INTRODUCTION

The regulation of integrin-dependent adhesion and de-adhesion is important in leukocyte cell-cell and cell-matrix interactions in immunity and inflammation. Serpentine receptors of the G_i-linked chemoattractant receptor subfamily have been implicated in leukocyte adhesion regulation and are thought to play essential roles in controlling leukocyte trafficking and homing in vivo. These receptors stimulate an amplified and branching cascade of second messengers triggered through either  or subunits of heterotrimeric GTP-binding proteins (1). The small GTP-binding protein rho has recently been identified as a critical element in the signaling cascade responsible for fast integrin-dependent leukocyte adhesion (2). Chemoattractants stimulate very rapid guanine nucleotide exchange on the small G-protein RhoA, and inhibition of rho by C3 transferase inhibits agonist-triggered integrin activation. The pathway linking chemoattractant receptors to rho activation is still unknown but seems to be independent of diacylglycerol (DAG)¹-dependent protein kinase C isozymes (PKC) (2).

To explore further the regulation of chemoattractant to integrin signaling, we have assessed the effect of cAMP, a potent inhibitor of several leukocyte proinflammatory activities such as NADPH oxidase activation, granule exocytosis in neutrophils, and leukocyte transendothelial migration (3-5). We report that intracellular cAMP, acting through protein kinase A (PKA), abrogates the proadhesive response of lymphoid cells and of neutrophils to chemoattractant but not to phorbol ester stimulation. This inhibitory effect is associated with blockade of chemoattractant-induced guanine nucleotide exchange on the small GTP-binding protein RhoA, suggesting that cAMP-dependent PKA acts as a negative modulator or "gate" on the chemoattractant to rho to integrin signaling pathway.

EXPERIMENTAL PROCEDURES

PBS, fMLP, PMA, Bt₂cAMP, theophylline, control rabbit antibody to mouse immunoglobulin, Triton X-100, deoxycholate, SDS, benzamidine, leupeptin, pepstatin, aprotinin, soybean trypsin inhibitor, phenylmethylsulfonyl fluoride, EGTA, EDTA, dithiothreitol, GTP, GDP, and human fibrinogen were purchased from Sigma; fetal calf serum (FCS), RPMI 1640, phosphate-free RPMI 1640, and dialyzed FCS were purchased from Irvine; [³²P]orthophosphate was from Amersham Corp.; rabbit polyclonal anti-RhoA, which recognizes the sequence KDLRNDEHTRRELA, was from Santa Cruz Biochemicals; Trisacryl-protein A beads were from Pierce; polyethyleneimine cellulose plates were from Fisher.

L1/2 cells (mouse pre-B lymphocytes) were stably transfected with human formyl peptide receptor (fPR) or with human IL-8 receptor A (RA) as described (6, 7). Vascular cell adhesion molecule-1 (VCAM-1) was purified from mouse spleens as described previously and used to coat 18-well glass slides by dilution below the critical micellar concentration and incubation overnight at 4 °C (6). Coated wells were blocked for 10 min with FCS. 8 × 10⁴/well (4 × 10⁶/ml in RPMI 1640, buffered with CO₂ to pH 7.4) transfectant cells were added to the coated wells, incubated for 10 min at 37 °C, and then stimulated by addition of the agonists for 1 min (100 nM fMLP or 100 ng/ml IL-8) and 20 min (100 ng/ml PMA). After rapid washing in cold PBS the cells were fixed in cold PBS, 1.5% glutaraldehyde for 1 h. The number of adherent cells in 0.2 mm² was counted using an inverted microscope with ×20 magnification and NIH-Image 1.56 as cell-counting software. Background binding in the absence of added agonist was determined for each condition, was minimal (less than 2% of stimulated adhesion), and was subtracted from agonist-stimulated adhesion for data presentation.

