|Home | About | Journals | Submit | Contact Us | Français|
Neutrophils undergo spontaneous apoptosis, but their survival can be extended during inflammatory responses. αMβ2 is reported either to delay or accelerate neutrophil apoptosis, but the mechanisms by which this integrin can support such diametrically opposed responses are poorly understood. The abilities of closely related αMβ2 ligands, plasminogen and angiostatin, derived from plasminogen, as well as fibrinogen and its two derivative αMβ2 recognition peptides, P1 and P2-C, differed markedly in their effects on neutrophil apoptosis. Plasminogen, fibrinogen, and P2-C suppressed apoptosis via activation of Akt and ERK1/2 kinases, while angiostatin and P1 failed to activate these prosurvival pathways and did not prevent neutrophil apoptosis. Using cells transfected with αMβ2 or its individual αM or β2 subunits, and purified receptors and its constituent chains, we show that engagement of both subunits with prosurvival ligands is essential for induction of the prosurvival response. Hence, engagement of a single integrin by closely related ligands can induce distinct signaling pathways, which can elicit distinct cellular responses.
Neutrophils (polymorphonuclear leukocytes, or PMNs)3 are terminally differentiated cells with a very short half-life, being committed to programmed cell death (apoptosis) (1, 2). PMN apoptosis is critical in the clearance of such cells from inflamed tissue, a process that supports the resolution of protective inflammatory responses. However, in some inflammatory diseases, such as arthritis, meningitis, peritonitis, hypersensitivity reactions, and acute coronary syndromes, PMNs not only accumulate but also survive for extended times, leading to injury in the affected tissues (3–5). These observations support the hypothesis that constitutive apoptosis, also referred to as spontaneous or passive apoptosis, is a default fate of PMNs, which can be delayed or accelerated depending on the balance of proapoptotic and anti-apoptotic signals. Agents that promote PMN responsiveness, such as GM-CSF, IL-8, LPS, and fibrinogen (Fg) delay PMN apoptosis (6 –9). These agents promote PMN survival by activating intracellular signaling pathways, such as MAPK, particularly ERK1/2, and PI3K/Akt pathways (10, 11). In contrast, Fas ligand or TNF-α promote PMN death via interaction with their receptors on the cell surface (12, 13).
Growing evidence demonstrates a role of integrin αMβ2 (CD11b/CD18, Mac-1) in regulation of PMN apoptosis. This member of the β2 integrin family is a heterodimer composed of a unique αM (CD11b) subunit noncovalently linked to the common β2 subunit (CD18), which is shared by other leukocyte integrins. αMβ2 is indispensable for diverse PMN functions crucial for innate immunity, including adhesion, transmigration across the endothelium, phagocytosis, and the oxidative burst (reviewed in Ref. 14). The role of αMβ2 as a key regulator of the PMN lifespan depends on its capacity to either delay or accelerate apoptosis. For example, PMN exposure to immobilized or soluble αMβ2 ligands, Fg and ICAM-1, Ab crosslinking of αMβ2 on these cells, or αMβ2-dependent PMN transmigration through endothelium induces survival signals (9, 15, 16). On the other hand, engagement of αMβ2 with activating Abs in the presence of proapoptotic agents, TNF-α or anti-Fas Ab, accelerates PMN apoptosis (15). Phagocytosis of complement-opsonized targets including Escherichia coli also induces PMN death in an αMβ2-dependent manner (17, 18). Studies of β2-deficient mice demonstrate that they exhibit a neutrophilia. This increase in PMN in the β2-deficient mice has been ascribed to up-regulation of antiapoptotic Bcl-XL, which delays apoptosis (19), but others have suggested that neutrophilia is the result of increased granulocytosis induced by elevated levels of IL-17 and G-CSF without effects on apoptosis (20, 21). Thus, the mechanisms of αMβ2-dependent regulation of PMN apoptosis are complex and unresolved.
Several recent observations point to a potential role of plasminogen (Plg), a major fibrinolytic plasma protein, in PMN apoptosis. First, regulation of the inflammatory response by Plg is observed in vivo. PMN recruitment induced by biopolymer implants is attenuated in Plg−/− mice as compared with wild-type mice (22). Additionally, Plg gene expression is up-regulated by inflammatory cytokines, resulting in an increase in circulating levels of Plg (23, 24). Second, plasmin (Plm), the active enzymatic form of Plg, degrades extracellular matrix proteins, leading to detachment and apoptosis of smooth muscle and neuronal cells as well as fibroblasts (25–27). On the other hand, with nonadherent cells such as monocytes, Plg binding capacity is significantly increased on late apoptotic cells (28) and Plm inhibits TNF-α-induced apoptosis in monocytes via a PAR-1-dependent manner (29).
It has been previously demonstrated by us (30) and others (31, 32) that Plg and its short derivative Ang(1– 4), angiostatin composed of four kringle domains of Plg, are ligands of αMβ2 integrin. As both Plg and αMβ2 are important modulators of leukocyte survival, in the present study we examined how interactions of Plg and Ang(1– 4) with αMβ2 influence PMN apoptosis. Our results led us to compare the effects of other αMβ2 ligands on PMN apoptosis. Ultimately, our studies have identified a molecular and cellular basis to explain how occupancy of the same receptor, αMβ2, by closely related ligands can elicit prosurvival responses in PMN.
