Mapping Part of the Functional Epitope for Ligand Binding on the Receptor for Urokinase-type Plasminogen Activator by Site-directed Mutagenesis*

The urokinase-type plasminogen activator receptor (uPAR) is a glycolipid anchored multidomain member of the Ly-6/uPAR protein domain superfamily. Studies by site-directed photoaffinity labeling, chemical cross-linking, and ligand-induced protection against chemical modification have highlighted the possible involvement of uPAR domain I and particularly loop 3 thereof in ligand binding (Ploug, M. (1998) Biochemistry 37, 16494–16505). Guided by these results we have now performed an alanine scanning analysis of this region in uPAR by site-directed mutagenesis and subsequently measured the effects thereof on the kinetics of uPA binding in real-time by surface plasmon resonance. Only four positions in loop 3 of uPAR domain I exhibited significant changes in the contribution to the free energy of uPA binding (ΔΔG ≥ 1.3 kcal mol−1) upon single-site substitutions to alanine (i.e. Arg53, Leu55, Tyr57, and Leu66). The energetic impact of these four alanine substitutions was not caused by gross structural perturbations, since all monoclonal antibodies tested having conformation-dependent epitopes on this domain exhibited unaltered binding kinetics. These sites together with a three-dimensional structure for uPAR may provide an appropriate target for rational drug design aimed at developing new receptor binding antagonists with potential application in cancer therapy.

The urokinase-type plasminogen activator receptor (uPAR) 1 is a multifunctional membrane glycoprotein primarily involved in the regulation of pericellular proteolysis due to its high affinity interaction with the growth factor-like module of urokinase-type plasminogen activator (uPA) (1), but which has also been implicated in the promotion of cell adhesion due to its vitronectin and integrin binding properties (2), in signal transduction (3) and chemotaxis (4,5). However, the uPA-uPAR interaction per se is intimately coupled to the latter "nonproteolytic" functions of uPAR, since this interaction either elicits or modulates these events. Cell surfaces expressing uPAR constitutes favored microenvironments for uPA-mediated plasminogen activation (6). Consequently, the uPA-uPAR interaction represents an attractive molecular target for the development of small receptor binding antagonists that may prove useful during treatment of certain diseases in which uPAR has been implicated, i.e. cancer invasion and metastasis (7)(8)(9)(10)(11)(12)(13).
uPAR is a glycosylphosphatidylinositol-anchored plasma membrane protein (14) having an extracellular part composed of three homologous domains belonging to the Ly-6/uPAR protein domain family, 2 as reviewed (15). This family is dominated by single domain proteins among which the glycolipid-anchored members are found primarily in mammalians (i.e. CD59, Ly-6, E48, and ThB) whereas the secreted members belong to either reptiles or amphibians (i.e. ␣-neurotoxins, fasiculins, cardiotoxins, and xenoxins). Intriguingly, a similar phylogenetic relationship exists for the few family members identified so far containing two Ly-6/uPAR domains, i.e. the glycolipid anchored RoBo-1 (16) and metastasis-associated C4.4 (17) isolated from rat versus the secreted phospholipase A2 inhibitor isolated from cobra blood (18). Comparison of the three-dimensional protein structures available for several single domain members of this protein family reveals a "three finger" consensus structure consisting of 3 loops, a central 3-stranded ␤-sheet, and a globular, disulfide-rich core (19 -21). The individual domains of uPAR (numbered I, II and III) are thought to adopt a similar "three finger fold" (15).
The involvement of uPAR domain I (residues 1-87) 3 in uPA binding is demonstrated by several lines of evidence. First, uPAR domain I can be specifically cross-linked to a receptor binding derivative of uPA using two different chemical crosslinking reagents (23,24). Second, the uPA-uPAR interaction can be inhibited competitively by monoclonal antibodies recognizing epitopes on uPAR domain I (25)(26)(27). Third, uPAR domain I is also a target for the specific photoinsertion from a small peptide antagonist of uPA binding (28). Specific proteolytic cleavage after Tyr 87 , situated in the linker region between domains I and II, is, however, also accompanied by a Ͼ1,500fold reduction in the affinity for uPA (⌬⌬G Ͼ 4 kcal/mol), which clearly emphasizes the necessity of the multidomain structure of uPAR for high affinity ligand binding (27,28). Previous attempts at dissecting the ligand interaction site on uPAR have highlighted the importance of loop 3 in uPAR domain I, since Arg 53 , Tyr 57 , and Leu 66 reside at the receptor-ligand interface as assessed by either photoaffinity labeling (29)  tromethane (30). Guided by these data we have performed an alanine scanning analysis of this region (residues 47-70) of uPAR by site-directed mutagenesis enabling us to discriminate between structural and functional epitopes of ligand-binding, i.e. residues present at the interface versus those residues also making a productive contribution to the free energy of ligandbinding (31)(32)(33).

