uPA (urokinase-type plasminogen activator) is a potential therapeutic target in a variety of pathological conditions, including cancer. In order to find new principles for inhibiting uPA in murine cancer models, we screened a phage-displayed peptide library with murine uPA as bait. We thereby isolated several murine uPA-binding peptide sequences, the predominant of which was the disulfide-bridged constrained sequence CPAYSRYLDC, which we will refer to as mupain-1. A chemically synthesized peptide corresponding to this sequence was found to be a competitive inhibitor of murine uPA, inhibiting its activity towards a low-molecular-mass chromogenic substrate as well as towards its natural substrate plasminogen. The Ki value for inhibition as well as the KD value for binding were approx. 400 nM. Among a variety of other murine and human serine proteases, including trypsin, mupain-1 was found to be highly selective for murine uPA and did not even measurably inhibit human uPA. The cyclic structure of mupain-1 was indispensable for binding. Alanine scanning mutagenesis identified Arg6 of mupain-1 as the P1 residue and indicated an extended binding interaction including the P5, P3, P2, P1 and P1′ residues of mupain-1 and the specificity pocket, the catalytic triad and amino acids 41, 99 and 192 located in and around the active site of murine uPA. Exchanging His99 of human uPA by a tyrosine residue, the corresponding residue in murine uPA, conferred mupain-1 susceptibility on to the latter. Peptide-derived inhibitors, such as mupain-1, may provide novel mechanistic information about enzyme–inhibitor interactions, provide alternative methodologies for designing effective protease inhibitors, and be used for target validation in murine model systems.

INTRODUCTION

Serine proteases of the trypsin family (clan SA) have many physiological and pathophysiological functions. There is therefore extensive interest in generating specific inhibitors to be used for pharmacological interference with their enzymatic activity. Moreover, serine proteases are classical subjects for studies of catalytic and inhibitory mechanisms (for a review, see [1]). One interesting member of the trypsin family of serine proteases is uPA (urokinase-type plasminogen activator), which catalyses the conversion of the zymogen plasminogen into the active protease plasmin through cleavage of the Arg15–Val16 bond [2]. uPA has a catalytic serine protease domain with a trypsin-like fold. Besides the serine protease domain, uPA has an N-terminal extension consisting of a kringle domain and an epidermal growth factor domain, which binds to the cell-surface-anchored uPAR (uPA receptor) [3]. uPA-catalysed plasmin generation participates in turnover of extracellular matrix proteins in physiological tissue remodelling (for a review, see [4]). Abnormal expression of uPA is responsible for tissue damage in several pathological conditions, including rheumatoid arthritis, allergic vasculitis and xeroderma pigmentosum. In particular, abnormal expression of uPA is a key factor for the invasive capacity of malignant tumours (for reviews, see [5,6]).

As with other serine proteases, there has been extensive interest in generating specific inhibitors of the enzymatic activity of uPA. The plasminogen activation activity of huPA (human uPA) can be inhibited specifically by monoclonal antibodies [7]. Several inhibitory monoclonal antibodies have epitopes encompassing the 37- and 60-loops [8]. Another type of protein protease inhibitor is the dimeric non-specific serine protease inhibitor ecotin, binding two serine protease molecules per ecotin dimer through interactions with both the protease active site and an exosite. With some success, ecotin has been converted into a high-affinity huPA inhibitor by engineering each of the two interaction sites [9]. Moreover, several classes of low-molecular-mass organochemical inhibitors of huPA have been synthesized. The binding modes for many of these were studied by X-ray crystal structure analyses. An important feature of such inhibitors is an arginine analogue inserting into the S1 pocket of uPA. However, the challenge has been to achieve selectivity for uPA over other serine proteases with P1 arginine specificity, by exploiting small variations in the subsite geometry of the S1 pocket and its surroundings [1015].

A number of murine cancer model systems are available (for a review, see [16]), and it is of great interest to use these for validation of uPA as a therapeutic target. It is a general assumption that the tumour biological function of uPA may rely on uPA produced either by the cancer cells themselves or by stromal cells. The availability of inhibitors specific for huPA and muPA (murine uPA) respectively, will allow evaluation of the relative importance of cancer and host uPA in xenotransplanted models with human tumours growing on immunodeficient mice. However, there is presently a lack of specific inhibitors of muPA. The only available specific inhibitors of muPA are polyclonal antibodies [17]. A series of naphtamidine-based inhibitors targeted to the 60-loop and S1β pocket of huPA had only very weak inhibitory effects on muPA [18]. A 4-amidinobenzylamine-derived peptide inhibitor did inhibit rat uPA with a Ki of 19 nM, but lacked general specificity, inhibiting trypsin and uPA with almost equal potency [11].

