The GTPase Rap1 Regulates Phorbol 12-Myristate 13-Acetate-stimulated but Not Ligand-induced β1Integrin-dependent Leukocyte Adhesion*

Leukocyte migration from bloodstream to tissue requires rapid, coordinated regulation of integrin-dependent adhesion and de-adhesion. In a previous study we demonstrated that inhibition of protein geranylgeranylation inhibited phorbol ester-stimulated avidity modulation of β1 integrin in several leukocyte cell lines. Both RhoA and Rap1 require post-translational modification by geranylgeranylation for full function. In this report we identify Rap1, not RhoA, as a critical geranylgeranylated protein mediating phorbol ester-stimulated β1 and β2integrin-dependent adhesion of Jurkat cells. Overexpression of the Rap1-specific GTPase-activating protein, SPA-1, or inactivated form of Rap1 (N17Rap1) blocked phorbol ester-stimulated adhesion of Jurkat cells to fibronectin (α4β1) and ICAM-1 (αLβ2). With high concentrations of fibronectin as ligand, Jurkat cells adhered spontaneously without phorbol ester stimulation. Unlike the phorbol ester-stimulated adhesion, adhesion induced by high density ligand was not dependent upon Rap1 activation or actin cytoskeleton reorganization. Thus, the “inside-out” adhesion signal induced by phorbol ester and the “outside-in” signal induced by high density ligand involve different pathways.

Leukocyte migration from bloodstream to tissue requires rapid, coordinated regulation of integrindependent adhesion and de-adhesion. In a previous study we demonstrated that inhibition of protein geranylgeranylation inhibited phorbol ester-stimulated avidity modulation of ␤ 1 integrin in several leukocyte cell lines. Both RhoA and Rap1 require post-translational modification by geranylgeranylation for full function. In this report we identify Rap1, not RhoA, as a critical geranylgeranylated protein mediating phorbol ester-stimulated ␤ 1 and ␤ 2 integrin-dependent adhesion of Jurkat cells. Overexpression of the Rap1-specific GTPase-activating protein, SPA-1, or inactivated form of Rap1 (N17Rap1) blocked phorbol ester-stimulated adhesion of Jurkat cells to fibronectin (␣ 4 ␤ 1 ) and ICAM-1 (␣ L ␤ 2 ). With high concentrations of fibronectin as ligand, Jurkat cells adhered spontaneously without phorbol ester stimulation. Unlike the phorbol ester-stimulated adhesion, adhesion induced by high density ligand was not dependent upon Rap1 activation or actin cytoskeleton reorganization. Thus, the "inside-out" adhesion signal induced by phorbol ester and the "outsidein" signal induced by high density ligand involve different pathways.
The essential role of leukocyte integrin receptors in cell-cell and cell-substrate adhesion in the inflammatory and immune systems is well established. The adhesive capacity of leukocyte integrins is highly regulated. Integrin receptors in a low adhesive state do not mediate strong adhesion to other cells or ligands. However, when leukocytes are appropriately activated, there is often a detectable increase in integrin adhesiveness within a few seconds to minutes. Some activation stimuli induce a measurable change in integrin receptor affinity, whereas others mediate their effects without altering affinity but instead utilize post-receptor events involving cytoskeletondependent clustering of receptors that serve to increase overall adhesivity. Increases in integrin adhesivity produced by postreceptor events without changes in receptor affinity have been defined as increased avidity. Post-receptor events also regulate "outside-in" signaling in which integrins transduce information from the exterior to the interior of the cell, engaging classic signaling pathways that control growth, differentiation, apoptosis, and cytokine expression.