Glutamic Plg was isolated from normal human plasma by affinity chromatography on lysine-Sepharose followed by gel filtration (33). Ang(1– 4), Ang(1–3), and Ang(1–5) and BSA were from Calbiochem. mAbs to the αM subunit (44a and 904), to the β2 (TS1/18 and IB4) and to MHC class I (W6/32) were from American Type Culture Collection (ATCC). mAb 24 was kindly provided by N. Hogg (34). Fas receptor-activating mAb was from Upstate Biotechnology. Human Fg was from Enzyme Research Laboratories, and neutrophil inhibitory factor (NIF) was a gift from Corvas International. Peptides corresponding to sequences in the fibrinogen γ-chain, P1, Fgγ(190)GWTVFQKRLDGSV(202) and P2-C, Fgγ(385)-KIIPFNRLTIG(395), were synthesized on an Applied Biosystems model 430A peptide synthesizer and purified on HPLC as described (35). Methyl-β-cyclodextrin (MβCD) (36), cytochalasin D, 2,3-butanedione 2-monoxime (BDM), 1-(5-isoquinolinylsulfonyl)-2-methylpiperazine dihydrochloride (H7) were from Sigma-Aldrich, and myosin light chain kinase inhibitor peptide 18 and calpain inhibitor III were from Calbiochem.
PMNs were isolated from human peripheral blood of healthy volunteers drawn into sterile acid citrate dextrose (1/7 (v/v) 145 mM sodium citrate (pH 4.6) and 2% dextrose). Isolation was performed by density gradient centrifugation onto Ficoll-Hypaque (Pharmacia), followed by dextran sedimentation of erythrocytes and hypotonic lysis of residual erythrocytes.
Two constitutively active mutants of αMβ2 were generated: αM(Ile316→Gly)β2 (37) and αM(D294C/Q311C)β2 (38) using the QuikChange Multi Site-Directed Mutagenesis kit (Stratagene) according to the manufacturer’s instructions. Cytoplasmic tails were removed from αM and β2 cDNAs by replacing the codons for Lys1120 in αM and Leu729 in β2 with stop codons by PCR to generate αM(1119Δ) and β2(728Δ), which were subsequently cloned to pcDNA3.1 vector using a TOPO cloning kit (In-vitrogen). The αM/αLβ2 chimera cDNA encoding the extracellular/transmembrane regions of αM and cytoplasmic tail of αL (39) was recloned to pcDNA3.1 vector using a TOPO kit.
Human erythroleukemic K562 cells were transiently transfected with 10 μg of pcDNA 3.1 (Invitrogen) containing respective cDNAs for wild-type or mutant αM and/or β2 or vector alone (mock-transfected) using Cell Line Nucleofector Kit V (Amaxa) and the O-17 program. Forty-eight hours after transfection, receptor expression and activation status were analyzed routinely by flow cytometry (FACS) using a FACSCalibur instrument (BD Biosciences) and mAbs 44a, 904, IB4, TS1/18, and 24 recognizing the activation-dependent epitope of the integrin. The data were analyzed with the CellQuest program (BD Biosciences).
PMNs or K562 cells expressing various variants of αMβ2 or mock cells (48 h after transfection) were resuspended in HBSS buffer containing 1 mM CaCl2, 1 mM MgCl2, and 0.1% BSA (pH 7.4) at density of 1 × 106 PMNs/ml or 1.5 × 106 K562 cells/ml and incubated in the absence or presence of increasing concentrations of human Plg, Ang(1– 4), Ang(1–5), NIF, Fg (0 –10 μM), and P1, P2-C, P2-Cscr peptides (0 –500 μM) for 4 –24 h at 37°C in suspension with rotation. In inhibition experiments, PMN were pretreated for 20 min at 37°C with 10 μM Akt inhibitor from Calbiochem, 50 μM ERK1/2 inhibitor, PD 98059 (Calbiochem), or 50 nM NIF before addition of αMβ2 ligands, and these inhibitors were present for the 16 h of incubation. The inhibitors of integrin clustering MβCD (5 mM), myosin L chain kinase (MLCK) inhibitor peptide 18 (0.5 μM), H7 (300 μM), BDM (20 mM), cytochalasin D (10 μM), and calpain inhibitor III (10 μM) were added simultaneously with αMβ2 ligands and cell mixtures were incubated for 16 h. Cell apoptosis was estimated using annexin V FITC apoptosis detection kit (Calbiochem), according to the manufacturer’s instructions, and analyzed by FACS. As defined, cells that were annexin V+propidium iodide (PI)− were in early apoptosis, and cells that were annexin V+PI+ were in late apoptosis. Alternatively, cell pellets, obtained by centrifugation, were lysed in Chaps Cell Extract Buffer (Cell Signaling Technology), and 40 μg of total protein from each sample was analyzed by Western blot using Abs recognizing cleaved products of caspase-3 (BD Biosciences), caspase-8 (Cell Signaling Technology), or to bax (BD Biosciences). Anti-GAPDH (Chemicon International) was used as a loading control. Blots were developed using the secondary HRP-conjugated goat anti-rabbit or anti-mouse IgG (Calbiochem) and SuperSignal West Pico chemiluminescent substrate (Pierce).
