Vasodilator-Stimulated Phosphoprotein Regulates Inside-Out Signaling of β₂ Integrins in Neutrophils Free

Modulation of the affinity and avidity of β₂ integrins for its endothelial ligands is a key step in chemokine- and chemoattractant-induced arrest and leukocyte trafficking (1, 2). Chemoattractants and chemokines signal through Gi protein-coupled receptors and, thereby, control rapid activation of leukocyte β₂ integrins (inside-out signaling). However, the intracellular signaling mechanisms involved in inside-out signaling of β₂ integrins remain largely elusive.

The monomeric GTPase Rap1 is a key regulator of integrin adhesiveness (3). Bacterial products, such as LPS or fMLP, activate Rap1 in macrophages (3) and polymorphonuclear leukocytes or neutrophils (PMNs) (4), respectively, and, thereby, control β₂ integrin-dependent functions of these phagocytic cells.

A hallmark of Rap1, and other monomeric GTPases, is its ability to cycle between an inactive GDP- to an active GTP-bound state. The GDP–GTP switch is catalyzed by guanine nucleotide exchange factors (GEFs) (5). C3G, CalDAG-GEFI-III, and Epacs are well-characterized GEFs for Rap1 (5). In the GTP-bound state, Rap1 interacts with effectors to control rapid activation of integrins (6, 7).

The essential role of Rap1 in the regulation of integrin avidity and affinity to endothelial ligands is supported by the discovery of a human genetic deficiency of leukocyte adhesion to endothelium: leukocyte adhesion deficiency (LAD)-III (8). LAD-III syndrome is due to a lack of expression of CalDAG-GEFI. Hence, in LAD-III patients, PMNs do not adhere to venules and do not migrate to the site of inflammation (8); this explains why these patients have recurrent bacterial infections. However, not all LAD-III patients have impaired expression of CalDAG-GEFI. This is exemplified by a recent finding that an LAD-III syndrome can also be due to loss of expression of kindlin 3, a protein that is essential for conformational activation of integrins (9).

In human PMNs, several signaling pathways are activated in response to fMLP, a potent chemoattractant, including PI3K, Src tyrosine kinases, protein kinase C family members, phospholipase Cβ, MAPKs, and Ca²⁺ release (10). fMLP also induces a moderate increase in NO and guanosine-3′,5′-cyclic monophosphate (cGMP) in human PMNs (11, 12), but concentrations of cGMP must be elevated in specific cellular compartments (11), thus explaining the ability of this second messenger to enhance PMN functions, including degranulation and chemotaxis (12). In PMNs, NO is produced by NO synthases (i.e., endothelial NO synthase and inducible NO synthase [iNOS]) in response to fMLP (13). This gas binds to the reactive heme group of soluble guanylyl cyclases and, thereby, enhances the production of cGMP. In PMNs, one of the downstream effectors of cGMP is the cGMP-dependent protein kinase (cGKI) (14). Thus, exogenous NO regulates PMN functions, such as chemotaxis and exocytosis of granules through a mechanism dependent on cGKI (12, 15).

Vasodilator-stimulated phosphoprotein (VASP) belongs to a family of proteins that includes the Drosophila protein enabled (Ena), its mammalian ortholog Mena, and the Ena-VASP like protein Evl (16, 17). VASP, Mena, and evl are made up of a central proline-rich domain and N- and C-terminal domains named Ena-VASP homology domains 1 and 2 (EVH1 and EVH2). Ena-VASP homology domain 1 interacts with a proline-rich sequence (FPPPP motif), which is found in zyxin and vinculin. The central core of Ena/VASP contains proline-rich stretches that bind to profilin, as well as proteins containing SH3 or WW domains. EVH2 domain has a coiled-coil domain that binds to filamentous and globular actin (16). Ena/VASP controls actin polymerization by directly binding to monomeric and F-actin and, thereby, promotes actin filament nucleation, bundling, and elongation. Hence, VASP regulates many biological functions that are dependent on actin, such as adhesion and migration (18). Human VASP has three sites that are phosphorylated by protein kinase A and cGKI: Ser¹⁵⁷, Ser²³⁹, and Thr²⁷⁸ (19). Ser²³⁹, which is contained in the EVH2 domain, is preferentially phosphorylated by cGKI (19). In addition, AMP-activated protein kinase phosphorylates VASP on Thr²⁷⁸ (20). Phosphorylation of Ser²³⁹ of VASP by cGKI inhibits actin anticapping and filament-bundling activities of VASP (21); however, this issue is controversial (17).

