Protein Kinase C-dependent in VivoPhosphorylation of Prourokinase Leads to the Formation of a Receptor Competitive Antagonist*

We recently reported that in vivophosphorylation of urokinase-type plasminogen activator on Ser138/303 prevents its catalytic-independent ability to promote myelomonocytic cell adherence and motility. We now show that Ca2+ activated, phospholipid-dependent protein kinase C from rat brain phosphorylates in vitro a peptide corresponding to prourokinase residues 133–143 (DGKKPSSPPEE) and the full-length molecule on Ser138/139. The in vivoinvolvement of the protein kinase C isoenzyme family is supported by the finding that inhibition of kinase C activity prevents prourokinase phosphorylation on Ser138/303 in A431 human carcinoma cells. Conversely, a short treatment of A431 cells with phorbol myristate acetate increases the extent of phosphorylated prourokinase and, concomitantly, affects its function; under these conditions, the capability of prourokinase to up-regulate U937 monocyte-like cell adherence is severely impaired, although receptor binding is unaltered. By the aid of a “phosphorylation-like” variant (Ser138to Glu) we show that modification of Ser138 is sufficient to confer to prourokinase the antagonistic properties observed following in vivo stimulation of protein kinase C activity. These observations provide the first evidence that protein kinase C directs the formation of a receptor competitive antagonist by regulating the in vivo phosphorylation state of prourokinase.

Urokinase-type plasminogen activator (uPA) 1 catalyzes the conversion of plasminogen to active plasmin, a trypsin-like enzyme responsible for the lysis of fibrin and the degradation of many extracellular matrix components (1). This uPA-dependent proteolytic cascade is regulated by a complex network of interactions between specific domains of the protease and other macromolecules, such as plasminogen activator inhibitors (PAI-1, PAI-2 and others), a specific receptor (uPAR) and extracellullar matrix components (vitronectin) (2)(3)(4)(5)(6)(7).
Urokinase consists of an NH 2 -terminal region sharing considerable homology with epidermal growth factor, a central kringle and a short proteolytically sensitive region which precedes a large carboxyl-terminal catalytic domain (8,9). Whether soluble or receptor-bound, prourokinase (pro-uPA) is a zymogen that undergoes extracellular activation via the cleavage of the Lys 158 -Ile 159 bond, thereby yielding a two-chain active urokinase capable of reacting with the inhibitors (10,11). Recent evidence indicates that receptor-bound pro-uPA may fulfill additional functions besides those strictly dependent on its catalytic activity. Ligation of uPAR with uPA or with its noncatalytic amino-terminal fragment ATF (amino acids 1-135) stimulates intracellular biochemical pathways leading to a cellular response, which may involve changes in gene expression, protein phosphorylation, adhesion, migration, and metabolism. In particular, ligand-dependent uPAR activation leads to increased motility or adherence of myelomonocytic cell lines (12,13).
Steadily increasing evidence indicates that pro-uPA and uPAR syntheses are subjected to spatial and temporal regulation by oncogene activation, hormones, growth factors, and tumor promoters (14 -16). In particular, protein kinase C (PKC), which belongs to an ubiquitous family of key regulatory isoenzymes in cell growth, differentiation, adhesion, carcinogenesis, and metastasis, regulates the uPA system at multiple levels (17,18). It is known that PKC activation induces pro-uPA mRNA synthesis through a composite polyoma enhancer activator 3/activator protein 1 site located about 2 kilobase pairs upstream of the transcription initiation site in a variety of cell types, such as macrophages, keratinocytes, endothelial cells, and neurons, suggesting a highly conserved mechanism (19). Also, phorbol 12-myristate 13-acetate (PMA) stimulates uPAR synthesis in the U937 monocytic cell line and in migrating keratinocytes of wounded cultures (3,20,21). Taken together, these observations point to a complex network of interactions that link in vivo protein kinase C activation and regulation of urokinase function and localization.