By screening a phage-displayed peptide library with huPA as bait, we previously isolated a cyclic, disulfide-bridged constrained peptide, referred to as upain-1. It inhibits huPA competitively, with a Ki value of approx. 2 μM [19]. Site-directed mutagenesis [19] and X-ray crystal structure analysis of the upain-1–huPA complex [20] demonstrated that this peptide forms an extended interaction surface with huPA, involving not only the active site but also several surface loops. The extended interaction surface can account for the high specificity of this inhibitor, being highly selective for huPA among a number of human serine proteases. Remarkably, upain-1 did not measurably inhibit muPA [19].

With the purpose of isolating a similar inhibitor of muPA to be used in murine cancer models, we have now screened a phage-displayed peptide-library using muPA as the bait. We have thereby isolated a highly specific inhibitor of muPA.

EXPERIMENTAL

Serine proteases

Active muPA was purchased from Molecular Innovation. Acitive huPA was from Wakamoto Pharmaceutical. An expression construct in the vector pcDNA3.1(−), harbouring full-length muPA, with an His6 (hexahistidine) tag extension at the C-terminus, was generated by standard methods from a construct kindly provided by Kasper Almholt (Finsen Laboratory, Copenhagen, Denmark). Wild-type and mutant recombinant huPA and muPA were expressed in HEK (human embryonic kidney)-293T cells [8].

Human Glu-plasminogen and human plasmin was from American Diagnostica. Human aPC (activated protein C) was from Maxygen. Human plasma kallikrein was a gift from Inger Schousboe (Department of Medical Biochemistry and Genetics, University of Copenhagen, Copenhagen, Denmark). Human thrombin was a gift from Dr John Fenton (New York State Department of Health, Albany, NY, U.S.A.). Human Factor VIIa was from Enzyme Research. Human tPA (tissue plasminogen activator) was from Genentech. This enzyme was acquired as a mixture of one-chain and two-chain molecules and was converted into the active two-chain form with immobilized plasmin [21]. Bovine β-trypsin was from Roche Applied Sciences. It was purified from a tosylphenylalanylchloromethylketone-treated commercial preparation by chromatography on soybean trypsin inhibitor–Sepharose [22]. Murine aPC, murine Factor IXa, murine Factor Xa, murine thrombin, murine tPA, murine plasma kallikrein and murine plasmin were purchased from Molecular Innovation.

Antibodies

The monoclonal antibodies MA-H77B6 and MA-H77A10 against the A-chain of muPA were purchased from Molecular Innovation. HRP (horseradish peroxidase)-conjugated monoclonal anti-M13 phage antibody directed against the major phage coat protein g8p was from Amersham Biosciences. HRP-conjugated rabbit anti-mouse IgG and HRP-conjugated swine anti-rabbit IgG was from Dako. Monoclonal anti-huPA antibody mAb 6 was as described previously [8].

Miscellaneous materials

All enzymes used for DNA technology were from New England Biolabs. Oligonucleotides were purchased from DNA Technology or MWG Biotech AG. Expression vector pET20b(+) was from Merck Novagen. pcDNA3.1(−) was from Invitrogen. All DNA constructs and mutations were verified by sequencing.

The chromogenic protease substrates S-2444 (pyro-EGR-p-nitroaniline), S-2288 (H-D-IPR-p-nitroaniline), S-2403 (pyro-EFK-p-nitroaniline), S-2238 (H-D-Pro-piperidine-Arg-p-nitroaniline), S-2222 (benzyl-IEGR-p-nitroaniline), S-2366 (pyro-EPR-p-nitroaniline), S-2765 (benzyloxycarbonyl-D-RGR-p-nitroaniline) and S-2302 (H-D-PFR-p-nitroanline) were from Chromogenix. SpectrozymeFVIIa (methansulfonyl-D-cyclohexylalanyl-butyl-Arg-p-nitroaniline) was from American Diagnostica. VLK-AMC (H-D-VLK-7-amido-4-methylcoumarin) was from Bachem. PAB (para-aminobenzamidine) and amiloride was from Sigma–Aldrich.

Synthetic peptides were purchased from Peptide Protein Research.