Because the NO/cGKI pathway regulates adhesion-dependent functions of PMNs, we wanted to know whether this pathway plays a role in the activation of Rap1 and β₂ integrins in leukocytes. Furthermore, we analyzed whether VASP, a substrate of cGKI, was a regulator of C3G, a GEF for Rap1.

The anti-Rap1 pAb polyclonal Ab (pAb) (sc-65), mouse monoclonal anti-C3G Ab (sc-17840), VASP small interfering RNA (siRNA) (sc-29516), and scrambled siRNA (sc-37007) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Benzamidine; PMSF; HBSS without calcium chloride, magnesium chloride, or phenol red (HBSS; H 6648); 10× HBSS without calcium chloride, magnesium chloride, or NaHCO₃ (HBSS 10×; H 4641), and fatty acid-free BSA (A8806) were obtained from Sigma-Aldrich (Dorset, U.K.). Dextran, Ficoll-Hypaque, and Percoll were from GE Healthcare Biosciences (Uppsala, Sweden). All electrophoresis reagents were from Bio-Rad (Hercules, CA). Sterile water for injection and 0.9% normal saline (solution for infusion) were from Hameln Pharmaceuticals (Gloucester, U.K.) and Baxter Healthcare (Thetford Norfolk, U.K.), respectively. All other chemicals were of analytical grade and came from Sigma-Aldrich.

Venous blood was collected from healthy donors, after obtaining informed consent. This study was approved by the Office for Research Ethics Committees Northern Ireland (Ref 07/NIR03/86).

Blood was collected from healthy donors (50 ml) and immediately poured into sterile Falcon tubes (Sarstedt, Numbrecht-Rommelsdorf, Germany) containing 4% sodium citrate (prepared with water for injection) to prevent coagulation. The blood was transferred to a 75-ml sterile tissue-culture flask (kept vertically) and diluted with 100 ml ice-cold 0.9% normal saline (solution for infusion) and 25 ml 10% dextran (prepared in 0.9% normal saline). The flask was kept at 4°C until the RBCs had sedimented (∼30 min). The supernatant was collected and transferred to fresh sterile Falcon tubes, which were spun at 190 × g for 10 min in a refrigerated centrifuge. The pellet was collected, and erythrocytes were lysed for 30 s with 6 ml sterile water (water for injection). The lysis was stopped by adding 2 ml PBS containing 9% NaCl and 2 ml Ringer’s modified phosphate buffer (120 mM NaCl, 4.9 mM KCl, 1.7 mM KH₂PO₄, 1.2 mM MgSO₄ 7 H₂O, 8.3 mM Na₂HPO₄ 2H₂O, 10 mM glucose [pH 7.3]). The cell suspension was loaded on the top of Ficoll-Hypaque (15 ml) and subjected to centrifugation (390 × g for 30 min). Thereafter, the pellet was recovered and washed twice with 50 ml Ringer’s modified phosphate buffer. Finally, the cells were resuspended in a calcium-containing medium (136 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 5.0 mM NaHCO₃, 1.1 mM CaCl₂, 5.5 mM glucose, 20 mM HEPES [pH 7.4]). The cell suspension consisted of ∼97% PMNs (22).

VASP-deficient mice were generated by homologous recombination as described (23). The VASP knockout (KO) mice were back-crossed for nine generations to C57BL/6 and kept in a professional breeding facility (Charles River, Kisslegg, Germany) under standard conditions. The control mice emerged from the back-crosses of heterozygous from the same parents as the VASP-KO mice. Therefore, they can be regarded as genetically identical, except for the VASP locus. Mice were genotyped to confirm deletion of the VASP gene: genotyping was carried out at the breeding facility (Charles River) and at the animal facility of Queen’s University Belfast prior to experimentation. Animal work was approved by the Queen’s University Belfast ethical review committee and performed under a U.K. Home Office license (PPL2531b).

