Urokinase-type Plasminogen Activator Receptor (CD87) Is a Ligand for Integrins and Mediates Cell-Cell Interaction*

,

The receptor for urokinase-type plasminogen activator (uPA), 1 uPAR/CD87, is a glycoprotein (M r 35,000 -65,000) composed of 283 amino acid residues. It is anchored to the plasma membrane by a glycosyl phosphatidylinositol (GPI) linkage (reviewed in Refs. 1 and 2). uPAR is the cellular receptor for urokinase, a serine protease that is constitutively or inducibly secreted by most uPAR-expressing cells. Receptor-bound uPA can convert plasminogen to plasmin, which mediates pericellular proteolysis of extracellular matrix proteins in the path of cellular invasion. uPAR is expressed by activated leukocytes, endothelial cells, fibroblasts, and different types of cancer cells (reviewed in Refs. 2 and 3 and references therein). Expression of uPAR has been shown to correlate with the prognosis of many human cancers (2). In murine tumor models, expression or administration of uPAR antagonists has a marked inhibitory effect on the metastatic ability of cancer cells (4) and on the growth of the primary tumor (5), and the down-regulation of uPAR leads to dormancy of carcinoma cells in vivo (6,7). Thus, uPAR expression has been implicated in cancer progression. In addition, it appears that uPAR is up-regulated on cells that are in motion. Migratory/chemotaxis-inducing stimuli (e.g. vascular endothelial growth factor, fibroblast growth factor, plateletderived growth factor, and interleukins) up-regulate uPAR in endothelial cells, smooth muscle cells, and leukocytes in vitro, whereas unstimulated cells do not have detectable expression of uPAR (reviewed in Refs. 2 and 3 and references therein). Thus, uPAR may play a role in leukocyte recruitment, angiogenesis, and tumor metastasis.
uPAR has been reported to associate with many signaling molecules and to mediate signal transduction (8 -10). In recent reports the binding of uPA to uPAR in tumor or endothelial cells has been shown to activate the mitogen-activated protein kinases, extracellular regulated kinase 1 and 2 (11)(12)(13). However, a major question is how uPAR mediates cellular signaling, because the molecule has no transmembrane structure. The existence of one or more hypothetical "transmembrane adapter molecules" that connects uPAR and signaling molecules inside cells has been proposed (14).
It has been shown that the ␤1, ␤2, and ␤3 integrin receptor families interact with uPAR using immunocoprecipitation, immunocolocalization, and resonance energy transfer approaches (15)(16)(17). The uPAR-integrin interaction may be significant, because many integrin receptors activate intracellular signals coupled to the pathways used by both receptor and nonreceptor tyrosine kinases (18 -20). Integrin-and receptor tyrosine kinase-mediated signals may complement each other to fully activate cell survival and proliferation pathways (21,22). It has been proposed that uPAR forms cis-interactions with integrins on the same cell surface as an integrin-associated protein (reviewed in Ref. 1). However, it has not been established that this is the dominant mode of interaction responsible for signal transduction events.
In the present study, we designed experiments to examine the uPAR-integrin interaction in detail using isolated domains derived from recombinant soluble uPAR (suPAR) and cells expressing recombinant ␤1 or ␤3 integrins. These studies establish that uPAR binds to integrins in a manner that is very similar to that of known integrin ligands (e.g. vascular cellular adhesion molecule-1 (VCAM-1)). Additionally, we have found that uPAR interacts with integrins on apposing cells (transinteraction). These unexpected findings may help to clarify the complex role of uPAR in signal transduction, inflammation, and cancer.
CHO cells expressing the three domain forms of human uPAR (designated HuPAR-CHO) were prepared by transfecting full-length human uPAR cDNA (provided by L. A. Miles, The Scripps Research Institute) into a pCDNA3 vector (Invitrogen) together with a plasmid containing a neomycin-resistant gene. Cells were selected with G-418 (0.7 mg/ml medium). Approximately 50% of cells stably expressed uPAR after selection with mAb 3B10. CHO cells stably expressing uPAR were sorted by FACStar (Becton-Dickinson) to obtain cells homogeneously expressing uPAR at a high level.