MATERIALS AND METHODS
Chemicals and Reagents-Pfu DNA polymerase (EC 2.7.7.7) was from Stratagene (La Jolla, CA). HPLC-purified DNA oligonucleotides were purchased from DNA Technology (Aarhus, Denmark). Active twochain uPA (EC 3.4.21.31) was purchased from Serono (Aubonne, Switzerland), recombinant pro-uPA expressed in Escherichia coli was a generous gift from Dr. D. Saunders (Grü nenthal, Germany), while the amino-terminal fragment of uPA (ATF) was kindly provided by Dr. A. Mazar (Ångstrom Pharmaceuticals, San Diego, CA). Murine anti-uPAR monoclonal antibodies were produced at the Finsen Laboratory (clones R2, R3, and R9) as described previously (25 Construction of Expression Vectors-An expression vector, encoding a secreted truncated soluble form of human uPAR (denoted suPAR and encompassing amino acids 1-283), 3 was constructed utilizing two separate PCR performed by standard PCR techniques using Pfu DNA polymerase (Fig. 1). In the first reaction, generating a domain I encoding fragment, a SpeI site was introduced in front of the non-coding 5Ј-end, and a XhoI site was introduced at the 3Ј-end by an A 3 G nucleotide shift at position 394. 3 In the second reaction, generating a domain II ϩ III fragment, a XhoI site was introduced into the 5Ј-end at position 394 as above, and a TAA stop codon, followed by an EcoRI site, was introduced at position 962 in the 3Ј-end. Finally, the two PCR products from these reactions were cut with the appropriate restriction enzymes, ligated, and cloned into the NheI (SpeI compatible) and EcoRI sites of the mammalian expression vector pCI-neo (Promega) as a SpeI-EcoRI fragment giving the plasmid pCI-neo/suPAR.
Mutations were introduced into suPAR domain I by PCR using a one-tube based modification of the megaprimer procedure which employs Pfu DNA polymerase (34). In brief, the procedure requires three oligonucleotide primers: two flanking primers, which are upstream and downstream of the mutation site, and one mutagenic primer (Table I) designed to contain at least 10 perfectly matched bases at both the 5Јand 3Ј-end. To that end, uPAR domain I was flanked by SpeI and XhoI sites as above and cloned into pBluescript (Stratagene), as shown in Fig. 1. This plasmid (pBluescript/shuPAR-D1) was used as a template with T3 and T7 primers as flanking primers for the production of PCR products containing the desired mutation. Mutated domain I PCR fragments were cloned as cassettes into the NheI and XhoI sites of pCI-neo/ suPAR depleted for wild-type domain I. The suPAR-W129A mutant in domain II was constructed using pCI-neo/suPAR wild-type DNA as template in a site-directed mutagenesis PCR using flanking primers residing in domain I and III, respectively. The W129A mutant PCR fragment was cloned as a cassette into XhoI and ApaI sites of pCI-neo/ suPAR depleted for wild-type domain II. Prior to transfection PCRgenerated sequences of all constructs were confirmed by DNA sequencing.
Transfection, Expression, and Purification of suPAR Mutants-Chinese hamster ovary cells were transfected with expression vectors using a calcium phosphate precipitation procedure (Stratagene). After G418 selection, clones were picked, propagated, and the conditioned medium tested for the production of recombinant suPAR by enzyme-linked immunosorbent assay (35) using anti-uPAR monoclonal antibody R2 recognizing domain III as detecting antibody. The cells were maintained as monolayers at 37°C in a humid atmosphere with 5% CO 2 in minimum essential ␣-medium with ribo-and deoxyribonucleotides and GlutaMAX I (Life Technologies, Inc.) supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 800 g/ml G418. The harvest fluid contained between 0.01 and 0.5 g/ml recombinant protein (Table I).
