Soluble urokinase-type plasminogen activator receptor inhibits cancer cell growth and invasion by direct urokinase-independent effects on cell signaling.

The urokinase-type plasminogen activator receptor (uPAR) is released from human cancers and is readily detected in blood. In animal models, soluble uPAR (SuPAR) antagonizes cancer progression; however, the mechanism by which SuPAR functions in vivo remains unclear. It is generally thought that SuPAR scavenges uPA and prevents its interaction with membrane-anchored uPAR. In this study, we demonstrate a novel molecular mechanism by which SuPAR may inhibit cancer progression. We show that SuPAR has the potential to directly and in a uPA-independent manner block the signaling activity of membrane-anchored uPAR. Whether SuPAR inhibits signaling is cell type-specific, depending on the state of the endogenous uPA-uPAR signaling system. In uPAR-deficient cells that lack endogenous uPAR signaling, including uPAR-/-murine embryonic fibroblasts and human embryonal kidney 293 cells, SuPAR functions as a partial signaling agonist that activates ERK/mitogen-activated protein kinase. By contrast, in cells with potent autocrine uPA-uPAR signaling systems, including MDA-MB 231 breast cancer cells and low density lipoprotein receptor-related protein-1-deficient murine embryonic fibroblasts, SuPAR substantially decreases ERK activation. The mechanism probably involves competitive displacement of membrane-anchored uPAR-uPA complex from signaling adaptor proteins. As a result of its effects on cell signaling, SuPAR blocks cell growth and inhibits cellular invasion of Matrigel. Cleavage of SuPAR by proteinases increases its signaling agonist activity and reverses its inhibitory effects on growth and invasion. Thus, proteolytic cleavage represents a molecular switch that neutralizes the anticancer activity of SuPAR.

The multifunctional, glycosylphosphatidylinositol-anchored membrane protein uPAR 1 is a receptor for both urokinase-type plasminogen activator (uPA) and vitronectin that regulates diverse aspects of cell physiology, including cell growth, apoptosis, cell adhesion, and migration (1,2). uPAR-associated uPA readily activates plasminogen and thereby initiates a cell surface proteinase cascade, which culminates in metalloproteinase activation and degradation of extracellular matrix proteins, particularly in association with leading lamellipodia in migrating cells (3)(4)(5). uPAR laterally associates with integrins within the same plasma membrane and may regulate the state of integrin activation (6 -9). Furthermore, uPAR activates at least two distinct cell signaling responses that are controlled by its two ligands.
uPA binding to uPAR activates a number of pathways, including the Ras-extracellular signal-regulated kinase (ERK) pathway, that control cancer cell migration, growth, and invasion (10,11). Vitronectin binding to uPAR activates the small GTPase, Rac1 (12,13). Because uPAR is glycosylphosphatidylinositol-anchored, adaptor proteins are necessary to transmit signaling responses, and in this capacity, multiple proteins have been implicated, including integrins, the epidermal growth factor receptor, FPR-like receptor-1/lipoxin A4 receptor, caveolin, and gp130 (11, 14 -19). We have presented evidence that the multiprotein uPAR signaling receptor complex is dynamic in composition (19). The function of uPA may be to alter the conformation of uPAR so that association with transmembrane adaptor proteins is favored (20,21).
Mature uPAR consists of three homologous domains joined by flexible linker sequences. The principal uPA-binding sequence is located in the most N-terminal domain (D1); however, secondary binding sites are contributed by D2 and D3 (22)(23)(24). D2 contains vitronectin recognition sequences, which may be supported by sequences in the other domains (25)(26)(27). Intact uPAR is released from the plasma membrane by glycosylphosphatidylinositol-specific phospholipase D (28). Soluble uPAR (SuPAR) also may be generated by alternative splicing of the uPAR mRNA (29). Resnati et al. (30) showed that SuPAR functions as a chemoattractant after it is cleaved between D1 and D2 ϩ D3 by any of a number of proteinases, including uPA, plasmin, or chymotrypsin. Cleavage exposes the active epitope, SRSRY, which remains intact and associated with D2, depending on the actual peptide bond that is cleaved. Cleaved SuPAR (CSuPAR) activates the Src family member hck and induces changes in the cytoskeleton and relocation of integrins to the leading edge of cellular migration, mimicking the responses observed when cells are treated with uPA (30 -32). We reported that SuPAR promotes MCF-7 cell migration, again mimicking the response observed with uPA (33). Similarly, Aguire Ghiso et al. (11) reported that SuPAR activates ERK in uPAR antisense RNA-expressing epidermoid carcinoma cells. In the MCF-7 and epidermoid carcinoma cell systems, SuPAR cleavage apparently was not necessary.
