VLA-4 Integrin Cross-linking on Human Monocytic THP-1 Cells Induces Tissue Factor Expression by a Mechanism Involving Mitogen-activated Protein Kinase*

Adhesion molecules such as VLA-4 are important not only for monocyte adhesion to extracellular matrix proteins, but also for subsequent cell activation. Monocyte adherence to fibronectin or engagement of VLA-4 has been demonstrated to stimulate production of potent inflammatory mediators such as tumor necrosis factor- (cid:97) , interleukin-1, and the procoagulant tissue factor protein. However, the intracellular signaling cascades leading to gene expression have not been elucidated. Using the human monocytic THP-1 cell line, VLA-4 cross-linking by monoclonal antibodies directed against its (cid:97) 4 and (cid:98) 1 subunits produced a time-dependent in- crease in tyrosine phosphorylation of a broad range of cellular proteins. Using Western blot analysis directed against the phosphorylated form of the extracellular signal-related kinase (ERK) mitogen-activated protein (MAP) kinase proteins, as well as immunoprecipitation and in vitro kinase assays, we found that VLA-4 cross- linking increased ERK1/ERK2 tyrosine phosphorylation and activity. In conjunction, integrin cross-linking also increased NF- (cid:107) B nuclear translocation and 4-h expression of tissue factor. Inhibition of tyrosine kinase activ- ity with genistein (10 (cid:109) g/ml) as well as selective MAP kinase inhibition with the MEK-1 inhibitor PD98059 abolished the VLA-4-dependent ERK tyrosine phosphorylation, RPMI and HBSS media. Fetal calf serum (FCS) was from HyClone. The following antibodies were used in the integrin engagement studies: mouse IgG1 anti-CD49d (mAb HP2.1 (Immuno- tech) and mAb 44H6 (Serotec)), CD29 (mAb K20 and mAb Lia1.2 (Immunotech)), goat F(ab (cid:57) ) 2 anti-mouse IgG (Immunotech), mouse IgG1 anti-CD45 mAb 1214 (PDI Bioscience), and negative mouse IgG1 (Se-rotec). Inhibitory anti-tissue factor antibody (mAb 4509) was obtained from American Diagnostica. tional 20 min at room temperature. DNA-protein complexes were re- solved on a 5% non-denaturing polyacrylamide (60:1 cross-link)/Tris glycine gel and autoradiographs prepared by exposure at (cid:50) 70 °C using a Kodak X-Omat film. To demonstrate specificity of the protein-DNA complex, 125 M excess of unlabeled probe or mutated TF (cid:107) B probe was added to the nuclear extract before adding the radiolabeled probe. The sequences of the plus strands of the oligonucletides used were as follows.

The integrin family of surface adhesion molecules plays a key role in leukocyte recruitment to areas of extravascular inflammation. These heterodimeric integral membrane proteins are important not only for the adhesion to and transmigration across endothelial barriers, but also for adhesive inter-actions with extracellular matrix proteins (1)(2)(3). While initially considered important only for their adhesive properties, recent studies have suggested that integrin engagement can initiate signal transduction pathways contributing to cellular activation (4 -8).
Monocyte recruitment to extravascular sites is an important component of the host response to a variety of stimuli including bacterial infection, tumor deposits, and atherosclerotic plaques. In this location, the surface expression as well as release of a number of macrophage products serve to coordinate the local inflammatory response. Fibrin deposition induced by macrophage tissue factor (TF) 1 expression is known to contribute significantly to the development of this response. Within the vascular space, adherence of monocytes to the endothelium stimulates expression of TF on monocytes, a process that likely contributes to local microvascular thrombosis (9,10). Furthermore, at extravascular sites, products of both the coagulation and fibrinolytic cascades contribute to the generation of the inflammatory response through their interaction with infiltrating cells. Indeed, strategies aimed at reducing fibrin deposition or precluding TF expression have been shown to mitigate the full expression of both the local and systemic inflammatory response depending on the model system studied (9,(11)(12)(13)(14)(15).