The suPAR mutants were purified by immunoaffinity chromatography from the conditioned media as described previously (36) followed by reversed-phase HPLC using a Brownlee Aquapore C4 column and a linear gradient (40 min) from 0 to 70% (v/v) acetonitrile in 0.1% (v/v) trifluoroacetic acid at a flow rate 250 l min Ϫ1 . Purified mutants were dissolved in phosphate-buffered saline after solvent evaporation and were either quantified spectrophotometrically using E 280 nm 1% ϭ 9.2 (37) or by quantitative amino acid analysis (mutants containing tryptophan substitutions). Purity was assessed by SDS-PAGE followed by silver staining and judged to be Ͼ95%.
Verification of Mutations in suPAR Domain I by MALDI-MS-To ascertain that the desired mutation in suPAR domain I had occurred at the protein level and to analyze the consequence of the mutations on the processing of the glycan moiety attached to Asn 52 (38), the NH 2 -terminal domain I (residues 1-87) was liberated from suPAR by limited proteolysis using chymotrypsin and subsequently purified by size exclusion chromatography using 0.1 M NH 4 HCO 3 as solvent (36). The mass of purified domain I was determined by MALDI-MS before and after treatment with N-glycanase (5 milliunits/l for 60 min at 37°C) and compared with that of wild-type uPAR domain I (38), see Table I.
Real Time uPA Binding Kinetics of suPAR Mutants Measured by Surface Plasmon Resonance-The kinetics of the interaction between suPAR (wild-type as well as mutants) and immobilized uPA, pro-uPA, or various monoclonal anti-uPAR antibodies were measured in realtime by surface plasmon resonance using a BIAcore 2000™ equipment (Pharmacia Biosensor, Uppsala, Sweden). A carboxymethylated dextran matrix (CM5 sensor chip) was preactivated with N-hydroxysuccinimide/N-ethyl-NЈ-[3-(diethylamino)propyl)carbodiimide according to the manufacturers recommendations. Random amine coupling of the respective ligands was achieved by subsequent injection of uPA, pro-uPA (each 20 g/ml), or a monoclonal anti-uPAR antibody (50 g/ml) in 10 mM sodium acetate, pH 5.0, at a flow rate of 5 l min Ϫ1 for 6 min. Sensorgrams (resonance units versus time) were recorded by the BIA- core 2000 TM at a flow rate of 10 l min Ϫ1 at 5°C, using 8 different concentrations of suPAR in the range of 2 to 200 nM in running buffer (10 mM HEPES, 150 mM NaCl, pH 7.4, including 0.005% surfactant P-20). All analyses were performed at least in triplicate using several purified preparations of the individual uPAR mutants and different sensor chips coupled with uPA (ranging from 1000 to 3000 resonance units). Sensor chips were regenerated at the end of each run by injection of 0.1 M acetic acid containing 0.5 M NaCl.
Data obtained from parallel mock coupled flow cells (derivatized in the presence of buffer only) served as blank sensorgrams for subtraction of changes in the bulk refractive index. The sensorgrams obtained were analyzed by nonlinear least squares curve fitting using BIAevaluation 3.0 software (Pharmacia Biasensor) assuming single-site association and dissociation models. In brief, values for k off were determined from data collected during the dissociation phase (dR/dt ϭ Ϫk off R) after which the corresponding values for k on were derived from the following equation assuming pseudo-first order conditions: dR/dt ϭ k on [suPAR]-(R max Ϫ R) Ϫ k off R, as described in detail (39), where R is the recorded surface plasmon resonance signal.
Differences in the Gibbs free energy (⌬⌬G) of uPA binding between mutant uPAR (⌬G mut ) and wild-type uPAR (⌬G wt ) were calculated by the following equation, where R is the gas constant (R ϭ 1.99 cal/mol K), T is the absolute temperature (T ϭ 278 K), and K d is the equilibrium binding constant derived from the kinetic rate constants determined by surface plasmon resonance (K d ϭ k off /k on ).
Miscellaneous Analyses-Matrix-assisted laser desorption mass spectrometry was performed on a linear time of flight instrument (Voy-ager TM , Perseptive Biosystems, MA) equipped with a 1.2-m flight tube and a 337-nm nitrogen laser. Sample was desorbed from an ␣-cyano-4hydroxycinnaminic acid matrix after sample deposition by the sandwich method (40). Spectra were calibrated internally by addition of 0.5 pmol of ␣-cobratoxin (7,821.04 Da), which yielded mass assignments of deglycosylated, wild-type domain I (9,754.07 Da) with a mass accuracy better than 0.05% (Ϯ5 Da).