Human PMNs were isolated from venous blood from healthy adult volunteers as previously reported (5). The entire isolation procedure was conducted at 4 °C, using lipopolysaccharide-free medium. 18-well glass slides were coated for 60 min at 37 °C with 10 µg/well human fibrinogen in lipopolysaccharide-free water. 5 × 10⁴/well PMNs (2.5 × 10⁶/ml in RPMI 1640, 20 mM HEPES, pH 7.3) were added to the coated glass, incubated for 10 min at 37 °C, and then stimulated by addition of the agonists for 1 min (100 nM fMLP or 100 ng/ml IL-8) or 10 min (100 ng/ml PMA). After the treatment the PMN were washed and resuspended in RPMI 1640. Background (no agonist) adhesion was 51 ± 7 cells/0.2 mm² and was subtracted.

Inhibition of Up-regulation of Neutrophil ₂-Integrin Expression by 4,4-Diisothiocyanostilbene-2,2-disulfonic Acid (DIDS)

Human PMNs (2.5 × 10⁶/ml in RPMI 1640, 20 mM HEPES, pH 7.3) were pretreated for 20 min at 37 °C with 250 µM DIDS (Sigma). The cells were then used in adhesion assay as described above. Alternatively, ₂-integrin expression was evaluated by fluorescence-activated cell sorter in buffer (control) or DIDS-treated PMNs, stimulated at 37 °C with 100 nM fMLP for 1 min, and stained with the anti-CD18 mouse monoclonal antibody IB4, as described (8).

L1/2 cell transfectants were incubated overnight at 37 °C in phosphate-free RPMI 1640, 10% dialyzed FCS and labeled with 0.2 mCi/ml [³²P]orthophosphate for 4 h in the same medium. Cells were resuspended at 4 × 10⁷/ml in PBS, 1 mM CaCl₂, 1 mM MgCl₂, 1 mg/ml BSA and stimulated with the appropriate agonists at 37 °C while stirring. 2 × 10⁷ cells (0.5 ml of the suspension) were lysed on ice in 0.5 ml of 100 mM HEPES buffer, pH 7.4, 2% Triton X-100, 1% deoxycholate, 0.1% SDS, 300 mM NaCl, 10 mM MgCl₂, 2 mM EGTA, 2 mg/ml BSA, 20 mM benzamidine, 20 µg/ml leupeptin-pepstatin-aprotinin-soybean trypsin inhibitor, 2 mM phenylmethylsulfonyl fluoride. Nuclei were pelleted and lysates were adjusted to 500 mM NaCl. After preclearing, the samples were immunoprecipitated with 2 µg of rabbit anti-RhoA polyclonal antibody recognizing the sequence KDLRNDEHTRRELA (119-132) or rabbit anti-mouse Ig negative control for 60 min at 4 °C, followed by 4 µl of Trisacryl-protein A beads for 90 min. The beads were washed 10 times in 50 mM HEPES buffer, pH 7.4, 500 mM NaCl, 0.1% Triton X-100, 0.005% SDS, and the nucleotides were eluted in 5 mM EDTA, 2 mM dithiothreitol, 0.2% SDS, 0.5 mM GTP, 0.5 mM GDP for 30 min at 68 °C (9). Separation of eluted nucleotides was on polyethyleneimine cellulose plates run in 0.75 M KH₂PO₄, pH 3.5, as described (9). Radioactive spots, determined by autoradiography with X-Omat AR films (Kodak), were scraped off the plates and counted in a scintillation -counter. Alternatively the cells were resuspended at 5 × 10⁷/ml in Ca²⁺/Mg²⁺-free PBS, 1% pluronic F-68 (Sigma), and 60 µCi/ml GTP³⁵S. The cells were syringe-loaded through a tuberculin syringe with a 30-gauge needle (14). After 14 passes through the needle, 0.5-1% of added radioactivity was incorporated into the cells. The cells were washed twice, resuspended at 4 × 10⁷/ml in PBS, 1 mM CaCl₂, 1 mM MgCl₂, 1 mg/ml BSA, and after 10 min of recovery at 37 °C, stimulated with the appropriate agonists at 37 °C while stirring and processed as above. Radioactivity was detected with a Molecular Dynamics PhosphorImager 445 SI after 2 days of exposure.