In breast cancer and other forms of malignancy, high levels of uPA and uPAR have been correlated with cancer progression and a poor prognosis (5,34). In humans, SuPAR is normally found in the plasma; however, the level of SuPAR may be dramatically increased in patients with various forms of malignancy, including ovarian cancer, non-small cell lung cancer, leukemia, colon cancer, and breast cancer (35)(36)(37)(38)(39)(40)(41)(42)(43). These studies provide compelling evidence that soluble uPAR is present in the microenvironment of malignant cells. In xenograft animal model experiments, SuPAR reduces the growth and metastasis of MDA-MB 231 breast cancer cells and OV-MZ-6#8 ovarian cancer cells (44,45). The mechanism was assumed to involve SuPAR functioning as a scavenger for uPA, like a high affinity antibody, because alternative mechanisms by which SuPAR may counteract cancer progression had not been identified (44 -46).
The purpose of the present investigation was to test the hypothesis that SuPAR inhibits cancer progression by directly regulating cell signaling, as opposed to "uPA scavenging." To test this hypothesis, we studied the effects of human SuPAR on murine embryonic fibroblasts (MEFs). Human SuPAR does not bind murine uPA, eliminating uPA scavenging as a possible mechanism (47). Our results demonstrate that SuPAR may activate or inhibit ERK phosphorylation, depending on the state of the autocrine uPA-uPAR signaling system. In cells that have a highly activated autocrine signaling system, including low density lipoprotein receptor-related protein (LRP-1)-deficient MEFs and MDA-MB 231 breast cancer cells, SuPAR inhibits ERK activation, cell growth, and Matrigel TM invasion. The most likely mechanism involves competition with membrane-anchored uPAR-uPA complex for signaling adaptor proteins. Proteinase cleavage serves as a molecular switch, converting SuPAR from a partial agonist into a full agonist and thereby eliminating its function as an anticancer agent.

MATERIALS AND METHODS
Reagents and Antibodies-Recombinant human SuPAR was from R & D Systems, Inc. (Minneapolis, MN). CSuPAR was prepared by treating SuPAR with 2.0 nM chymotrypsin for 7 h at 37°C. The chymotrypsin was inactivated with 1.0 mM PMSF, as described (22). Chymotrypsin, poly-L-lysine, sodium orthovanadate, and protease inhibitor mixture were from Sigma-Aldrich. Antibody that specifically detects phosphorylated ERK1 and ERK2 was from Cell Signaling Technologies (Beverly, MA). Polyclonal antibody that recognizes total ERK was from Zymed Laboratories Inc. (San Francisco, CA). Horseradish peroxidaseconjugated antibodies specific for rabbit IgG was from Amersham Biosciences.
ERK Phosphorylation-MEF-2, A1 MEFs, and MDA-MB 231 cells were plated in 60-mm dishes and cultured in serum-free medium. After 18 h, the cultures were washed and then treated with recombinant human SuPAR or CSuPAR, as indicated. The cells were extracted in 1% (v/v) Nonidet P-40, 10 mM HEPES, 150 mM NaCl, 2 mM EDTA, pH 7.5, containing protease inhibitor mixture and sodium orthovanadate (1 mM). The protein concentration in each extract was determined by bicinchoninic acid assay (Sigma-Aldrich). Equal amounts of cell extracts were subjected to SDS-PAGE on 10% slabs, electrotransferred to polyvinylidene difluoride membranes, and probed with specific antibodies for phosphorylated ERK and total ERK. Control experiments were performed in which PMSF alone or PMSF-treated chymotrypsin was added to cultures. Changes in ERK phosphorylation were not observed.