Qualitative assessments of the relative ligand-binding affinities of selected suPAR single-site mutants were obtained after preincubation of 0.5 nM suPAR mutant with 1.0 nM 125 I-labeled ATF followed by covalent cross-linking of the preformed complexes using N,NЈ-disuccinimidyl suberate (41). The complex formation was visualized by autoradiography after SDS-PAGE of reduced and alkylated samples.

RESULTS AND DISCUSSION
Choice, Design, and Expression of Soluble uPAR Mutants-In the present study we have selected non-glycine residues in loop 3 of uPAR domain I (residues 47-70) as the primary target for site-directed mutagenesis, since this region previously has been implicated in ligand-binding by either photoaffinity labeling (Arg 53 and Leu 66 ) (28,29), ligand-induced protection against chemical modification (Tyr 57 ) (30) or enzymatic deglycosylation (Asn 52 ) (38). In addition, the only two tryptophan residues present in uPAR (Trp 32 and Trp 129 ) were also mutagenized, since tryptophan residues generally participate so prevalently in protein-protein interactions (33,42), and a surface exposed aromatic/hydrophobic patch on uPAR correlates to the vacancy of the high-affinity uPA-bind-  ing site as demonstrated by 8-anilino-1-naphthalene sulfonate fluorescence spectroscopy (27).
We have chosen to perform the present alanine scanning mutagenesis analysis using a truncated, secreted uPAR variant (denoted suPAR), having uPA binding properties indistinguishable from those of the wild-type glycolipid-anchored receptor present on the cell surface (27). This truncated receptor can conveniently be produced by Chinese hamster ovary cells after transfection with a DNA carrying a deletion corresponding to the COOH-terminal signal peptide responsible for the attachment of the glycolipid anchor (14,37,43). A cassette expression vector was therefore constructed in such a manner that single-site mutations in suPAR domain I could be introduced without the necessity to confirm the DNA sequence encoding the entire receptor for each mutant created (Fig. 1). Expression of recombinant "wild-type" suPAR (encompassing residues 1-283) as well as mutants thereof were accomplished after stable transfection of Chinese hamster ovary cells and the secreted receptor proteins were isolated from the harvest fluid by immunoaffinity chromatography using the high-affinity monoclonal antibody R2 specific for uPAR domain III (25) followed by reversed-phase HPLC. The purity of these protein preparations were Ͼ95% as judged from silver-stained gels after SDS-PAGE of reduced and alkylated samples (data not shown).
To reassure that the correct mutation had been introduced into the purified protein and to reveal any impact thereof on the carbohydrate processing of Asn 52 , suPAR domain I of each individual mutant was excised by limited proteolysis (36). Analysis by MALDI-MS before and after removal of the Nlinked carbohydrate on domain I by N-glycanase treatment revealed that the masses recorded were compatible with the respective alanine replacements and the carbohydrate processing at Asn 52 was similar to that of the wild-type suPAR (38), the only exceptions being those point mutations that compromised the glycosylation motif itself i.e. Asn 52 3 Gln and Thr 54 3 Ala where no glycosylation were detected (Table I).
