ϩ at 2 ϫ 10 6 /ml. For adhesion experiments in the presence of Mn 2ϩ , the cells were washed in HBSS with 2 mM EGTA once, HBSS 2ϩ or HBSS ϩ 0.5 mM Mn 2ϩ once, and resuspended in HBSS 2ϩ or HBSS ϩ 0.5 mM Mn 2ϩ . Cells were treated with wortmannin or LY294002 at the indicated concentration or Me 2 SO as a control for 15 min at 37°C or with pertussis toxin (2 g/ml) or control buffer in HBSS ϩ 1% human serum albumin for 2 h at 37°C. For antibody inhibition experiments, cells were incubated with 10 or 25 g/ml of the appropriate antibody for 15 min at RT. 1 ϫ 10 5 cells were added per well to Immulon 2 plates coated with BSA and a 1:25 dilution of rabbit anti-BSA to form IC or 5% FCS as described (64). For PMA or fMLP-stimulated adhesion, PMA (50 g/ml final), fMLP (100 nM final), or Me 2 SO control was added to the cells after allowing them to settle onto FCS-coated wells for 6 min at RT. The cells were incubated at 37°C for the indicated time. The fluorescence (485 nm excitation and 530 nm emission wavelengths) was measured using a fMax fluorescence plate reader (Molecular Devices, Sunnyvale, CA) before and after washing twice with 150 l of phosphate-buffered saline. Percent adhesion was calculated by dividing the fluorescence after washing by the fluorescence before washing. In preliminary experiments, fluorescence was shown to be linearly related to cell number (data not shown).
Flow Cytometry-Purified PMN (4 ϫ 10 6 /ml in HBSS 2ϩ ) were treated with wortmannin (100 nM) or Me 2 SO for 15 min at 37°C. For experiments with pertussis toxin, 1 ϫ 10 7 cells/ml were incubated with pertussis toxin (2 g/ml) or control buffer for 2 h at 37°C and then washed. 2 ϫ 10 6 cells were then treated with 30 l insoluble IC (IIC) prepared exactly as described (65), fMLP (100 nM), C5a (50 nM), IL-8 (100 nM), or PMA (50 g/ml) at 37°C for 10 min, placed on ice, washed once with ice-cold wash buffer (phosphate-buffered saline, 1% FCS, 0.1% sodium azide), and resuspended in 100 l of wash buffer plus primary antibody (25 g/ml). Cells were incubated with primary antibody for 40 min on ice and then washed twice. After incubation with fluorescein isothiocyanate-conjugated F(abЈ) 2 sheep anti-mouse IgG secondary antibody at a 1:50 dilution in 200 l of wash buffer for 20 min on ice, cells were washed twice, and the relative fluorescence of gated PMN was measured using a EPICS XL (Coulter, Miami, FL) flow cytometer. For Mn 2ϩ experiments, cells were treated with wortmannin, washed, resuspended in HBSS 2ϩ or HBSS ϩ 0.5 mM Mn 2ϩ , and incubated for 10 min at 37°C and then placed on ice. Primary antibody was added directly to the cells (25 g/ml) for 40 min on ice, washed twice, and incubated with secondary antibody as above. All washes were done with HBSS containing appropriate divalent cations.
Pak1 Kinase Assays-Purified PMN were suspended at 1 ϫ 10 7 cells/ml in HBSS 2ϩ . After pretreatment with wortmannin, LY294002, pertussis toxin, or control buffer as described above, 7.5 ϫ 10 6 cells were added to 6-well plates coated with IC or FCS as described (66) or stimulated in suspension with fMLP (100 nM). After incubating at 37°C, the cells were lysed by adding cold 2ϫ lysis buffer (1% Nonidet P-40, 150 mM NaCl, 5 mM EGTA, 50 mM NaF, 5 mM sodium pyrophosphate, 1 mM NaVO 4 , 5 mg/ml leupeptin and aprotinin, and 1 mM diisopropyl fluorophosphate, final concentration) for 30 min on ice. Pak1 was immunoprecipitated from the lysates with 5 l of rabbit anti-Pak1 antiserum and 40 l of a 1:1 slurry of Protein A-Sepharose for 2 h at 4°C. The immunoprecipitates were washed four times with lysis buffer and two times with reaction buffer (25 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 ). Kinase reactions were performed with the Pak1 immunoprecipitates by adding 30 l of reaction buffer with 2.5 g of MBP to the beads, incubating for 10 min at RT, followed by 10 l of reaction buffer containing 100 M cold ATP and 0.5 Ci of [␥-32 P]ATP (4500 Ci/mmol) for a final ATP concentration of 25 M. The reactions were incubated for 20 min at 30°C, after which the reaction was stopped with 50 l of SDS-polyacrylamide gel electrophoresis sample buffer containing 10% SDS. Phosphorylation of MBP was detected by SDS-polyacrylamide gel electrophoresis, transfer to polyvinylidene difluoride membranes, and autoradiography. For each experiment, Pak1 protein was immunoblotted using anti-Pak1 antiserum (1:1000) primary antibody, horseradish peroxidase-conjugated goat anti-rabbit antiserum (20 g/ml) secondary antibody, and enhanced chemiluminescence substrate (ECL, Pierce) to ensure that equivalent amounts of kinase protein were added to each in vitro kinase reaction.

