Urokinase-type Plasminogen Activator-induced Monocyte Adhesion Requires a Carboxyl-terminal Lysine and cAMP-de-pendent Signal Transduction*

Urokinase-type plasminogen activator (u-PA) or its amino-terminal fragment (ATF) containing the u-PA receptor (u-PAR) binding domain, is known to promote monocyte adhesion. In the present study, U937 monocyte adhesion to a plastic surface was used to investi- gate the mechanism of its promotion by u-PA and ATF. Adhesion was found to be inhibited by cycloheximide or actinomycin D, implicating protein synthesis and gene expression in u-PA-induced monocyte adhesion. Adhe- sion was prevented by 2 (cid:42) -deoxyadenosine 3 (cid:42) -monophos-phate, indicating that a cAMP-dependent pathway of signal transduction was involved. This concept was supported by the complementary finding that u-PA-induced adhesion was greatly promoted by forskolin, cholera toxin, or 8-bromo-cAMP, which by themselves induced little adhesion. Furthermore, similar to many other cAMP-dependent activities, cGMP diminished u-PA-in- duced adhesion. When u-PA or ATF was treated with immobilized carboxypeptidase B, its proadhesive effect was abolished, implicating the involvement of carboxyl-terminal lysine residues (Lys 158 on u-PA and Lys 135 on ATF). Moreover, when a carboxyl-terminal lysine analog was added, the proadhesive effect of carboxypeptidase B-treated u-PA or ATF was restored. In conclusion, the present study indicates that u-PA- or acid analysis, 90% performed as described previously (1) with modifications. cells at 10 6 cells/ml labeled with 1 3 H]thymidine in the medium human trans- forming factor- ,25-dihydroxyvitamin D 3 (52 n M for h. H-labeled by and CO , three times. cells buffer glycerol, 0.2% sodium and (1 ml of lysate was into ml formula-989 and

Urokinase-type plasminogen activator (u-PA) 1 is a well characterized, highly restricted serine protease that converts plas-minogen to plasmin. It contains three domains, an epidermal growth factor (EGF)-like domain, a kringle, and a protease domain. The protease domain catalyzes plasminogen activation and is involved in many biological processes including cell migration, tissue remodeling, ovulation, tumor metastasis, and fibrinolysis. In addition, u-PA has been shown to induce monocyte adhesion (1), to be involved in cell focal adhesion (2), and to promote cell growth (3,4). These latter functions are independent of the protease domain of u-PA. Instead, they are related to the occupation of a u-PA receptor (u-PAR) found on the surface of many cells (1,3), which is mediated by the EGF domain of u-PA (5).
The u-PAR is a glycosylphosphatidylinositol-anchored protein that does not have a transmembrane domain or intracellular kinase domain (6). Therefore, a transmembrane protein partner has been implicated in u-PAR-mediated signal transduction (7-9) but remains to be identified.
It has been postulated that the physical binding of u-PAR (10,11) or u-PA (12) to vitronectin accounts for the proadhesive effect. However, a number of observations are inconsistent with this hypothesis. First, according to the u-PA-vitronectin binding mechanism, an excess of free u-PA should inhibit u-PAinduced cell adhesion (12), but such an inhibition was not found (1,10). Second, u-PA-induced adhesion was found to be timedependent (1), which is not characteristic of physical interaction but rather suggests a biochemical process. Third, when a vitronectin-coated surface instead of plastic was used (1,10), the promotion of cell adhesion by u-PA (3 nM) was reduced from ϳ40to 4-fold (1,10). Moreover, the promotion of vitronectin binding to phorbol myristate acetate-primed U937 cells by u-PA (10 nM) was reported to be only about 2-fold, as was the promotion of soluble u-PAR binding to immobilized vitronectin (11). Since u-PA promotion of monocyte adhesion was more than 50-fold (1), a direct physical interaction between vitronectin and u-PAR/u-PA appears to be responsible for only a minor part of u-PA-induced adhesion. Finally, low temperature was found to inhibit u-PA-induced adhesion, suggesting a metabolic process (10).
In the present study of u-PA-or ATF-induced monocyte adhesion, evidence for cAMP-dependent signal transduction was found. A carboxyl-terminal lysine residue was also found to be required for this process.