Whether leukocyte integrin adhesivity is regulated primarily by affinity or avidity modulation is still somewhat controversial (1). For a given leukocyte cell type, different activation stimuli may modulate integrin adhesivity by one or the other mechanism. For example, the functional activity of ␤ 1 integrins on human T-cells can be regulated by treatment with certain divalent cations or activating monoclonal antibodies (mAbs), 1 which directly increase the affinity of ␤ 1 integrins for their ligands, presumably by altering receptor conformation (2). In contrast, the protein kinase C (PKC) activator phorbol 12myristate 13-acetate (PMA) generally promotes ␤ 1 -dependent leukocyte adhesion by targeting events that occur following receptor occupancy without significantly affecting the affinity of the receptor for the ligand (3). Moreover, for the same stimulus, integrin receptors of one subfamily may respond differently from those of another. For example, Weber et al. (4) noted that there was differential regulation of ␤ 1 and ␤ 2 integrin adhesiveness in eosinophils stimulated by the same chemoattractant. ␤ 2 integrin-mediated adhesion was dependent upon affinity modulation, whereas ␤ 1 integrin-mediated adhesion involved post-receptor events. For leukocyte ␤ 1 integrins, the regulation of adhesivity may be mediated predominantly by post-receptor events such as diffusion/clustering in the membrane and subsequent cytoskeletal interactions rather than by affinity modulation (3). Activation of leukocyte ␤ 2 integrins also involves avidity modulation, although affinity mechanisms clearly apply (5). Although a number of cytoplasmic protein regulators of integrin affinity have been identified and characterized in a variety of cell types (1), the signal transduction pathways involved in avidity modulation in leukocytes have not been fully elucidated. One current model of avidity modulation proposes that leukocyte integrins are loosely restrained in the plasma membrane in an inactive, non-clustered state by interaction of the ␤-subunit cytoplasmic tail with cortical actin cytoskeleton (1). Regulatory proteins such as Mac-MARCKS (6) and L-plastin (7) maintain the cortical cytoskeleton. Diverse stimuli trigger activation of the integrin receptors by "inside-out" signaling, resulting in activation of PKC, phosphorylation of L-plastin and MacMARCKS, increases in intracellular Ca 2ϩ , and activation of calpain. These signals promote the release of integrin receptors from the cytoskeletal constraints, allowing diffusion of the receptors within the membrane. With increased diffusion there is clustering of receptors upon contact with immobilized ligand, augmenting cellular avidity for the ligand-coated surface. Subsequently, other signaling components such as cytohesin-1 (8), Rack1 (9), and GTPases are recruited to the clusters, leading to adhesion strengthening, actin stress fiber formation, spreading, acquisition of "activation" epitopes, and assembly of signaling molecules. All of these events may take place without significant changes in receptor affinity.
Several Ras/Rho small GTPases have also been reported to regulate leukocyte integrin avidity. H-Ras and Rac were shown to be involved in stimulated ␤ 2 integrin-dependent leukocyte adhesion (10). Previous studies (11) have established a critical role for RhoA in integrin clustering, adhesion, and spreading in a wide variety of adherent cell types. However, studies of leukocyte adhesion using C3 exoenzyme to inactivate RhoA specifically have yielded conflicting results. Blockade of RhoA function by the C3 exoenzyme was reported to inhibit ␤ 2 integrindependent stimulated neutrophil adhesion (12). However, in other studies treatment with C3 exoenzyme did not affect ␤ 2 integrin-dependent adhesion of neutrophils (13) or JY lymphocytic cells (14) or ␤ 1 integrin-dependent adhesion of peripheral blood T-cells (15) or U937 monocytic cells (16). Finally, several recent studies have implicated Rap1, a small GTPase with 53% amino acid sequence homology to K-Ras, in stimulated ␤ 1 (17)(18)(19) and ␤ 2 (10, 19 -21) integrin-dependent leukocyte adhesion, as well as ␤ 3 -dependent platelet adhesion (22).
In a previous study (23) we demonstrated that inhibition of protein geranylgeranylation inhibited phorbol ester-stimulated avidity modulation of ␤ 1 integrin in several leukocyte cell lines. Both RhoA and Rap1 require post-translational modification by geranylgeranylation for full function. In this report we identify Rap1, not RhoA, as a critical geranylgeranylated protein mediating phorbol ester-stimulated ␤ 1 -dependent adhesion of Jurkat T-cells. We show further that adhesion to fibronectin (FN) stimulated by PMA and "spontaneous" adhesion induced by binding to a high concentration of FN involve different mechanisms. Unlike PMA-stimulated adhesion, adhesion induced by binding to high density ligand is not dependent upon Rap1 activation, actin cytoskeleton reorganization, PKC activation, or tyrosine phosphorylation.