Mice were sacrificed by CO₂ intoxication, and the femur and tibia from both hind legs were removed and freed of soft tissue attachments. The extreme distal tip of each extremity was cut off. A Kendall monoject insulin syringe containing HBSS-modified medium (1× HBSS without calcium chloride, magnesium chloride, or phenol red supplemented with 2.5 g/l BSA and 15 mM HEPES [pH 7.4]) was used to flush out the bone marrow from the bones. After dispersing cell clumps, the cell suspension was centrifuged (400 × g for 10 min) and resuspended in 3 ml HBSS-modified medium. The cell suspension was then loaded on the top of a two-layer Percoll gradient (62%, 55%) and subjected to centrifugation (1500 × g for 30 min). Cells at the interface between 62% and 55% were collected with a sterile Pasteur pipette and washed with 50 ml HBSS-modified medium. Cells were then spun down (400 × g for 10 min). The RBCs in the resulting pellets were lysed by adding 5 ml hypotonic Gey’s balanced salt solution. After 5 min, the reaction was stopped by adding 50 ml HBSS-modified medium. PMNs were collected and resuspended in calcium-containing medium. A total of 3–5 × 10⁶ PMNs were isolated per mouse (70% purity) (24).

Easy Grip petri dishes (Sarstedt) were incubated overnight at 4°C with 20 μg/ml fibrinogen or 1% BSA in PBS. The wells were then blocked for 30 min with 1% BSA in PBS and then washed twice with PBS and once with calcium-containing medium. PMNs (0.5 × 10⁶) were allowed to adhere to fibrinogen- or BSA-coated dishes at 37°C in the presence of stimuli. After 20 min, nonattached cells were removed by aspiration, and plates were washed three times with PBS. Adherent cells were fixed for 30 min with 3.4% paraformaldehyde. Thereafter, cells were washed with PBS and stained with crystal violet (15 min, 0.1% crystal violet in 10% methanol). The plates were extensively washed with PBS. The stain was eluted with 0.1% SDS. OD of the elution was measured at 570 nm.

Undifferentiated PLB-985 cells were grown at 37°C in an atmosphere of 5% CO₂ in RPMI 1600 medium (supplemented with 10% FCS) and l-glutamine (20 mM) to a density of 5 × 10⁵ cells/ml. Differentiation into PMN-like cells was carried out by culturing undifferentiated PLB-985 cells for 5 d in antibiotic-free RPMI 1600 medium supplemented with 5% FCS, l-glutamine (20 mM), and 1.25% DMSO. Differentiated or undifferentiated PLB-985 cells were resuspended in calcium-containing medium (10 × 10⁶ cells/ml).

PMNs were lysed in a buffer composed of 50 mM Tris-HCl (pH 7.5), 1% Triton X-100, 100 mM NaCl, 10 mM MgCl₂, 5% glycerol, 1 mM Na₃VO₄, and protease inhibitors (1 mg/ml leupeptin, 2.5 mM benzamidine, 1 mM PMSF). The GST-RalGDS-RBD fusion protein (4) was coupled to glutathione-Sepharose beads for 1 h and then the beads were added to the clarified PMN lysates. After 1 h, the beads were collected by centrifugation and washed three times with lysis buffer. The beads were resuspended in Laemmli sample buffer and boiled under reducing conditions. The precipitated proteins were subjected to 12% SDS-PAGE and transferred to polyscreen polyvinylidene difluoride (PVDF) membranes. The membranes were blocked in PBS supplemented with 0.2% Tween 20 and 5% milk and then incubated for 1 h overnight in a cold room with an anti-Rap1 pAb (1/2000 dilution of anti-Rap1). Thereafter, the membranes were washed in PBS supplemented with 0.2% Tween 20 and subsequently incubated for 1 h with peroxidase-conjugated anti-rabbit IgG (1:10,000) in blocking buffer. Ab binding was visualized by ECL. The ECL solution was prepared by mixing 6 ml luminol solution (1.25 mM 5-amino-2,3-dihydro-1,4-ptalazinedione in 0.1 M Tris [pH 8.5]) with 30 μl 68 mM p-coumaric acid in DMSO and 2 μl 30% H₂O₂ solution. Stocks of luminol solution and p-coumaric acid were kept at 4°C in a light-protected bottle or in Eppendorf tubes at −20°C, respectively.

The protein content was estimated according to Schaffner and Weissmann (25).

PLB-985 cells were collected and resuspended in ice-cold disruption buffer made of 100 mM Tris-HCl (pH 7.4), 5 mM EDTA, 5 mM EGTA, 50 mM NaCl, 5 mM NaF, 1 mM Na₃VO₄, 1 mg/ml leupeptin, 2.5 mM benzamidine, and 1mM PMSF and then disrupted by sonication. The suspension was collected and the nuclei, heavy membrane fraction, and undisrupted cells were sedimented by centrifugation at 10,000 × g for 10 min. The resulting supernatant was centrifuged at 100,000 × g for 1 h at 4°C. The obtained pellet, which contains a marker for plasma membrane (HLA-1) (26), was resuspended in the disruption buffer. Aliquots of the suspension were mixed with Laemmli buffer containing 1 mM DTT and boiled. The proteins were separated on 12% SDS-PAGE, and immunoblot analysis was performed using anti-C3G Abs (1 μg/ml), followed by peroxidase-conjugated anti-mouse IgG (1:10,000) and the detection of Ab binding by ECL.