RESULTS

To study the intracellular regulation of chemoattractant-induced lymphocyte adhesion and rho activation, we have used mouse lymphoid L1/2 cells transfected with human (fPR) or with IL-8RA as a model. Agonist stimulation of these cells induces rapid adhesion to VCAM-1. Triggered binding occurs within seconds and is mediated by activation of the integrin ₄₁ (CD49d/CD29) (6, 7).

The second messenger cAMP regulates a number of signal transduction pathways (10). To evaluate the effect of cAMP on rapid chemoattractant-triggered adhesion, we pretreated L1/2 transfectants with Bt₂cAMP, a permeable analog of cAMP. Bt₂cAMP treatment inhibited IL-8 or fMLP-induced adhesion in a dose-dependent manner. In contrast, adhesion induced by the phorbol ester PMA, an activator of DAG-dependent PKCs, was not affected by cAMP pretreatment (Fig. 1A). The inhibitory effect of Bt₂cAMP was not due to metabolic release of butyrate or to contamination of Bt₂cAMP with butyrate because butyrate itself (200 µM) had no effect on binding (see control data in Fig. 1B). The most prominent effector of cAMP is PKA. Pretreatment with specific PKA inhibitors, H89 or HA1004, blocked the inhibitory effect of Bt₂cAMP, completely restoring agonist-induced VCAM-1 binding in response to fMLP and IL-8 (Fig. 1B). We conclude that cAMP through its effector PKA inhibits chemoattractant activation of the lymphocyte integrin ₄₁.

To ask if cAMP might regulate chemoattractant signaling to integrins in other settings as well, we assessed the effect of Bt₂cAMP on fMLP- and IL-8-induced adhesion of human neutrophils to the _M2 (CD11b/CD18) integrin ligand fibrinogen. As shown in Fig. 2, Bt₂cAMP effectively inhibited fMLP- and IL-8-induced but not PMA-induced neutrophil adhesion. Although chemoattractants can induce mobilization of neutrophil integrins from intracellular pools, adhesion triggering under conditions of optimal agonist stimulation, as used here, is mediated by activation of pre-existing membrane integrins (2). We confirmed this observation in our system by blocking integrin up-regulation with DIDS, an anion channel blocker that prevents granule fusion with the plasma membrane (11). DIDS effectively prevented increased staining of stimulated neutrophils with anti-₂-integrin monoclonal antibody IB4 but had no effect on chemoattractant-stimulated neutrophil adhesion (data not shown). The results suggest that cAMP may be a general modulator of the rapid chemoattractant-induced activation of leukocyte integrins.

Previous studies have shown that the small GTP-binding protein rho is an important intracellular mediator of integrin triggering both through chemoattractant receptors (2) and also through PMA-activated PKC (2, 12). The ability of cAMP to inhibit chemoattractant but not PMA-induced leukocyte adhesion (Figs. 1A and 2) suggested that PKA might act upstream of rho, blocking a mechanism of rho activation specifically triggered by G-protein-linked chemoattractant receptors. On the other hand, Bt₂cAMP has no effect on fMLP- or IL-8-triggered elevation in intracellular calcium in transfectants or in neutrophils (data not shown); thus it does not inactivate the chemoattractant receptor itself. To test the effect of elevation of intracellular cAMP on chemoattractant-induced rho activation, we evaluated guanine nucleotide exchange on RhoA, the predominant rho protein in lymphocytes (12). Rho small G-proteins have high intrinsic GTPase activity so that GDP/GTP exchange on RhoA is followed rapidly by conversion of bound GTP to GDP. This rapid hydrolysis precludes detection of their GTP-bound form in vivo (13); we therefore assessed accumulation of ³²P-labeled GDP on immunoprecipitated RhoA as a measurement of stimulated rho guanine nucleotide exchange activity, as previously reported (2, 14). Transfected L1/2 cells were labeled with radioactive phosphate, and the accumulation of ³²P-labeled GDP was measured. As shown in Fig. 3A, the amount of ³²P-labeled GDP bound to RhoA, which is very low in resting cells, was increased 6-8-fold by stimulation with fMLP or IL-8, as reported previously (2). Preincubation of leukocytes with Bt₂cAMP inhibited agonist-induced accumulation of ³²P-labeled GDP on RhoA in a dose-dependent manner, up to 85% for fMLP or 83% for IL-8. To confirm this finding, transfected L1/2 cells were loaded with GTP³⁵S, an hydrolysis-resistant radioactive analog of GTP (2). As shown in Fig. 3B, stimulation of cells with either fMLP, IL-8, or PMA triggered binding of GTP³⁵S to RhoA. In contrast RhoA did not bind GTP³⁵S in non-stimulated cells, as previously reported (2). Preincubation of leukocytes with Bt₂cAMP inhibited fMLP and IL-8-induced accumulation of GTP³⁵S on RhoA. However, PMA-induced accumulation of GTP³⁵S on RhoA was unaffected. Importantly, Bt₂cAMP treatment had no effect on the quantity of RhoA protein immunoprecipitated from stimulated cells (Fig. 3C), implying that the reduction of ³²P-labeled GDP or GTP³⁵S bound to RhoA in Bt₂cAMP-treated cells is due to a decrease of RhoA guanine nucleotide exchange activity. Thus, cAMP inhibits chemoattractant-induced rapid activation of RhoA.