K562 cells or PMNs stimulated with 1 nM PMA were incubated in the absence or presence of the αMβ2 ligands in HBSS buffer containing 1 mM CaCl2, 1 mM MgCl2, and 0.1% BSA (pH 7.4) for 30 min at 37°C and then lysed in ice-cold lysis buffer (10 mm Tris (pH 7.5), 5 mm EDTA, 50 mm sodium pyrophosphate, 50 mm NaF, 50 mm NaCl, 0.5% Triton X-100, 0.1% SDS, 1% Nonidet P-40, 0.1 mm Na3VO4, and 1 mm PMSF. In the inhibition experiments, PMNs were pretreated for 20 min at 37°C with 10 μM Akt inhibitor or 50 μM ERK1/2 inhibitor (PD 98059, Calbiochem), 50 nM NIF, F(ab′)2 fragments of anti-αM mAb (clone 44a) or anti-MHC-1 (cloneW6/32) (20 μg/ml) before addition of Plg. Cell lysates were clarified by centrifugation, and protein concentration was measured in supernatants using the Bio-Rad DC protein assay (Bio-Rad Laboratories). Equal amounts of total protein from cell lysates were analyzed by Western blots using PathScan Multiplex Western Cocktail I, anti-phospho-Akt, anti-Akt, anti-phospho-ERK1/2, anti-ERK1/2, or anti-actin Abs (Cell Signaling Technology) and were developed as described above. Since such phosphorylation responses occur rapidly after integrin engagement and often diminish with time, we used a low dose of PMA (1 nM) as a stimulus of integrin activation to enhance binding of αMβ2 ligands, whose interactions with PMN (except NIF) are dependent on receptor activation. Other experiments such as apoptosis and integrin clustering assays did not require presence of PMA, since there was enough time (4 –16 h) for the integrin-ligand interactions to occur, more likely due to increasing αMβ2 activation.
PMNs were incubated in the absence or presence of the αMβ2 ligands as described under “Apoptosis assay” for 4 h at 37°C, fixed with 4% paraformaldehyde for 15 min at 22°C, and stained with anti-αM mAb (10 μg/ml) (clone OKMI) and Alexa 488-conjugated goat anti-mouse IgG (1/500). The inhibitors of integrin clustering (final concentrations as specified above) were added simultaneously with the ligands. Following washing with HBSS, the cells were cytospun onto Superfrost Plus microscope slides and mounted using Vectashield mounting medium (Vector Laboratories). The samples were observed using a Leica DMR microscope equipped with ×5/0.12 numeric aperture (NA), ×10/0.4 NA, ×20/0.5 NA, or ×40/0.7 NA objective lenses (Leica Microsystems). Images were photographed with a Qimaging Retiga ExiFas camera using Image Pro 5.1 software (Media Cybernetics).
Non-tissue culture-treated 96-well Falcon plates (BD Biosciences) were coated with 100 μl of Plg, Ang(1– 4) at 0.2 μM, or with Fg (1 μg/ml) in PBS overnight at 4°C and then postcoated with 0.5% polyvinylpyrrolidone (PVP) for 1 h at room temperature (40). Control wells were coated with PVP only. Before use, the plates were rinsed three times with PBS. Transfected K562 cells were harvested, washed with HBSS, and resuspended in DMEM F-12. The cells were seeded at 1.5–2 × 105 cells/well onto the assay plates and incubated at 37°C for 30 min. The plates were washed three times with HBSS and adherent cells were quantified using the Cy-Quant cell proliferation assay kit (Molecular Probes) according to the manufacturer’s instructions. Data are presented as relative fluorescence units (RFU) ± SEM of three independent experiments. PMNs, which were stimulated with 1 nM PMA, or transfected K562 cells were incubated with soluble human Alexa 488-conjugated Plg or Ang(1– 4) in HBSS buffer containing 1 mM CaCl2 and 1 mM MgCl2 for 40 min at 37°C followed by two washings with the same buffer. In the inhibition experiments, the cells were pretreated with function-blocking mAbs to αM or β2 subunit (20 μg/ml) or NIF (20 nM) for 20 min at 37°C.
αMβ2, αM, and β2 recombinant integrin subunits were purified as previously described (35). These were coated onto 96-well Immulon 4HBX microtiter plates (Dynatech Laboratories) in TBS containing 10 mM n-octyl-β-D-glucopyranoside overnight at 4°C and postcoated with 0.5% PVP for 1 h at 37°C. The αMβ2 ligands were biotinylated using EZLink sulfo-NHS-LC-biotin (Pierce) according to the manufacturer’s instructions. Next, increasing concentrations (0 –10 μM) of biotinylated ligands were added to the respective wells and incubated for 2 h at 37°C in TBS containing 10 mM n-octyl-β-D-glucopyranoside and 1 mM CaCl2/1 mM MgCl2. The bound ligands were detected using alkaline phosphatase-conjugated ImmunoPure avidin (Pierce Chemicals) and para-nitrophenylphosphate (Pierce Chemicals) as the substrate, and absorbance at 405 nm (A405) was measured. In binding isotherm studies, the Kd values of Plg and Ang(1– 4) binding to αMβ2 and its subunits were estimated using the SigmaPlot software (SPSS) in which a one-site binding equation was used to fit the data.
The data are expressed as means ± SEM. To determine the significance of differences between two groups, a two-tailed Student’s t test was performed using the SigmaPlot software program (SPSS); p < 0.05 was considered significant.