PLB-985 cells were differentiated with 1.25% DMSO for 4 d in antibiotic-free medium. The cells were then nucleofected using the Human B Cell Nucleofector kit (cat no.VPA-1001; Lonza, Wokingham, U.K.). A total of 10 × 10⁶ cells were added to 100 μl the nucleofection solution (18.2 μl the supplement mixed with 81.8 μl nucleofection buffer). The cell suspension (100 μl) was combined with 8 μl (0.8 μM) scrambled siRNA or a pool of three siRNAs targeting VASP (sense 1, CCUCUACUUGACUUGGAAUTT; sense 2, GAAGGAGGGAAUUUCACAUTT; and sense 3, CACCUUUAGCUUCUUGAAATT). For rescue experiments, siRNAs targeting VASP along with 4 μg wild type VASP or VASPSer235Ala (cloned in plasmid murine stem cell virus) were added to 100 μl the nucleofection solution. The cell/DNA suspension was transferred into cuvettes, which were inserted into the cuvette holder of the Nucleofector. Nucleofection was performed with the U-15 program (Lonza). The reaction was stopped by adding 500 μl prewarmed RPMI 1600 medium. The cells were transferred to a sterile flask containing 10 ml antibiotic-free medium. After 24 h of incubation at 37°C, cells were collected, spun down (190 × g for 5 min), resuspended in 10 ml fresh medium, and left overnight.

It is well established that fMLP activates Rap1 in human PMNs independently of PI3K, Ca²⁺ signaling, protein kinase C, or phospholipase Cβ (4). Because human PMNs produce cGMP in response to fMLP (11–13), we hypothesized that the NO/cGKI pathway may be involved in the regulation of Rap1 in response to the chemoattractant. To investigate this hypothesis, human PMNs were pretreated with pharmacological agents that block the different enzymes of the NO/cGKI pathway, after which the amounts of GTP-bound Rap1 were determined using the GST-RalGDS pull-down assay (27). The components tested were N-(3-(aminomethyl)benzyl)acetamidine (1400W) and 8-(4-chlorophenylthio)guanosine 3′,5′-cyclic monophosphothioate, Rp-isomer triethylammonium salt (Rp-8pCPT-cGMPS), which are well-characterized inhibitors of iNOS (28) and cGKI (29), respectively. The results indicated that stimulation of human PMNs with fMLP (1 μM for 1 min) increased the amounts of GTP-bound Rap1; this was prevented when PMNs were pretreated with 1400W (10 μM for 20 min) or Rp-8pCPT-cGMPS (1 μM for 20 min) (Fig. 1A). Because activation of β₂ integrins in phagocytic cells depends on loading GTP on Rap1 (3), we measured the adhesion of PMNs onto fibrinogen (a ligand for β₂ integrins) as a read-out for the functional activation of β₂ integrins. Indeed, static condition assays reflect LFA-1 affinity changes in lymphocytes (30). We found that PMNs adhered to fibrinogen-coated, but not BSA-coated, plates in response to fMLP (2.3-fold increase over controls) (Fig. 1B), indicating that PMN adhesion was dependent on β₂ integrins. Furthermore, we found that adhesion of PMNs, induced by fMLP, was partly, but significantly, abolished if the leukocytes had been pretreated with 1400W (10 μM for 20 min) or Rp-8pCPT-cGMPS (1 μM for 20 min) (Fig. 1B). To further confirm that fMLP-induced activation of Rap1 was dependent on cGKI, we performed experiments with DT-3, a highly specific membrane-permeable peptide blocker of cGKI (31). We found that pretreatment of human PMNs, isolated from two blood donors, with DT-3 (1 μM for 45 min) totally blocked fMLP-induced activation of Rap1 (Fig. 1C). In parallel, DT-3 blocked, in a dose-dependent manner, the adhesion of PMNs to a fibrinogen-coated surface in response to fMLP (Fig. 1D). Based on the use of 1400W, Rp-8pCPT-cGMPS, and DT-3, our results strongly indicated that fMLP induces a cGKI-dependent activation of Rap1 and β₂ integrins.