EXPERIMENTAL PROCEDURES
Diisopropyl Fluorophosphate (DFP) Treatment of u-PA-Two-chain u-PA was prepared from recombinant single-chain u-PA (Farmitalia, Milan, Italy), as described previously (13) and was incubated with 5 mM DFP (Sigma) at room temperature for 2 h and then exhaustively dialyzed. No amidolytic activity was detected after the DFP treatment, using S2444 (Kabi, Franklin, OH), the chromogenic substrate for u-PA as described previously (14).
DFP-treated u-PA and the ATF (American Diagnostics, Greenwich, CT) were used in all the experiments, thereby avoiding any complication by the anti-adhesive effect of PAI-1 against enzymatic u-PA (1).
Carboxypeptidase B (CPB) Treatment of u-PA and ATF-To remove the carboxyl-terminal lysine of u-PA and ATF, 2 nM u-PA or ATF in 0.1 M HEPES, pH 7.4, 0.1 M NaCl, and 0.25 M ZnCl 2 was incubated with immobilized CPB (60 mg/ml, including the bead weight) (Calbiochem) at 37°C for 6 h. At the end of the reaction, CPB was removed by centrifugation. To test the possibility that CPB beads alone might influence cell adhesion, HEPES buffer was incubated with CPB beads (60 mg/ml) alone for 6 h at 37°C. The supernatant after centrifugation was used as the control for the cell adhesion assay. As described previously (15), a small portion of sample was precipitated with 10% * This study was supported by a grant from Neurex, Inc., Menlo Park, CA. 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 U.S.C. Section 1734 solely to indicate this fact.
trichloroacetic acid, and its supernatant was subjected to amino acid analysis, which showed that more than 90% of the COOH-terminal lysine was removed by the CPB treatment.
The cell adhesion assay was performed as described previously (1) with modifications. U937 cells at 10 6 cells/ml were labeled with 1 Ci/ml [ 3 H]thymidine in the culture medium containing human transforming growth factor-␤1 (1 ng/ml) and 1␣,25-dihydroxyvitamin D 3 (52 nM) for 24 h. The 3 H-labeled cells were collected by centrifugation at 350 ϫ g for 4 min at 4°C. After washing three times with serum-free RPMI 1640, cells were resuspended in RPMI 1640 containing 10% bovine fetal serum at 10 6 cells/ml. Test reagents were added to the cell suspension, and 300 l of cells/well were placed in duplicate in 24-well plates (Costar Corp., Cambridge, MA). After 17 h of incubation (37°C) in 5% CO 2 , the contents of each well were removed using a Pasteur pipette, and wells were gently rinsed with phosphate-buffered saline three times. Adherent cells were lysed in lysis buffer containing 10% glycerol, 0.2% sodium dodecyl sulfate, and 0.2% Triton X-100 (1 ml/well). Each 1 ml of lysate was added into 10 ml of formula-989 high flash point mixture and counted in a liquid scintillation counter.
Ligand Binding Analysis-U937 cells (2 ϫ 10 6 /ml) in RPMI 1640 containing 0.4% bovine serum albumin (Sigma) were incubated with 8 nM 125 I-u-PA in the absence or presence of 100 nM competitors (ATF, CPB-ATF, u-PA, or CPB-u-PA) for 2 h at 4°C. At the end of the incubation, the free ligands were removed by centrifugation, and the cells were washed with phosphate-buffered saline three times. The radioactivity bound to the cells was counted with a ␥ counter. (1), u-PA substantially induced adhesion of U937 cells. The adhesion was found to be essentially abolished by cycloheximide (an inhibitor of protein synthesis) and actinomycin D (an inhibitor of RNA synthesis), suggesting that u-PA-induced monocyte adhesion required gene expression and protein synthesis, which are usually the final events of signal transduction (Fig. 1). Moreover, u-PA-induced adhesion was prevented by 2Ј-deoxyadenosine 3Ј-monophosphate, an inhibitor for adenylyl cyclase, implicating a cAMP-dependent pathway of signal transduction ( Fig. 2A). This was confirmed by the complementary findings that u-PA-induced adhesion was greatly promoted by forskolin (a direct activator of adenylyl cyclase) ( Fig. 2A) or by cholera toxin (an indirect activator for adenylyl cyclase through modification of G-protein) (Fig. 2B). More directly, 8-Br-cAMP was also shown to greatly promote u-PA-induced adhesion (Fig.  2C). Forskolin (up to 150 M), cholera toxin, (16.7 ng/ml), or 8-Br-cAMP (up to 5 mM) alone was shown to induce only a modest adhesion (Fig. 2), indicating that elevation of cAMP is indispensable but not sufficient for monocyte adhesion.

U-PA-induced Monocyte Adhesion and cAMP-dependent Signal Transduction-Consistent with previous reports
The findings indicate that the signal of u-PA binding is amplified by an increase in cAMP by adenylyl cyclase, which triggers the other events involved in monocyte adhesion such as RNA and protein synthesis. The activity of adenylyl cyclase is known to be usually regulated by G-protein. This would suggest that the transmembrane partner protein of u-PAR may interact with G-protein. Similar to many other cAMP-dependent activities, cGMP was found to be an antagonist of cAMP in these experiments, diminishing u-PA-induced adhesion (Fig.  2C), suggesting that u-PA-induced monocyte adhesion was regulated by the opposing effects of cAMP and cGMP.
It is noteworthy that cAMP elevation by u-PA was previously reported to inhibit adhesion in cultured mesangial cells (16,17). This suggests that the function of cAMP-dependent signal transduction by u-PA may be cell type-specific. Furthermore, in the case of mesangial cells in contrast to U937 cells, adhesion was dependent on the enzymatic activity of u-PA (17).
A tyrosine kinase (7) and a serine kinase (8,9) have been reported to be activated by u-PA occupation of u-PAR. However, using their corresponding inhibitors, herbimycin A (2 M) for protein tyrosine kinases and H7 (10 M) for serine kinases, no significant effects were seen in the present study, even at the highest concentrations of the inhibitors possible without