PI 3-Kinase Activity Is Required for Fc␥R-induced Activation
of ␣ M ␤ 2 -dependent Adhesion-PI 3-kinase activity is required for adhesion of a variety of cell types to fibronectin (37,38), agonist-induced aggregation and up-regulation of activated ␣ IIb ␤ 3 expression in platelets (39 -41), and PDGF-induced chemotaxis in PMN (48), suggesting that PI 3-kinase activity is important for integrin activation. Fc␥R-induced phagocytosis in PMN is blocked by wortmannin (46), suggesting that PI 3-kinase activity may be required for Fc␥R-induced signal transduction and effector functions in PMN. We used two phar-macologic inhibitors of PI 3-kinase, wortmannin and LY294002, to test the hypothesis that PI 3-kinase activity is required for Fc␥R-induced, ␤ 2 integrin-dependent adhesion to IC in PMN.
Adhesion of control PMN to IC was maximal by 15 min and sustained for up to 40 min (Ref. 8 and Fig. 1A). PMN pretreated with wortmannin or LY294002 initially adhered to IC-coated surfaces identically to control cells, even at inhibitor concentrations up to 1 and 200 M, respectively (Fig. 1, A and C, and data not shown). However, after 10 min, adhesion of both wortmannin-and LY294002-treated PMN to IC decreased until there was no specific, IC-dependent adhesion by 40 min (Fig. 1, A and C). Similar kinetics of adhesion were obtained with PMN pretreated with wortmannin for 30 min. Non-adherent, wortmannin-treated PMN excluded trypan blue dye, demonstrating that they were viable. Although the wortmannin and LY294002-treated PMN cells initially (Ͻ10 min) spread as well as control cells on IC, after 10 min spreading decreased until by 40 min the cells were completely round (data not shown). In contrast, PMA-induced adhesion to, and spreading on, FCS were not affected at any dose of either PI 3-kinase inhibitor (Fig. 1, B and D). Wortmannin and LY294002 inhibited sustained (Ͼ10 min) adhesion to IC in a dose-dependent manner, with IC 50 of 5 nM and 8 M, respectively (Fig. 1, B and D). The IC 50 for inhibition of sustained adhesion for each compound is comparable to the IC 50 for inhibition of in vitro PI 3-kinase activity in anti-p85 immunoprecipitates from pretreated PMN in our system and others (Ref. 33 and data not shown), and published doses that inhibit fMLP-induced respiratory burst activity, in vitro PI 3-kinase activity, and PIP 3 accumulation in PMN (32,33,36,47) and is 6-fold less than that reported to inhibit Fc␥R-mediated phagocytosis in PMN (46).
Previous work in our lab had shown that adhesion of PMN to IC-coated surfaces occurs in two phases (8). Initial (Ͻ10 min) adhesion is independent of ␤ 2 integrins as shown by PMN from patients with leukocyte adhesion deficiency or normal PMN treated with anti-␤ 2 antibodies. Sustained adhesion (Ͼ10 min), however, requires the ␤ 2 integrin ␣ M ␤ 2 . The kinetics of adhesion of wortmannin-treated PMN were identical to PMN treated with anti-␤ 2 and anti-␣ M F(abЈ) 2 antibody fragments ( Fig. 2 and data not shown). This demonstrates that PI 3-kinase inhibitors specifically block the ␣ M ␤ 2 -dependent phase of PMN adhesion to IC. Other inhibitors of adhesion, including cytochalasin, bromophenacyl bromide, W7, myosin light chain kinase inhibitors ML-9 and KT 5926, or the myosin inhibitor 2,3-butanedione monoxime, inhibited both phases of PMN adhesion (data not shown). Initial adhesion was inhibited partially by anti-Fc␥RII antibody Fab fragments and completely by a combination of anti-Fc␥RII and anti-Fc␥RIII antibodies (data not shown).