Integrin αMβ2 is not only involved in PMN adhesion and migration but also controls PMN apoptosis and may accelerate or suppress cell death in specific settings (reviewed in Ref. 41). We previously demonstrated that αMβ2 interacts with Plg by enhancing Plg activation on the PMN surface (30). Fg, also a ligand of αMβ2, has been shown to delay apoptosis via its engagement of the integrin (9), and we first sought to determine whether these two interrelated plasma protein ligands exerted similar effects on PMN survival. In parallel, we determined if a shorter Plg fragment Ang(1– 4), which is composed of the first four kringle domains of Plg and has antiangiogenic and antiinflammatory properties (32, 42), behaved similar to its parent molecule with respect to PMN apoptosis. As an initial monitor of apoptosis, we measured FITC-conjugated annexin V binding and PI staining by FACS. Increasing concentrations of these αMβ2 ligands were added to human PMNs and incubated for 4 or 16 h at 37°C with gentle rotation. After 4 h in the absence of αMβ2 ligands, 20 ± 5% were in early apoptosis (annexin V+PI−) (Fig. 1A). The percentage of early apoptotic cells was reduced by Plg in a dose-dependent manner; at 1 μM Plg, apoptosis was only 3.5 ± 1.5%. The protective effect of Plg was similar to that observed with Fg; that is, 5 ± 2% (Fig. 1A, left panel). In contrast to intact Plg, Ang(1– 4) failed to protect PMNs against apoptosis at any concentration tested, even as high as 20 μM. Additionally, NIF, a high-affinity ligand of αMβ2, known to block αMβ2-dependent PMN functions (43), also failed to reduce apoptosis. Two peptides derived from Fg γ-chain have been shown to interact with αMβ2. These peptides, designated P1γ(190)GWTVFQKRLDGSV(202) and P2-Cγ(385)KIIPFNRLTIG(395), both interact with the I domain within the αM sub-unit of αMβ2 (35, 44). In view of the distinct effects of Plg and Ang(1– 4) on PMN spontaneous apoptosis, we compared the effects of the two Fg peptides on cell survival. As shown in Fig. 1A (right panel), the P2-C peptide reduced PMN apoptosis in a dose-dependent manner while the P1 and scrambled P2-C peptides did not have any prosurvival effect (17–22 ± 5% of early apoptotic cells).
At 4 h only 2–3% of PMNs were in late apoptosis (annexin V+PI+). At 16 h, a robust increase of not only early (annexin V+PI−) but also of late apoptotic PMNs was observed. The data in Fig. 1B examine all annexin V+ cells and demonstrate that different αMβ2 ligands still exert distinct effects on PMN survival. Plg, Fg, and P2-C reduced PMN apoptosis in a dose-dependent manner (20 –30 ± 4 –10% apoptotic cells) as compared with PMNs incubated without αMβ2 ligands (90 ± 20% apoptotic cells). On the other hand, Ang(1– 4), NIF, P1, and P2-Cscr did not rescue PMNs from apoptosis (80 –90 ± 5–15% apoptotic cells). In the presence of the prosurvival ligands, the percentage of apoptotic PMNs still increased from 4 to 16 h (3–5% apoptotic PMNs at 4 h vs 20 –30% apoptotic PMNs at 16 h), suggesting that the prosurvival αMβ2 ligands delay rather than prevent apoptosis.
Additionally, the αMβ2 prosurvival ligands reduced not only spontaneous, but also Fas-induced PMN apoptosis. When PMNs were incubated with mAb activating the Fas receptor for 4 h, 50 ± 8% PMNs were in late and early apoptosis (annexin V+), whereas in the presence of Plg, Fg, or P2-C, only 4 –15 ± 2–5% of the treated PMNs were apoptotic. The ligands that did not affect spontaneous PMN apoptosis also failed to reduce the Fas-dependent apoptosis (data not shown). In control experiments, when Plg was added to apoptotic PMNs, it did not compete with annexin V-FITC binding at any concentration tested (0 –10 μM) (data not shown).
Another characteristic phenotype of apoptotic cells is their activation of caspases, a family of cysteine proteases that coordinates the structural dismantling of the cell (45). Activation of caspase-8 and caspase-3, which regulate PMN spontaneous apoptosis, and cleavage of Bax, a proapoptotic member of the Bcl-2 family, were assessed. As shown on Fig. 1C, the 22-kDa Bax and caspase-8 fragments, as well as the 18-kDa caspase-3 fragment, were detected in the lysates of PMNs incubated for 16 h in the buffer alone or in the presence of Ang(1– 4), the P1 peptide, and control P2-Cscr, while these apoptotic markers were absent in cells treated with the prosurvival αMβ2 ligands, Plg, Fg, and P2-C. These results provide independent corroboration of the differential effects of the various αMβ2 ligands on PMN survival.
Activation of Akt and/or ERK1/2 has been implicated in antiapoptotic signaling in PMNs and other cells (9, 15, 46). Thus, we sought to determine whether the prosurvival αMβ2 ligands selectively induced phosphorylation of these kinases. As shown in Fig. 2A, Plg, Fg, and P2-C triggered robust Akt phosphorylation as compared with untreated cells, while Akt phosphorylation was negligible in the presence of Ang(1– 4), P1, and P2-Cscr. Similarly, Plg, Fg, and P2-C induced ERK1/2 activation, whereas Ang(1– 4), P1, and control P2-Cscr did not (Fig. 2B). The data shown in Fig. 2 are representative of at least three experiments performed with PMNs from different donors and distinguish the prosurvival ligands from those that do not prolong PMN survival.
Activation of Akt and ERK1/2 by the prosurvival ligands of αMβ2 suggests that these kinases may be directly involved in the delay of PMN spontaneous apoptosis. Indeed, when PMNs were incubated with Plg and an Akt inhibitor, a phosphatidylinositol ether analog, the Plg-mediated PMN protection from apoptosis was completely abrogated (32 ± 10% with Plg, 85 ± 22% (n = 6) with Plg + Akt inhibitor), indicating a critical role of Akt in PMN survival (Table I). Additionally, complete inhibition of ERK1/2 activity with PD 98059 (as shown on the Western blot in Fig. 2D) also reduced Plg-dependent PMN survival, but to a lesser extent than the Akt inhibitor (64 ± 14% (n = 6) apoptotic cells). Consistent with these effects on apoptosis, these inhibitors significantly blocked Plg-dependent Akt and ERK1/2 activation in PMNs as assessed by Western blots (Fig. 2, C and D).