Next, we examined whether fMLP also activated Rap1 through the NO/cGKI pathway in the myeloid cell line PLB-985. To this end, nondifferentiated PLB-985 myelomonoblasts were differentiated into mature PMN-like cells by culturing them for 5 d in DMEM medium containing 1.25% DMSO. We ensured that immature PLB-985 cells had differentiated into mature PMN-like cells by measuring the amount of p47^phox (a component of the NADPH oxidase) in whole-lysate extracts. Indeed, p47^phox is a good marker of granulocytic differentiation because its expression correlates with functional assembly of the NADPH oxidase (32). As shown by Western blot analysis, nondifferentiated immature PLB-985 cells expressed low amounts of p47^phox, whereas high levels of p47^phox were detected in differentiated PLB-985 cells (Fig. 2A). We found that pretreatment of differentiated PLB-985 cells with 1400W (10 μM for 20 min); 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ; 1 μM for 20 min), an inhibitor of NO-sensitive guanylyl cyclases (33); or Rp-8pCPT-cGMPS (1 μM for 20 min) partly or totally blocked fMLP-induced activation of Rap1 (Fig. 2B). In parallel, these inhibitors totally blocked fMLP-induced adhesion of these cells to fibrinogen-coated plates (Fig. 2C). Similarly, the DT-3 peptide blocked fMLP-induced activation of Rap1 (Fig. 2D), as well as adhesion to fibrinogen (Fig. 2E). Thus, differentiated PLB-985 cells represent an excellent alternative model to human PMNs in which to study the regulation of Rap1 by the NO/cGKI pathway.

Because fMLP triggers the production of cGMP in PMNs (11–13), we next tested whether the cGMP analog 8-(4-chlorophenylthio)guanosine-3′,5′-cyclic monophosphate (8-pCPT-cGMP) could trigger adhesion and activation of Rap1 in human PMNs and differentiated PLB-985 cells. We found that 8-pCPT-cGMP induced the adhesion of human PMNs (Fig. 3A, left panel) and differentiated PLB-985 cells (Fig. 3B, left panel) to fibrinogen-coated plates in a dose-dependent manner. Maximal effects were observed with concentrations of 1–10 μM for human PMNs (1.4-fold increase over controls) and 10–50 μM for differentiated PLB-985 cells (2.8-fold increase over controls). In parallel, we observed that stimulation of human PMNs (Fig. 3A, right panel) or differentiated PLB-985 cells (Fig. 3B, right panel) with 8-pCPT-cGMP (10 μM) (a dose that induced maximal adhesion of these cells to fibrinogen) led to activation of Rap1. Altogether, these results demonstrated that cGMP reproduces the effects of fMLP in terms of activation of Rap1 and β₂ integrins.

Next, we sought to identify the downstream target of cGKI implicated in fMLP-induced activation of Rap1. Because VASP is a substrate of cGKI (19), we investigated whether VASP plays a role in this process. We knocked down the expression of endogenous VASP in differentiated PLB-985 cells by nucleofection using a pool of three target-specific 20–25-nt VASP siRNAs, after which the activation of Rap1 in response to fMLP or a cGMP analog was determined. Differentiated PLB-985 cells were used for this purpose because human PMNs are not amenable for transfection. It is relevant to use differentiated PLB-985 cells because they exhibit similar responses as human PMNs to a cGMP analog or fMLP, in terms of activation of Rap1 and β₂ integrins (Figs. 1, 2). We used Western blot analysis to verify that expression of VASP was decreased consecutive to nucleofection with siRNAs targeting VASP. In all experiments, VASP protein levels were lower in cells nucleofected with VASP siRNAs than in cells nucleofected with scrambled siRNA. However, the extent to which it was decreased varied between experiments (compare levels of VASP in Fig. 4A, left panel, and Fig. 4B, left panel). We found that fMLP (1 μM) or 8-pCPT-cGMP (10 μM) activated Rap1 in cells nucleofected with scrambled siRNA (control cells) (Fig. 4A, 4B, left panels). In contrast, knocking down the expression of VASP in PLB-985 cells resulted in constitutive activation of Rap1. Thus, resting cells in which VASP had been knocked down had similar levels of GTP-bound Rap1 as did cells nucleofected with a scrambled siRNA and stimulated with 8-pCPT-cGMP. The more we knocked down the expression of VASP in PLB-985 cells, the more GTP-bound Rap1 was found in resting cells (Fig. 4A, 4B). Surprisingly, we observed that fMLP reduced the level of GTP-bound Rap1 in cells in which VASP had been knocked down (Fig. 4A, left panel). This result may be explained by the fact that, in PMNs, Rap1 oscillates between a GDP- and GTP-bound form in response to fMLP (M. Koney-Dash and K. Dib, personal communication), as described in thrombin-stimulated platelets (34). Therefore, if basal activity of Rap1 is already high as a result of the knock-down of VASP, stimulation with fMLP would reduce, rather than augment, the level of GTP-bound Rap1 to allow rapid cycling of Rap1 between its GTP- and GDP-bound forms.