Transfection Protocol-The pLXSN-SPA-1, pLXSN-SPA-1-⌬GRD, and the pLNSX plasmids were gifts of Dr. N. Minato (Kyoto University, Kyoto, Japan) (27). The plasmids were transfected into the ecotropic packaging line, PE501 (a gift of A. D. Miller, Fred Hutchinson Cancer Research Center, Seattle, WA), by calcium phosphate precipitation. The viral supernatants were harvested and used to transduce the amphotropic line PA317 (ATCC) in the presence of 9 mg/ml Polybrene (Sigma). Polyclonal and monoclonal retrovirus-producing cell lines were ob-tained by selection in 1 mg/ml G418 (Invitrogen). Retroviral supernatants from the PA317 cell lines were used to infect Jurkat cells. Following selection in 1 mg/ml G418, the expression level of SPA-1 was determined with anti-SPA-1 antibody (gift of N. Minato). Polyclonal retrovirus-infected Jurkat cells were used in order to avoid artifacts because of retroviral integration.
Labeling with [ 3 H]Geranylgeraniol Pyrophosphate-2.0 ϫ 10 6 Jurkat cells were first treated with 10 M lovastatin for 2 days to deplete the endogenous isoprenyl groups. They were then incubated with or without 10 M of the PGGT-I inhibitor (GGTI-298) in the presence of 25 Ci of [ 3 H]geranylgeraniol pyrophosphate for 2 days, a period determined previously (23) as sufficient to restore integrin function. Following labeling in this manner, the cells were lysed in SDS sample buffer, and the lysate was analyzed on 12% SDS-PAGE. The [ 3 H]geranylgeraniollabeled bands were visualized by autoradiography (28).
Assay for ADP-ribosylation of RhoA-Cells (2.0 ϫ 10 6 ) were treated with 10 g/ml C3 transferase (Calbiochem) as described previously (29). After overnight incubation, the cells were washed with cold PBS and lysed in 250 l of cold lysis buffer (2 mM MgCl 2 , 0.1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 80 g/ml benzamidine in 50 mM HEPES, pH 7.4), and the lysate was then sonicated. Aliquots (100 g of protein) of the lysate were incubated with 32 P-NAD (PerkinElmer Life Sciences) and 10 g of C3 transferase at 37°C for 30 min before Laemmli buffer was added. The samples were heated for 3 min at 95°C and run on SDS-PAGE. [ 32 P]ADP-ribosylated proteins were detected by autoradiography with a PhosphorImager (Amersham Biosciences).
Adhesion Assay-Leukocyte adhesion to FN or to recombinant ICAM-1 was performed as follows. Human FN (Collaborative Research, Inc., Bedford, MA) as noted or 100 g/ml ICAM-1 (R & D Systems, Minneapolis, MN) was coated onto a 96-well Pro-Bind TM assay plate (Falcon, BD Biosciences) by incubating overnight at 4°C. The plate was then blocked with 3% bovine serum albumin in PBS at room temperature for 1 h. Immediately before use, plates were washed three times with PBS. After centrifugation for 7 min at 300 ϫ g, the infected Jurkat cells were resuspended in 1 ml of phenol red-free medium and then labeled by incubation with 5 l of the fluorescent dye calcein-AM (1 mg/ml in Me 2 SO, Molecular Probes, Eugene, OR) for 30 min at room temperature in the dark. The cells were then washed twice with phenol red-free medium. After incubation with PMA or the ␤ 1 -activating mAb 8A2 (2 g/ml) for 30 min in control medium at room temperature, cells (ϳ1 ϫ 10 5 /well) were added to triplicate wells. After incubation for 30 min at 37°C, the total population of cells in the well was analyzed using a fluorescence plate reader (Perspective Biosystems, Framingham, MA). Unbound cells were removed by washing the plate three times with phenol red-free medium, and the plate was then reanalyzed to determine the fluorescence of bound cells. After subtraction of background, the percent adherence was calculated as the emission at 530 nm of bound cells divided by the emission of total cells.