fMLP-and PMA-induced Adhesion Is PI 3-Kinase-independent-PI 3-kinase could be required for integrin-dependent adhesion because it is involved in an Fc␥R-initiated signaling pathway which results in an alteration in ␣ M ␤ 2 avidity or because it is involved in the cytoskeletal rearrangements which are required for increased adhesion. To distinguish these possibilities, we assessed the role of PI 3-kinase in PMA and fMLP-induced PMN adhesion. PMN do not adhere to surfaces coated with FCS in the absence of ␣ M ␤ 2 activation but adhere strongly when stimulated with agonists (Fig. 3, A and B). In contrast to its effect on sustained PMN adhesion to IC, wortmannin had no effect on either PMA or fMLP-induced adhesion. fMLP (100 nM)-stimulated adhesion to FCS was maximal by 3 min, decreased by 10 min, but remained above base-line adhesion for at least 30 min (Fig. 3A). Wortmannin treatment had no significant effect on fMLP-induced adhesion at any time point. In confirmation of earlier reports (32,33,47), 10 nM wortmannin completely inhibited fMLP-induced respiratory burst activity in PMN (data not shown). Furthermore, wortmannin pretreatment inhibited PI 3-kinase activity in anti-p85 and anti-p110␥ immunoprecipitates with IC 50 of 5 and 20 nM, respectively, demonstrating the efficacy of wortmannin treatment (data not shown). Likewise, PMA (50 ng/ml) induced significant adhesion by 3 min which continued to increase until 30 min and was unaffected by wortmannin (Fig. 3B). Identical results were obtained for fMLP-and PMA-induced adhesion to the ␣M␤2 ligand fibrinogen (data not shown). Like wortmannin, LY294002 had no effect on fMLP-or PMA-induced adhesion (data not shown). The PKC inhibitor Gö6976 inhibited fMLP, PMA, and IC-induced adhesion (data not shown).
The fMLP receptor is a seven transmembrane receptor associated with a pertussis toxin (PT)-sensitive G protein, as are many receptors implicated in integrin regulation (9,11,(67)(68)(69). To determine whether the PI 3-kinase-independent pathway for adhesion required the G protein, the effect of PT on fMLP-induced adhesion was assessed. PT pretreatment completely inhibited fMLP-induced adhesion but had no effect on IC-or PMA-induced adhesion (Fig. 3C). To determine whether this difference in sensitivity to PI 3-kinase inhibitors reflected a difference in mechanism or a difference between FCS and IC-coated surfaces, the effect of fMLP and PMA on wortmannin sensitivity of sustained adhesion to an IC-coated surface was assessed (Fig. 4, A and B). Stimulation of wortmannin-treated PMN adherent to IC for 7 min with PMA or fMLP induced sustained adhesion that was equivalent to control PMN at 40 min (Fig. 4, A and B). Furthermore, both fMLP and PMA restored the spread phenotype of wortmannin-treated PMN on IC (data not shown). These data demonstrate that PI 3-kinase activity is not required for adhesion per se and support the hypothesis that PI 3-kinase is an essential component of the signaling cascade initiated by Fc␥R ligation resulting in ␣ M ␤ 2 activation. Whereas PMA directly activates PKC and thus could bypass wortmannin inhibition by this mechanism, fMLP binds to a heterotrimeric G protein-coupled receptor which initiates several signaling cascades of its own. Thus, fMLPinduced adhesion to FCS or IC-coated surfaces is a wortmannin-and LY294002-insensitive pathway for integrin regulation.