Screening of phage-displayed random peptide libraries for peptides binding to muPA

Monoclonal antibodies (5 μg/ml) against the A-chain of muPA (MA-H77B6 or MA-H77A10) were immobilized in Maxisorp Nunc-Immuno Tubes (Nunc) overnight at 4 °C in 0.1 M NaHCO3/Na2CO3 (pH 9.6). Non-specific binding was blocked by incubation for 1 h at room temperature (25 °C) with HBS (Hepes-buffered saline; 140 mM NaCl and 10 mM Hepes, pH 7.4) containing 5% non-fat skimmed milk powder. muPA (200 nM) was then incubated in the tubes for 1 h at room temperature in blocking buffer [HBS containing 5% (w/v) non-fat dried skimmed milk powder], followed by 1 h incubation with 1011 colony-forming units from each of two phage-displayed peptide libraries, in the formats CX7C and CX8C, kindly provided by Dr Erkki Koivunen (Department of Biosciences, University of Helsinki, Finland). The theoretical diversity of the library was >109 unique peptide inserts [23]. Subsequent to ten washes with HBS, the bound phage particles were eluted with 1 ml of 10 mM HCl/glycin (pH 2.2). The eluted phage particles were neutralized with 0.5 ml of 1 M Tris (pH 8.0) and propagated in Escherichia coli JM109 cells overnight at 37 °C. The bacterial cells were removed by centrifugation (8500 g for 15 min at 4 °C) and the secreted phage particles were precipitated by adding 0.25 vol of 2.5 M NaCl and 20% poly(ethylene glycol) 2000, incubated for 30 min at 0 °C, and centrifuged (15500 g for 15 min at 4 °C). The pellet were resuspended in HBS with 10% glycerol. Four successive rounds of selection were performed. Alternating antibodies were used for subsequent rounds of selection in order to avoid enrichment of antibody-binding phage particles.

Expression of the mupain-1 peptide sequence in fusion with the N-terminal domains of the phage coat protein g3p

A DNA fragment encoding the first two domains of the phage coat protein g3p (D1 and D2, residues 1–219) was amplified from the phage fUSE5 [24] with the PCR primers fUSEfwd (5′-CATGCCATGGGCTCGGCCGACGGGGC-3′) and fUSEbck2 (5′-GTACCTCGAGGCCGCCAGCATTGACAGG-3′), using the Pfu Turbo DNA polymerase (Stratagene).The generated PCR product was purified with the QIAquick PCR Purification kit (Qiagen) and ligated into the E. coli expression vector pET20b(+) (Merck Novagen) via NcoI and XhoI restriction sites. The resulting vector will be referred to as pETD1D2. Using the same approach, but with the mupain-1 phage as a template for the PCR reaction, a vector was created for expression of the mupain-1 sequence fused to D1D2 of g3p (MGASADGACPAYSRYLDC-GAAG-g3p1–219-LEHHHHHH, the mupain-1 sequence is underlined). This fusion protein will be referred to as mupain-1–D1D2. Expression vectors for derivatives of mupain-1–D1D2 were generated by site-directed mutagenesis using pETmupain-1–D1D2 as a template. The fusion proteins were expressed from cultures of E. coli BL21(DE3)pLysS (Merck Novagen) containing the relevant plasmids and purified by Ni+ chelate affinity chromatography [25,26], subjected to size-exclusion chromatography on Superdex 75 (Amersham Bioscience) equilibrated in HBS, and finally concentrated with Centricon centrifugal filter devices (Millipore).

ELISA for measuring phage particles binding to muPA

Unless otherwise stated, the buffer used was HBS supplemented with 5% non-fat skimmed milk powder.

The relative concentration of muPA variants in conditioned medium was determined by a quantitative ELISA in which muPA was captured on Ni-NTA (Ni2+-nitrilotriacetate) HisSorb™ Strips (Qiagen) via the His6 tag and detected with a monoclonal antibody (MA-H77B6). The procedure was otherwise as described below.

For the antibody-based ELISAs, a monoclonal antibody to be immobilized on the solid phase (2.5 μg/ml in 100 mM NaHCO3/Na2CO3, pH 9.6) was coated in the wells of a 96-well Maxisorp plate (Nunc), followed by blocking with HBS containing 5% non-fat skimmed milk powder. The wells were incubated with 20 nM muPA for 1 h at room temperature. After washing with HBS, the wells were incubated with phage particles (∼109 colony-forming units/ml) for 1 h. In some cases, up to 3 mM PAB or amiloride were added together with the phage particles. The wells were then incubated for 1 h with a 5000-fold dilution of HRP-conjugated anti-M13 monoclonal antibody. The wells were developed by the addition of 0.5 mg/ml o-phenylenediamine (100 μl) (KemEn Tech) in 50 mM citric acid (pH 5.0), supplemented with 0.03% H2O2. When suitable colour had developed, the reactions were quenched with 50 μl of 1 M H2SO4. The A492 of the wells was read in a microplate reader. When testing the binding of mupain-1 phage to variants of muPA, conditioned medium from muPA-expressing HEK-293T cells was used directly in the ELISA. Each muPA variant was captured via the His6 tag on Ni-NTA HisSorb™ Strips (Qiagen). To ensure capture of equal amounts of the muPA variants on the solid phase, an ELISA with a monoclonal anti muPA anti-body (MA-H77B6, 1 μg/ml) instead of the phage particles and a HRP-conjugated rabbit anti-mouse antibody (1:2000, DAKO) instead of the anti-M13 antibody was performed in parallel.