Urokinase function is also subjected to post-translational control; we found that the human proenzyme undergoes intra-cellular serine phosphorylation in A431 human carcinoma cells, resulting in a reduction of its sensitivity to the inhibitor PAI-1 (22)(23)(24). According to other reports, phosphorylated urinary urokinase activates plasminogen with a greater catalytic efficiency and is neither inhibited by PAI-1 nor by PAI-2 (25). Receptor-bound urokinase is phosphorylated on tyrosine and serine residues in a human metastatic carcinomatous cell line (26). Others have shown that urokinase from human urine and from HT1080 fibrosarcoma cells contains phosphotyrosine residues, although no functional effects have been reported as yet (27,28). Recent data from this laboratory indicate that two phosphorylation sites are located within the A and B chains of pro-uPA (Ser 138/303 ) from A431 human carcinoma cells and that pro-uPA phosphorylation renders the protease unable to activate uPAR-dependent signaling in myeloid cells (29). Interestingly, the nonsignaling serine phosphorylated pro-uPA binds to uPAR with unaltered affinity, such as a naturally occurring competitive antagonist (29). In this work, we attempted to shed light on the generation of such a molecule by analyzing the kinase pathway that directs pro-uPA phosphorylation on Ser 138/303 in vivo. First, we report that protein kinase C from rat brain is able to directly modify Ser 138 and/or Ser 139 in vitro. Second, we show that in vivo modulation of protein kinase C activity regulates the extent of pro-uPA phosphorylation on Ser 138/303 in the A431-P1 cell line. Finally, we present evidence that PKC-dependent in vivo phosphorylation of pro-uPA, as well as the replacement of Ser 138 with a glutamic acid residue, markedly inhibit pro-uPA signaling ability, yet do not alter receptor binding.
Cell Culture and Labeling Conditions-The A431-P1 cell line is a stable clone of A431 cells harboring pRSV-uPA, a plasmid containing the human uPA gene driven by the Rous sarcoma virus promoter (23). Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) in a 5% CO 2 atmosphere. For metabolic labeling, A431 cells were seeded at a density of 2 ϫ 10 6 /10-cm dish and grown for 24 h in 10 ml of DMEM with 5% FBS. Then, the culture medium was removed and substituted with phosphate-free or methionine-free DMEM containing 5% dialyzed FBS. After 6 h, the medium was replaced with 2.5 ml of either phosphate-free DMEM containing 600 Ci of [ 32 P]orthophosphate or methionine-free DMEM containing 250 Ci of [ 35 S]methionine/cysteine, and labeling was allowed to proceed for the indicated time periods. Treatment of cells with PMA or with the kinase inhibitors (H-7, bisindolylmaleimide, myristoylated PKC (19 -27) peptide, PD-98059, and SB-203580) was performed as specified in the figure legends.
Affinity Purification of Pro-uPA-Pro-uPA was purified by immunoaffinity chromatography of the A431-P1 cell conditioned medium with 5B4 monoclonal antibody, according to a previously described procedure with minor modifications (4,30). Purification of histidine-tagged mutant pro-uPA (His-pro-uPA 138E ) was performed from the conditioned medium of HeLa-stable transfectants, as described previously (29). In both cases, degradation and dephosphorylation was inhibited by including in all buffers the following inhibitors: 20 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 100 mM NaF, 10 mM orthovanadate, 0.5 M NaCl, 0.01% Tween 20. Immunoaffinity purified pro-uPA was quantitated by enzyme-linked immunosorbent assay and analyzed by 12.5% polyacrylamide gel electrophoresis under reducing conditions followed by autoradiography (31).
Purification of Protein Kinase C and in Vitro Phosphorylation Assay-Protein kinase C was purified from rat brain about 200-fold to a specific activity of 1,200 units/mg (32). 15 g of pro-uPA or related proteins were phosphorylated with 2 units of PKC in 50 mM Tris, 2 mM EGTA, 7 mM MgCl 2 , 4 mM CaCl 2 , 0.1 mM dithiothreitol, 150 g/ml phosphatidylserine, 200 ng/ml PMA, 1 M ATP, and 200 Ci of [␥-32 P]ATP/ml for 30 min at 37°C (33). The in vitro phosphorylation assay of synthetic peptides was carried out for 3 h. Protein kinase C from Promega (Madison, MO) or from Boehringer Mannheim were employed under the same conditions with similar results.