Kinetic Analysis of the Interaction between uPA and Purified Single-site suPAR Mutants-To study the impact of these single-site alanine mutations on the binding kinetics of the uPA-uPAR interaction, uPA, or its zymogen, pro-uPA was immobilized on a biosensor chip by random amine coupling and the interaction with various suPAR mutants measured in real-time by recording the changes in surface plasmon resonance upon injection of suPAR followed by buffer. Typical binding curves (sensorgrams) obtained at high receptor concentration (100 nM) are shown in Fig. 2A for wild-type suPAR and 4 selected mutants. Single-site mutations affecting the uPA-uPAR interaction by destabilizing the receptor-ligand complexes are readily recognized in such sensorgrams as they exhibit significantly increased off-rates, as illustrated in Fig. 2B. The kinetic rate constants of uPA binding were subsequently determined for each suPAR mutant after analysis of several receptor concentrations in the range of 2-200 nM (Table II). From the sche- matic representation of these data (Fig. 3A) it is evident that 5 of the 24 single-site alanine mutants exhibited off-rates greater than 5 times that of the wild-type. However, in common with other protein-protein interactions studied by alanine scanning mutagenesis (32, 44 -46), only moderate effects were observed on the association rate constants (less than 2.5-fold, see Fig.  3B). These data show that depletion of all possible interactions involving side chain atoms beyond the ␤-carbon of 4 of these 5 residues (Arg 53 , Leu 55 , Tyr 57 , or Leu 66 ) in uPAR domain I by single-site alanine substitutions caused more than a 10-fold increase in the equilibrium dissociation constant (K d ) corresponding to a change in the free energy on uPA binding of ⌬⌬G Ն 1.3 kcal/mol (Table II). A comparable change in the free energy of ligand binding (⌬⌬G) was also observed for the interaction between these suPAR mutants and immobilized pro-uPA (data not shown), which is in accordance with the location of the epitope for receptor binding on the structurally autonomous, growth factor-like module of uPA (28,(47)(48)(49). Three of the receptor residues critical for uPA binding (Arg 53 , Tyr 57 , and Leu 66 ) have previously been assigned to the structural epitope on uPAR for ligand binding, i.e. present at the receptor-ligand interface (28 -30). The data obtained in this study by alanine scanning therefore demonstrates that these residues together with Leu 55 also form part of the functional epitope for uPA binding, as they contribute significantly to the binding energetics of the receptor-ligand interaction.
Smaller effects were also consistently observed for the interaction between uPA and uPAR mutants carrying either a Thr 51 3 Ala or Asn 52 3 Gln mutation, both of which caused an approximately 6-fold increase in the K d (⌬⌬G ϳ 1 kcal/mol). From a structural point of view the involvement of Thr 51 in uPA binding is interesting, since it is located at a position normally engaged in the formation of an otherwise strictly conserved disulfide bond in the Ly-6/uPAR domain family, but which is uniquely absent from the NH 2 -terminal domains of both uPAR and C4.4 (17). The structural importance of this disulfide bond is emphasized by the severe folding problems encountered when it is removed by site-directed mutagenesis in the single domain proteins CD59 (50) and -bungarotoxin (51). It has been speculated previously that the lack of this particular disulfide bond in uPAR domain I could be of functional consequence for the receptor (15).
Enzymatic removal of the carbohydrate from Asn 52 in uPAR domain I has previously been reported to have a moderate impact on the receptor binding kinetics for uPA causing a 5-fold increase in the K d (38). Consistently, the two mutations introduced in the present study preventing subsequent glycosylation of Asn 52 (i.e. Asn 52 3 Gln and Thr 54 3 Ala) do also cause a moderate increase in the K d for uPA binding (Table II). Neither of the two tryptophan residues in uPAR seems critical for uPA binding.
To validate the data obtained by surface plasmon resonance, a selection of suPAR mutants, excluding those involving lysine replacements, were subjected to a semiquantitative cross-linking analysis (41). Complexes between uPAR and 125 I-labeled ATF were allowed to form at receptor concentrations close to the K d of the interaction for the wild-type proteins (i.e. 0.5 nM) before addition of a homobifunctional chemical cross-linker to covalently stabilize the preformed complexes. As shown by SDS-PAGE and autoradiography, alanine substitutions of any of the 4 amino acids previously assigned to the functional epitope for uPA binding by surface plasmon resonance (i.e. Arg 53 , Leu 55 , Tyr 57 , and Leu 66 ) consistently led to reduced complex formation as compared with the wild-type (Fig. 4). Likewise, a number of non-alanine suPAR mutants tested by chemical cross-linking also revealed complex levels that were in agreement with the kinetic data obtained by surface plasmon resonance (Fig. 4 versus Fig. 5A).
Further Mutations of the Functional Epitope on uPAR for uPA Binding-To further characterize the molecular interactions of the individual side chains of the key residues assigned to the functional epitope for uPA binding, we also mutated Arg 53 , Leu 55 , and Tyr 57 to other residues than alanine. uPAR mutants having Tyr 57 replaced with either Trp or Phe demonstrated uPA binding kinetics indistinguishable from those of the wild-type receptor (compared with the ⌬⌬G of 1.3 kcal/mol for the alanine mutation) highlighting the importance of an aromatic side chain at position 57 and also demonstrating that the hydroxyl group of Tyr 57 is energetically dispensable for the functional epitope (Table II and Fig. 5A). In contrast, introduction of either negatively or positively charged side chains at this position (Glu or Lys) renders the uPA-uPAR complex very unstable, causing large changes in the free energy of uPA binding (Table II).