PI 3-Kinase-dependent and -independent Up-regulation of ␣ M ␤ 2 Integrin-When PMN are activated two processes occur that can potentially affect adhesion. Cell surface ␤ 2 integrin expression increases as integrins stored in intracellular granules are released to the cell membrane by degranulation. Integrins already present on the cell surface also are activated, causing an increase in avidity for ligand which is independent of any increase in receptor expression. To determine whether one or both of these steps was affected by the distinct fMLP and IC signaling pathways and to determine whether one or both was required for sustained adhesion, we examined these processes independently. We first tested whether Fc␥R-induced up-regulation of ␤ 2 integrin expression is affected by wortmannin using flow cytometry to detect surface expression of ␤ 2 on PMN stained with anti-␤ 2 integrin antibody F(abЈ) 2 .
Stimulation of PMN in suspension with IC, fMLP, or PMA all caused a significant increase in expression of ␤ 2 (Fig. 5A). Wortmannin (100 nM) inhibited IC-induced ␤ 2 expression but had no effect on fMLP and PMA-induced ␤ 2 expression (Fig.  5A). Interestingly, cytochalasin D also specifically inhibited ICbut not fMLP-or PMA-induced up-regulation of ␤ 2 expression (Fig. 5A). Cytochalasin D also inhibits the increase in [Ca 2ϩ ] i (where [Ca 2ϩ ] i means intracellular calcium concentration) induced by IC but not by fMLP (70). Since increased [Ca 2ϩ ] i is required for enhanced ␣ M ␤ 2 expression (71), this may be the mechanism for the inhibitory effect of cytochalasin D. Identical results were obtained for ␣ M ␤ 2 expression specifically using the ␣ M -specific antibody, OKM1 (data not shown). Thus, sensitivity to cytochalasin D is another distinguishing feature of the signaling pathways activated by IC and fMLP. IC, fMLP, or PMA did not increase expression of HLA (Fig. 5B) or ␣ L ␤ 2 integrin (LFA-1, as measured by expression of ␣ L , data not shown); in fact PMA caused HLA and ␣ L expression to decrease. Although cytochalasin D slightly decreased HLA and ␣ L expression in

FIG. 3. fMLP-and PMA-induced adhesion to FCS is PI 3kinase-independent.
A and B, PMN were treated with control Me 2 SO, wortmannin (wort) (100 nM), anti-␤ 2 mAb IB4 F(abЈ) 2 (10 g/ ml), or the control anti-IAP mAb B6H12 F(abЈ) 2 (10 g/ml) for 10 min at 37°C prior to measurement of adhesion to FCS-coated wells. Vehicle control (A and B), 100 nM fMLP (A), or 50 ng/ml PMA (B) were added to the appropriate wells to activate ␣ M ␤ 2 -mediated adhesion. Adhesion was measured as described in Fig. 1 after the indicated times. C, PMN were treated with control buffer, wortmannin (100 nM) or LY294002 (25 M), or PT(2 g/ml) prior to measurement of adhesion to IC (30 min) or fMLP-or PMA-induced adhesion to FCS (3 min). Wortmannin and LY294002 significantly inhibited adhesion to IC but did not affect fMLP-or PMA-stimulated adhesion to FCS (Ͻ0.05). PT significantly inhibited fMLP-but not PMA-or IC-induced adhesion (p Ͻ 0.05). The data are representative of three separate experiments. PI 3-Kinase-dependent and -independent Pathways for ␣ M ␤ 2 Activation-We next examined the role of PI 3-kinase in activation of surface-expressed ␣ M ␤ 2 . We used a monoclonal antibody (CBRM1/5) that specifically recognizes a neoepitope on activated ␣ M ␤ 2 to detect activated ␣ M ␤ 2 on the surface of stimulated PMN (61). IC, fMLP, and PMA all increased expression of the CBRM1/5 epitope (Fig. 6A and data not shown). However, wortmannin and cytochalasin D inhibited the increased expression of activated ␣ M ␤ 2 only on PMN stimulated with IC and not on fMLP-or PMA-stimulated PMN (Fig. 6A and data not shown). Thus, both Fc␥R-induced up-regulation of ␣ M ␤ 2 expression and activation are PI 3-kinase-dependent and also require the actin cytoskeleton. In contrast, fMLP and PMAinduced up-regulation of ␣ M ␤ 2 expression and activation are PI 3-kinase-independent and cytochalasin-insensitive.