Reduction, Carboxymethylation and Enzymatic Fragmentation of in Vitro Phosphorylated 32 P-pro-uPA-600 g of in vitro phosphorylated recombinant 32 P-pro-uPA were treated with 15 g/ml plasminogen (containing about 10% plasmin) for 5 h at 37°C and precipitated with trichloroacetic acid. 32 P-pro-uPA was resuspended at 2 mg/ml in 6 M guanidine HCl, 0.1 M Tris-HCl, pH 8.5, and incubated with an excess of 5 M dithiothreitol over the total number of cysteines for 4 h at 40°C under N 2 flux. Carboxymethylation was performed with 15 M iodoacetic acid for 1 h at 4°C under N 2 flux. The samples were then desalted on a C8 RP-HPLC column with a 3-cm RP-HPLC Guard Holder pre-column using a 60-min linear 0 -100% acetonitrile gradient in 0.1% trifluoracetic acid (flow rate 1 ml/min); absorbance of the eluted fractions was monitored at 280 nm, and radioactivity was determined by measuring Cerenkov radiation. Radioactive fractions were subjected to sequencing.
Cleavage of 32 P-Pro-uPA with Plasmin-In vitro 32 P-phosphorylated pro-uPA was recovered by centrifugation following precipitation with trichloroacetic acid and subjected to enzymatic digestion with plasmin for 30 min at 37°C, in the presence of 1.4 g/ml plasminogen (containing about 10% plasmin) in the buffer employed for the in vitro phosphorylation reaction. The reaction with plasmin was stopped with 25 g/ml aprotinin, and the products were analyzed by SDS-PAGE, 15% gel, under reducing conditions.
Peptide Synthesis and Purification-Peptides DGKKPSSPPEE and KENSTDYPEWQLK were synthesized on an automatic solid phase peptide synthesizer and purified by RP-HPLC on a Nova Pack C18 column equilibrated with 0.1% trifluoroacetic acid in water using a Waters 501 apparatus. Peptides were eluted using a 60-min linear gradient of 0 -80% acetonitrile in 0.1% trifluoroacetic acid and were detected by absorbance at 215 nm. In vitro phosphorylated 32 P-peptides were purified by RP-HPLC as described above. The radioactivity associated with each fraction was determined on a liquid scintillation spectrometer.
Engineering and Tagging of Pro-uPA-Histidine tagged pro-uPA variant S138E (His-pro-uPA 138E ) was obtained, stably expressed in HeLa cells, and purified as described previously (29).
Adhesion Assays-Exponentially growing U937 cells were diluted to 0.4 ϫ 10 6 cells/ml in RPMI, 10% FBS and treated with 1 ng/ml TGF-␤, 50 nM dihydroxyvitamin D 3 in the presence of 10% FBS for 20 h. Then, 1 ϫ 10 5 cells/sample were incubated in 24-multiwell plates with 0.2 nM of the indicated effectors (unless otherwise specified) for 30 min at 37°C. Nonadherent cells, harvested by pipetting, and adherent cells, removed with 0.05% trypsin, were counted in a hemocytometer. The number of adherent cells is expressed as percentage of the total cell number and represents an average from three different experiments performed in duplicate.
Chemotaxis Assays-The assays were performed using Boyden chambers with 5-m pore size polycarbonate filters coated with collagen type I, according to Resnati et al. (13). Briefly, 1 ϫ 10 5 U937 cells were applied to the upper compartment in serum-free RPMI. Effectors were diluted in culture medium at the indicated concentrations and added to the lower compartment; the chambers were incubated at 37°C for 90 min. Then the filters were removed, fixed, and stained. The cells on the lower side of the filter were counted and reported as a percentage of the basal random migration in the absence of chemoattractant.