Recent work has defined a role for integrin engagement in the induction of monocyte TF expression, as well as other immediate early genes such as interleukin-1␤ (16), interleukin-8, and tumor necrosis factor, and (8) transcription factors IB (11), c-Jun, and c-Fos (17). While monocytes are endowed with a variety of surface integrins, engagement of very late antigen 4 (VLA-4) appears to consistently induce gene expression (16 -23). For example, ligation of VLA-4 by monoclonal antibody in both human peripheral blood monocytes (PBM) and in the monocytic THP-1 cell line promotes TF expression, whereas engagement of ␤ 2 integrins has little effect (20). The intracellular signaling mechanisms leading to the VLA-4-mediated induction of TF as well as other inflammatory genes appears to involve the induction of tyrosine phosphorylation (16,18,24). Recent studies by Lin and colleagues have implicated a possible signaling role for Syk tyrosine kinase in this process (18). In human monocytic THP-1 cells, VLA-4 engagement caused prominent tyrosine phosphorylation as well as activation of Syk tyrosine kinase, an effect that occurred in concert with the induction of the interleukin-1␤ gene. * This work was supported by the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Integrin engagement through interaction with extracellular matrix proteins has been shown to contribute to the regulation of cellular growth and differentiation and to modulate tumor behavior (1). In fibroblasts, this process involves tyrosine phosphorylation and activation of MAP kinase through a cascade involving Raf-1 and MEK (1,(25)(26)(27). While this cascade is known to be activated in macrophages in response to various proinflammatory stimuli (28 -35), its contribution to the ␤ 1 integrin-induced activation of inflammatory cell gene expression is unknown. In the present studies, engagement of VLA-4 on the surface of THP-1 monocytic cells and on human monocytes, both through integrin cross-linking and attachment to a fibronectin substratum, was shown to induce tyrosine phosphorylation and activation of the ERK1/ERK2 MAP kinases. This effect occurred in parallel with nuclear translocation of NF-B and stimulation of TF on the surface of these cells. Furthermore, we took advantage of the existence of a novel selective inhibitor of MEK1 to define the contribution of this pathway to integrin-induced TF expression. This agent, PD98059, caused a dose-dependent inhibition of tyrosine phosphorylation and activation of MAP kinase in response to VLA-4 engagement and concomitantly prevented NF-B translocation and TF induction. Considered together, these studies support a contribution of the MAP kinase pathway to integrin-induced gene expression in cells of monocyte/macrophage lineage.

Cell Preparation
Human monocytic THP-1 cells (ATCC) were propagated in RPMI/ 10% FCS/penicillin/streptomycin at 37°C, 5% CO 2 . Human PBM were isolated from the blood of normal healthy donors by centrifugation over a Ficoll-Hypaque gradient at 400 ϫ g for 20 min. The mononuclear layer was aspirated, washed twice and resuspended in RPMI/2% FCS/Lglutamine. This cell population contained 25-35% monocytes as assessed by Wright's stain and CD14 expression (flow cytometry with fluorescein isothiocyanate-conjugated anti-CD14 Ab; Becton-Dickinson), with Ͼ96% viability by trypan blue exclusion and propidium iodide uptake.

Cell Activation
For integrin engagement studies, THP-1 cells were suspended in RPMI/2% FCS/L-Gln at 5 ϫ 10 6 cells/ml. Surface CD29, CD49d, and CD45 antigens were ligated with monoclonal antibody for 25 min at 15 g/ml and 4°C, washed twice in cold RPMI, and then cross-linked with 5 g/ml goat anti-mouse F(abЈ) 2 for 25 min at 4°C. The concentration of primary antibody was shown to be saturating by flow cytometry (data not shown). Cells were washed twice in cold RPMI, resuspended in RPMI/2% FCS/L-Gln and incubated at 37°C, 5% CO 2 for times ranging from 1 min to 4 h. Reactions were stopped by placing the cells on ice. For cell adhesion studies, six-well polystyrene culture plates were coated either with 1 mg/ml poly-L-lysine or 0.05 mg/ml fibronectin for 1 h at 25°C. Poly-L-lysine was cross-linked under UV light for 2 h; all wells were then washed twice in cold RPMI and blocked for 1 h with RPMI/ 10% FCS at 37°C. THP-1 (1 ϫ 10 6 ) or PBM (0.5 ϫ 10 6 ) were layered onto the coated wells and allowed to settle for 25 min at 4°C, and then brought to 37°C for up to 4 h before being placed on ice. In inhibition studies, THP-1 cells or human PBM were preincubated in the presence of 1 and 10 M PD98059 or 10 g/ml genistein for 45 min at 4°C.