Determination of the inhibition constant (Ki)

In order to determine the Km, app and Vmax, app values for S-2444 hydrolysis by muPA at various inhibitor concentrations, 4 nM muPA was incubated with various concentrations of mupain-1 peptide (0–2 μM) in HBS supplemented with 0.1% BSA at 37 °C for 15 min prior to the addition of the chromogenic substrate, S-2444. Each inhibitor concentration was combined with a series of S-2444 concentrations over the range 0–12 mM. For each inhibitor concentration ([I0]), the initial velocity, monitored as an absorbance of 405 nm, were determined at several substrate concentrations. The initial velocities V were plotted against the substrate concentrations. According to standard Michaelis–Menten kinetics, the Vmax, app and the Km, app values were determined at each inhibitor concentration by a non-linear fit to the equation (eqn 1):

 
formula
(1)

The following equation (eqn 2) is expected to apply for competitive inhibition according to Michaelis–Menten kinetics:

 
formula
(2)

In eqn (2), [S]0 and [I]0 are the total substrate and inhibitor concentrations respectively; Ki is the inhibition constant; Km is the Michaelis constant for S-2444 under the assay conditions. From eqn (1), Km, app can be defined as (eqn 3):

 
formula
(3)

In the case of competitive inhibition, Km, app, according to eqn (2), is expected to have a linear relationship to [I]0, whereas Vmax, app will be independent of [I]0. The Ki values can thus be estimated from the slope of the line relating Km, app to [I]0.

For routine determination of Ki values for the inhibition of purified recombinant protease under equilibrium inhibition conditions, a fixed concentration of the protease was pre-incubated in HBS with 0.1% BSA at 37 °C at pH 7.4, unless otherwise stated, with various concentrations of mupain-1 peptide (0–50 μM) or mupain-1–D1D2 (0–150 μM) for 15 min prior to the addition of the appropriate chromogenic substrate. The following protease–substrate combinations were used: muPA (4 nM) and S-2444 (750 μM); huPA (4 nM) and S-2444 (47 μM); human tPA (2.0 nM) and S-2288 (300 μM); murine tPA (2.0 nM) and S-2765 (250 μM); human plasmin (2.0 nM) and S-2403 (125 μM); murine plasmin (2.0 nM) and S-2366 (200 μM); human thrombin (0.5 nM) and S-2238 (50 μM); murine thrombin (0.5 nM) and S-2238 (100 μM); bovine β-trypsin (2.0 nM) and S-2222 (50 μM); murine β-trypsin (2.0 nM) and S-2222 (200 μM); human aPC (8.5 nM) and S-2366 (300 μM); murine aPC (8.5 nM) and S-2366 (500 μM); human Factor VIIa (10 nM) and SpectrozymefVIIa (500 μM); human Factor Xa (0.5 nM) and S-2765 (100 μM); murine Factor Xa (0.5 nM) and S-2765 (500 μM); human plasma kallikrein (4.0 nM) and S-2302 (300 μM); murine plasma kallikrein (4.0 nM) and S-2302 (125 μM); murine Factor IXa (5 nM) and SpectrozymefIXa (1250 μM). The initial velocities were monitored as changes in the absorbance at 405 nm. As above, the Km values for these proteases were determined according to standard Michaelis–Menten kinetics. The Ki values were subsequently determined from the non-linear regression analyses of plots for Vi/V0 against [I]0, using an [S]0 near the Km value (eqn 4):

 
formula
(4)

In this equation, Vi and V0 are the reaction velocities in the presence and absence of inhibitor respectively.

The Ki values for inhibition of different huPA variants by mupain-1 were determined essentially as described above, except for the following modifications: 20 μl of conditioned HEK-293T cell medium was used for each huPA variant instead of the purified protease and an S-2444 concentration near the Km.