Detection of Small G-protein Activation-Small G-protein activation was detected with an activation assay kit (Upstate Biotechnology, Inc., Lake Placid, NY). Briefly, cells were lysed with Mg 2ϩ lysis/wash buffer (MLB) containing 25 mM HEPES, pH 7.5, 150 mM NaCl, 1% CA-630, 10 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM sodium orthovanadate, and 25 mM sodium fluoride. The lysates were incubated at 4°C for 30 min with Raf-1-PBD-agarose to detect Rap1 or Ras activation or with PAK-1-PBD-agarose for Rac. After washing three times with MLB, the pellet was subjected to SDS-PAGE. The gels were then analyzed by immunoblotting with an anti-Rap1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), anti-Ras antibody (Upstate Biotechnology, Inc.), or anti-Rac antibody (Upstate Biotechnology, Inc.). Because Raf-1-PBD has a much higher binding efficiency for Ras than it has for Rap1 (30), in the Rap1-GTP loading experiment, the cell lysate was pre-cleared by immunoprecipitation of Ras with the anti-Ras antibody.
␤ 1 Integrin Receptor Clustering-An assay to detect stimulated ␤ 1 integrin diffusion/clustering was developed based upon the report from Rehemtulla et al. (31). Briefly, control or treated cells were incubated with the cross-linker 3,3Ј-diothiobis(sulfosuccinimidyl propionate) (DTSSP) (Pierce), which covalently links molecules within 12 Å. Following cross-linking, cells were lysed, and limiting anti-␤ 1 mAb was added. More integrin receptor was immunoprecipitated if ␤ 1 integrin receptors were either freely diffusible, allowing "capture" by the crosslinker, or clustered. In contrast, with excess anti-␤ 1 mAb equivalent amounts of integrin receptor were immunoprecipitated, regardless of receptor mobility or clustering. Jurkat cells (2.0 ϫ 10 6 ) were untreated or treated with 100 ng/ml PMA or with cytochalasin D (1 g/ml) for 30 min at room temperature before adding 1.25 mg/ml of DTSSP. After cross-linking for 3 h at room temperature, DTSSP was removed by washing the cells three times with PBS. Immunoprecipitation was performed with the anti-␤ 1 mAb 8A2 at a concentration of 1 g/ml.
Immunoprecipitates were subjected to SDS-PAGE, and immunoblotting was then performed with an anti-␤ 1 integrin mAb (Calbiochem).
Significance was determined using two-tailed Student's t-test with unequal variance, using Microsoft Excel (Redmond, WA).

A 20-30-kDa Geranylgeranylated Protein Is a Candidate
Regulator of PMA-stimulated Jurkat Adhesion-We have shown previously (23) that protein geranylgeranylation was necessary for PMA-stimulated, ␤ 1 integrin-dependent leukocyte adhesion. Because stimulated adhesion was blocked by pretreatment with the PGGT-1 inhibitor GGTI-298, we first determined the effect of that inhibitor on protein geranylgeranylation in Jurkat cells. In the absence of GGTI-298, two bands with apparent molecular weights between 20 and 30 kDa were labeled by [ 3 H]geranylgeraniol pyrophosphate (data not shown). Treatment with the PGGT-1 inhibitor GGTI-298 markedly reduced labeling of the more slowly migrating band. RhoA is a 22-kDa geranylgeranylated protein that has been implicated in ␤ 2 integrin-dependent adhesion (32). Our previous study (23) had shown that GGTI-298 inhibited RhoA protein geranylgeranylation in Jurkat cells at the same dose as it inhibited PMA-stimulated cell adhesion. Consequently, we examined the involvement of the Rho signaling pathway in PMAstimulated ␤ 1 integrin-dependent Jurkat cell adhesion. Treatment with C3 exoenzyme in vivo significantly reduced the amount of RhoA protein labeled with 32 P in vitro, demonstrating that the C3 transferase had penetrated the intact Jurkat cells and had ADP-ribosylated RhoA (data not shown). However, treatment with C3 exoenzyme at a concentration at which most RhoA protein was ADP-ribosylated in vivo did not inhibit PMA-stimulated adhesion to FN (Fig. 1). In addition, the RhoA kinase inhibitor Y-27632 did not reduce PMA-stimulated adhesion. These results indicated RhoA and RhoA kinase activation were not required for PMA-stimulated, ␤ 1 integrindependent Jurkat cell adhesion to FN (Fig. 1).