To determine whether the PI 3-kinase-independent pathway for integrin regulation required the G protein, the effect of PT on fMLP-induced CBRM1/5 was assessed. PT (2 g/ml) completely inhibited fMLP-induced CBRM1/5 expression but had no effect on CBRM1/5 expression induced by IIC (Fig. 6A). We  A and B). Pertussis toxin significantly inhibited fMLP-, C5a-, and IL-8-induced CBRM1/5 binding but had no significant effect on IIC-induced CBRM1/5 binding (A and B). PMA-induced CBRM1/5 binding was unaffected by any inhibitor (data not shown).
next assessed the mechanisms by which ligands for other seven transmembrane receptors regulate integrin avidity. Like fMLP, C5a and IL-8 increased CBRM1/5 binding to PMN. For both C5a and IL-8, CBRM1/5 binding was unaffected by wortmannin but was inhibited by PT (Fig. 6B). Thus, ligation of these different G protein-coupled receptors activated ␣ M ␤ 2 in PMN in a wortmannin-insensitive manner.
Activation of ␣ M ␤ 2 Avidity Is Necessary and Sufficient for Sustained Adhesion to IC-Whereas both Fc␥R-induced increases in ␤ 2 expression and activation of the ␣ M ␤ 2 activation epitope require PI 3-kinase activity, it is well established that receptor activation is required, but increased receptor expression is not necessary for ␣ M ␤ 2 adhesion in other systems (50 -52, 61). To determine the requirements for sustained adhesion to IC, we tested the effect of CBRM1/5 (61) on adhesion. Like wortmannin and the anti-␤ 2 antibody, CBRM1/5 inhibited sustained adhesion to IC but had no effect on the initial ␤ 2independent adhesion (Fig. 7) Thus, activated ␣ M ␤ 2 is necessary for sustained adhesion.
To determine whether increased ␣ M ␤ 2 expression is required, PMN were treated with the ion channel blocker 4,4Јdiisothiocyanostilbene-2,2Ј-disulfonic acid, which prevents increased ␤ 2 expression by inhibiting degranulation (52). 4,4Ј-Diisothiocyanostilbene-2,2Ј-disulfonic acid had no effect on sustained adhesion to IC (data not shown). Moreover, sustained adhesion to IC was perfectly normal in the absence of extracellular Ca 2ϩ , even in cells treated with the intracellular Ca 2ϩ chelator, 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid (data not shown). Since these cells have very low [Ca 2ϩ ] i , up-regulation of receptor expression is deficient (71). These data suggested that increased receptor expression was not required for sustained adhesion to IC. To examine this definitively, we used Mn 2ϩ to induce ␣ M ␤ 2 activation. Mn 2ϩ activates the high avidity state of many integrins, including ␣ M ␤ 2 (72), without inside-out signaling, possibly by altering the conformation of the receptor to increase ligand affinity. Treatment of PMN with Mn 2ϩ increased expression of the CBRM1/5 epitope (Fig. 8A) but had no effect on total ␤ 2 integrin expres-sion (Fig. 8B). Mn 2ϩ -induced expression of the CBRM1/5 epitope was unaffected by wortmannin (Fig. 8A). Mn 2ϩ treatment rescued sustained adhesion of wortmannin-treated PMN to IC (Fig. 8C). The combination of anti-␤ 2 and CBRM1/5 antibodies completely inhibited Mn 2ϩ -induced adhesion to IC and FCS, indicating that the mechanism of adhesion to these substrates remained ␤ 2 integrin-dependent. Thus, ␣ M ␤ 2 activation is necessary and sufficient for sustained adhesion to IC. These data also show that wortmannin does not inhibit adhesion to IC by affecting ␤ 2 integrin outside-in signaling.