u-PA-induced Monocyte Adhesion 30283
interfering with cell growth. Therefore, u-PA-induced monocyte adhesion appears to be independent of these protein kinases.
Increased protein expression of u-PAR by 2-fold and u-PA by 5-fold in U937 monocytes was observed to be induced by u-PA or ATF (10 nM) (data not shown). This finding was consistent with previous reports that u-PAR expression was up-regulated in endothelial cells by activation of adenylyl cyclase (18) and in LLC-PK1 kidney cells by cAMP elevation (19). The increase in u-PAR or u-PA expression was probably insignificant (maximally 2-fold) as an explanation for the induced adhesion based on an increase in physical binding sites. However, it could have a significant effect if it represents a positive feedback mechanism in u-PA(R)-induced signaling. In the presence of 1 nM exogenous u-PA or ATF, there was little increase of u-PAR and u-PA expression (data not shown), while a Ͼ10-fold increase in adhesion was seen (Fig. 1). This finding suggests that other cell surface proteins (probably some adhesive integrins) were responsible for the induced adhesion and that they were synthesized or activated as a consequence of u-PA(R) signaling.
u-PA-induced Monocyte Adhesion Was Abolished by CPB Treatment-The u-PA-induced monocyte adhesion was abolished when u-PA was replaced by CPB-pretreated u-PA (Fig.  3A). Carboxyl-terminal analysis showed that CPB treatment of u-PA resulted in removal of a carboxyl-terminal lysine. Based on the amino acid sequence of two-chain u-PA (20), this was the carboxyl-terminal lysine 158 of the A-chain, implicating this residue was critically involved in u-PA-induced monocyte adhesion.
ATF-induced Monocyte Adhesion Was Abolished by CPB Treatment-The ATF (residues 6 -135), consisting of the EGF and kringle domains, was previously shown to induce comparable U937 cell adhesion as u-PA (1). Since the ATF also has a carboxyl-terminal lysine residue (Lys 135 ) (20), the effect of CPB treatment on ATF-induced cell adhesion was evaluated. A dose-dependent proadhesive effect by ATF was seen, as previously reported (1), which was abolished by CPB pretreatment of the ATF (Fig. 3B).
Receptor Binding of ATF or U-PA Was Unaffected by CPB Treatment-Since the proadhesive effect of u-PA or ATF is dependent on receptor binding, the effect of CPB treatment on this property was evaluated. Receptor binding was measured as competition by u-PA, CPB-u-PA, ATF, and CPB-ATF (100 nM) of 125 I-u-PA (8 nM) binding to U937 cells. The CPB treatment was found not to affect receptor binding, since competition was unaffected by this treatment (Fig. 4), indicating that this was not the mechanism for the CPB effect.
Carboxyl-terminal Lysine Involvement Was Independent of cAMP Elevation-The loss of proadhesive activity by CPB treatment was not restored when adenylyl cyclase activated by forskolin (100 M) or 8-Br-cAMP (up to 2 mM) was added into the culture medium, indicating that the role of the carboxylterminal lysine in cell adhesion was independent of cAMP elevation. However, the proadhesive effect of CPB-pretreated u-PA was essentially restored by a carboxyl-terminal lysine analog, EACA (Fig. 5). This was not due to an independent effect of EACA, since no significant adhesion was obtained by EACA alone (0.4 mM) or by EACA plus 8-Br-cAMP (2 mM). These findings indicated that u-PA-induced monocyte adhesion requires a carboxyl-terminal lysine in addition to u-PAR occupation and cAMP elevation.
In conclusion, the present study indicates that u-PA-induced monocyte adhesion involves cAMP-dependent signal transduction and new protein synthesis, which are triggered by u-PAR u-PA-induced Monocyte Adhesion 30284 occupancy. The presence of a carboxyl-terminal lysine residue was found to be additionally required. Therefore, it is postulated that when u-PA or ATF binds to u-PAR, its transmembrane protein partner is activated, which may interact with G-protein-activating adenylyl cyclase, initiating cAMP-dependent signal transduction and new protein synthesis. The resulting cell adhesion probably occurs mainly due to expression of adhesive integrins. The role of a carboxyl-terminal lysine in this sequence of reactions remains to be determined. However, since kringle domains are well known to interact with carboxylterminal lysines in the fibrinolysis-related proteins, like plasminogen and plasminogen activators, it may be speculated that the carboxyl-terminal lysine involved in cell adhesion also interacts with a kringle-carrying protein. Two families of transmembrane proteins that are tyrosine kinases have been found to have one or more kringle domains (21,22). Kringle domains are not common among transmembrane proteins.