Since αMβ2 was identified as a major Plg receptor on human PMNs, we sought to verify that this integrin participates in Plg-dependent PMN survival. As shown in Table I, F(ab′)2 fragments of an anti-αM mAb, but not of the control anti-MHC-1 mAb, completely inhibited Plg-dependent PMN survival to the level observed in the absence of Plg (Plg + anti-αM: 80 ± 20% vs Plg + anti-MHC-1: 30 ± 10%). Additionally, these reagents substantially reduced Plg-induced activation of Akt and ERK1/2 (Fig. 2, C and D), while the control anti-MHC-1 mAb did not. These results indicate that the engagement of integrin αMβ2 with Plg triggers activation of Akt and ERK1/2, which subsequently delays PMN spontaneous apoptosis. Of note, NIF, which does not have a protective effect on PMNs (Fig. 1, A and B) and inhibits Plg binding to αMβ2 (30), also abrogated the protective effects of Plg (Table I).
In view of previous findings that indicated involvement of αMβ2 clustering in regulation of PMN apoptosis (15), we analyzed αMβ2 clustering in the presence of its various ligands.
As shown in Fig. 3A, the αMβ2 macroclusters, as defined by Kim et al. (47), were detected on the PMN surface in the presence of the antiapoptotic ligands, Fg, Plg and P2-C, while Ang(1– 4) and P1 did not support αMβ2 clustering. With these latter ligands or with PMNs incubated in the absence of ligands, αMβ2 was more uniformly distributed with more intense staining along the rims of the cells. Thus, the ligands that prolong PMN survival trigger αMβ2 clustering, whereas ligands that do not do so also fail to induce integrin macroclustering. Next, we utilized several inhibitors of integrin clustering to assess the role of this response in PMN survival: 1) MβCD, a disruptor of membrane lipid rafts, which are pivotal in integrin clustering (36); 2) inhibitors of MLCK activity, MLCK inhibitor peptide 18 and H7, which inhibit phosphorylation of MLCK (48); 3) modifiers of the cytoskeleton: BDM, an inhibitor of myosin ATPase activity, and actin-myosin interactions (48); 4) cytochalasin D, an disrupter of the actin cytoskeleton (49); and 5) calpain inhibitor III, which releases integrin from its cytoskeleton constraint (49, 50). In preliminary experiments, all of these inhibitors exerted their anticipated effects on αMβ2; that is, they blocked or significantly reduced Plg-dependent αMβ2 clustering in PMNs as assessed by fluorescence microscopy (Fig. 3B). Most of the inhibitors tested did not modify PMN apoptosis in the absence of αMβ2 ligands, and all completely blocked PMN survival dependent on engagement of αMβ2 with Fg, Plg, and P2-C, but they did not affect apoptosis of cells incubated with P1 (Table II). In control experiments, Plg and Fg binding to αMβ2, when they were added to PMNs together with the clustering inhibitors, was not inhibited as assessed by FACS (Table III). Taken together, our results indicate that engagement of αMβ2 with the ligands that protect PMNs from apoptosis, but not with the ligands failing to do so, leads to αMβ2 clustering, which is a crucial step in αMβ2-dependent PMN survival.
A mechanism was sought to account for the differential effects of specific ligands on PMN apoptosis, focusing on Plg and Ang(1– 4) as being representative of ligands that do or do not alter PMN survival. Human erythroleukemic K562 cells, which do not express αMβ2, were transiently transfected with the cDNAs encoding for the separate αM or β2 subunits or cotransfected with both cDNAs. In previous studies conducted in HEK cells (35), we have shown that this transfection strategy can lead to cell lines expressing the separate subunits or the αMβ2 heterodimer. The generation of such cells in the K562 background was successful and provided cells that undergo spontaneous apoptosis. As assessed by FACS, cells expressing αMβ2 or the original αM or β2 subunits were reactive with appropriate mAbs: αM cells stained with mAbs to the αM subunit (clones 44a, 904) but not with mAbs to the β2 subunit (clones IB4, TS1/18); β2 cells stained with the mAbs to the β2 subunit but not with mAbs to the αM subunit; and αMβ2 cells stained with both sets of mAbs. None of the mAbs reacted with the mock-transfected cells (data not shown). The integrin activation status on these cells was analyzed by FACS using mAb 24, which recognizes an activation-dependent epitope in the αM subunit (34). In the presence of Ca2+/Mg2, the αMβ2 cells (mean fluorescence intensity of 92.6 ± 20) and αM cells (95.7 ± 22) were reactive whereas the mock cells were not (mean fluorescence intensity of 4.28 ± 1.2). Integrin activation was further confirmed in functional studies. The αMβ2, the αM, and the β2 cells bound soluble Fg and adhered to immobilized Fg in the αM- or β2-specific manner; that is, these interactions were inhibited by the appropriate function-blocking mAbs to the αM or the β2 subunit to the level observed with the mock cells (data not shown). Thus, αMβ2 and its subunits are at least partially activated on the K562 cells.