It is well documented in the literature that human uPA has a Differences in the free-energy changes are calculated as the difference between ⌬G of mutant and wild-type receptor (⌬⌬G ϭ ⌬G mutant Ϫ ⌬G wild-type ) and were derived from the kinetic rate constants determined by surface plasmon resonance and calculated as RT 1n(K d (mut) /K d wt ).
b When analyzed by the Biaevaluation 3.0™ software the binding profiles observed for these mutants did not yield a perfect fit to a 1:1 Langmuir binding. Nevertheless, the best association and dissociation rate constants derived from such fitting attempts are shown. a very low affinity for mouse cells and vice versa (52,53). Since Leu 55 is the only residue that differs between man and mouse among those assigned so far to the uPA binding epitope on human uPAR, we also changed Leu 55 to Met, the equivalent residue present in the murine uPAR sequence (Fig. 5B). However, this conservative replacement did not affect the binding kinetics to human uPA, excluding Leu 55 as a major determinant of the observed species specificity.
Finally, we created two additional substitutions at position 53 (Arg 53 3 Leu or Lys) both of which provide reasonable isosteric replacements of the relatively long aliphatic side chain of Arg 53 , but only the Lys substitution retains the additional potential for electrostatic interactions that is provided by the guanido group of Arg 53 . As shown in Table II and Fig. 5A a lysine substitution at Arg 53 proved more permissible for highaffinity ligand binding than the corresponding alanine or leucine substitutions (⌬⌬G ϳ 0.5 kcal/mol versus 1.4 kcal/mol). It is therefore possible that the guanido group of Arg 53 may FIG. 3. Kinetic rate constants determined by surface plasmon resonance for the interaction between immobilized uPA and various single-site alanine mutants of uPAR. Panel A shows the dissociation rate constants determined for the various suPAR mutants. The broken line corresponds to a 5-fold enhancement in the dissociation rate constant compared with that of the wild-type suPAR. Panel B shows the corresponding association rate constants determined in parallel. Data is transferred from Table II. FIG. 4. Chemical cross-linking analysis of preformed complexes between ATF and soluble uPAR mutants. Purified uPAR mutants (0.5 nM) and 125 I-labeled ATF (1 nM) were incubated at 4°C for 60 min before the formed uPAR⅐ATF complexes were conjugated chemically by addition of 1 mM disuccinimidyl suberate. Shown is the autoradiogram obtained after SDS-PAGE of these samples.

FIG. 5. Secondary mutations introduced at positions in uPAR domain I previously identified as part of the functional epitope for ligand binding by alanine scanning mutagenesis.
Panel A, dissociation rate constants for the interaction between purified, soluble uPAR mutants and immobilized uPA were determined by surface plasmon resonance at 5°C. Panel B, sequence alignment of loop 3 in human and murine uPAR domain I as well as human CD59. Note that CD59 contains a consensus disulfide bond in this region uniquely lacking in uPAR domain I, but not in uPAR domains II and III (15,36). Residues assigned to the functional epitope for ligand binding by alanine scanning mutagenesis are highlighted. Residues critical for the inhibitory properties of the glycolipid anchored CD59 on the assembly of autologous complement membrane attack complex determined by alanine scanning mutagenesis are also highlighted (55).
FIG. 6. Assessment of the overall structural integrity of the various receptor mutants relative to the wildtype uPAR probed by monoclonal antibodies. The interaction between various immobilized, monoclonal anti-uPAR antibodies (R3, R5, and R9) and soluble uPAR mutants were measured by surface plasmon resonance. The dissociation rate constants are shown. The bars represent the standard deviations obtained after at least five independent determinations. The asterisks highlight residues belonging to the functional epitope determined here for uPA binding. contribute to ligand binding via electrostatic interactions.