Cell Growth-Cell growth was measured by MTT assay, using the Cell Proliferation Kit I (Roche Applied Sciences). The cells were plated in 96-well plates and cultured for 18 h in serum-containing medium. The cultures were then washed with serum-free medium and treated with SuPAR, CSuPAR, or vehicle in serum-free medium for 48 h. MTT hydrolysis was determined, as directed by the manufacturer and detected on the basis of the absorbance at 570 nm. In each experiment, parallel cultures were analyzed prior to the 48-h incubation with SuPAR, CSuPAR, or vehicle. MTT hydrolysis is proportional to the number of viable cells in a culture.
Matrigel™ Invasion-Matrigel TM invasion chambers with polymerized growth factor-reduced Matrigel TM were purchased from BD Biosciences. MDA-MB 231 cells were plated in vitronectin-coated wells and cultured in serum-free L-15 medium for 18 h. The cells were then treated with SuPAR, CSuPAR, or vehicle as indicated, released from monolayer culture, and transferred to the upper wells of the Invasion Chambers. SuPAR or CSuPAR were added to both chambers when the cells had been pretreated with either reagent. The lower chamber also contained 10% fetal bovine serum. The cells were allowed to invade for 24 h, at which time the Matrigel TM and cells that were associated with the top surfaces of the membranes were removed using cotton swabs. Cells that penetrated through the Matrigel TM to the underside surfaces of the membranes were fixed and stained with 0.1% crystal violet, as previously described (50). The dye was eluted with 10% acetic acid, and the absorbance at 570 nm was determined.

Regulation of ERK Activation by
SuPAR-There is considerable evidence that SuPAR interacts directly with transmembrane adaptor proteins to elicit cellular responses that are equivalent to those caused by membrane-associated uPAR-uPA complex (17,30,51). To eliminate the possibility that SuPAR scavenges endogenously produced uPA and thereby indirectly affects cell signaling, we took advantage of the strict species specificity in uPA action (47) and studied the response of MEFs to human SuPAR. A1 MEFs are uPAR-deficient murine cells, prepared from uPAR gene knock-out embryos (13). Because uPAR is absent, these cells cannot establish autocrine uPA-uPAR signaling. When A1 cells were cultured in serum-free medium for 18 h, the basal level of ERK phosphorylation was very low, irrespective of whether the cells were plated on vitronectin, fibronectin, type-1 collagen, or poly-L-lysine (Fig.  1). Treatment with 10 nM human SuPAR for 2 min activated MEF-2 cells are murine and LRP-1-deficient. These cells have increased levels of uPA and uPAR and establish autocrine signaling in serum-free medium so that the basal level of activated ERK is elevated (13,52). Similarly, MDA-MB 231 breast cancer cells express large amounts of uPA and uPAR, which form an autocrine signaling loop to maintain an elevated level of activated ERK under serum-free conditions (53). In both of these cells lines, SuPAR (10 nM, 2 min) actually decreased ERK phosphorylation, and this result was extracellular matrix protein-independent. One explanation for these results is that SuPAR scavenges uPA and prevents uPA binding to membrane-anchored uPAR. However, in experiments with MEF-2 cells, human SuPAR cannot scavenge murine uPA, eliminating this mechanism in favor of a model in which SuPAR suppresses ERK activation by direct interaction with the cells.
The Effects of SuPAR on ERK Activation Are Sustained-ERK activation by uPA may be highly transient. In MCF-7 breast cancer cells, the response is sustained for 5 min or less; however, this duration of response is sufficient to induce longlived effects on cell migration (10). Cell growth and the rate of apoptosis are not affected by uPA in MCF-7 cells, consistent with the work of others demonstrating the requirement for sustained ERK phosphorylation for cells to enter the S phase (54,55). Fig. 2A shows that ERK activation by SuPAR in uPAR-deficient A1 MEFs is sustained through 60 min. As a second cell culture model system in which the cells do not express uPAR, we studied HEK 293 cells (17). Again, SuPAR activated ERK, and the response was sustained. Suppression of ERK activation by SuPAR, in LRP-1-deficient MEF-2 cells, was also sustained, without apparent change through 60 min (the time course of the complete experiment).