Measurement of Procoagulant Activity
At 4 h, THP-1 cells (0.5 ϫ 10 6 ) were sedimented at 1000 ϫ g for 3 min. The cell pellet was resuspended at 10 6 cells/ml RPMI, freeze-thawed at Ϫ70°C, and procoagulant activity (PCA) measured by single-stage recalcification clotting assay (36). PCA was expressed as milliunits/10 6 cells by comparison to rabbit brain thromboplastin. In a typical experiment, ␤ 1 integrin cross-linking shortened clotting times from 70 s to 56 s, representing an increase in PCA from 74 milliunits/10 6 cells to 376 milliunits/10 6 cells. In the cell adhesion studies, THP-1 or PBM were harvested using a rubber policeman over ice after a 4 h incubation, washed, and resuspended at 10 6 cells/ml RPMI for PCA assessment. PCA was attributed to TF expression on the basis of a complete reversal of PCA with inhibitory anti-TF antibody (see "Results") and failure of FVII-deficient serum to clot (data not shown).

Immunoprecipitations and Immune Complex Kinase Assays
Phosphotyrosine and ERK Immunoprecipitations-Cells (3 ϫ 10 6 ) were lysed as above and the postnuclear supernatant precleared with Protein G-Sepharose (Pharmacia Biotech Inc.). Cellular proteins were immunocomplexed using polyclonal anti-phosphotyrosine antibody (Transduction) or anti-ERK-1 and -ERK-2 antibody (SCB) for 1 h at 4°C. Protein G-Sepharose was added and incubated at 4°C for 1 h. The resulting immune complexes were washed five times with cold phosphate-buffered saline/0.01% Tween 20, and then separated from beads by 2 ϫ Laemmli buffer, 0.1 M DTT and boiling at 100°C for 5 min. Beads were then sedimented by ultracentrifugation and the supernatant collected for Western blot analysis.
ERK-2 Immune Complex Kinase Assays-ERK-2 immunocomplexes were washed with five changes of cold phosphate-buffered saline/0.01% Tween 20, and then incubated for 30 min at 30°C with 20 g of ultra-pure myelin basic protein (

Electrophoretic Mobility Shift Assay
5 g of nuclear extract protein were preincubated with the nonspecific DNA competitor poly(dI-dC) (5 g, Pharmacia) for 10 min at room temperature. 32 P-Radiolabeled probe containing 2 NF-B sites derived from the human immunodeficiency virus-1 enhancer (HIV-ENH), or containing the TF-specific NF-B site (49), was incubated for an addi-

VLA-4 Induces Monocytic Tissue Factor via MAP Kinase
tional 20 min at room temperature. DNA-protein complexes were resolved on a 5% non-denaturing polyacrylamide (60:1 cross-link)/Tris glycine gel and autoradiographs prepared by exposure at Ϫ70°C using a Kodak X-Omat film. To demonstrate specificity of the protein-DNA complex, 125 M excess of unlabeled probe or mutated TF B probe was added to the nuclear extract before adding the radiolabeled probe. The sequences of the plus strands of the oligonucletides used were as follows.

Statistical Analysis
The data are represented as the mean and standard error of the indicated number of experiments. Where representative studies are shown, these are indicative of at least three equivalent studies. Statistical comparisons were made for continuous data using one-way ANOVA with post hoc Tukey. For comparisons of experiments involving two treatment groups only, the two-tailed Student's t test was employed. Fig. 1A demonstrates that cross-linking of either the ␤ 1 or ␣ 4 subunits of VLA-4 induces a time-dependent increase in phosphotyrosine accumulation in THP-1 cells. The increase occurred within 1 min, reached a maximum at 5-10 min, and persisted for ϳ30 min. While receptor ligation alone caused a small increase in phosphotyrosine residues, cross-linking markedly amplified this effect (Fig. 1B). This effect was not due to the addition of the secondary antibody per se, since this did not induce phosphotyrosine accumulation (Fig. 1C). We also compared the effect of cross-linking using antibodies directed against different epitopes of the ␣ 4 ␤ 1 integrin. As shown in Fig.  1D, cross-linking of ␣ 4 ␤ 1 with the inhibitory antibody Lial 1.2 directed against the ␤ 1 subunit caused a similar pattern of phosphotyrosine accumulation compared with the non-inhibitory antibody K20, although the magnitude was somewhat less. HP2/1, an inhibitory antibody directed against the ␣ 4 subunit caused similar accumulation of phosphotyrosine residues, while use of 44H6, also directed against ␣ 4 , failed to induce tyrosine phosphorylation despite surface binding equivalent to mAb HP2/1. (data not shown, as assessed by flow cytometry, Coulter Epics MCL). Finally, the effect was not due to the process of cross-linking of surface antigen per se, since neither cross-linking with mAb 1214 (Fig. 1C) nor 44H6 induced an increase in tyrosine-phosphorylated proteins (Fig. 1D). As for 44H6, mAb 1214 had surface binding characteristics equivalent to K20 and HP2/1 (data not shown, as assessed by flow cytometry).