Methods
Production of suPAR in Drosophila S2 Cells-cDNAs encoding soluble wild-type uPAR (amino acids 1-277) and soluble domains 2 and 3 (amino acids 88 -277) were generated by polymerase chain reaction using pTracer-full-length uPAR as a template. The fragments were digested with BglII and Xho and sub-cloned into the expression vector (pMT/BiP/V5, Invitrogen). Soluble suPAR domains were expressed in Drosophila Schneider S2 cells (DES system, Invitrogen) as described by the manufacturer.
Purification of suPAR Expressed in S2 Cells-Small scale preparations of suPAR and variants were purified from the medium using a polyclonal anti-uPAR antibody affinity column. For large scale preparations, S2 cell culture supernatants were filtered, and the medium were loaded onto a 40-g hydroxyapatite column equilibrated with 10 mM K 2 HPO 4 , pH 7.0. The column was then eluted with a gradient of 10 -200 mM K 2 HPO 4 , pH 7.0, and the suPAR-containing fractions were identified by Western blotting. Fractions containing monomeric suPAR eluted between 50 and 120 mM K 2 HPO 4 . Fractions containing monomeric suPAR were pooled, concentrated, and further purified using C8 reverse-phase HPLC. Crude suPAR (20 mg of total protein) was loaded in a volume of 2 ml onto a semi-preparative (10 ϫ 250 cm) C8 column and eluted at a flow rate of 4 ml/min with a linear gradient of 0 -70% solvent B, where solvent A was 100% H 2 O, 0.1% trifluoroacetic acid, and solvent B was 100% acetonitrile, 0.1% trifluoroacetic acid. suPAR eluted as a single broad peak under these conditions with a retention time of ϳ27 min. SDS-polyacrylamide gel electrophoresis analysis of suPAR purified in this manner demonstrated a single major peak at 35 kDa under nonreducing conditions. Also observed was a slight laddering effect with three to four minor higher molecular weight species. These were determined to be SDS-stable aggregates of suPAR, which disappeared when SDS-polyacrylamide gel electrophoresis was performed under reducing conditions. Matrix-assisted laser desorption ionization-time of flight mass spectrometry revealed a single broad peak with an average mass of 34,797 Da. The predicted mass based on the amino acid sequence is 30,672 Da, indicating the presence of about 4 kDa in glycosylation. Expression levels of suPAR were typically 30 mg/liter determined by enzyme-linked immunosorbent assay. Purification yielded about 10 -12 mg of pure suPAR protein per liter of culture supernatant.
Enzymatic Digestion of Wild-type suPAR-suPAR was digested with chymotrypsin to generate the soluble D1 and D2D3 fragments. Chymotrypsin was added to suPAR (1 mg/ml in phosphate-buffered saline) at a final molar ratio of suPAR:chymotrypsin of 1000:1, and the digest was allowed to proceed for 2 h at room temperature. The digest was quenched by addition of Pefablock (100 M final concentration). The D1 and D2D3 fragments were separated using C8 reverse phase-HPLC. The D1 and D2D3 fragments eluted with a retention time of ϳ27 and 25 min, respectively. The purified D1 fragment was sequenced, and a single N terminus was observed beginning with RS (representing the two extra amino acids present in this construct) followed by LR, amino acids 1 and 2 in the mature suPAR sequence. Sequencing of the D2D3 fragment revealed a single N terminus as well, beginning with amino acids SRS, corresponding to amino acids 88 -90 in the mature suPAR sequence. Matrix-assisted laser desorption ionization-time of flight mass spectral analysis revealed a single peak for the D1 fragment (mass ϭ 11,041 Da) and several peaks for the D2D3 fragment (mass ϭ 23,846; 23,691; 22,936; 22,800). These were presumed to be glycosylation isoforms, although additional digestion of the D2D3 fragment from the C terminus could not be excluded. However, no consensus chymotryptic cleavage site that could result in the observed mass differences is present at the C terminus of the D2D3 fragment.