To assess whether VASP also played a role in the regulation of Rap2, a close relative of Rap1, the blots were stripped of Rap1 Abs and analyzed by Western blot analysis using an anti-Rap2 Ab. The results show a similar pattern of regulation of Rap2 in response to fMLP or 8-pCPT-cGMP (Fig. 4A, right panel, and Fig. 4B, right panel). Also, we found that 8-pCPT-cGMP (10 μM) induced adhesion to fibrinogen of differentiated PLB-985 cells nucleofected with a scrambled siRNA (1.5-fold increase over controls). No such effect was observed with PLB-985 cells nucleofected with VASP siRNAs, which is consistent with the concept that Rap1 is constitutively active in these cells (Fig. 4A, 4B).

We next conducted experiments using bone marrow-derived PMNs from wild type (^+/+) or VASP KO mice (^−/−) with a similar genetic background (C57 BL/6). To obtain sufficient cells to carry out Rap1 pull-down and adhesion assays, PMNs were isolated from five mice and pooled. PMNs from VASP^+/+ or VASP^−/− mice (8 × 10⁶) were stimulated or not with fMLP (10 μM for 30 s), after which cells were lysed and GST-RalGDS pull-down assays were carried out. We used a high dose of fMLP (10 μM), a common procedure to study the regulation of mice PMN functions (24), because mice express low-affinity binding receptors for fMLP (35). We could not detect any basal level of GTP-bound Rap1 in unstimulated PMNs from VASP^+/+ mice, even though the blot had been overexposed (Fig. 5A, left panel). In addition, stimulation of VASP^+/+ PMNs with fMLP (10 μM for 30 s) led to activation of Rap1 (Fig. 5A, left panel). In contrast, we detected basal Rap1 activity in PMNs from VASP^−/− mice and the level of GTP-bound Rap1 was reduced in VASP^−/− PMNs upon stimulation with fMLP (10 μM, 30s) (Fig. 5A, right panel), similarly to what we observed in PLB-985 cells in which VASP had been knocked down (Fig. 4A). We also found that VASP^+/+ PMN adhered to fibrinogen-coated plates in response to fMLP (1.4-fold increase over controls), whereas the chemoattractant did not induce such an effect in VASP^−/− PMN (Fig. 5B).

To investigate whether VASP is phosphorylated by cGKI in response to fMLP, human PMNs or differentiated PLB-985 cells were stimulated with fMLP (1 μM, 1 min) after which cells were lysed. Levels of phospho-VASP were determined by Western blot analysis using an anti-phosphoSer²³⁹ VASP Ab, which recognizes human VASP phosphorylated on Ser²³⁹ (36). We found that fMLP induced phosphorylation of VASP on Ser²³⁹ (the main phosphorylation site of cGKI) in human PMNs (Fig. 6A, left panel) and differentiated PLB-985 cells (Fig. 6A, right panel). Next, we investigated whether phosphorylation of VASP by cGKI was required for fMLP-induced activation of Rap1. To this end, differentiated PLB-985 cells were nucleofected with cDNAs cloned in the plasmid murine stem cell virus encoding for murine GFP-tagged wild type VASP or VASPSer235Ala mutant (the position of Ser²³⁹ in human VASP corresponds to position 235 in mouse) (37), together with siRNAs targeting endogenous VASP. Nucleofection efficiency was assessed by flow cytometry. Upon nucleofection, we observed two populations of cells that corresponded to dead (population 1) and living cells (population 2). The high level of dead cells was due to the nucleofection procedure. After selecting gating for population 2 on the forward angle and 90° light scatter plot, graphs indicating mean fluorescence were generated. We found that both cDNAs were equally expressed in the cells (∼50% nucleofection efficiency) (Fig. 6B, left panel). The results showed that fMLP activated Rap1 in cells in which wild type VASP, but not VASP Ser235Ala, had been overexpressed (Fig. 6B, right panel). In parallel, cells expressing wild type VASP adhered to fibrinogen in response to fMLP (1.6-fold increase over unstimulated control cells), whereas in contrast, no such effect was observed in cells expressing VASPSer235Ala mutant (Fig. 6C).