We also explored the SuPAR concentration dependence of ERK activation and deactivation. SuPAR, at a concentration of 10 pM, consistently activated ERK in A1 MEFs (Fig. 2B). Maximum ERK activation was observed with SuPAR at concentrations from 0.1 to 10 nM. Similarly, in experiments with MEF-2 cells, inhibition of ERK activation was consistently observed with 10 pM SuPAR. These results indicate that fairly low concentrations of SuPAR are sufficient to directly regulate cell signaling.
Because SuPAR induced sustained changes in ERK activation, we undertook experiments to examine the potential of SuPAR to regulate cell growth. We measured cell growth using the MTT assay, which determines the total number of viable cells. As shown in Fig. 3, SuPAR increased A1 MEF proliferation by 2.2-fold in 48 h (p Ͻ 0.05, n ϭ 8). By contrast, in MEF-2 and MDA-MB 231 cells, SuPAR essentially blocked cell growth over the same time period. Thus, the effects of SuPAR on cell growth correlate with its effects on the level of ERK activation. SuPAR may also have promoted apoptosis in MEF-2 and MDA-MB 231 cells. This effect would have been indistinguishable from growth inhibition by MTT assay.
SuPAR Cleavage by Proteinase Reverses Its Inhibitory Activity-Chymotrypsin cleaves SuPAR in the linker region between D1 and D2 ϩ D3, revealing the sequence, SRSRY, as the new N terminus of D2. Because of this activity, chymotrypsin serves as a model of proteinases that activate SuPAR for cell signaling (30,31). We performed experiments to determine whether cleavage of SuPAR by chymotrypsin reverses its activity as an inhibitor of ERK activation and cell growth. Fig. 4A shows that chymotrypsin completely cleaved SuPAR, generating new bands with the anticipated mobilities of D1 and D2 ϩ D3.
CSuPAR and SuPAR activated ERK equivalently in uPARϪ/Ϫ A1 MEFs, indicating that the proteinase-cleaved product was active (Fig. 4B). In control experiments, the chymotrypsin, which was used to cleave SuPAR, had no effect on ERK activation in A1 MEFs, when added alone (data not shown). The chymotrypsin was treated with PMSF before incubation with the cells, following the protocol used to prepare CSuPAR.
In MEF-2 cells and MDA-MB 231 cells, SuPAR and CSuPAR yielded different results. After incubation with cells for 2 min, SuPAR decreased ERK phosphorylation as already described; however, in multiple experiments, CSuPAR either had no effect on or slightly increased ERK phosphorylation. Equivalent results were obtained when the time of incubation with CSuPAR was extended (data not shown). Because CSuPAR did not inhibit ERK activation, in MEF-2 and MDA-MB 231 cells, we tested its effects on cell growth. As anticipated, CSuPAR did not inhibit growth of the MEF-2 and MDA-MB 231 cells (Fig.  5). Thus, SuPAR cleavage neutralizes its ability to function as an inhibitor of ERK activation and cell growth.
SuPAR Regulates MDA-MB 231 Cell Invasion-As a model system to test the effects of SuPAR on MDA-MB 231 cell invasion, we used the Matrigel TM invasion assay. SuPAR has already been shown to inhibit cancer progression in animal models (44,45), and our in vitro model system allowed us to compare SuPAR and CSuPAR without potentially confounding variables encountered in vivo, such as differential clearance rates or local proteolysis.
Monolayer cultures of MDA-MB 231 cells were treated with SuPAR or CSuPAR for 18 h. The cells were then dissociated and transferred to invasion chambers. As shown in Fig. 6, SuPAR caused a 30% decrease in invasion, which was highly significant (p Ͻ 0.01, n ϭ 8). By contrast, invasion was unaltered when the cells were treated with CSuPAR. In separate experiments, we pretreated MDA-MB 231 cells with SuPAR or CSuPAR for 6 h, instead of 18 h, before adding the cells to invasion chambers. Equivalent results were obtained. However, when cells were added to invasion chambers without preincubation, SuPAR decreased invasion by less than 5% (data not shown). These results suggest that SuPAR may slowly induce changes in the phenotype of MDA-MB 231 cells, which counteract the ability of these cells to invade. Because SuPAR affected invasion minimally when a preincubation period was not included, it is unlikely that differences in cell growth contributed significantly to the invasion results observed.