With these characterizations established, the interaction of these cells with Plg and Ang(1– 4) was characterized. Cells expressing the αMβ2 heterodimer or the αM subunit alone adhered to Plg and Ang(1– 4) (Fig. 4A). In contrast, cells expressing the β2 subunit alone adhered to Plg but not to Ang(1– 4). This difference was confirmed by FACS. As shown in Fig. 4B, the K562 cells expressing the β2 subunit bound soluble Plg but failed to bind soluble Ang(1– 4). K562 cells expressing αMβ2 or the αM subunit alone bound both ligands. Finally, we validated this distinction in recognition specificity in direct binding studies using αMβ2, αM, or β2 purified from HEK293 cell lines (35). Using solid phase assays in which the integrin or its subunits were immobilized, various concentrations of biotinylated Plg or Ang(1– 4) were added, and the binding isotherms were analyzed to determine the affinity of each ligand for each receptor component. As summarized in Table IV, both Plg and Ang(1– 4) bound to αMβ2. The affinity of the intact integrin was ~10-fold higher for Plg than for Ang(1– 4). When the binding of these ligands to separate integrin subunits was analyzed, Plg interacted with the individual αM and β2 subunits with similar affinities. In contrast, Ang(1– 4) was recognized only by the αM and not by the β2 subunit. The affinity of Ang(1– 4) for the αM subunit was similar to its affinity for the heterodimeric receptor. Taken together, these data demonstrate a distinct interaction mechanism for Plg and Ang(1– 4) with αMβ2: Plg interacts with both integrin subunits, while the αM subunit predominates in recognition of Ang(1– 4).
We sought to determine whether this difference in recognition specificity extended to Fg and its peptides, which also exert differential effects on PMN survival. The binding affinities of these ligands for αMβ2 and its individual subunits (see Table IV) indicate that Fg and P2-C peptide engage both integrin subunits, while P1 is recognized predominantly by the αM subunit of the αMβ2 heterodimer. Thus, the prosurvival ligands Plg, Fg, and P2-C interacted with the β2 subunit while the ligands that did not affect spontaneous apoptosis, Ang(1– 4) and P1, did not. Falling into the latter category was NIF, which previously has been demonstrated to interact exclusively with the I domain within the αM subunit (51), and consistently NIF bound very poorly to the β2 subunit (Table IV) and failed to prolong PMN survival.
Having observed these recognition differences, we sought to determine whether and how these ligands regulate survival of K562 cell lines. The cells were incubated in the presence or absence of Plg, Ang(1– 4), Fg, and the two Fg peptides in serum-free HBSS buffer for 24 h, and the percentage of early and late apoptotic cells was measured by FACS. As shown on Fig. 5A, all tested cells, αMβ2, αM, and β2, underwent spontaneous apoptosis (56 – 67%, SEM = 10 –16%, n = 3) in the absence of αMβ2 ligands. Apoptosis of the αMβ2 cells was blocked by Plg (5 ± 2% apoptotic cells, n = 3) but not by Ang(1– 4) (55 ± 10% apoptotic cells, n = 3). Fg and P2-C had the same effect as Plg, prolonging the survival of the αMβ2 cells (3 ± 1% and 4 ± 2% apoptotic cells, n = 3), whereas P1 (56 ± 7% apoptotic cells, n = 3) and the P2-Cscr peptide (59 ± 12% apoptotic cells, n = 3), like Ang(1– 4), failed to do so. However, none of tested ligands prevented apoptosis of the αM cells, the β2 cells, and the mock cells (55– 65% apoptotic cells, SEM = 8 –17%, n = 3).
These results were corroborated when the activation of caspase-3, caspase-8, and Bax were analyzed in cell lysates. As shown on Fig. 5, B–D, the cleavage products of these caspases and Bax were not detected in the αMβ2 transfectants treated with Plg, Fg, and P2-C (Fig. 5, B–D, row 1), while they were present in untreated as well as in Ang(1– 4), P1, and P2-Cscr-treated αMβ2 cells. The mock-transfected cells and cells expressing the single integrin subunits were not protected from apoptosis by any of tested ligands, and cleaved caspase-3, caspase-8, and Bcl-2 protein Bax were found in all these cell lysates.
Since interaction of the prosurvival ligands with αMβ2 induced extensive phosphorylation of Akt and ERK1/2 in PMNs, which was critical to the survival response (see above), we examined activation of these kinases in the K562 cells expressing the various integrin subunits. As in PMN, robust Akt phosphorylation was triggered by Plg, Fg, and P2-C, but only in the K562 cells expressing αMβ2 cells and not in K562 cells expressing the individual αM or β2 subunits (Fig. 6A) or mock-transfected cells. However, Ang(1– 4) and P1, which are recognized predominantly by the αM subunit but not by the β2 subunit, failed to induce Akt activation, not only in the αMβ2 cells but also in the three other cell types tested. In contrast to Akt, ERK1/2 activation was strongly induced by Plg, Fg, and P2-C not only in the αMβ2 cells but also in the αM cells, indicating that engagement of the αM integrin subunit with these ligands is sufficient for induction of ERK1/2 activation. ERK1/2 phosphorylation was poorly induced by these ligands in mock cells and the β2 cells and by Ang(1– 4) and P1 in all transfected K562 cells. Moreover, activation of both kinases was negligible in all untreated cells. Thus, ERK1/2 is necessary for the prosurvival ligands of αMβ2 to be optimally protective, but is not sufficient to induce a survival response.