Recruitment of GEFs to the plasma membrane is associated with their activation (5). Therefore, we investigated whether cGKI and VASP were involved in the recruitment of C3G to the plasma membrane. To this end, differentiated PLB-985 cells, pretreated or not with DT-3, were stimulated with fMLP (1 μM for 1 min), after which cells were collected; crude membrane fractions (100,000 × g pellet), which contain HLA-1, a marker of PMN plasma membrane (26), were prepared. Proteins (5–10 μg) were subjected to 12% SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was then subjected to immunoblotting with anti-C3G Abs. We confirmed that gels were equally loaded with proteins by staining the nitrocellulose membrane with Ponceau red prior to immunoblotting. The results showed that fMLP induced recruitment of C3G to the plasma membrane-enriched fraction, and this effect was blocked by DT-3 (Fig. 7A). We also found that the cGMP analog induced redistribution of C3G to the membrane in PLB-985 cells nucleofected with scrambled siRNA (Fig. 7B). In contrast, silencing VASP in these cells led to stimuli-independent membrane recruitment of C3G to the membrane (Fig. 7B). We also found that expression of VASPSer235Ala mutant in PLB-985 cells abrogated fMLP-induced translocation of C3G to the plasma membrane-enriched fraction (Fig. 7C). Thus, phosphorylation of VASP by cGKI triggers membrane recruitment of C3G.

In the current study, we aimed at clarifying the mechanisms by which fMLP regulates inside-out signaling of β₂ integrins. In particular, we investigated the intracellular signaling mechanisms implicated in fMLP-induced activation of Rap1, a monomeric GTPase that controls activation of β₂ integrins in phagocytic cells (3).

In human PMNs, it is well established that fMLP activates Rap1 and promotes adhesion; however, the nature of the signals controlling GTP loading of Rap1 and activation of integrins remains unknown (4). For instance, although the Src tyrosine kinases Hck and Fgr are rapidly activated in response to fMLP, chemokine-induced activation of β₂ integrins is unaffected in PMNs isolated from Hck^−/− Fgr^−/− mice (38). In addition, calcium and DAG, the two secondary messengers that bind to and activate CalDAG-GEFI, a GEF for Rap1, are not involved in fMLP-induced activation of Rap1 in human PMNs (4). In the current study, we made the novel observation that the NO/cGKI pathway is a key regulator of inside-out signaling of β₂ integrins. This was demonstrated by showing that blocking the NO/cGKI pathway with inhibitors prevented fMLP-induced activation of Rap1 and adhesion, whereas a cGMP analog reproduced the effects of fMLP in terms of activation of Rap1 and β₂ integrins. These data may explain previous findings showing that the NO/cGKI pathway enhances chemotaxis and degranulation (12, 15), which are PMN functions dependent on β₂ integrins (39). In keeping with our work, it was shown in nonleukocytic cells that NO and cGMP are major regulators of osteoclast attachment and mobility (40). In contrast, other investigators found that NO donors prevented PMN adhesion to endothelium (41). This latter result may be explained by the fact that NO donors generate supraphysiological levels of NO and cGMP, leading to augmented levels of cAMP (42), a secondary messenger that blocks functional activation of PMN β₂ integrins (43).