Affinity-purified Anti-suPAR-A polyclonal antibody against suPAR was prepared as described (38). Briefly, recombinant suPAR was expressed in SP2/0 cells and purified using a single-chain uPA-Sepharose column. Serum was collected from rabbits immunized with purified suPAR, and the IgG was obtained by a 50% ammonium sulfate precipitation. This material was dialyzed (10,000ϫ) against phosphate-buffered saline and further purified using a suPAR-Sepharose column to generate affinity-purified anti-suPAR IgG.
Cell-Cell Binding Assay-Integrin-transfected, mock-transfected (vector only), or parental K562 cells were labeled with 2Ј,7Ј-bis-(2carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (Molecular Probes, Eugene, OR) according to the manufacturer's instructions. The labeled cells (10 5 cells/well) were added to the monolayer of parent, mock-transfected, or HuPAR-CHO cells and incubated for 30 min at 37°C. After the wells were rinsed with medium to remove unbound cells, bound cells were quantified by assaying fluorescence (excitation 485 nm, emission 530 nm) using an FL500 microplate fluorescence reader (Bio Tek Instruments, Winooski, VT). Antibodies were used at the same concentrations as described above. Data are shown as means Ϯ S.D. of three independent experiments.

Binding of Recombinant suPAR to ␤1 and ␤3
Integrins-To study whether and how integrins are involved in uPAR-mediated signal transduction, we first asked whether recombinant soluble uPAR fragments interact with ␤1 and ␤3 integrins. It has been proposed that the activity of suPAR requires chymotrypsin cleavage between the N-terminal domains 1 and 2 (14). We used several different suPAR fragments that were expressed in Drosophila S2 cells (domain 1, domains 2 and 3, or full-length suPAR containing domains 1, 2, and 3, designated D1, D2D3, D1D2D3, respectively) and domains 2 and 3 expressed in CHO cells. Unexpectedly, these suPAR fragments supported adhesion of both ␣4and ␤3-CHO cells expressing recombinant human ␣4/hamster ␤1 (␣4␤1) and hamster ␣v/ human ␤3 hybrid (␣v␤3), respectively, but not CHO cells (Fig.  1A). Parental CHO cells express endogenous ␣5␤1, ␣v␤1, and ␣v␤5 but not ␣4␤1 or ␣v␤3 integrins (42). It appears that suPAR expressed in Drosophila and CHO cells supports adhesion of ␣4and ␤3-CHO cells at comparable levels. We used suPAR D2D3 expressed in CHO cells (43) for further charac-terization of suPAR-integrin interaction throughout this study. The rationale of using D2D3 is that removal of D1 leaves uPAR D2D3 that is capable of signal transduction without uPA (14), and thus suPAR D2D3-integrin interaction is likely to be biologically relevant.
We further tested the specificity of interaction using antibodies against uPAR. We found that antibodies against uPAR effectively blocked suPAR binding to ␣v␤3 (Fig. 1E). We then for ␤3-CHO cells. E, the effect of affinity-purified anti-uPAR antibody on ␣v␤3-suPAR D2D3 interaction was studied. Polyclonal anti-uPAR or control rabbit IgG was included in the assay medium at 3.4 g/ml. SuPAR D2D3 was used at a coating concentration of 5 g/ml. F, the capability of ␣-bungarotoxin, a protein structurally similar to uPAR, to support ␣v␤3-mediated cell adhesion was studied as a function of substrate concentration. SuPAR D2D3 and ␣-bungarotoxin were coated up to 5 g/ml.
tested whether ␣-bungarotoxin, which has a "three-finger protein" structure similar to the uPAR structure (44), "nonspecifically" binds to ␣v␤3. The toxin showed only weak, if any, affinity to ␣v␤3 (Fig. 1F). These results suggest that suPAR binding to integrins is specific to the uPAR sequence and structure.