Induction of Tyrosine Phosphorylation by Cross-linking VLA-4 -
Endotoxin is known to cause tyrosine phosphorylation in monocytic cells (31,34). The observed effect of cross-linking was unlikely to be due to endotoxin contamination of the system. Polymyxin B did not reverse the stimulation of phosphotyrosine accumulation, and heating of the primary antibody prior to incubation with the cells completely prevented induction of tyrosine phosphorylation. Furthermore, as indicated above, neither primary nor secondary antibody alone mimicked the stimulatory effect.
Tyrosine Phosphorylation and Activation of ERK1/ERK2 following VLA-4 Cross-linking-While cross-linking induced a pattern of tyrosine phosphorylation encompassing a broad range of proteins, cross-linking with K20 caused some degree of persistent phosphorylation in the 42-50-kDa range. To determine whether p44 ERK1/p42 ERK2 MAP kinases might be among the candidate substrate proteins, cell lysates were probed first for phosphotyrosine residues, and the blots stripped and reprobed for p42 ERK2 and p44 ERK1. Both ERK proteins comigrated with an area of VLA-4-induced persistent tyrosine phosphorylation (data not shown). Importantly, the absolute amount of ERK protein in the cell lysates did not change over time following integrin engagement.
To conclusively identify the ERK proteins as targets of the VLA-4-induced tyrosine phosphorylation, two approaches were used. First, immunoprecipitation studies were carried out at 5 min after cell stimulation. The upper panel of Fig. 2A shows a blot of immunoprecipitated tyrosine phosphoproteins probed with anti-ERK2 Ab under each of the treatment conditions. Integrin cross-linking with K20 markedly enhances the amount of tyrosine-phosphorylated ERK2 protein compared with control cells, while engagement alone causes a slight increase. A similar effect is observed when cell lysates are sedimented with either anti-ERK1 Ab or anti-ERK2 Ab and then probed with anti-phosphotyrosine Ab (Fig. 2A, middle and  lower panels, respectively). Similar results were obtained when FIG. 1. VLA-4 ligation and crosslinking induces tyrosine phosphorylation. A, time course. THP-1 surface VLA-4 was cross-linked using either mAb HP2/1 or K20 and the anti-mouse F(abЈ) 2 fragment. The cells were incubated for the times indicated, lysed, and the lysates probed for phosphotyrosine residues by Western blot. B, cross-linking versus ligation. VLA-4 was engaged by various mAb either by cross-linking or simple ligation, and cell lysates probed for phosphotyrosine residues. C, controls. The phosphotyrosine accumulation induced by crosslinking the ␤ 1 (K20) or ␣ 4 (HP2/1) VLA-4 subunits was compared with that induced by cross-linking CD45 (mAb 1214), or by incubating cells in the presence of goat anti-mouse F(abЈ) 2 or negative mouse IgG 1 . Cells were lysed at 5 min in each case. D, epitope-specific PTK activation. VLA-4 was engaged by cross-linking ␤ 1 (K20, Lia1.2) and ␣ 4 (HP2/1, 44H6) subunits. THP-1 cells were lysed 5 min after cross-linking and probed for phosphotyrosine accumulation.