SPA-1 Specifically Inhibits PMA-stimulated Rap1 Activation and Cell Adhesion-The geranylgeranylation of Rap1 was also inhibited by GGTI-298 at the same time course and dose as GGTI-298-inhibited cell adhesion (23). We investigated the role of Rap1 by transducing cells with the Rap1-specific GTPaseactivating protein (GAP), SPA-1 (27). Tsukamoto et al. (27) demonstrated previously that overexpression of SPA-1 blocked Rap1 activation through its GAP activity in 293T cells and that a mutant SPA-1 without a GAP-related domain, ⌬GRD, was without effect. As shown in Fig. 2, PMA activated Rap1 from the GDP-to GTP-binding form. Compared with the vector control, overexpression of SPA-1 suppressed Rap1 activation in both resting and PMA-stimulated cells. However, the mutant SPA-1-⌬GRD enhanced production of Rap1-GTP. SPA-1 did not inhibit the GTP loading of Ras or Rac GTPases (Fig. 2).
Overexpression of SPA-1 significantly reduced PMA-stimulated Jurkat cell adhesion to FN (Figs. 1 and 3A). There was some increased adhesion with PMA compared with basal adhesion in the SPA-1-overexpressing cells, but maximal adhesion was much lower than that of vector control or that of SPA-1-⌬GRD (Fig. 3B). The SPA-1-⌬GRD mutant moderately increased cell adhesion compared with vector control (Fig. 3B), consistent with its effect on Rap1-GTP binding in vivo (Fig. 2). Cells overexpressing an inactivated form of Rap1, N17Rap1, evidenced reduced adhesion in response to PMA (Fig. 3C). In contrast to its effect on PMA stimulation, overexpression of SPA-1 had no effect on adhesion induced by the activating mAb 8A2 (Fig. 3A).
Overexpression of SPA-1 Does Not Inhibit Cytochalasin Dinduced Adhesion or ␤ 1 Integrin Diffusion/Clustering-Previous reports (33) showed that low doses of cytochalasin D induced cell adhesion, whereas higher doses inhibited it. We found that overexpression of SPA-1 did not block cytochalasin D-induced adhesion (Fig. 4). As shown in Fig. 5, PMA treatment of Jurkat cells markedly increased the amount of ␤ 1 integrin immunoprecipitated by limiting amounts of anti-␤ 1 mAb in the presence of cross-linker, indicating an increase in diffusion, clustering, or both. Cytochalasin D treatment also markedly increased ␤ 1 integrin immunoprecipitated in resting cells, consistent with dissolution of basal cytoskeletal restraints, allowing increased diffusion and capture by cross-linker. SPA-1 did not block cytochalasin D-or PMA-induced integrin diffusion/ clustering as detected by this technique (Fig. 5).
The Inhibitory Effect of SPA-1 Is Dependent Upon the Density of Ligand-As shown in Fig. 7A, the adhesive behavior of Jurkat cells differed dramatically with the concentration of FN coating the wells. PMA stimulation markedly increased adhesion of vector control and SPA-1-⌬GRD cells, but not of SPA-1 cells, to wells coated with 1 g/ml FN. When the concentration of FN used to coat the wells was increased from 1 ("low") to 2 ("high") g/ml, the adhesion of unstimulated vector control cells increased from 5 to 65% and unstimulated SPA-1-⌬GRD cells from 16 to 79% and even that of the SPA-1 cells increased from 1 to 51%. Although overexpression of SPA-1 reduced both unstimulated and PMA-stimulated adhesion to low FN, SPA-1 had little effect on the adhesion of SPA-1 cells to high FN. At 1 g/ml FN the PMA-stimulated adhesion of SPA-1 cells was only 22%, and those of the vector control and SPA-1-⌬GRD were 74 and 86%, respectively. For PMA-stimulated SPA-1 cells, adhesion increased to 88% at 2 g/ml FN, which was comparable with that of vector control (89%) and SPA-1-⌬GRD (93%) cells.
We next examined Rap1 activation at a high concentration of FN. Rap1 was not activated in any of the cells adherent to high FN, based on the GTP-loading assay. In the same experiment PMA did induce Rap1 activation in control and SPA-1-⌬GRD but not in SPA-1 cells (Fig. 7B). Of note, SPA-1-⌬GRD also increased basal Rap1 activation (Fig. 7B). This is consistent with the enhanced basal adhesion observed in SPA-1-⌬GRDtransfected cells (Fig. 3A).