SPR (surface plasmon resonance) analysis

The equilibrium binding constants (KD) for the binding of mupain-1–D1D2 and mupain-1 to muPA were determined by SPR on a BIACORE T100 instrument (Biacore) equipped with a CM5 censor chip. Two different formats were used. In the first, an anti-muPA antibody against the A-chain (MA-H77A10) was immobilized to a density of 2000 RU/mm2 (where RU is response unit). muPA (100 nM) in running buffer [30 mM Hepes (pH 7.4), 135 mM NaCl and 1 mM EDTA) supplemented with 0.05% Tween 20 and 0.1% BSA, was injected at a flow rate of 30 μl·min−1 for 30 s until a capture level of ∼250 RU/mm2. Mupain-1–D1D2 or mupain-1 peptide, at various concentrations, were injected at 30 μl·min−1 at 25 °C in running buffer for 30 s. After discontinuation of the injection, dissociation of bound ligand was monitored for another 120 s. Binding was expressed in relative RUs as the response obtained from the flow cell containing immobilized ligand minus the response obtained from a reference flow cell in which the ligand (muPA) had been omitted. In the second format, mupain-1–D1D2 was coupled to a density of 1000 RU/mm2. muPA (15–0.5 μM, in 2-fold dilution steps in running buffer supplemented with 0.05% Tween 20 and 0.1% BSA) was injected at a flow rate of 30 μl·min−1 at 25 °C for 30 s. After discontinuation of the injection, dissociation of bound ligand was followed for another 120 s. Binding was expressed in relative RUs as the response obtained from the flow cell containing immobilized ligand minus the response obtained from a reference flow cell in which D1D2 has been immobilized to a density of 1000 RU/mm2. In both formats the association rate constants (kon values), the dissocation rate constants (koff values) and the equilibrium binding constant (KD) were estimated by global fitting to a 1:1 binding model, using the BIACORE T100 evaluation software.

Measurements of the binding of mupain-1 peptide variants, in which single amino acids have been substituted with an alanine residue as well as a linear version with the cysteine residue residues replaced by a serine residue, to muPA were performed as described above. However, in most of these cases the association and dissociation reactions were too fast for determination of the rate constants. The equilibrium binding constants (KD) were therefore determined by fitting the steady-state binding values to a single binding site saturation model using the equation (eqn 5):

 
formula
(5)

In this equation, RUsteady state equals the steady-state binding measured in RU. Rmax is maximum binding in RU.

In order to estimate the relative binding of mupain-1 to a series of muPA variants, conditioned medium from cells transfected with the corresponding muPA variant cDNA were applied to a chip at a flow rate of 30 μl·min−1 for 11 min. The binding of mupain-1–D1D2 to each variant was scored as the steady-state binding level compared with the muPA capture level and expressed relative to the binding of mupain-1–D1D2 to muPA wild-type in the same experiment.

Measurements of the binding between mupain-1 peptide to variants of huPA were performed using conditioned medium from cells transfected with the corresponding huPA variant cDNA. huPA was captured on the monoclonal anti-huPA antibody mAb6, immobilized on a CM5 chip to a density of ∼4500 RU/mm2. Conditioned medium containing huPA was injected at a flow rate of 30 μl·min−1 for 120 s until a capture level of ∼950 RU/mm2. Mupain-1 peptide in a concentration range 0–50 μM was injected in running buffer [30 mM Hepes (pH 7.4), 135 mM NaCl and 1 mM EDTA supplemented with 0.05 % Tween 20 and 0.1 % BSA] at a flow rate of 30 μl·min−1 and a contact time of 30 s. After discontinuation of the injection, dissociation of bound ligand was followed for 60 s and the KD was determined by fitting the steady-state binding values to a single binding site saturation model.

Investigation of substrate behaviour of mupain-1

To determine whether wild-type mupain-1–D1D2 behaved as a substrate toward closely related serine proteases, mupain-1–D1D2 was incubated at a final concentration of 2.5 μM in the presence of 50 nM bovine β-trypsin, human uPA, human aPC, human tPA, human thrombin, human Factor VIIa, human plasma kallikrein, human Factor Xa, murine tPA, murine thrombin, murine plasmin, murine Factor IVa, murine Factor Xa, murine plasma kallikrein and murine trypsin respectively, in HBS at 37 °C for different time periods (0–120 min). The reactions were quenched by adding 0.1 vol of 0.1 M HCl. The reaction products were analysed by SDS/PAGE under reducing conditions and Coomassie Blue staining.

Plasminogen activation assay in vitro

muPA (0.25 nM) was pre-incubated with increasing concentrations of mupain-1 (0–10 μM) in HBS containing 0.1% BSA for 15 min at 37 °C. Plasminogen (0.5 μM) and the chromogenic plasmin substrate S-2251 (500 μM) were added and the change in absorbance at 405 nm was recorded at regular intervals for a time period of 120 min.

Plasminogen activation assay in the presence of WEHI-3 cells

WEHI-3 cells, grown in DMEM (Dulbecco's modified Eagle medium) supplemented with 10% (v/v) fetal bovine serum (Cambrex Bio Science), 4.5 g/l glucose, 10 units/ml penicillin, 100 μg/ml streptomycin and 2 mM L-glutamine (Cambrex Bio Science), were washed extensively in PBS, and resuspended in TBS [50 mM Tris and 100 mM NaCl (pH 7.4)] with 0.1% BSA at a density of 107 cells/ml. The cells were pre-incubated for 1 h at 37 °C with 50 nM muPA, and subsequently washed to remove unbound muPA. They were then aliquoted into the wells of 96-well microtitre plates (Nunc) and pre-incubated for 15 min at room temperature with mupain-1. Plasminogen (0.5 μM) and VLK-AMC (200 μM) were added to a total volume of 200 μl. The fluorescence of the wells was monitored at 2 min intervals in a Spectromax Gemini fluorescence plate reader (Molecular Devices) using an excitation wavelength of 390 nm and an emission wavelength of 480 nm.