VLA-4 Induces Monocytic Tissue Factor via MAP Kinase
cell surface integrins were cross-linked with mAb against the ␣ 4 subunit (data not shown). ERK phosphorylation induced by VLA-4 cross-linking was also evaluated by Western blot studies using an antibody specific to ERK phosphorylated on tyrosine residues. Cross-linking of VLA-4 by mAb against the ␤ 1 subunit induced a time-dependent increase in ERK tyrosine phosphorylation, particularly of the p42 ERK2 (Fig. 2B). This effect occurred early, peaked at 30 min, and persisted for at least 60 min. This was accompanied by a slight retardation in the electrophoretic mobility, consistent with phosphorylation of ERK2. Integrin ligation alone caused a somewhat delayed rise in phospho-ERK accumulation, although cross-linking induced more phosphorylation at any given time point.
Having demonstrated tyrosine phosphorylation of ERK1/ ERK2 proteins, studies were performed to evaluate their level of activation. Using MBP as a substrate, Fig. 3 (A and B) shows that cross-linking of ␣ 4 and ␤ 1 subunits of VLA-4, respectively, cause a time-dependent increase in immunoprecipitated ERK2 activity, the magnitude of which is similar to that seen with 1 g/ml LPS. Similar results were seen with immunoprecipitated ERK-1, although the increase in activity stimulated by VLA-4 engagement was less, approximately 2-3-fold (data not shown). As expected from the pattern of induction of phosphotyrosine accumulation, cross-linking mAb Lia1.2 markedly increased immunoprecipitated ERK2 activity, while cross-linking 44H6 or mAb 1214 had little to no effect (data not shown).
Adherence to Fibronectin Stimulates Tyrosine Phosphorylation of ERK-To determine whether adhesion to a physiological substratum might induce a similar activation of ERK, human cells were plated onto fibronectin or poly-L-lysine substrata. Fig. 4A shows the pattern of accumulation of tyrosine-phosphorylated proteins in response to adhesion to the indicated substratum. In particular, proteins with molecular masses corresponding to 42, 72, 85, and 95 kDa were tyrosinephosphorylated in fibronectin-exposed cells but not in cells adhered to poly-L-lysine. This pattern generally corresponded to that observed for VLA-4 cross-linking by antibodies in suspended, although the pattern was less complex (compare with Fig. 1). Tyrosine phosphorylation of ERK following adhesion was also evaluated by immunoprecipitation. Cell lysates were recovered at varying time points after adhesion to fibronectin or poly-L-lysine, and ERK2 was immunoprecipitated and probed with anti-phosphotyrosine antibody. As shown in Fig.   4B, there was there was a time-dependent rise in phosphorylated ERK in fibronectin-exposed cells, but not in those adhered to poly-L-lysine. Total ERK protein after immunoprecipitation did not differ between substrata (data not shown).
PD98059 Inhibits Both ERK Activation and Tyrosine Phosphorylation-Recent studies have reported the development of a synthetic inhibitor of the MAP kinase pathway. This compound, PD98059, was shown to specifically block the activation of MEK1, the upstream activator of ERK1/ERK2, without an effect on other protein kinases including the stress activated protein kinases and p38 (37,38). Having demonstrated that VLA-4 cross-linking was able to induce activation of ERK1/ ERK2, we used this agent to discern the role of this pathway in stimulating gene expression in macrophages in response to integrin engagement. First, the effect of PD98059 on VLA-4stimulated ERK activation was studied. PD98059 prevented ERK-2 activation following VLA-4 cross-linking in a dose-dependent manner, with complete inhibition at 10 M (Fig. 5, A  and B). Binding studies with the flow cytometer and fluorescein isothiocyanate-labeled secondary Ab demonstrated that PD98059 did not influence overall binding of antibody to integrin (data not shown). Consistent with its known mechanism of FIG. 2. VLA-4-induced ERK tyrosine phosphorylation. A, THP-1 cells were lysed 5 min after engagement of VLA-4 by mAb K20, following which ERK1, ERK2, or proteins phosphorylated on tyrosine residues were immunoprecipitated. Western blotting was performed as indicated. Similar results were obtained with mAb HP2/1. B, following VLA-4 engagement THP-1 cells were lysed at the times indicated and prepared for Western blotting with Ab specific to phosphorylated ERK. Note that while both ligation (binding) and VLA-4 cross-linking induce the same pattern of ERK phosphorylation, cross-linking induces considerably more at any given time. All studies were replicated a minimum of three times. FIG. 3. Activation of ERK2 by VLA-4 engagement. The VLA-4 integrin on THP-1 cells was engaged by cross-linking the ␣ 4 (mAb HP2.1) or ␤ 1 (mAb K20) subunit, and the cells incubated at 37°C for the times indicated. For comparison, cells were treated with 1 g/ml LPS for 30 min. Following cell lysis, ERK-2 protein was immunoprecipitated as described under "Experimental Procedures," and the immunocomplexes allowed to phosphorylate ultra-pure MBP for 30 min at 30°C. Proteins were then separated on 15% SDS-PAGE. A and B, time course of ERK-2 activation by VLA-4 engagement; the upper panels are autoradiograms of the 20-kDa [ 32 P]MBP band, and the lower panels represent the radioactivity of the corresponding bands as determined by phosphoimager. The data shown are representative of two independent experiments for each antibody.