The Outside-in Signal Is Distinct from the Inside-out Signal-PMA-stimulated cell adhesion was dependent upon the activation of Rap1 (Fig. 3A), whereas adhesion induced by high FN was independent of Rap1 activation (Fig. 7A). The inactivated form of Rap1, N17Rap1, also failed to inhibit unstimulated adhesion to high density of FN (5 g/ml). Unstimulated adhesion of control cells was 12% to low FN and 30% to high FN. Adhesion of N17Rap1 cells to low FN was 4.5 versus 28% to high FN (means of eight replicates). These results suggested that PMA-stimulated and high density ligand-induced adhesion involved distinct mechanisms. This was supported by the observation that PMA-stimulated adhesion was inhibited by cytochalasin D, whereas adhesion to high density FN was not affected (Fig. 8A).
Additional studies were performed to determine the signaling pathways involved in Jurkat cell adhesion at high density FN, which was not dependent upon Rap1 or cytoskeletal reorganization. Both PMA-stimulated adhesion to low FN and unstimulated adhesion to high FN were blocked by deoxyglucose and azide (Fig. 8B), indicating that both required active cell metabolism. PMA-stimulated adhesion to low FN was reduced by treatment with the tyrosine kinase inhibitor, genistein, and the PKC inhibitor, staurosporine, whereas neither agent affected unstimulated adhesion to high FN (Fig. 8B). DISCUSSION We had determined previously (23) that a 20 -30-kDa protein(s) whose geranylgeranylation was catalyzed by PGGT-1 was a candidate to regulate PMA-stimulated, ␤ 1 integrindependent Jurkat cell adhesion. Geranylgeranylation of both RhoA and Rap1 proteins was inhibited when PMA-stimulated cell adhesion was blocked by lovastatin or the PGGT-1 inhibitor GGTI-298 (23). However, treatment of cells with C3 exoenzyme, which ADP-ribosylates RhoA protein and prevents RhoA interaction with its downstream protein, did not inhibit PMAstimulated adhesion to FN (Fig. 1). This suggested that RhoA activation was not required for PMA-stimulated, ␤ 1 integrindependent leukocyte adhesion. That conclusion was further supported by the observation that inhibition of RhoA kinase, Vector control cells (control), SPA-1-infected cells (SPA-1), and SPA-1-⌬GRD-infected cells (⌬GRD) were treated with or without 100 ng/ml PMA in suspension or adhered to wells treated with 4 g/ml FN without PMA (ϩFN). Rap1-GTP was detected by pull-down assay with Raf-1-RBD and anti-Rap1. Because Raf-1-RBD has higher affinity for Ras than for Rap1, cell lysate was pre-cleared with anti-Ras antibody before the pull-down assay was performed. the downstream effector of RhoA, with the specific inhibitor Y-27632 did not inhibit PMA-stimulated ␤ 1 integrin-mediated adhesion (Fig. 1). In addition, we have shown that RhoA activation is not required for stimulated ␤ 2 integrin-dependent neutrophil adhesion but is instead involved in the process of de-adhesion (34). With RhoA excluded as the geranylgeranylated protein mediating PMA-stimulated Jurkat adhesion, we focused on Rap1. Tsukamato et al. (27) first implicated Rap1 in cell adhesion with the observation that overexpression of SPA-1, a Rap1-specific GAP, blocked granulocyte-colony stimulating factor-induced promyelocytic 32D cell adhesion to culture dishes. Reedquist et al. (18) reported that Rap1 was involved in CD31-induced, ␤ 1 integrin-dependent adhesion of Jurkat cells. Rap1 was also shown to be the activation signal for PMA-stimulated, LFA-1-dependent adhesion of Jurkat cells to ICAM-1 (10), for lipopolysaccharide-induced, ␤ 2 integrindependent adhesion and spreading of macrophages (21), and for erythropoietin-or interleukin-3-induced, ␤ 1 integrin-dependent adhesion of 32D cells (17). Constitutively activated Rap1, V12Rap1, was recently shown to increase ␤ 1 and ␤ 2 integrindependent adhesion of lymphocytes (19). Consistent with these studies, we found that overexpression of the Rap1-specific GAP, SPA-1, in Jurkat cells markedly reduced PMA-stimulated, ␣ 4 ␤ 1 -mediated adhesion to FN, a process dependent upon avidity modulation. The effect of SPA-1 on cell adhesion was further confirmed with the inactivated form of Rap1, N17Rap1 ( Fig. 3C) (10). Both N17Rap1-and SPA-1-transfected Jurkat cells exhibited significantly lower adhesion when stimulated with PMA (Fig. 3, A and C). The results with N17Rap1 confirm that SPA-1 is involved in cell adhesion by regulating Rap1 activation. In contrast, the transfection of SPA-1 did not inhibit Jurkat cell adhesion to FN by the ␤ 1 integrin-activating mAb 8A2, which directly modulates ␤ 1 integrin affinity (Fig. 3A). Because the overexpression of SPA-1 also reduced ␤ 2 integrindependent Jurkat cell adhesion to ICAM-1, we conclude that Rap1 activation is involved in inside-out activation of both ␤ 1 and ␤ 2 integrins by PMA in Jurkat cells.