Pak1 Is Activated by IC and fMLP-The small GTPases Rac and Cdc42 have been found to be important for regulation of the actin cytoskeleton and formation of integrin complexes in several cell types (73). We investigated the possibility that Rac and/or Cdc42 regulated integrin activation in PMN by examining the activation of the Rac1/Cdc42 effector Pak1 which has been implicated in the regulation of the actin cytoskeleton and formation of focal adhesions (54). Both fMLP and adhesion to IC activated Pak1 in PMN (Fig. 9, A and B). The kinetics of Pak1 activation in response to fMLP and IC was identical to the kinetics of adhesion induced by these stimuli (Fig. 9, C and D). Pak1 activation induced by fMLP or adhesion to IC was unaffected by pretreatment with anti-␤ 2 F(abЈ) 2 , indicating that Fc␥R ligation activates Pak1 independently of ␤ 2 integrins, suggesting that Pak1 activation may be a component of inside-out signaling (Fig. 9E). Importantly, we found that ICinduced Pak1 activation was dependent on PI 3-kinase activity and independent of a PT-sensitive G protein (Fig. 9, A and B). In contrast, fMLP-induced Pak1 activation was inhibited completely by PT but was unaffected by wortmannin (Fig. 9, A and  B). Both wortmannin and LY294002 inhibited IC-induced Pak1 activation with IC 50 of 5 nM and 5 M (data not shown), identical to the IC 50 for inhibition of sustained adhesion to IC. These data demonstrate that while Fc␥R and the fMLP receptor induce initially distinct signals, the signaling pathways initiated by these receptors converge on the Rac/Cdc42-activated kinase Pak1. DISCUSSION A hallmark of neutrophil activation is the requirement for cell adhesion to achieve full functional capacity. The ␤ 2 integrins, particularly ␣ M ␤ 2 , have a central role in this adhesiondependent activation, as shown by the multiple, profound functional defects of PMN from patients with leukocyte adhesion deficiency (3)(4)(5)(6)(7)(8). Many lines of investigation have demonstrated that unactivated PMN are poorly adherent, and integrin-mediated adhesion is rapidly and reversibly induced by activation stimuli (11,12,14). This regulation, which has been called inside-out signaling, is a key event in PMN transendothelial migration, in motility through extracellular matrix, and in induction of effector functions at the site in inflammation. Whereas ␣ and ␤ chain domains required for inside-out signaling have been studied in some detail for several integrins (74 -77), and integrin clustering and alterations in association with cytoskeleton have been described as consequences of inside-out signaling (19 -22), the molecular details of the signal transduction cascade involved in integrin activation remain unclear.
We have identified two inside-out signaling pathways for activation of ␣ M ␤ 2 integrin-dependent adhesion, an Fc␥R-induced, PI 3-kinase-dependent pathway and an fMLP-induced PI 3-kinase-independent pathway. PMA-induced adhesion is also PI 3-kinase-independent, either because PMA-activated PKC is downstream of PI 3-kinase or because it is in the PI 3-kinase-independent pathway. Gö6976, an inhibitor of the classical calcium-dependent PKC, inhibits adhesion to IC as well as fMLP-and PMA-induced adhesion (data not shown), suggesting that PKC is in a common pathway. Although we cannot definitively rule out the possibility that the effect of wortmannin and LY294002 in our experiments is a result of inhibition of other enzymes, the fact that two inhibitors with distinct mechanisms of action cause the same biologic effect argues strongly for specificity. Furthermore, the wortmannin and LY294002 IC 50 for inhibition of adhesion and Pak1 activity induced by IC were 5 nM and 5 M, respectively. These values agree with our own and published values for inhibition of PI 3-kinase activity in PMN and other cells (32, 33, 36, 40, 41, 45, 47, 78 -80) and are 30-and 10-fold less than values reported for myosin light chain kinase and PI 4-kinase, respectively (34,35).
Recently, several groups have suggested an important role for PI 3-kinase in inside-out signaling by lymphocyte receptors including the IL-2 receptor, CD2, CD7, and CD28 (38,42,43,45), by Fc␥RIIA or thrombin receptors on platelets (39 -41), and by stem cell factor receptors on mast cells (37). These studies generally have had as their integrin targets ␤ 1 and ␤ 3 integrins, which may differ significantly from leukocyte ␤ 2 integrins in their mechanisms for regulation (23,67). The finding that a PI 3-kinase-independent pathway exists for activation of ␤ 1 -dependent adhesion by the PDGF receptor (44) and that PMA and ␤ 1 integrin-activating antibodies induce adhesion in wortmannin-treated HL60 cells (38) suggest that PI 3-kinase is an element in the signal transduction cascade for ␤ 1 integrin activation. Our demonstration that ␤ 2 integrin activation by the physiologic agonist fMLP as well as PMA can occur in PMN in which PI 3-kinase has been inhibited further establishes that PI 3-kinase is a component of inside-out signal transduction and is not required for adhesion itself.