FIG. 4. Induction of PTK activity and ERK phosphorylation by matrix adhesion.
A, human PBM were allowed to adhere either to poly-L-lysine (PL) or fibronectin (FN). Cell lysates taken at the times indicated were probed with anti-phosphotyrosine Ab. B, human PBM were allowed to adhere to fibronectin or poly-L-lysine for the times indicated, when they were lysed and ERK-2 was immunoprecipitated. The resulting blot was probed with anti-phosphotyrosine Ab. In parallel experiments it was determined that over the first 30 min, Ͼ70% of all cells of the human PBM/lymphocyte layer that bound to the fibronectin or poly-L-lysine matrices were monocytes by CD14 analysis on the flow cytometer.

VLA-4 Induces Monocytic Tissue Factor via MAP Kinase
action, PD98059 also inhibited the increase in phospho-ERK which occurs in response to VLA-4 cross-linking (Fig. 5C). In addition, this agent prevented the slight mobility shift observed with activation of ERK. The inhibitory effect of PD98059 occurred without altering cell viability, as assessed by propidium iodide uptake, and also had no effect on total cellular levels of the predominant isoform ERK2 (data not shown). In a similar fashion, the tyrosine kinase inhibitor genistein (10 g/ml) also decreased ERK2 activation (Fig. 5, A  and B). Fig. 5D demonstrates that genistein causes a global reduction in the level of tyrosine phosphoproteins following cross-linking of ␤ 1 , while PD98059 has little effect on the overall pattern of phosphorylation. Considered together, these studies demonstrate that both tyrosine kinase inhibition as well as inhibition of MEK-1 activation are able to prevent ERK activation in response to VLA-4 cross-linking.
PD98059 and Genistein Inhibit VLA-4-stimulated Monocytic Tissue Factor Expression-Consistent with previous reports (20,23), VLA-4 engagement by antibody cross-linking or adhesion to fibronectin caused marked increases in monocyte PCA (Fig. 6, A and B, respectively). The ability to stimulate PCA correlated with the increase in phosphotyrosine accumulation in response to treatment. In Fig. 6A, PCA was increased following cross-linking with K20, HP2/1, and Lia1.2, but not 44H6 or CD45 (compare with Fig. 1). Furthermore, the increase did not occur following engagement with primary Ab alone or secondary antibody alone. The induction of PCA was comparable to that observed following stimulation with LPS (1 g/ml). The effect of adhesion to fibronectin was observed both in freshly plated human blood monocytes as well in THP-1 cells (Fig. 6B). PCA was attributed to tissue factor induction on the basis of inhibitory studies with anti-TF Ab. In a typical experiment, control cells had clotting times of 67 Ϯ 5 s (112 Ϯ 58 PCA milliunits), while cells cross-linked with mAb K20 exhibited induction of PCA (56 Ϯ 1 s, 393 Ϯ 50 milliunits). The ability of cross-linked cells to shorten clotting time was completely eliminated by inhibitory anti-TF Ab (to 66 Ϯ 1 s, 118 Ϯ 1 milliunits), but not by mouse anti-human IgA Ab (Jackson). Furthermore, this effect was not due to LPS contamination, since heated mAb did not induce PCA and polymyxin B (50 g/ml) FIG. 6. Tissue factor induction by integrin engagement and matrix adhesion. A, the VLA-4 integrin on THP-1 cells was engaged either with mAb alone (ligation), or with mAb cross-linked with F(abЈ) 2 secondary Ab, and cells were harvested after 4 h for PCA determination. B, THP-1 cells or human PBM were allowed to adhere to poly-Llysine or fibronectin matrices for 4 h, and then harvested for PCA assessment. The cumulative results of three separate experiments are shown; data: mean Ϯ S.E., n ϭ 6 -8/group. ***, p Ͻ 0.001 versus control (ANOVA with Tukey); **, p Ͻ 0.0001 versus control (Student's T-test).