Effect of Cations and ␤1 Integrin Activation and Inactivation on uPAR Binding-Ligand binding to integrins is tightly regulated by activation/inactivation of integrins through insideout signal transduction (45). ␤3-CHO cells significantly adhere to suPAR D2D3 in the presence of Mg 2ϩ but not in the presence of Ca 2ϩ (Fig. 2A). CHO cells do not adhere to suPAR in either condition. These results suggest that ␣v␤3 requires Mg 2ϩ for binding to suPAR D2D3 under the conditions employed. Although ␣2-CHO, ␣3-CHO, and CHO cells do not strongly adhere to suPAR D2D3 under the assay conditions used, it is still possible that binding of ␣2␤1, ␣3␤1, and ␣5␤1 may require more activation. Therefore, we studied whether Mn 2ϩ (0.1 mM), which universally activates integrins, facilitates uPAR binding to these integrins ( Fig. 2A). We found that ␣5␤1-deficient B2 variant CHO cells did not significantly adhere to suPAR D2D3 in the presence of Mn 2ϩ , but parental CHO cells did. We found that B2 cells expressing human ␣3 or ␣4 (designated ␣3or ␣4-B2 cells, respectively) also adhered to suPAR D2D3 in the presence of Mn 2ϩ . ␣2-B2 cells did not adhere to suPAR D2D3 under any conditions tested. Ca 2ϩ did not significantly support adhesion of these cells. These results suggest that ␣3␤1, ␣4␤1, and ␣5␤1 but not ␣2␤1 bind to suPAR D2D3 in an activationdependent manner.
We then asked whether the activation status of ␤1 integrins that is required for uPAR binding to natural integrins is similar to the requirements for the recombinant integrins that we studied. To answer this question, we studied uPAR binding to nonrecombinant integrins on Jurkat human T-cell leukemia cells (␣4␤1ϩ, ␣5␤1ϩ). We can stimulate or suppress ␤1 integrin-ligand interaction from outside cells using anti-␤1 antibodies (e.g. TS2/16, activating; AIIB2, inhibiting) (reviewed in Ref. 46). We found that adhesion of Jurkat cells to suPAR D2D3 is stimulated by TS2/16 and blocked completely by AIIB2 and SG73 (anti-␣4 mAb) and partially by KH72 (anti-␣5 mAb) (Fig.  2B). These results suggest that nonrecombinant ␣4␤1 and ␣5␤1 in Jurkat cells may interact with suPAR D2D3 in an activation-dependent manner.
Does uPAR Share Common Binding Sites in Integrins with Known Integrin Ligands?-We then asked whether uPAR competes with VCAM-1, a known ␣4␤1 ligand, for binding to ␣4␤1. To address this question, we examined the adhesion of ␣4-B2 cells to suPAR D2D3 in the presence of soluble VCAM-1. We used B2 cells to completely eliminate the contribution of ␣5␤1 in uPAR binding. We found that adhesion of ␣4-B2 cells to suPAR D2D3 was blocked by VCAM-1 in a dose-dependent manner (Fig. 3A) but not by irrelevant ligand fibrinogen. These results suggest that the inhibitory effect of VCAM-1 is specific to ␣4␤1-suPAR D2D3 interaction, and thus suPAR and VCAM-1 compete for binding to ␣4␤1. We also studied the effect of the GRGDS peptide, a widely distributed integrinbinding motif, on uPAR-␣v␤3 interaction (Fig. 3B). We found that GRGDS peptide completely blocked adhesion of ␤3-CHO cells to suPAR D2D3, but control GRGES peptide did not, suggesting that uPAR and the ligand-derived RGD motif compete for binding to ␣v␤3. Thus it is highly likely that the uPAR-binding sites in these integrins may overlap with those of known integrin ligands.
We next studied whether uPAR binding is inhibited by mutations in these integrins that block binding of known ligands. We have previously reported that the mutation to Ala of several amino acid residues, Tyr-187, Trp-188, and Gly-190 in ␣4 (designated Y187A, W188A, and G190A mutations, respectively), blocks VCAM-1 and fibronectin connecting segment-1 (CS-1) peptide binding to ␣4␤1 (47). These residues are located within the putative ligand-binding site in ␣4, and the corresponding residues in ␣IIb or ␣5 have also been reported to be critical for ligand binding to ␣IIb␤3 or ␣5␤1 (47,48). We studied whether these ␣4 mutations block adhesion of ␣4␤1 to suPAR D2D3. We found that these mutations blocked uPAR-␣4␤1 interaction, but several other mutations did not (Fig. 4). The overall effect of these mutations on uPAR binding to ␣4␤1 is similar to their effect on VCAM-1 or CS-1 binding to ␣4␤1. These results suggest that the uPAR-binding site in ␣4 overlaps with those for VCAM-1 and CS-1.