G protein-coupled receptors can activate PI 3-kinase and use it to stimulate PIP 3 production, respiratory burst activity, and activation of Raf-1 and ERK1/2 (33,47,81). In other cells, trimeric G protein-linked receptors activate the p110␥ isoform of PI 3-kinase by both the ␣ and ␤␥ subunits of the heterotrimeric G protein (31). The wortmannin sensitivity of this isoform may be less than the classical p85/p110 heterodimeric kinase, but the IC 50 for wortmannin is still less than 50 nM (31). Indeed, in our system, wortmannin pretreatment inhibited PI 3-kinase activity in anti-p110␥ immunoprecipitates with an IC 50 of 20 nM (data not shown). Wortmannin has been shown to inhibit completely fMLP-induced PIP 3 production in PMN with an IC 50 of 5 nM (32,33). Wortmannin inhibited fMLP-induced respiratory burst activity in our system with an IC 50 of 2 nM (data not shown). Thus, activation of PI 3-kinase by fMLP is entirely wortmannin-sensitive in PMN; hence, wortmannin insensitivity demonstrates that PI 3-kinase activation is not required for the activation of ␣ M ␤ 2 . This second, PI 3-kinaseindependent, pathway for activating ␣ M ␤ 2 can be initiated by several ligands for seven transmembrane receptors, including fMLP, C5a, and IL-8. This result is consistent with the finding that wortmannin does not inhibit fMLP-or IL-8-induced chemotaxis (48). Interestingly, wortmannin does inhibit PDGFinduced chemotaxis in PMN (48). These data suggest that Fc␥R and PDGF receptor, which activate tyrosine kinase cascades, utilize PI 3-kinase for integrin activation, whereas G proteinlinked receptors do not. Thus, whether PI 3-kinase is involved in integrin activation may depend on whether the initial stimulus initiates a tyrosine kinase cascade. However, PI 3-kinase Each point represents the mean Ϯ S.E. relative fluorescence from three separate experiments. Mn 2ϩ significantly increased CBRM1/5 binding (p Ͻ 0.05) (A) but had no effect on total ␤ 2 expression (B). Wortmannin pretreatment did not significantly effect CBRM1/5 or anti-␤ 2 binding. Mn 2ϩ had no significant effect on anti-HLA binding (data not shown). C, adhesion to IC after 40 min of PMN pretreated with wortmannin or control was measured in the presence of anti-␤ 2 mAb IB4 (10 g/ml), CBRM1/5 (25 g/ml), both anti-␤ 2 and CBRM1/5 (10 and 25 g/ml, respectively), or the control anti-HLA W6/32. Adhesion was quantitated as in Fig. 1. These data are representative of three separate experiments.
FIG. 9. Pak1 activation induced by adhesion to IC is PI 3-kinase-dependent, whereas fMLP-induced Pak1 activity does not require PI 3-kinase. A, PMN were treated with wortmannin (W) (100 nM) and allowed to adhere to IC or FCS for 10 min or stimulated in suspension with control (C) Me 2 SO or fMLP (100 nM) for 1 min. The kinase activity of immunoprecipitated Pak1 was assayed using MBP as a substrate (A, top). Western blot analysis of the Pak1 immunoprecipitates with anti-Pak1 antiserum demonstrates that equivalent amounts of protein were used for the kinase reactions (A, bottom). B, PMN were treated with control buffer, wortmannin (100 nM), or PT (2 g/ml) and allowed to adhere to IC or FCS for 10 min or were stimulated in suspension with fMLP (100 nM) for 1 min. The kinase activity of immunoprecipitated Pak1 was assayed as in A. Phosphorylation of MBP was quantitated by densitometry. Each point represents the mean Ϯ S.E. of three separate kinase reactions. C, PMN were allowed to adhere to IC or FCS, and Pak1 kinase activity after various times was assayed as in A. Phosphorylation of MBP was quantitated by densitometry. D, PMN were treated with 100 nM fMLP or vehicle control in suspension, and Pak1 activity was quantitated after various times as in C. E, after treatment with Me 2 SO control, wortmannin (Wort) (100 nM), anti-␤ 2 mAb IB4 F(abЈ) 2 , or the control anti-IAP mAb B6H12 F(abЈ) 2 PMN were allowed to adhere to IC for 10 min, and (E, top) Pak1 kinase activity was assessed as in A. Western blot of Pak1 immunoprecipitates demonstrate that equivalent amounts of Pak1 protein were used for the kinase reactions (E, bottom). These data are representative of three separate experiments.
is not absolutely required for all tyrosine kinase-initiated integrin activation, because PDGF receptor mutants that cannot bind PI 3-kinase but are able to bind phospholipase C␥ are perfectly capable of activating ␤ 1 integrins in mast cells (44). At physiologic concentrations of PDGF, PDGF receptor-initiated activation of Erk1 and Erk2 is inhibited by wortmannin, but at higher concentrations of PDGF, activation of Erk1/2 can occur by a PI 3-kinase-independent, phospholipase C␥-and PKC-dependent pathway, suggesting that PI 3-kinase activity is critical at low but not high signal strength (82).