ERK activation may affect cell migration, which represents a possible mechanism by which SuPAR may regulate Matrigel TM invasion; however, the relationship between ERK activation and cell migration is conditional, depending on the cell type, integrin expression profile, and migration substratum (14, 56 -59). To determine whether SuPAR affects MDA-MB 231 cell migration, we performed experiments using Transwell migration chambers, precoated with 20% serum, as previously described (58,59). SuPAR or CSuPAR (10 nM) were added to lower chamber or to both chambers. Significant changes in cell migration were not observed. DISCUSSION The prognostic value of uPA as a marker of cancer aggressiveness has been reproducibly documented, especially in breast cancer (34). In fact, uPA levels in breast cancer may have more value than traditional prognostic indices, such as tumor size and grade (60,61). Understanding the mechanism by which uPA promotes cancer progression is an important objective. uPA activates a cell surface proteinase cascade that facilitates cellular penetration of tissue boundaries, such as basement membranes (5); however, there is considerable evidence that this mechanism is not exclusively responsible for the effects of uPA on cancer progression in vivo. For example, in tumor xenograft experiments and in clinical trials, plasminogen activator inhibitor-1 frequently emerges as a cancer promoter, which is inconsistent with the "proteinase cascade model" because plasminogen activator inhibitor-1 is the principal inhibitor of uPA proteolytic activity (61)(62)(63).
It is now recognized that uPA regulates multiple activities associated with membrane-anchored uPAR, which may impact on cancer progression, including vitronectin binding, lateral association with integrins, and cell signaling (2). An important property of uPA is its ability to establish an autocrine signaling pathway with uPAR, which serves as a major determinant of the basal level of activated ERK in cancer cells (11,53). The potency of the autocrine uPA-uPAR signaling pathway probably reflects the fact that low fractional occupancy of uPAR with uPA is sufficient to trigger signaling (64). Ligated uPAR undergoes conformational change (20,21) that by an unclear mechanism alters its interaction with adaptor proteins so that signaling is initiated.
SuPAR is released by cancer cells and is thus available to influence the activity of the uPA-uPAR system. In in vitro experiments, proteolytically cleaved SuPAR directly interacts with uPAR adaptor proteins at the cell surface, triggering cellular responses that are equivalent to those elicited by membrane anchored uPAR-uPA complex (17). All of the previously reported effects of SuPAR (cleaved and uncleaved) on cell signaling and cell physiology would be expected to promote cancer cell aggressiveness. Thus, a paradox emerges because, in vivo, SuPAR antagonizes cancer progression. This paradox provides support for an alternative hypothesis in which SuPAR antagonizes cancer progression in vivo by scavenging uPA and by inhibiting uPA binding to membrane-anchored uPAR (44 -46).
In this report, we present evidence for the first time demonstrating that SuPAR may antagonize cancer progression by direct, uPA-independent effects on cell signaling. Our results support a model in which uncleaved SuPAR functions as a partial agonist that triggers cell signaling but not as effectively as membrane-anchored uPAR-uPA complex. When membraneanchored uPAR-uPA complex is present, the complex competes with SuPAR for common adaptor protein-binding sites, which are necessary for cell signaling. In A1 MEFs and HEK 293 cells, which lack uPAR, competition with SuPAR is not possible and thus uncleaved SuPAR activates ERK. Similarly, in MCF-7 cells, uPA is not expressed at a sufficient level to establish autocrine signaling and, as a result, uncleaved SuPAR promotes cell migration in this cell line, an activity that has been attributed to ERK activation (33). By contrast, in the aggressive MDA-MB 231 breast cancer cell line and in MEF-2 cells, autocrine uPA-uPAR signaling is highly activated (13,53). Because SuPAR is less effective at triggering signaling than membrane-anchored uPAR-uPA complex, competitive displacement of the membrane-anchored complex results in a decrease in the level of activated ERK. Murine uPA does not bind to human uPAR (47), precluding the alternative model in which SuPAR inhibits ERK activation by binding uPA produced endogenously by the MEF-2 cells (52).