Previous studies (35) have shown that low dose cytochalasin D stimulates ␤ 1 and ␤ 2 integrin-dependent leukocyte adhesion, presumably by releasing cytoskeletal restraints and allowing diffusion of integrin receptors in the membrane with subsequent ligand-induced clustering. We found that overexpression of SPA-1 did not inhibit cytochalasin D-induced adhesion (Fig.  4). Like PMA, cytochalasin D also induced ␤ 1 integrin receptor mobility in the plasma membrane, and overexpression of SPA-1 did not block this effect induced by either PMA or cytochalasin D (Fig. 5). Thus, blockade of Rap1 by overexpression of SPA-1 did not interfere with the process of integrin release from the cytoskeleton. PMA stimulated a dose-dependent adhesion of both the vector control and the SPA-1-⌬GRD-infected cells to FN (Fig. 3B). Interestingly, SPA-1-⌬GRD-infected cells exhibited greater Rap1 activation and increased adhesion when com- FIG. 9. Different signaling pathways are involved in PMA-stimulated adhesion to low density ligand and spontaneous adhesion to high density ligand. A, PMA-stimulated adhesion to low density FN. PMA stimulation triggers inside-out signaling involving two steps. In the first step, MacMARCKS phosphorylation leads to integrin receptor release from cytoskeletal restraints, allowing increased integrin mobility and ligand-induced clustering. Low dose cytochalasin D can also trigger this step. In the second step, Rap1 activation facilitates formation of a stable adhesion complex with ligand-bound integrin. High dose cytochalasin D inhibits this step. B, unstimulated adhesion to high density ligand. With high density ligand there is spontaneous adhesion without exogenous stimulation. This outside-in signal does not require Rap1 activation or reorganization of the actin cytoskeleton. pared with vector control (Figs. 3 and 7), consistent with overexpressed SPA-1-⌬GRD functioning as a dominant-negative mutant of endogenous SPA-1. However, with overexpression of SPA-1 there was an initial stimulation at the very low dose of 1 ng/ml PMA but no further stimulation at concentrations up to 100 ng/ml PMA. Taken together, these results support the suggestion by van Kooyk and Figdor (1) that PMA modulates leukocyte integrin-dependent adhesion by a multistep process. In the first step, similar to cytochalasin D, PMA activation loosens actin cytoskeletal restraints, thereby allowing diffusion of integrin receptors. This suggestion is supported by the crosslinking experiment. Both PMA and cytochalasin D increased capture of ␤ 1 integrin by a cross-linker, as assessed by immunoprecipitation with a limiting amount of anti-␤ 1 integrin mAb, consistent with greater diffusion. Notably, overexpression of SPA-1 did not block this effect of cytochalasin D or PMA. Zhou and Li (6) reported that PMA induced integrin receptor diffusion through activation of PKC and phosphorylation of Mac-MARCKS. Because SPA-1 overexpression has no effect on the initial diffusion, we speculate that PMA-stimulated phosphorylation of MacMARCKS and Rap1 activation are either two independent signals or that Rap1 activation is downstream of MacMARCKS phosphorylation. The second step in the process of PMA stimulation of integrin-dependent adhesion involves integrin clustering, cytoskeletal rearrangement, and a complex assembly to promote efficient binding to ligand. Because overexpression of SPA-1 blocked PMA-stimulated Rap1 activation and stable adhesion, but did not inhibit ␤ 1 integrin diffusion/ clustering, we propose that activation of Rap1 plays a role in this second step.