Stability of mupain-1 in serum

Mupain-1 (400 μM) was incubated in 77% fetal bovine serum or 77% mouse serum (Dako Denmark A/S) at 37 °C for 0, 1, 2 or 20 h. After the incubation, 1/15 volume of 1 M glycine/HCl (pH 3) and 1/15 volume of 1 M HCl was added to inactivate α2-macroglobulin. The acid treatment was followed by neutralization with 4 vol. of 30 mM Hepes (pH 7.4), 135 mM NaCl, 1 mM EDTA and 0.1% BSA. Aliquots of various volumes of incubation mixture were then mixed with a fixed concentration of muPA (4 nM). The remaining activity of muPA was determined by an S-2444 assay.

RESULTS

Selection of a peptide binding to muPA

For isolating muPA binding peptides, we used a phage-displayed random peptide library with peptide sequences in the formats CX7C and CX8C, where X denotes random natural amino acids and the flanking cysteine residues are oxidized, resulting in constrained circular peptides. The theoretical diversity of the library was >109 unique peptide inserts. As bait, we used muPA immobilized on a monoclonal anti-muPA antibody with an epitope in the A-chain. Four rounds of selection were performed. Among 24 individual phage clones showing muPA-binding in ELISA, all had sequences which were variations on a theme that included a centrally placed arginine residue. One of the two most common sequences, displayed on six phage clones out of the 24, had the sequence CPAYSRYLDC. This sequence was chosen for further analysis and will be referred to as mupain-1. The other common sequence, CPLYNRMIGC, will be referred to as mupain-2.

The selected peptide sequence is a competitive inhibitor of muPA

A peptide corresponding to the mupain-1 sequence, with oxidized cysteine residues, was synthesized chemically. Also, to be able to analyse muPA binding of the mupain-1 sequence in its native state, i.e. at the N-terminus of the phage coat protein g3p, the mupain-1 sequence was expressed in fusion with the two N-terminal domains of g3p, D1 and D2. The fusion protein will be referred to as mupain-1–D1D2.

The mupain-1 peptide and mupain-1–D1D2 inhibited the peptidolytic activity of muPA against the chromogenic substrate S-2444, whereas the fusion partner D1D2, without the mupain-1 sequence, showed no or only weak inhibitory activity, as no inhibition could be measured with D1D2 concentrations up to 150 μM. By determining the rate of S-2444 hydrolysis at several substrate and several inhibitor concentrations, an apparent Km (Km, app) could be determined at each inhibitor concentration. The Km, app increased linearly with the inhibitor concentration, whereas Vmax, app changed only negligibly (Table 1), clearly in agreement with the expected inhibition mode of a competitive inhibitor (eqn 3). We therefore concluded that mupain-1 competitively inhibits muPA.

Table 1
Mupain-1 is a competitive inhibitor of muPA activity

Initial rates of muPA-catalysed hydrolysis of various concentrations of S-2444 in the presence of various concentrations of mupain-1 were estimated by monitoring the time course of the absorbance at 405 nm. The Km, app and the Vmax, app values were determined by non-linear regression analysis of rate–substrate data pairs on the basis of eqn (1). Values are means±S.D. for two independent determinations.

[Mupain-1] (μM) Km, app (mM) Vmax, app (10−8 M/s) 
2.6±0.2 3.9±0.2 
0.25 3.0±0.4 3.5±0.2 
0.50 3.9±0.6 3.7±0.3 
5.5±0.4 3.5±0.2 
8.6±0.5 3.4±0.1 
[Mupain-1] (μM) Km, app (mM) Vmax, app (10−8 M/s) 
2.6±0.2 3.9±0.2 
0.25 3.0±0.4 3.5±0.2 
0.50 3.9±0.6 3.7±0.3 
5.5±0.4 3.5±0.2 
8.6±0.5 3.4±0.1 

In our analysis of inhibition kinetics, we routinely used plots of Vi/V0 against inhibitor concentration at a single substrate concentration near the Km value and analysis by non-linear regression curve-fitting according to eqn (4) and independently determined Km values for muPA-catalysed S-2444 hydrolysis.

The Ki value for inhibition of muPA by the mupain-1–D1D2 construct was significantly higher (10-fold) than the corresponding values for the chemically synthesized peptide, suggesting that presentation of the peptide sequence in a protein scaffold considerably changed the affinity for the muPA target.

The Ki value did not depend on the pH (results not shown).