Cell growth and Matrigel TM invasion were inhibited by Su-PAR in MDA-MB 231 and MEF-2 cells. These changes in cell physiology have been previously linked to the basal level of ERK activation (65,66). The extent of cell growth inhibition (nearly 100%) was unanticipated. This result may be partially explained if SuPAR promoted apoptosis in MDA-MB 231 and MEF-2 cell, in addition to inhibiting cell proliferation, as has been previously described for uPA-specific antibodies (53). Alternatively, compensatory signaling pathways that support cell The cells that were treated with SuPAR or CSuPAR were maintained in the presence of the same agent in the invasion chamber. The bottom chamber also contained 10% fetal bovine serum. Invasion was allowed to occur for 24 h at 37°C. Cells that penetrated to the underside of the membrane were fixed and stained with Crystal Violet. Cell invasion is expressed as a percentage of that observed with untreated control cells (Con) (mean Ϯ S.E., n ϭ 8). The asterisk indicates a highly significant difference compared with the control (n ϭ 8, p Ͻ 0.01). In cell migration experiments, MDA-MB 231 cells were pretreated with Su-PAR or CSuPAR (10 nM) for 15 min. The cells were then allowed to migrate through serum-coated Transwell membranes for 6 h. Migration of cells to the underside of each membrane is expressed as a percentage of that observed with control cells (Con), which received no treatment (mean Ϯ S.E., n ϭ 4). growth may be down-regulated in MDA-MB 231 and MEF-2 cells, because of the potency of the uPA-uPAR autocrine signaling system.
The linker region between D1 and D2 is sensitive to proteolysis. When SuPAR is cleaved so that the sequence SRSRY (amino acids 88 -92) remains intact, the product activates cell signaling similarly to uPA (30,31). Resnati et al. (17) demonstrated direct binding of a uPAR fragment (amino acids 88 -274), which is equivalent to CSuPAR, to the uPAR adaptor protein, FPR-like receptor-1/lipoxin A4 receptor, expressed in HEK 293 cells. Membrane-anchored D1-deficient uPAR loses its ability to interact with integrins (67). These results suggest that proteolytic cleavage of uPAR alters its binding partners, probably by inducing uPAR conformational change. We have shown that proteolytic cleavage of SuPAR substantially increases its signaling agonist activity and, as a result, CSuPAR does not inhibit ERK activation, cell growth, or Matrigel TM invasion, even in cells that have a potent uPA-uPAR autocrinesignaling pathway. We propose a model in which CSuPAR displaces membrane anchored uPAR-uPA complex from critical adaptor protein-binding sites, similarly to SuPAR; however, in this case, the membrane-anchored complex is replaced by an equally effective signaling initiator. Proteinases, including uPA, that cleave SuPAR may function as a molecular switch, converting the partial agonist, SuPAR, into the complete agonist, CSuPAR. Proteolysis of SuPAR eliminates a potentially effective inhibitor of tumor growth and invasion. Strategies that block the ability of proteinases to convert SuPAR into CSuPAR may have therapeutic efficacy.
There is evidence that paracrine uPA-uPAR signaling may be established in breast cancer (5). In this case, inflammatory cells express and secrete uPA locally, which is available to bind uPAR, expressed by malignant epithelium (68). Paracrine pathways eliminate the need for malignant epithelium to express uPA and increase the likelihood that signaling through uPAR occurs in cancer. Our results demonstrating efficacy of SuPAR at inhibiting ERK activation, even when present at low picomolar concentration, suggest that SuPAR directly regulates cancer cell signaling in vivo. Furthermore, our studies demonstrate that it may be possible to enhance the anticancer activity of SuPAR by preventing CSuPAR formation.