DISCUSSION

We have shown that elevation of intracellular levels of cAMP blocks chemoattractant stimulation of ₄₁-integrin activation in lymphoid cells, and ₂-integrin triggering in neutrophils. The effect is mediated by PKA, and this PKA-dependent inhibition occurs downstream of heterotrimeric G-protein activation but upstream of the small GTPase RhoA, a critical mediator of chemoattractant to integrin signaling (2). Recent studies have highlighted the importance of intracellular cAMP as a gating element in a number of different signaling pathways (10), including mitogen-activated protein kinase activation and cellular proliferation stimulated through growth factor receptors (24-26), and long range patterning during development mediated by the diffusible morphogen Sonic Hedgehog (27). Our results expand this concept to include cAMP and its effector, PKA, as gating elements in chemoattractant stimulation of rho and of rho-dependent integrin activity leading to leukocyte adhesion.

An independent example of the negative role of cAMP on rho-dependent signaling has been previously suggested. In a study of human NK cells, cAMP inhibited spontaneous rho-dependent slow cell movement. In that model, PKA phosphorylation of active (GTP-bound) RhoA induced gradual rho-guanine dissociation inhibitor mediated translocation of GTP-RhoA from the plasma membrane to the cytosol (15), thus terminating rho signaling over several minutes. This contrasts with the inhibition of chemoattractant-stimulated RhoA GDP/GTP exchange by cAMP, reported here, which allows cAMP to prevent the initiation of rho signaling, thus blocking the rapid rho-dependent triggering of integrins by chemoattractants. Thus, it appears that cAMP can be a negative modulator of rho through two separate mechanisms, either by preventing rapid rho activation, as shown here, or by terminating an already active rho-signaling pathway, increasing the capability of rho-guanine dissociation inhibitor to bind rho. Moreover, our data show for the first time that cAMP can inhibit a small GTP-binding protein-dependent pathway by blocking the activation of the GTPase itself (Fig. 4).

In addition to triggering integrin activation rho mediates cytoskeletal remodeling (16), and in both of these roles it is thought to be important to cell trafficking. The inhibitory activity of PKA on rho activation in leukocytes may thus help explain the ability of some cAMP-elevating drugs to inhibit leukocyte transendothelial migration in vitro and recruitment and homing in vivo (17-21), phenomena that are dependent on chemoattractants and integrins. The effect may also permit cross-talk between pro-adhesive and anti-adhesive heterotrimeric G protein-linked receptors, potentially contributing, for example, to the inhibition of chemoattractant-induced neutrophil migration by adenosine, prostaglandin E1, or ₂-adrenergic receptors (22, 23), G_s-linked serpentine receptors that activate adenylyl cyclase to produce cAMP.