The Asp-130 residue of ␤1 (42,49) and the corresponding Asp residues in ␤2 (50), ␤3 (Asp-119) (51), and ␤6 (52) have been reported to be critical for ligand binding. We tested whether mutation of Asp-119 to Ala in ␤3 affected suPAR D2D3 binding to ␣v␤3. The wild type or Asp-119 to Ala (D119A) dominantnegative mutant of human ␤3 was transiently expressed in CHO cells, and the ability of cells to adhere to suPAR D2D3 was determined. Wild-type or mutant ␤3 is expressed as hamster ␣v/human ␤3 hybrid. We found that cells transiently expressing the ␣v␤3(D119A) mutant did not adhere to suPAR D2D3, although cells expressing wild-type ␣v␤3 did (Fig. 5A). The levels of adhesion (ϳ22%) of wild-type ␤3 are lower than in Fig. 1 but are substantial, considering that only 35% of added cells express human ␤3. Note that the nonfunctional ␤3(D119A) mutant was expressed on the surface of cells at a comparable level.
uPAR-Integrin Trans-interaction Supports Cell-Cell Adhesion-uPAR has been proposed to interact with integrins on the same cells (cis-interaction) (reviewed in Ref. 1). Because the present results suggest that uPAR may interact with integrins as a ligand, it is possible that uPAR may interact with integrins on apposing cells (trans-interaction), analogous to the interaction between VCAM-1 and ␣4␤1. To determine whether uPAR interacts in-trans with integrins, we tested whether K562 erythroleukemic cells expressing recombinant ␣4␤1 (designated ␣4-K562 cells) interact with unsorted CHO cells expressing human uPAR (the three-domain form, designated HuPAR-CHO cells) (Fig. 6A). We found that ␣4-K562 cells showed significantly greater adhesion to HuPAR-CHO cells than to parental CHO cells and that this adhesion was blocked by SG73 (anti-␣4). Mock-transfected and parental K562 or CHO cells generated essentially the same results, and the data with parental K562 or CHO cells are shown. Endogenous ␣5␤1 in K562 cells and endogenous hamster uPAR in CHO cells (53) may explain the background binding of K562 cells to CHO cells. We obtained essentially the same results with K562 cells expressing recombinant ␣9␤1 and ␣v␤3 (designated ␣9and ␣v␤3-K562 cells, respectively) and cloned HuPAR-CHO cells (Fig. 6, B and C). Y9A2 and 7E3 blocked binding of ␣9and ␣v␤3-K562 cells to cloned HuPAR-CHO cells, respectively, suggesting that these interactions are specific to the respective integrins. These results suggest that GPI-anchored uPAR specifically interacts with several integrins on the apposing cells and supports cell-cell interaction. DISCUSSION uPAR as a Ligand for Several Non-I Domain Integrins-The present study for the first time establishes that uPAR is a FIG. 3. Competition between uPAR and VCAM-1 for binding to ␣4␤1 (A) or between uPAR and RGD peptide for binding to ␣v␤3 (B). A, the effect of soluble VCAM-1 on ␣4␤1-suPAR D2D3 interaction was studied. The concentration of suPAR D2D3 used for coating was 5 g/ml. The ␣5␤1-deficient CHO B2 variant cells expressing human ␣4/hamster ␤1 (a receptor for VCAM-1) (␣4-B2) were used instead of ␣4-CHO cells. Fibrinogen, an irrelevant ligand, was used as a negative control. VCAM-1 and fibrinogen were added at the final concentrations of 1, 5, or 10 g/ml in the incubation medium. The numbers in parentheses are concentrations of proteins (g/ml). Fg, fibrinogen. B, the effect of GRGDS or GRGES peptide on ␣v␤3-suPAR interaction was studied with ␤3-CHO cells. The concentration of suPAR D2D3 used for coating was 5 g/ml. GRGDS or GRGES peptide was added at 10 or 100 M final concentration in the incubation medium. The numbers in parentheses are concentrations of proteins (M).