Mupain-1 inhibits muPA-catalysed activation of plasminogen in vitro and on cell surfaces

The effect of mupain-1 on the plasminogen activation activity of muPA was tested in a coupled plasminogen activation assay. muPA was incubated with the peptide prior to the addition of plasminogen and a chromogenic substrate for plasmin (S-2251). Figure 1(A) illustrates how increasing amounts of mupain-1 inhibit the plasminogen activation activity of muPA in a dose-dependent manner.

The effect of mupain-1 on the plasminogen activation activity of muPA

Figure 1
The effect of mupain-1 on the plasminogen activation activity of muPA

(A) muPA (0.25 nM) was incubated for 15 min without (●) or with 0.1 μM (○), 0.25 μM (▼), 0.5 μM (▽), 1 μM (■), 2.5 μM (□), 5 μM (♦) or 10 μM (♦) mupain-1, after which plasminogen (0.5 μM) and S-2251 (500 μM) were added. The reaction was monitored spectrophotometrically at 405 nm. (B) Effect on cell-surface-associated plasminogen activation. Aliquots of WEHI-3 cell suspension (107 cells/ml) that had been saturated with muPA were incubated at room temperature for 15 min without addition (○) or in the presence of 0.1 μM (▼), 1 μM (▽), 5 μM (■), 20 μM (□), 100 μM (♦) mupain-1 or 100 μM mupain-1(R6A) (♦). Cells without muPA (●) served as a negative control. Plasminogen and the fluorogenic plasmin substrate VLK-AMC were added and fluorescence was measured over time.

Figure 1
The effect of mupain-1 on the plasminogen activation activity of muPA

(A) muPA (0.25 nM) was incubated for 15 min without (●) or with 0.1 μM (○), 0.25 μM (▼), 0.5 μM (▽), 1 μM (■), 2.5 μM (□), 5 μM (♦) or 10 μM (♦) mupain-1, after which plasminogen (0.5 μM) and S-2251 (500 μM) were added. The reaction was monitored spectrophotometrically at 405 nm. (B) Effect on cell-surface-associated plasminogen activation. Aliquots of WEHI-3 cell suspension (107 cells/ml) that had been saturated with muPA were incubated at room temperature for 15 min without addition (○) or in the presence of 0.1 μM (▼), 1 μM (▽), 5 μM (■), 20 μM (□), 100 μM (♦) mupain-1 or 100 μM mupain-1(R6A) (♦). Cells without muPA (●) served as a negative control. Plasminogen and the fluorogenic plasmin substrate VLK-AMC were added and fluorescence was measured over time.

The effect of mupain-1 on the plasminogen activation activity of muPA bound to uPAR on the surface of murine myelomonocytic WEHI-3 cells was determined by a coupled plasminogen activation assay. Aliquots of WEHI-3 cell suspension (107 cells/ml) that had been saturated with muPA were incubated at room temperature for 15 min without addition or in the presence of increasing amounts of mupain-1. Cells without muPA served as a negative control. Plasminogen and the fluorogenic plasmin substrate VLK-AMC were added to start the reaction and fluorescence was measured over time. Mupain-1 was found to inhibit the uPAR-bound muPA-catalysed plasminogen activation on the cell surface in a dose-dependent manner (Figure 1B).

The equilibrium binding constant for mupain-1–D1D2 and mupain-1 peptide binding to muPA

The equilibrium binding constant, KD, for the binding of mupain-1–D1D2 and mupain-1 peptide to muPA was determined using SPR with a BIACORE T100 instrument (Table 2). In one format, muPA was immobilized on a CM5 chip via a monoclonal anti-muPA antibody with an epitope in the muPA A-chain. The fusion protein (Figure 2A) or the peptide (Figure 2B) was injected at various concentrations. The rate constants for association and dissociation of the muPA–mupain-1–D1D2 complex and the muPA–mupain-1 complex were estimated by global fitting with BIACORE T100 Evaluation Software, assuming a simple 1:1 binding model. The KD values were then calculated as the koff/kon ratios. The estimated KD values agree excellently with Ki values from the enzyme kinetic measurements (see above). Using an alternative format, in which mupain-1–D1D2 was immobilized on the chip and muPA injected, the same KD value was found as with immobilized muPA, although the rate constants differed somewhat.

Table 2
KD values for mupain-1–muPA binding

A monoclonal anti-muPA antibody was immobilized on a CM5 sensor chip in a BIACORE T100. muPA (100 nM) was injected. After having obtained a stable muPA capture level, different concentrations of mupain-1 peptide (0–5 μM) or mupain-1–D1D2 (0–25 μM) were injected and the association and dissociation time courses were followed. In the inverse experimental setup, mupain-1–D1D2 was immobilized, muPA (0–15 μM) was injected and the association and dissociation was followed over time. The data was fitted to the simple exponential equations for association and dissociation and the apparent rate of association and dissociation were determined. Values are means±S.D. for the indicated number of experiments (shown in parentheses).