RESULTS
In Vitro Phosphorylation of Prourokinase by Protein Kinase C-We have previously reported that pro-uPA undergoes in vivo phosphorylation on Ser 138/303 (29). Analysis of the sequences surrounding the mapped serines suggests that Ser 138 is included in a recognition site for serine/threonine kinases such as mitogen-activated protein kinase kinase and casein kinase II (34). However, purified preparations of these kinases failed to phosphorylate in vitro pro-uPA from E. coli, under standard reaction conditions (not shown). On the contrary, when recombinant pro-uPA was incubated with rat brain PKC in the presence of 1 M ATP, 10 Ci of [␥-32 P]ATP, 150 mg/ml phosphatidylserine, and 200 ng/ml PMA, a specific phosphorylation reaction was observed. As shown in Fig. 1A, prourokinase incorporates [ 32 P]phosphate in the presence of PKC, whereas no phosphorylation occurs in its absence, and a dramatic reduction of the resulting specific activity was observed in the presence of H-7 or bisindolylmaleimide or a myristoylated PKC (19 -27) peptide. As expected, the mitogen-activated protein kinase kinase inhibitor PD-98059 did not inhibit pro-uPA phosphorylation by PKC. The Coomassie staining of the gel revealed equal amounts of recombinant pro-uPA in all samples (Fig. 1B). In control samples, phosphorylation of histone H1 yielded a strongly labeled band of about 35 kDa, whereas bovine serum albumin was not phosphorylated (not shown). Under the same conditions, prourokinase purified from the conditioned medium of the A431-P1 cell line overexpressing human pro-uPA (23) is also modified by PKC (not shown). The susceptibility of A431-P1 pro-uPA to in vitro phosphorylation is not surprising, as we have previously shown that about half of the secreted molecules are not phosphorylated, neither on A nor on B chain (24).
Localization of the in Vitro Phosphorylation Site(s)-To identify the protease domain(s) modified by PKC, in vitro phosphorylated 32 P-recombinant pro-uPA was subjected to limited proteolytic degradation with plasmin, which cleaves the Lys 158 -Ile 159 bond, thereby generating two fragments. Under these conditions, all the radioactivity previously incorporated in the intact 45-kDa protein was retained by the 18-kDa amino-terminal fragment ( Fig. 2A). Accordingly, incubation of preactivated recombinant pro-uPA with PKC results in the exclusive phosphorylation of the 18-kDa fragment ( Fig. 2A). In both cases, the stronger intensity of the 18-kDa fragments compared with the single-chain 45-kDa pro-uPA suggests that the PKC target region may undergo a conformational change following activation. To further restrict the analysis, ⌬125 (35) and a proteolytic fragment comprising the amino acids 1-135 (ATF) were employed as substrates (depicted in Fig. 2C). ⌬125 was highly susceptible to PKC-dependent phosphorylation, whereas ATF was slightly modified either in the presence of ⌬125 or in its absence (Fig. 2B). These results, taken together, strongly suggest that the major phosphorylation site is located between amino acids 135 and 158, as this region is included in ⌬125 and in the A chain of plasmin-cleaved pro-uPA but is not in the ATF. Analysis of the 135-158 region reveals three potential phosphate acceptors, two serine residues at 138 and 139 and a threonine residue at 152 (Fig. 2C). However, amino acid analysis of in vitro phosphorylated recombinant 32 P-pro-uPA exclusively showed the occurrence of 32 P-phosphoserine, suggesting that PKC indeed modifies Ser 138 and/or Ser 139 (not shown).
The latter possibility is in agreement with the results of an experiment in which recombinant pro-uPA was first phosphorylated with PKC in the presence of [␥-32 P]ATP, then extensively digested with plasmin, reduced, and carboxymethylated (see "Experimental Procedures"). The resulting peptides were fractionated by RP-HPLC, and the fractions containing radioactivity were subjected to sequencing. The most abundant radiolabeled peptide was eluted in fraction 114 (Fig. 3) and showed the amino-terminal sequence PSSPPEEL . . ., in agreement with Ser 138 / 139 being the predominant phosphate acceptor.
Confirmatory data were obtained with a peptide corresponding to pro-uPA residues 133-143 (DGKKPSSPPEE), which was included as a substrate in an in vitro phosphorylation reaction with PKC (see "Experimental Procedures"). In Fig. 4A, the position of the substrate peptide within the pro-uPA molecule is depicted. In this case, the 32 P-phosphorylated products were separated onto a HPLC column, and two radioactive fractions were obtained, the first eluted in the void volume of the column, corresponding to non incorporated [␥-32 P]ATP and the second containing the 32 P-phosphorylated peptide (Fig. 4B). Under the same conditions, peptide KENSTDYPEWQLK, which is a substrate for casein kinase II, did not exibit a second peak, indicating that it was not appreciably modified by PKC (not shown).