FIG. 4.
Critical residues in ␣4 for uPAR binding. CHO cells expressing different ␣4 mutants were tested for their ability to adhere to suPAR D2D3 (closed columns) and VCAM-1 (open columns). Mutation to Ala at positions 187 (Tyr), 188 (Trp), and 190 (Gly) within the putative ligand-binding sites of ␣4 blocks ␣4␤1-VCAM-1 binding; the data on VCAM-1 binding are from a previous publication (47) and are shown for comparison. The concentration of substrates used for coating was 3 g/ml for suPAR D2D3 and 0.25 g/ml for VCAM-1-Ck fusion protein. Goat anti-mouse Ck antibody was used to facilitate VCAM-1-Ck coating (47). wt, wild type. ligand for several ␤1 and ␤3 integrins. We demonstrated that 1) these integrins adhere to suPAR D2D3 in a dose-dependent and cation-dependent manner and 2) specific antibodies against these integrins or uPAR block these interactions. Another structurally related protein, ␣-bungarotoxin, showed only weak integrin binding relative to suPAR D2D3, suggesting that binding to integrins is a specific property of uPAR. We also demonstrated that the uPAR-binding sites in ␣4␤1 and ␣v␤3 overlap with the previously reported putative ligand-binding site in these integrins. This is based on the observations that 3) soluble ligands for ␣4␤1 and ␣v␤3 compete with suPAR D2D3 for binding to these integrins and 4) the known integrin mutations that block ligand binding to ␣4␤1 and ␣v␤3 block binding of suPAR D2D3 to these integrins as well. These results suggest that uPAR may directly compete with other ligands for binding to these integrins, rather than only operating indirectly by regulating the binding affinity of integrins to other ligands as an integrin-associated protein. These unexpected results are consistent with the observation that soluble uPAR blocks fibronectin binding to ␣5␤1 (17) but do not necessarily agree with the current view that uPAR interacts with integrins exclusively as an associated protein rather than as a ligand contains similar or distinct integrin-binding sites. Detailed binding kinetic studies using individual soluble domains would be required to address this question. Also, there are known differences in uPAR glycosylation among mammalian cells and between some mammalian tumor cells and S2 cells. It would be interesting to study the influence of glycosylation, if any, on avidity or specificity in future experiments.
We also demonstrated that uPAR may mediate cell-cell adhesion by trans-interaction with integrins (Figs. 6 and 7B). This broadens the currently held concept of uPAR-integrin interactions, in which uPAR is proposed to interact exclusively with integrins residing on the same cell (cis-interaction) as an "associated protein" that mediates signal transduction directly or through the mediation of a distinct transmembrane adapter protein. Thus, the current model holds that all of the molecules that are involved in the uPAR signaling pathway are located on the same cell surface as uPAR (Fig. 7A) (reviewed in Refs. 1 and 54). The present results do not rule out such uPAR-integrin cis-interactions. However, the present results predict that if uPAR interacts with integrins on the same cell surface (cisinteraction), uPAR may bind as a ligand rather than as an associated protein (Fig. 7C) and that signal transduction events are mediated through the engaged integrin.
It has been reported that association of uPAR with integrins induces tyrosine phosphorylation of focal adhesion kinase, paxillin, p130(cas), and mitogen-activated protein kinase (13). These signaling molecules have also been identified as being involved in signaling through integrins. If uPAR binds to integrins as a ligand, it is not difficult to imagine that GPI-anchored uPAR may transduce signals through the integrin pathway without the help of an additional putative transmembrane adapter molecule. Phosphorylation of these proteins may be induced through integrins on uPAR-integrin trans-interaction.