After cells bind to ligand(s) via integrin receptors, there is outside-in signaling, which engages classic signaling pathways controlling growth, differentiation, apoptosis, and cytokine expression (36). Previous studies (27,37) have also noted spontaneous ␣ 4 ␤ 1 -dependent adhesion to high density ligand. Our results suggest that adhesion to wells coated with the high concentration of FN involved a mechanism distinct from PMAstimulated inside-out signaling. With increasing concentrations of FN, spontaneous, unstimulated adhesion of Jurkat cells increased markedly for cells overexpressing SPA-1 as well as for control and SPA-1-⌬GRD-infected cells (Fig. 7A). Although overexpression of SPA-1 blocked PMA-stimulated adhesion to wells coated with lower concentrations of FN, it had little effect at the higher concentrations of FN (Figs. 3B and 7A). Notably, Tsukamato et al. (27) reported that adhesion of HeLa cells to high FN induced Rap1 activation, despite overexpression of SPA-1. Although their studies did not specifically address the integrin dependence of the adhesion process, it is likely that these cells utilized ␤ 1 integrin to adhere to FN. In our studies, however, overexpression of SPA-1 did not block Jurkat cell adhesion to high density FN, and adhesion to high density FN alone did not induce Rap1 activation. These results suggest that leukocyte adhesion induced by high density ligand in the absent of exogenous stimulation does not involve Rap1. In addition, the outside-in signal leading to spontaneous adhesion was different from the inside-out signaling triggered by PMA in that it was not blocked by cytochalasin D (Fig. 8A). Furthermore, the tyrosine kinase inhibitor, genistein, and the PKC inhibitor, staurosporine, significantly inhibited PMAstimulated adhesion but had much less or no effect on high FN-induced adhesion (Fig. 8B). Because mitogen-activated protein kinase, phosphatidylinositol 3-kinase, and Rho kinase have been reported to modulate integrin-dependent adhesion, we tested the effect of mitogen-activated protein kinase inhibitor, PD 98050, the phosphatidylinositol 3-kinase inhibitor, LY294002, and the RhoA kinase inhibitor, Y-27632, on both PMA-stimulated and FN-induced adhesion. None of these inhibitors inhibited either of these two pathways (data not shown). Hence, it is unlikely the activation of these kinases is essential for cell adhesion in this model system. Fig. 9 depicts a model for these two adhesive mechanisms. With PMA-stimulated adhesion (Fig. 9A), the integrins are first released from the cytoskeletal restraints, and their diffusion rate in the plasma membrane is increased. The increased diffusion accounts for the small initial increase in cell adhesion. PMA may activate this process through phosphorylation of MacMARCKS (6), whereas low dose cytochalasin D directly releases integrins from the cytoskeleton by inhibiting actin polymerization. In this initial step, the adhesion complex is not supported by a reorganized actin cytoskeleton, so the cells are not stably adhered. In the second step, integrin receptors are clustered; the phosphorylated MacMARCKS is dephosphorylated, and the adhesion complex is supported by a reorganized actin cytoskeleton, leading to stable strong adhesion. Thus, when actin cytoskeleton-dependent re-structuring was disrupted by the high concentration of cytochalasin D, cell adhesion was inhibited. Our studies indicate that Rap1 activation is required for this second step, perhaps in part modulating receptor affinity as suggested by Guerrero et al. (22).
Adhesion induced by outside-in signaling with high density ligand is clearly distinct from PMA-stimulated, inside-out signaling. In contrast to PMA-stimulated adhesion, it is not dependent upon Rap1 activation or reorganization of actin cytoskeleton. It was also not blocked by treatment with inhibitors of tyrosine kinase (genistein), phosphatidylinositol 3-kinase (LY-294002), PKC (staurosporine), mitogen-activated protein kinase (PD-98059) or Rho kinase (Y-27632).
It is surprising that such relatively small changes in ligand density in vitro (i.e. from 1 to 2 g/ml FN in this study or from 200 to 700 sites/m 2 of VCAM-1 in the study by Grabovsky et al. (37), so dramatically affect integrin-dependent adhesion. Unstimulated adhesion induced by high ligand density was nearly comparable with PMA-stimulated adhesion at low ligand density and was independent of Rap1 activation or cytoskeletal rearrangement. Because leukocytes might reasonably encounter high density ligand on the surface of endothelial cells or in tissue, further characterization of this novel pathway of adhesion is warranted.