In Vivo Phosphorylation of Pro-uPA and Dependence on PKC Activity-The activity of protein kinase C can be modulated in vivo by different effectors, which may consequently affect pro-uPA phosphorylation state. This possibility was tested in a set of experiments in which PKC activity was stimulated with 100 ng/ml PMA or down-regulated either by a prolonged treatment with 1 g/ml PMA or with the kinase inhibitor H-7. A431-P1 cells were metabolically labeled either with [ 32 P]orthophosphate to assess pro-uPA phosphorylation level or with [ 35 S]methionine to ensure an internal control of pro-uPA synthesis and secretion (23).
In the first experiment, cells prelabeled for 4 h were further incubated with fresh medium containing either [ 35 S]methionine or [ 32 P]orthophosphate with or without 100 ng/ml PMA. Aliquots of the resulting conditioned media were subjected to quantitative immunoprecipitation with anti-uPA antibody and separated by SDS-PAGE under reducing conditions. This analysis reveals single-chain pro-uPA and two-chain uPA deriving from partial serum-dependent proenzyme activation occurring in culture. As shown in Fig. 5A, after 30 min of incubation with PMA, the level of 35 S-pro-uPA/uPA is unaltered, whereas 32 Plabeled pro-uPA/uPA exhibit a marked increase. Quantitation of the resulting bands by autoradiogram scanning revealed a PKC-dependent 3-fold increase of overall phosphorylation; fur-thermore, the increased phosphorylation of both A and B chain suggests that modification of both Ser 138 and Ser 303 is dependent on PKC activation. In parallel experiments, we found that the addition of PMA does not stimulate the in vivo phosphorylation of His-pro-uPA 138E/303E (histidine-tagged pro-uPA variant in which Ser 138/303 are no longer available), confirming that PKC exclusively modulates phosphorylation of Ser 138/303 (not shown).
Down-regulation of PKC can be achieved by treating the cells with 1 g/ml PMA for 24 h. Therefore, A431 cells were either subjected to a 16-h treatment with 100 ng/ml PMA or to a 24-h treatment with 1 g/ml PMA. Samples of conditioned media from equal cell numbers were subjected to quantitative immunoprecipitation with 5B4 monoclonal antibody (Fig. 5B). Following PMA induction of PKC activity, there is again about a 3-fold increase in pro-uPA phosphorylation (Fig. 5B). On the contrary, pro-uPA phosphorylation is abolished if PKC activity is down-regulated, indicating that phosphorylation of Ser 138/303 is totally PKC-dependent (Fig. 5B). In the third experiment, the cells were incubated with 100 M H-7 or with diluents for 17 h, in the presence of each radiolabeled isotope (Fig. 5C). Furthermore, incubation of A431-P1 cells with bisindolylmaleimide or myristoylated PKC (19 -27) peptide, both specific inhibitors of PKC, results in a strong reduction of pro-uPA phosphorylation level. On the contrary, inhibition of p38 mitogenactivated protein kinase by SB-203580 (not shown) or inhibition of mitogen-activated protein kinase by PD-98059 (Fig. 5D) did not alter pro-uPA phosphorylation.