The trans-interaction model (Fig. 7B) predicts that tyrosine phosphorylation of proteins and the resulting induction of gene expression occur in the cells that have integrin receptors but not in the cells in which uPAR alone is expressed. In uPARintegrin-mediated cell-cell interaction, both of the apposing cells may express uPAR and integrins, and uPAR-integrin interaction may occur reciprocally; thus tyrosine phosphorylation might occur on both sides.
Implication of uPAR as an Integrin Ligand in Pathological Situations-It has been reported that down-regulation of uPAR makes a human epidermoid carcinoma Hep3 dormant in vivo (6, 7) as a result of a reduced proliferation rate rather than an increased apoptotic rate (7). It has been proposed that reduction of uPAR expression makes cells either incapable of responding to a signal, or unable to generate a sufficient signal, to propel them through G 0 /G 1 in vivo (10). However, cells expressing less than 50% of the normal level of uPAR grow indistinguishably from parental cells in culture (10). It is unclear why the altered uPAR expression level affects the proliferation of cancer cells in vivo but not in culture. We suspect that uPAR-integrin trans-interaction, which occurs during three-dimensional growth in tumor mass in vivo but may not occur in culture, may contribute to this discrepancy. Transinteractions between uPAR and integrins as described in this study have the potential to transduce proliferative signals within tumor masses through integrin-dependent pathways. It is possible that uPAR may also function as a ligand for integrin through cis-interactions (Fig. 7C), thereby providing autocrinetype proliferative signals. If this is the case, cancer cells that express high levels of uPAR may have the potential to grow faster than cells that express uPAR at lower levels.
Soluble uPAR is present in ascites of ovarian cancer patients (55). Soluble uPAR levels in plasma increase in patients with rheumatoid arthritis (56), ovarian cancer (57), and leukemia (58). Soluble uPAR levels correlate with resistance to chemotherapy in leukemia (58) and with tumor volume in animal models (59). It has also been shown that soluble uPAR induces extracellular regulated kinase activation when added to uPARlow Hep3 cells (10). The present study suggests that soluble uPAR might also interact as a ligand with integrins on leukemic cells, solid tumor cells, or inflammatory leukocytes and transduce proliferative signals (Fig. 7D). Thus binding of GPIanchored uPAR (trans-interaction) or soluble uPAR to integrins as a ligand is a potential therapeutic target in cancer, inflammation, or other pathological situations.
It has been reported that leukocyte rolling and adhesion dramatically increase in mesenteric postcapillary vessels in adjuvant-induced chronic vasculitis in rats (60). An anti-integrin ␣4 antibody significantly blocks this leukocyte rolling and adhesion. Interestingly, this ␣4-dependent interaction is not dependent on VCAM-1 or the CS-1 region of fibronectin. It has also been reported that platelet and endothelial uPAR is involved in the survival of platelets in the circulation in mice (61). Injection of tumor necrosis factor increases the number of platelets in the lung alveolar capillaries in wild-type mice, but platelet trapping is insignificant in mice deficient in uPAR. uPAR may be a candidate integrin ligand in these cases. It is possible that uPAR-integrin trans-interaction may be involved in these leukocyte-endothelium and platelet-endothelium interactions during inflammation.
In summary, we have shown that uPAR is a ligand for several integrins and mediates cell-cell adhesion through trans-interactions with these integrins. These findings also help to clarify how uPAR binding to integrins as a ligand transduces signals through integrin pathways. In addition, the present findings predict that interaction of uPAR with integrin FIG. 7. Models of interaction between uPAR and integrins. The current view of uPAR signaling is based on association of uPAR with integrins in-cis as an associated protein (A). We have shown that uPAR may interact with integrins in-trans (B). If uPAR interacts with integrins in-cis, uPAR may bind to the ligand-binding site of integrins (C). It has been shown that soluble uPAR may bind to integrins (D). The present study suggests that uPAR is a ligand but not an associate protein of integrins, suggesting that uPAR (GPI-anchored or soluble) binding to integrins may induce signal transduction through integrin pathways. as a ligand (cis or trans) may be involved in transduction of proliferative or activating signals in cancer and inflammatory cells. Additional studies will be required to clarify the role of uPAR-integrin interactions in these processes.