Knowing that only the ligands that engage both integrin subunits of αMβ2 protect K562 cells from apoptosis, we sought to establish the role of αM and β2 cytoplasmic tails in this process. Thus, in addition to wild-type αMβ2, we expressed mutant receptors with cytoplasmic tail truncations either within the αM (αM(1119Δ)β2) or β2 subunit (αMβ2(728Δ)). We also expressed the chimeric receptor, in which the cytoplasmic tail of the αM subunit was replaced with that of the αL subunit (αM/αLβ2), another member of the leukocyte β2 integrin family. As has been assessed by FACS, the expression levels of each of these αMβ2 mutants were similar to those of the wild-type αMβ2 integrin. The activation status of the truncated αMβ2 variants was ~30% higher than of wild-type αMβ2 based on staining with mAb 24 (data not shown). Additionally, as shown in Table V, the αMβ2 tailless mutants and chimeric αM/αLβ2 maintained their abilities to bind ligands; the binding of Alexa 488-labeled Plg or Fg was similar to that of wild-type αMβ2. However, K562 cells expressing αM(1119Δ)β2 or αMβ2(728Δ) lost their ability to survive even in the presence of the prosurvival ligands; for example, the percentage of apoptotic cells in the presence of Plg (~79 ± 10%, n = 5) was similar to that of mock cells (~78 ± 10%, n = 5) or αMβ2 cells without ligand present (72 ± 14, n = 5) (Table VI). K562 cells expressing the αM/αLβ2 chimeric receptor also underwent apoptosis in the presence of prosurvival ligands (76 ± 13%, n = 5), although their interactions with these ligands were not impaired. Additionally, no Akt activation could be detected in cells expressing the tailless variants of αMβ2 or chimeric αM/αLβ2 (Fig. 6B). These data indicate the crucial importance of both αM and β2 cytoplasmic tails in transmission of prosurvival signal and suggest integrin-dependent specificity in regulation of cell survival signals.
Next, we sought to determine the role of αMβ2 activation in K562 cell survival. Two constitutively active mutants of αMβ2 were αM(Ile316→Gly)β2 and αM(D294C/Q311C)β2. Both mutations stabilize the αM-I domain, the major ligand recognition site, in an open conformation, which results in significant enhancement of αMβ2 activation (37, 38). Indeed, K562 cells expressing these mutants showed enhanced adhesion to Plg as well as soluble Plg binding (3– 4-fold greater than cells bearing wild-type αMβ2), while their expression levels were comparable (data not shown). When apoptosis of these cells was compared in the absence of ligands, the percentage of apoptotic cells was similar in all cells, and in the presence of the prosurvival ligands, the observed subtle differences were not statistically significant ( p = 0.064 – 0.2073) (Table VI). Importantly, the ligands, which did not support survival in αMβ2 cells, also failed to protect the cells bearing the constitutively active variants of αMβ2. Thus, these results indicate that apoptosis is regulated by the integrin engagement with the specific ligands rather than by integrin activation per se.
The influence of integrin ligation on the survival of adherent cells is well documented (reviewed in Ref. 52), and growing evidence suggests that integrins may also participate in the survival of circulating cells, such as leukocytes. For example, engagement of the β7 integrins with ligands augments the survival of eosinophils (53), and interaction between α9β1 and vascular cell adhesion molecule-1 (VCAM-1) inhibits PMN apoptosis (54). The involvement of αMβ2 in control of PMN apoptosis also has been examined extensively (reviewed in Ref. 41), but the data are contradictory, showing that αMβ2 can either accelerate or delay PMN apoptosis (9, 15, 16, 19 –21, 55). Thus, molecular mechanisms governing the effects of αMβ2 on PMN apoptosis remain poorly understood. In this study, we demonstrate that different αMβ2 ligands, even very closely related ones, can exert different effects of PMN apoptosis, with a subset of ligands inducing a prosurvival response. Furthermore, we provide a mechanism for these differential effects: ligands that engage both subunits of αMβ2 prolong PMN life by inducing prosurvival signaling pathways, primarily Akt activation, while ligands that engage primarily the αM subunit are unable to induce the prosurvival signaling responses and do not delay PMN apoptosis (Fig. 7).
Our previous studies identified Plg, as well as its activator urokinase (uPA), as novel ligands of αMβ2. Their interactions with this integrin enhances uPA-dependent Plg activation on PMN surface, which, in turn, facilitates PMN migration and PMN-mediated fibrinolysis, activities that are directed toward the extracellular environment of the cell (30, 56). The present study demonstrates that interaction between Plg and αMβ2 on PMNs also induces intracellular signaling that results in a prosurvival response. Plg, in contrast to Ang(1– 4), its shorter four-kringle derivative, significantly inhibits spontaneous apoptosis of PMNs. This effect is abrogated when αMβ2 is blocked by F(ab′)2 fragments of function-blocking Abs to either the αM or β2 subunits and by NIF, a high-affinity ligand that blocks many αMβ2-mediated PMN responses (43, 51). Failure of Ang(1– 4) to prolong PMN survival is consistent with the data by Chavakis et al. (32), who showed that Ang(1– 4) has an antiinflammatory effect on leukocytes by inhibiting their migration and adhesion to αMβ2 ligands. These authors also demonstrated that Ang(1– 4) generation in vivo peaks at the last phase of wound healing when PMN apoptosis is needed to terminate the inflammatory response. Thus, it can be envisioned that Ang(1– 4) would bind to αMβ2 and compete with Plg and its other prosurvival ligands. Chavakis et al. (32) identified kringle 4 domain as directly interacting with αMβ2. However, in our experiments, not only Ang(1– 4) but also Ang(1–3) (data not shown), which lacks kringle 4, is recognized by this integrin. Indeed, Ang(1–3) recognition by αMβ2 is physiologically relevant, as biologically active Ang(1–3) is generated from Plg by neutrophil-derived elastase (57). Thus, PMNs have an intrinsic ability to induce their own apoptosis by generating angiostatin derivatives that compete with Plg and by degrading Plg, a prosurvival ligand. αMβ2 is not the only integrin interacting with Ang(1– 4); it is also recognized by α4β1, α9β1, and αVβ3, integrins that have also been implicated in cellular apoptosis.