In keeping with our previous results, PKC inhibitors do not alter pro-uPA secretion but they cause a strong reduction of the overall phosphorylation level, confirming that protein kinase C activity is responsible for most of pro-urokinase phosphoryla- Effects of PKC Activity Modulation on Receptor Binding and Signaling Ability of A431 Pro-uPA-Previous functional analysis of pro-uPA phosphorylated on Ser 138/303 revealed a severe impairment of its catalytic-independent ability to promote myelomonocytic motility and adherence (29). The above demonstrated PKC-dependence of such a modification leads to the prediction that in vivo direct stimulation of PKC activity should result in a reduction of pro-uPA signaling ability. Considering a PMA-dependent 3-fold increase in the overall phosphorylation level, we expect that all secreted pro-uPA will at least bear one phosphate group/molecule. To test the effect of PKC activation on pro-uPA function, subconfluent A431-P1 cells were either left unstimulated or subjected to a 16-h treatment with 100 ng/ml PMA. The protease was then purified from the cell-conditioned medium, quantitated, and tested in a receptor competition assay for binding of 125 I-ATF to U937 monocyte-like cells. As shown in Fig. 6A, both pro-uPA preparations exhibit comparable K d values for uPAR. This finding is in agreement with our previous results obtained with phosphorylated and unphosphorylated pro-uPA purified from A431-P1conditioned medium by Fe 3ϩ -chelated chromatography (29). The effect of PKC-dependent pro-uPA phosphorylation on uPAR-mediated signaling was tested in a U937 monocyte-like cell adhesion assay. These cells were "primed" with TGF-␤/ vitamin D 3 for 20 h and then subjected to an adhesion assay, in the presence of 10 nM pro-uPA purified from A431-P1 cells, either stimulated with 100 ng/ml PMA for 16 h or left unstimulated. In this experiment, prourokinase from untreated A431-P1 cells causes about 40% of the total cells to adhere to the culture dish. On the contrary, pro-uPA from PMA-treated A431-P1 cells is markedly hampered in its proadhesive ability, over a wide concentration range (Fig. 6B). Interestingly, a "phosphorylation-like" histidine-tagged pro-uPA variant, carrying a Ser to Glu substitution, which is expected to mimic the functional consequences of phosphorylation on Ser 138 , is similarly impaired at inducing cell adherence, up to 10 nM, although it binds to uPAR (Fig. 6, A and B). This result suggests that, although PKC regulates the overall phosphorylation level, modification of Ser 138 is mainly responsible for the dramatic inhibition observed. Confirmatory results were obtained in a U937 cell migration assay, employing pro-uPA as a chemoattractant; we found that pro-uPA from PMA-treated cells retains only 20 -30% of the chemotactic ability of pro-uPA from uninduced A431-P1 cells (not shown). These data indicate that protein kinase C activity down-regulates pro-uPA chemotactic and proadhesive ability without affecting uPAR binding. In addition, they show that a single amino acid substitution in pro-uPA mimics the functional effects observed following in vivo stimulation of PKC activity, therefore emphasizing the central regulatory role of PKC-dependent phosphorylation on Ser 138 . DISCUSSION The results presented in this study assign a role to PKC isoenzyme family in the in vivo regulation of prourokinase phosphorylation on Ser 138/303 . The modification of Ser 138 , in particular, results in a severe impairment of receptor-depend- ent pro-uPA ability to promote myeloid cell adherence, although it does not alter receptor binding. The latter site, which lies outside of the receptor binding domain, is a target of in vitro phosphorylation by rat brain protein kinase C. Analysis of the sequence surrounding Ser 138/139 reveals that, in agreement with the specificity requirements of PKC, basic residues are located near the serine phosphate acceptor site (34). On the other hand, despite consensus sites for casein kinase II and mitogen-activated protein kinase, these two kinases failed to appreciably modify recombinant pro-uPA in vitro (not shown). These data, together with the increased susceptibility of twochain uPA to phosphorylation by PKC as compared with the single-chain uPA, suggest that accessibility and secondary structure of the region surrounding Ser 138/139 may be relevant to PKC recognition. The lack of signaling by the pro-uPA variants carrying Glu 138 indicates that the aforementioned region, designated "connecting peptide" and including the zymogen activation site, although not required for binding, is a potential inhibitor of uPA signaling function by a PKC-mediated mechanism. Interestingly, secondary structure prediction, according to the Garnier et al. (36) method, generates a four-turns model for the 132-158 region, which is completely disrupted by the introduction of a negatively charged amino acid at 138. 2 The relevance of this region is also suggested by the finding that removal of Lys 135 or Lys 158 with carboxypeptidase A impairs the uPA proadhesive effect, yet does not affect receptor binding (37). Therefore, the possibility exists that the uPA-dependent signal can be delivered by a transient association of activated uPAR with a transmembrane receptor, which is modulated by a balance between negatively and positively charged amino acids at specific positions in the 135-158 region.