Many PMN responses to immune complexes require ␣ M ␤ 2 , including sustained adhesion (8,83). This study demonstrates that the recruitment of ␣ M ␤ 2 function requires PI 3-kinase activity from Fc␥R ligation. In our experiments, ␣ M ␤ 2 activation was measured by quantitating binding of the mAb CBRM1/5. CBRM1/5 recognizes a neoepitope induced in a subset of ␣ M ␤ 2 upon activation by agonists that induce adhesion (61). The characteristics of this subset remain unknown; however, it is clear that this subset of receptors is required for agonist-induced adhesion to ␣ M ␤ 2 ligand (61). Although adhesion to IC induced both increased surface expression of the integrin and the conformational change associated with binding of CBRM1/5, increased surface expression was not required for sustained adhesion. In contrast, the conformational change in ␣ M ␤ 2 recognized by CBRM1/5 was required for sustained adhesion to IC. This is similar to the conclusions about the role of ␣ M ␤ 2 in PMN-endothelial adhesion (52) and in PMN aggregation (50,51) and makes ␣ M ␤ 2 activation similar to activation of ␣ L ␤ 2 , another ␤ 2 integrin which exhibits regulated adhesion without changes in surface expression (84). Thus, while PI 3-kinase is involved in activating regulated secretion in PMN which results in increased plasma membrane expression of ␣ M ␤ 2 , its role in sustaining adhesion to immune complexes requires only induction of integrin activation. Sustained PMN adhesion to IC leads in turn to enhanced generation of LTB 4 (8), superoxide (83), and mediators of inflammation.
Our data demonstrate that Fc␥R and seven transmembrane receptor ligation induce distinct pathways that converge into a common pathway for activation of ␣ M ␤ 2 avidity. A potential effector of this common pathway is the Rac/Cdc42-activated kinase Pak1. Fc␥R-induced Pak1 activation is dependent on PI 3-kinase activity, whereas fMLP activation of Pak1 is independent of PI 3-kinase, consistent with their effects on ␣ M ␤ 2 activation. For both IC and fMLP, Pak1 activation is independent of ␣ M ␤ 2 ligation as assessed by antibody inhibition, suggesting it is upstream of integrin activation. Cdc42 and Rac regulate focal complex formation and adhesion in fibroblasts and the macrophage cell line Bac1.2F5 (85,86), consistent with the possibility that they regulate integrin avidity. Since there are several effector pathways initiated by both Rac and Cdc42, it is possible our data reflect a requirement for these small GTPases rather than for Pak1 itself. Our suggestion that Pak1 regulates integrin activation is supported by the finding that expression of an activated form of Pak1 in Swiss 3T3 cells causes large focal adhesions to form and actin accumulation in lamellipodia (54). Interestingly, PDGF and insulin receptorinduced actin cytoskeletal re-organization mediated by Rac are inhibited by wortmannin, while LPA and bombesin responses, which signal via G protein-linked receptors, are not (87), again suggesting that tyrosine-kinase pathways generally activate Rac and Pak1 by a PI 3-kinase-dependent mechanism while the G protein-dependent receptor-initiated pathway does not. It is intriguing as well that cytohesin-1, a cytosolic regulator of ␤ 2 -mediated adhesion which binds to the ␤ 2 cytoplasmic tail, is a guanine nucleotide exchange factor for Arf-1 (85,88). We suggest that PI 3-kinase is a necessary effector of tyrosine kinase-mediated but not seven transmembrane receptor-mediated integrin regulation in PMN. These cascades converge at activation of Pak1 through the small GTPases Rac and Cdc42. This model predicts that the rapid, reversible activation of integrin-mediated adhesion that is necessary for chemotaxis and transendothelial migration induced by chemoattractants is controlled by a distinct proximal pathway from that which activates the sustained, integrin-mediated adhesion necessary for PMN effector functions such as Fc␥R-mediated phagocytosis.