We hypothesized that one of the mechanisms underlying distinct PMN responses to Plg and Ang(1– 4) may be differences in their mode of αMβ2 engagement. Utilizing purified recombinant integrin and K562 cells expressing αMβ2 or its individual subunits, we found that engagement of both integrin subunits is a prerequisite for induction of prosurvival signals. Ang(1– 4) interacted only with the αM subunit and failed to inhibit apoptosis. This mechanism was generalized by experiments with Fg and its two αMβ2-binding peptides, P1 and P2-C (44, 58). Fg and P2-C exerted cytoprotective effects on PMN and αMβ2 K562 cells, while the P1 peptide failed to inhibit apoptosis in these cells. Fg and P2-C are recognized by the entire heterodimer, whereas P1 is recognized only by the αM subunit (35). Whitlock et al. proposed a model in which αMβ2 clustering inhibits spontaneous and TNF-α-induced apoptosis via activation of Akt and ERK1/2 pathways (15). In view of this model, we assessed αMβ2 clustering in the absence or presence of its ligands and found that cytoprotective ligands interacting with both integrin subunits, such as Plg, Fg, or P2-C, induced αMβ2 clustering while Ang(1– 4) and P1 did not. Furthermore, clustering was pivotal to the effects of the prosurvival ligands of αMβ2, as several clustering inhibitors blocked the prosurvival effects. According to Kim et al. (47), integrins do not cluster spontaneously, but only upon engagement of immobilized ligands. However, in our system ligands are presented in solution and they are still capable of αMβ2 clustering, which is consistent the data by Buensuceso et al. showing that soluble Fg induces αIIbβ3 clustering on platelets (59). Thus, engagement of both αMβ2 integrin subunits with cytoprotective ligands is prerequisite to the receptor clustering and subsequent transmission of prosurvival signals. This may be the case, as integrin interactions with cytoskeleton, which occur primarily via the β subunits, are critical for integrin clustering (60). Indeed, based on our data, both the cytoskeleton and cytoplasmic tails of both subunits of αMβ2 are pivotal in transmission of prosurvival signals. However, we cannot exclude the possibility that distinct patterns of engagement of the integrin subunits by the tested ligands may differentially regulate integrin activation since binding of an activating mAb to αMβ2 has been shown to prevent spontaneous apoptosis (15). The binding of all αMβ2 ligands tested (except NIF) is significantly enhanced by integrin activation. However, integrin activation by itself does not inhibit cell apoptosis in the absence of the ligands, as concluded from our experiments with cells expressing constitutively active αMβ2. Receptor engagement by the appropriate ligands is crucial for induction of the cytoprotective response.
The effect of Plg on PMN apoptosis did not require its conversion to active Plm, as diisopropyl fluorophosphate (DFP)-treated Plg recapitulated the prosurvival effect of untreated Plg. Additionally, we found that active Plm did not exert a cytoprotective function (data not shown). These observations contrast with the recent demonstration that Plm formation and its activation of PAR1 are necessary for induction of monocyte survival by Plg (29) and suggest cell type-specific mechanisms for Plg to influence apoptosis. Plg-dependent PMN survival was reduced by ε-aminocaproic acid, which blocks interaction of the lysine binding sites within the Plg kringles with cell receptors (30, 61).
Activation of Akt and ERK1/2 by engagement of αMβ2 with prosurvival ligands Plg, Fg, and P2-C is consistent with Whitlock et al. (15) and Rubel et al. demonstrating that these kinases are activated downstream of αMβ2 and other integrins (62) and are crucial for PMN survival. ERK1/2 inhibition has been shown to enhance resolution of inflammation by prolonging PMN lifespan in vivo (63). In our study, Plg-mediated PMN survival was completely reversed by Akt inhibition and was only partially reduced by ERK1/2 inhibition, suggesting that Akt is upstream of ERK1/2. This interpretation is supported by our finding that inhibition of Akt reduced Plg-dependent activation of ERK1/2 (data not shown), and by the fact that Akt directly phosphorylates MEK1 and MEK2, which are upstream activators of ERK1/2 (64). The pivotal role of Akt in PMN survival via ERK1/2 activation is further supported by our observation that Plg-dependent ERK1/2 phosphorylation occurs in the absence of Akt activation in αM cells but is not sufficient to rescue these cells from apoptosis. Complete blockade of PMN survival by Akt and only partial blockade by ERK1/2 inhibitors also indicate that there are other ERK1/2-independent downstream targets of Akt. Possible targets of Akt phosphorylation that could inhibit apoptosis include the proapoptotic protein Bad, activation of transcription factors CREB and NF-κB (46), or prevention of cytochrome c release (65).
PMN apoptosis is an important mechanism regulating the extent of an inflammatory response, making it self-limiting and preventing excessive tissue damage. At sites of inflammation, a complex interplay between proapoptotic and survival signals must control the outcome of the response. This study describes a novel mechanistic model by which various αMβ2 ligands may regulate PMN apoptosis and suggests that their subtle balance in the local microenvironment provides an important extrinsic checkpoint in the resolution of inflammation.
1Supported by National Institutes of Health Grant R01 HL17964 and P50 HL 081011 (E.F.P.) and by an American Heart Association Scientist Development Grant (E.P.).
3Abbreviations used in this paper: PMN, polymorphonuclear leukocyte; Ang(1– 4), angiostatin composed of four kringle domains; BDM, 2,3-butanedione 2-monoxime; Fg, fibrinogen; H7, 1-(5-isoquinolinylsulfonyl)-2-methylpiperazine dihydrochloride; MFI, mean fluorescence intensity; MLCK, myosin L chain kinase; NIF, neutrophil inhibitory factor; MβCD, methyl-β-cyclodextrin; PI, propidium iodide; Plg, plasminogen; Plm, plasmin; PVP, polyvinylpyrrolidone; RFU, relative fluorescence units.
The authors have no financial conflicts of interest.