In the in vitro experiments, no appreciable phosphorylation of Ser 303 by PKC takes place, suggesting several interpretations. It is possible that the conformation attained by the mature protein, either recombinant or secreted from A431 cells may not totally resemble that of intracellular pro-uPA (22). Alternatively, an unknown kinase may be responsible for the modification of Ser 303 in vivo, provided its activity is regulated by PKC. We cannot distinguish between these two possibilities, at the moment. Conversely, phosphorylation of Ser 138 , which plays a central role in the modulation of pro-uPA signaling ability, is totally PKC-dependent both in vitro and in vivo. These data support the hypothesis that PKC may directly modify Ser 138 in vivo. In agreement with this possibility, a direct role of pp60 Src and protein kinase C in pro-uPA phosphorylation by metastatic and tumor cells has been suggested by other authors (26). On the other hand, we have previously shown that pro-uPA serine phosphorylation does not occur in cell culture medium, but inside A431 cells (22). The analysis of these data poses an interesting topological dilemma, as most of the known PKC isoforms are cytosolic and pro-uPA is a secreted zymogen. However, extensive studies indicate localization of PKC in a variety of intracellular compartments different from the plasma membrane, including the Golgi complex, the perinucleus, the cytoskeleton, and the focal contacts of rat embryo fibroblasts (38). Among the atypical isoforms of PKC is the form, belonging to the subgroup, which has been detected over the luminal surfaces of acinar cells in pancreatic cells (39). Interestingly, an ecto-protein kinase with a catalytic specificity similar to PKC has been recently discovered on the cell surface of brain neurons (40). Given the growing information about the structure and physiology of PKC-related and PKC-like enzymes, we cannot exclude that a particular type will phosphorylate secreted proteins. Interestingly, pro-uPA susceptibility to in vitro phosphorylation by PKC is shared by other secreted proteins, such as the platelet coagulation factor Va and matrix vitronectin (41,42).
In any event, our results provide an important step toward understanding the formation and the occurrence of natural receptor antagonists, which may be shared by other ligandreceptor systems. In this study, we have shown that protein 2 C. Iaccarino and M. P. Stoppelli, unpublished results. kinase C in vivo dictates the conversion of an agonist into a receptor-competitive antagonist, which blocks the signaling by nonphosphorylated pro-uPA. This functionally resembles the case of IL-1␣ and IL-1␤, the unique cytokines for which a naturally occurring antagonist is known (IL-1RA); crystal structure of the IL-1RA/receptor complex has shown that, unlike the agonists, the "receptor trigger site" of the antagonist is not in direct contact with the receptor (43). In our system, the di-substituted variant is unable to mobilize the receptor, which may be due to a peculiar binding mode of phosphorylated pro-uPA (29). The possibility exsists that negatively charged residues at position 138 may confer to the ligand the capability to interact with uPAR domains D2 and/or D3 or with a negative regulator of cell adhesion. Alternatively, the negative effect may simply be due to the lack of a productive interaction between uPAR and an integrin-type receptor. Some interpretations are suggested by the analysis of intracellular signaling inhibitors which prevent the association of the TGF-␤ receptors (T␤R-I and T␤R-II), therefore providing a safeguard against leaky signaling (44,45). A model depicting the way agonists and antagonists may cooperate to control function is offered by Drosophila, in which the epidermal growth factor-like negative regulator Argos binds to the mammalian epidermal growth factor receptor homologue (Drosophila EGF receptor or DER), thereby preventing its activation by Spitz. In this way, the DER pathway is regulated by a balance between extracellular activating and inhibiting signals (46). The results presented here suggest that ligand-dependent uPAR signaling in vivo may result from a PKC-regulated ratio of phosphorylated versus nonphosphorylated pro-uPA forms. In addition, we have shown that phosphorylation of Ser 138 can be functionally mimicked by a Ser to Glu substitution in the protease, offering the unique possibility of selectively blocking the uPAR-dependent pathway in vivo.