To understand the mechanisms by which the chemoattractant fMLP activates β₂ integrins in PMNs, we next tried to identify the downstream target of cGKI implicated in the regulation of Rap1. cGKI regulates the function of several intracellular molecules and pathways, including VASP. Interestingly, platelets from VASP KO mice exhibit augmented adhesion to endothelial cells and blood vessels in vivo (44), as well as aggregation in vitro (45). Augmented adhesion and aggregation of platelets from VASP KO mice is associated with enhanced P-selectin and GPIIb-IIIa activation (44). Collectively, these results demonstrated that VASP negatively regulates platelet integrins. Furthermore, it was found recently that VASP KO mice, in comparison with wild type mice, have augmented levels of PMNs in the lung upon inhalation of LPS (46), suggesting that PMNs from VASP KO mice exhibit augmented adhesive and migratory capacity. However, it is still not known how the loss of VASP is linked to modulation of inside-out signaling of integrins, PMN recruitment to tissues, and platelet aggregation. We envisioned the possibility that VASP may be a regulator of Rap1, because this monomeric GTPase regulates the affinity of integrins for ligands. We proved this hypothesis by showing that knocking down VASP in resting PLB-985 cells, or in mice PMNs, led to constitutive stimuli-independent activation of Rap1 and β₂ integrins. This result implies that in resting PLB-985 cells, VASP prevents the loading of GTP on Rap1, probably because the majority of VASP is in a nonphosphorylated form. Thus, knocking down VASP removes the inhibitory effect exerted by VASP on GTP loading of Rap1. Signaling through fMLPRs must affect VASP in such a way that it would trigger activation of Rap1 and β₂ integrins. We showed that VASP becomes phosphorylated on Ser²³⁹ (the major phosphorylation site of cGKI) in human PMNs and differentiated PLB-985 cells in response to fMLP. Therefore, the crucial question we had to answer was whether phosphorylation of VASP by cGKI was required for activation of Rap1. We unequivocally gained support for this theory by showing that overexpression of the phosphodeficient VASP mutant VASPSer235Ala in PLB-985 cells completely prevented fMLP-induced activation of Rap1 and β₂ integrins. Our findings also shed light on the mechanisms by which platelet integrins are activated in response to stimuli. Indeed, because knocking down VASP in mice augments platelet integrin affinity for fibrinogen and, thereby, aggregation (44, 45), similarly to what we observed for β₂ integrins in PMNs, our results predict that phosphorylation of VASP by cGKI may be crucial for activation of platelet integrins. Thus, our findings support the model of Li et al. (47), showing that cGKI has an initial stimulatory, rather than inhibitory, role in platelet integrin activation.

It was shown that PMNs isolated from CalDAG-GEFI KO mice do not respond to LTB4, C5a, or platelet-activating factor in terms of activation of Rap1 (48). Because these ligands signal through Gi-coupled receptors similarly to fMLP, it is thought that fMLP activates Rap1 through CalDAG-GEFI. However, surprisingly, the investigators (48) did not provide any data implicating CalDAG-GEFI in the activation of Rap1 by bacterial fMLP. Furthermore, it was recently proposed that CalDAG-GEFI is not the sole GEF for Rap1 implicated in activation of platelet integrins (49). In addition, PMNs express low levels of CAlDAG-GEFI (50), and fMLP activates Rap1 in human PMNs independently of calcium or DAG (4). Altogether, these experiments argue against the involvement of CalDAG-GEFI in fMLP-induced activation of Rap1. We conclusively demonstrated that C3G is the GEF involved in the VASP-dependent regulation of Rap1. First, we demonstrated that cGKI controls recruitment of C3G to the membrane, in response to fMLP. Second, knocking down VASP in PLB-985 cells led to stimuli-independent membrane recruitment of C3G, which correlated with constitutive activation of Rap1. Third, overexpression of VASPSer235Ala mutant in PLB-985 cells prevented fMLP-induced recruitment of C3G to the membrane. In keeping with our findings, other groups reported that fibroblasts deficient for C3G display impaired adhesion and migration (51). We do not know how phosphorylation of VASP by cGKI controls the recruitment of C3G to the membrane, and we are trying to elucidate this mechanism.

In summary, we found that signaling through the NO/cGKI pathway is a prerequisite for activation of Rap1 and β₂ integrins, in response to bacterial fMLP. Furthermore, we proved that phosphorylation of VASP on Ser²³⁹ by cGKI is required for activation of Rap1 and inside-out signaling of β₂ integrins (Fig. 8). Such covalent modification of VASP may be required to generate an actin-dependent recruitment of C3G to the membrane, leading to activation of Rap1 and β₂ integrin-dependent host defense. Thus, we predict that a defect in chemokine- or chemoattractant-induced phosphorylation of VASP on Ser²³⁹ or impaired activation of C3G might also be responsible for a LAD-III syndrome.

We thank J.L. Bos and Frank Gertler for providing us with GST-RalGDS and VASP cDNA constructs, respectively; the blood donors; and Eilish Armstrong and Hazel Johnston for collecting blood. We acknowledge Urban Gullberg for the gift of PLB-985 cells. Candice Poux and Cheryl MacFarlane are acknowledged for help with flow cytometry. We are thankful to Len Stephens, Phil Hawkins, Karen Anderson, and Sue Kulkarni for teaching the procedure for isolation of mice PMNs. We are grateful to Reinhard Fässler for helpful comments.

Disclosures The authors have no financial conflicts of interest.