VLA-4 Induces Monocytic Tissue Factor via MAP Kinase
did not reverse the effect of cross-linking. Fig. 7 shows the effect of PD98059 and genistein on VLA-4induced PCA. At concentrations of PD98059 that preclude ERK activation in response to integrin cross-linking (10 M for HP 2/1 cross-linking and 1 M for K20 cross-linking), this inhibitor prevented induction of PCA (Fig. 7A). Similarly, genistein prevented the rise in PCA following cross-linking of VLA-4 ( Fig.  7A), at a concentration that caused near complete inhibition of MBP phosphorylation. Fig. 7B demonstrates the ability of PD98059 to abrogate the rise in PCA that occurs in response to attachment to a fibronectin substratum in both THP-1 cells and human monocytes.
VLA-4-induced NF-B Nuclear Translocation Is Dependent on ERK Activation-Previous studies have reported that VLA-4 engagement dramatically increases NF-B nuclear translocation and specific binding in THP-1 cells (20). Since NF-B binding to the promoter of the tissue factor gene is required for induction of gene transcription, we examined whether PD98059 might exert its effect through inhibition of NF-B translocation. Fig. 8 illustrates two representative studies. Cross-linking VLA-4 induces a dramatic increase in both HIV-ENH and TF-specific NF-B translocation. Preincubation of the cells with 10 M PD98059 largely abolished the HIV-ENH shift, and consistently effected a partial inhibition of the TF B shift. DISCUSSION Macrophage-mediated fibrin deposition via the surface expression of the procoagulant molecule tissue factor contributes significantly to the pathogenesis of both intravascular and extravascular inflammation. Induction of tissue factor occurs in response to a variety of proinflammatory stimuli including tumor necrosis factor, C3a, formyl peptides, as well as various bacterial species and their surface components (9,11). Recent studies including those reported here demonstrate that engagement of monocyte surface VLA-4 by specific antibody is able to induce tissue factor expression (20). The present results clearly demonstrate that tyrosine phosphorylation and activation of ERK1/ERK2 MAP kinase are involved in integrin-induced signaling pathway leading to tissue factor expression. Several lines of evidence support this conclusion. First, integrin aggregation through cross-linking causes tyrosine phosphorylation of these proteins. This was definitively shown by immunoprecipitation studies, as well as by experiments using an antibody directed against the phosphorylated form of ERK. Furthermore, adhesion of both THP-1 and human monocytes to fibronectin caused phosphorylation of ERK. Second, VLA-4 cross-linking caused a time-dependent activation of ERK1/ ERK2 MAP kinase as assessed by its ability to phosphorylate its substrate protein myelin basic protein. Finally, two strategies shown to prevent phosphorylation and activation of ERK precluded the induction of tissue factor, not only in response to cross-linking, but also following adhesion to fibronectin. These included the use of the tyrosine kinase inhibitor genistein as well as the novel specific inhibitor of the upstream activator of ERK, PD98059. Considered together, these findings invoke a role for the MEK-1/MAP kinase cascade in the integrin-induced activation of monocyte coagulation molecules.