Single-site Substitutions do Not Result in Gross Structural Perturbations within uPAR Domain I as Probed by Monoclonal
Antibodies-The structural integrity of the alanine mutants used to delineate the functional epitope for uPA binding was subsequently assessed by measuring their binding activities toward three different mouse monoclonal antibodies all having conformation-dependent epitopes on uPAR domain I (R3, R5, and R9). As shown in Fig. 6 binding of the four alanine mutants defining the functional epitope for uPA binding (i.e. Arg 53 , Leu 55 , Tyr 57 , and Leu 66 ) to domain I-specific monoclonal antibodies was essentially unchanged, indicating that the structural integrity of domain I is maintained in these mutants. A similar lack of effect was also observed with the non-alanine mutations (data not shown).
Part of the functional epitope for one of the monoclonal antibodies (R3) was identified during the present alanine scanning mutagenesis of uPAR, and being centered on Leu 61 and Lys 62 it is located in close sequence proximity to the uPAbinding site (Fig. 6, lower panel). This epitope mapping concurs with the functional properties of the R3-uPAR interaction: the antibody being a competitive inhibitor of uPA binding (25,54), preventing photochemical insertion of a photosensitive peptide antagonist of uPA binding (28), blocking the specific uPARinduced enhancement of ANS fluorescence (27) and not recognizing murine uPAR (for sequence comparison see Fig. 5B). Partly overlapping structural epitopes on human uPAR for uPA and R3 would be expected to confer severe steric hin-drance on mutual receptor binding of these protein ligands, even though their functional epitopes differ. A similar correlation between structural epitopes and the steric effects has been reported for a neutralizing antibody of the interaction between the vascular endothelial growth factor and its receptor (56). The two other monoclonal antibodies (R5 and R9) revealed no differences in their reactivities toward uPAR upon single-site alanine replacements of residues in loop 3 of domain I (Fig. 6,  middle and upper panel). Both monoclonal antibodies are noncompetitive inhibitors of uPA binding and are therefore also expected to possess structural epitopes that are non-overlapping with that for uPA (54). 4

CONCLUSIONS
Due to the involvement of the uPA system in tissue remodeling, including cancer invasion and metastasis (1,57,58), the high affinity interaction between uPA and uPAR represents an obvious target for experimental drug development (K d ϳ 0.6 nM; ⌬G ϭ Ϫ11.7 kcal/mol). To assist future attempts on rational design of low molecular weight uPAR antagonists, we have in the present study used site-directed mutagenesis to localize and characterize structural elements comprising the functional epitope on uPAR for uPA binding. Kinetic data obtained here by surface plasmon resonance for 33 single-site uPAR mutants have defined 4 positions in loop 3 of uPAR domain I as critically involved in uPA binding, i.e. Arg 53 , Leu 55 , 4 K. List, unpublished data. FIG. 7. Localization of the functional epitope for uPA binding on human uPAR. The primary sequence of human uPAR domain I is shown along with residues assigned to the functional epitope for uPA binding. For comparison the three-dimensional structure of the homologous, single domain human CD59 (20) is also shown with the residues in red corresponding to its functional epitope for complement inhibition determined by alanine scanning mutagenesis (55). The three loops of the individual Ly-6/uPAR type modules are indicated by Arabic numerals. The secondary structure of CD59 is highlighted as follows: ␤-sheets (green arrows), ␣-helix (red cylinder), and disulfide bonds (yellow). Tyr 57 , and Leu 66 (illustrated in Fig. 7). A comparison to some of the single domain members of the Ly-6/uPAR superfamily, for which three-dimensional structures as well as mutagenesis data are available, reveals that the position of the functional epitope in this domain family varies. The glycolipid anchored CD59 has a functional epitope for inhibition of autologous complement attack located on loop 3 (Fig. 7) (55), whereas the functional epitope on erabutoxin a for the nicotinic acetylcholine receptor contains 10 residues that are primarily located on loops 1 and 2 (19). The functional epitope on uPAR for uPA binding possibly includes additional residues besides those we have assigned to loop 3 of domain I. Accordingly, we have identified a composite structural epitope for a decamer peptide antagonist of uPA binding involving loop 3 of both uPAR domain I (Arg 53 and Leu 66 ) and domain III (His 251 ) (28,29). Furthermore, both uPA and the peptide antagonist require the full 3-domain receptor protein for a productive high-affinity interaction (27,28). We are therefore currently searching for additional residues of the functional epitope for uPA binding by single-site mutagenesis in both uPAR domains I and III.