The data presented here are consistent with those recently reported by Lin and colleagues showing that adhesion of THP-1 monocytes to fibronectin lead to the tyrosine phosphorylation of pp125 FAK , paxillin, and the nonreceptor tyrosine kinase Syk (18,39). Syk phosphorylation was also associated with its activation. At least two pathways leading to the activation of the Ras signaling cascade may link these effects to the activation of MAP kinase demonstrated in this study. First, phosphorylation of FAK creates an SH2-binding site for Grb2, resulting in localization of Grb2⅐SOS complexes and subsequent activation of Ras (40). In addition, in cultured mast cells, activated Syk has been shown to cause tyrosine phosphorylation of Shc in response to engagement of the Fc ⑀RI with antigen (41). Consequent association with Grb2 may lead to Ras activation. The precise signaling pathways leading to MAP kinase activation following integrin engagement, however, require further definition. While the activation of MEK-1/MAP kinase following integrin engagement in monocytic cells suggests the involvement of the Ras as well as the downstream kinase Raf in the signaling pathway, two lines of evidence suggest that the pathway does not align precisely along the lines suggested for growth factor-induced cell signaling. For example, a recent study using NIH 3T3 fibroblasts demonstrated activation of Raf-1, MEK-1, and MAP kinase following adhesion to fibronectin in a manner that was independent of Ras (42). Second, the studies reported by Lin and colleagues demonstrated that neither phosphorylation nor activation of Raf-1 occurred following ␤ 1 engagement in monocytes (18). Recent studies have suggested the existence of unidentified MEK activators that may not have been detected in these studies (43). Further studies are required to evaluate the upstream activation pathway leading to MAP kinase activation following ␤ 1 integrin engagement in cells of monocytic lineage.
Although integrin ligation by antibody without aggregation induces phosphorylation and activation of p42 ERK-2, crosslinking consistently induces the most phosphorylation, ERK activation, and procoagulant response ( Figs. 1 and 3). In addition, clustering by either adhesion-inhibiting antibodies as well as non-inhibitory antibodies induced comparable effects. These findings are consistent with previous reports demonstrating the need to aggregate integrin receptors in monocytes as well as in other cell types to achieve maximal tyrosine phosphorylation and gene expression (16,44,45). Considered together, the data suggest that the clustering of integrins plays a central role in initiating the signaling cascade leading to inflammatory gene induction. However, while integrin aggregation is necessary, it does not appear to be sufficient. The antibody 44H6 bound to THP-1 cells with equivalent affinity, yet was unable to induce either tyrosine phosphorylation or tissue factor expression when cross-linked. Particular VLA-4 epitopes as defined by mAb clones have already been suggested to play selective roles in adhesive functions (46). For example, Sato et al. demonstrated that mAb 8F2 (directed against the ␣ 4 VLA-4 subunit epitope "C") induced little phosphotyrosine accumulation in Jurkat T-cells, while two other "C"-specific clones did so strongly (24). From a physiological standpoint, the process of integrin aggregation using mAb is a rather artificial one. Monocyte adhesion to endothelial cells or extracellular matrix proteins in vivo would presumably result in both engagement of the integrin ligand as well as aggregation of integrin receptors. In this regard, adhesion of monocytes to fibronectin (Fig. 6) as well as interaction with endothelial cells has been shown to induce monocytic tissue factor expression (10,47). It should be noted that the increase in TF induced by THP-1 or human PBM adherence to culture plates coated with 50 g/ml fibronectin is only a moderate stimulus for TF expression when compared with that obtained with 1 g/ml LPS. However, the observed increase is consistent with the ability of fibronectin to induce phosphotyrosine accumulation and stimulation of immediate early genes, as reported by Lin et al. (18).
The promoter region of the tissue factor gene has a cis-acting integrin response element containing two AP-1 sites and a single B-like sequence (20,48,49). Mutation of either site resulted in reduced integrin responsiveness (20). In the present studies, induction of tissue factor expression following integrin cross-linking and TF B nuclear translocation were fully or partially precluded, respectively, by treatment with the MEK-1 inhibitor PD98059. HIV-ENH binding was abrogated by PD98059. Taken together, these data suggest an association between MAP kinase activation and NF-B translocation lead-ing to gene activation. Interestingly, in response to LPS, macrophages similarly exhibit both ERK activation and NF-B translocation to the nucleus. However, these two events do not appear to be causally related, although both contribute to gene activation (50,51). Similar dissociation has been demonstrated in response to cellular hypoxia (52). By contrast, overexpression of ERK-1 in Jurkat cells caused a marked increase in the DNA binding activity of NF-B (53). Together, these data suggest possible interaction between these two events, which may be cell-and stimulus-specific.
The present study is the first to describe a contributory role for the ERK pathway in the induction of adhesion-dependent inflammatory response in cells of monocyte/macrophage lineage. Since endothelial cell adhesion of monocytes via engagement of surface integrins is an early event in the mobilization of cells to sites of inflammation, it will be of interest to discern how inflammatory and antiinflammatory mediator molecules acting via this or other signaling cascades might interact with this activated pathway to modulate the inflammatory response.