On the solid phase In the fluid phase kon (μM−1s−1koff (s−1KD (μM) 
muPA Mupain-1 peptide 0.16±0.04 (4) 0.050±0.020 (4) 0.33±0.11 (4) 
muPA Mupain-1-D1D2 0.053±0.038 (3)* 0.066±0.033 (3) 14±2.3 (3)* 
Mupain-1–D1D2 MuPA 0.018±0.002 (3) 0.252±0.004 (3) 14±0.9 (3) 
On the solid phase In the fluid phase kon (μM−1s−1koff (s−1KD (μM) 
muPA Mupain-1 peptide 0.16±0.04 (4) 0.050±0.020 (4) 0.33±0.11 (4) 
muPA Mupain-1-D1D2 0.053±0.038 (3)* 0.066±0.033 (3) 14±2.3 (3)* 
Mupain-1–D1D2 MuPA 0.018±0.002 (3) 0.252±0.004 (3) 14±0.9 (3) 
*

Significantly different from the corresponding value for the mupain-1 peptide (P<0.01).

Association and dissociation curves for mupain-1–D1D2 and mupain-1 peptide binding to muPA, as studied by SPR

Figure 2
Association and dissociation curves for mupain-1–D1D2 and mupain-1 peptide binding to muPA, as studied by SPR

A monoclonal anti-muPA antibody was immobilized on a CM5 sensor chip in a BIACORE T100 machine. muPA (100 nM) was injected to a capture level of ∼250 RU. Mupain-1–D1D2 (0–25 μM) or mupain-1 (0–5 μM) were injected and the association and dissociation time courses were monitored. The kinetic parameters kon, koff, and KD were estimated by global fitting to a 1:1 binding model, using the BIACORE T100 evaluation software (Table 2).

Figure 2
Association and dissociation curves for mupain-1–D1D2 and mupain-1 peptide binding to muPA, as studied by SPR

A monoclonal anti-muPA antibody was immobilized on a CM5 sensor chip in a BIACORE T100 machine. muPA (100 nM) was injected to a capture level of ∼250 RU. Mupain-1–D1D2 (0–25 μM) or mupain-1 (0–5 μM) were injected and the association and dissociation time courses were monitored. The kinetic parameters kon, koff, and KD were estimated by global fitting to a 1:1 binding model, using the BIACORE T100 evaluation software (Table 2).

Determination of the protease specificity for mupain-1

We determined Ki values for the inhibition of several other related murine and human clan SA serine proteases that prefer a basic P1 residue, using mupain-1 peptide in competitive chromogenic substrate assays, each one with the corresponding optimal substrate for the respective protease. We observed no inhibition of murine or human tPA, aPC, plasmin, thrombin, plasma kallikrein, Factor VIIa, Factor IXa and Factor Xa by mupain-1 even at the highest concentrations tested, 150 μM. If one then assumes a lower limit for detection of approx. 10% inhibition, we estimated, on the basis of eqn (4), the Ki values for inhibition of these proteases. Mupain-1 inhibited β-trypsin with a Ki value 50–100-fold higher than that for muPA (Table 3). To further characterize the interaction of mupain-1 with uPA-related proteases, we investigated whether any of the tested proteases were able to cleave the mupain-1 sequence. Mupain-1–D1D2 (2.5 μM) was incubated with individual proteases (50 nM) at 37 °C for the time period 0–120 min, and the reaction products were analysed by SDS/PAGE. Under these conditions none of the tested proteases were able to cleave mupain-1 (results not shown).

Table 3
Reactivity of mupain-1 toward serine proteases related to muPA

Kinetic parameters were determined in assays with the indicated chromogenic substrates in HBS with 0.1% BSA at pH 7.4 and at 37 °C, except for human fVIIa, which was assayed in TBS at pH 8.1. The reported inhibition constants were determined from the combined data of three independent experiments and global evaluation by eqn (4). Values are reported as the fitted value±S.D. after correction for the competitive effect of the chromogenic substrate. The numbers in parentheses indicate the number of experiments.

Protease Substrate Km(μM) Ki(μM) 
muPA S-2444 1190±0.3 (3) 0.412±0.180 (7) 
Murine tPA S-2765 187±13 (3) >580* 
Murine plasmin S-2366 179±7 (3) >640* 
Murine thrombin S-2238 37±9 (3) >370* 
Murine β-trypsin S-2222 55±8 (3) 45±3 (3)* 
Murine aPC S-2366 584±39 (3) >730* 
Murine Factor IXa SpectrozymefIXa >1250† >675* 
Murine Factor Xa S-2765 78±23 (3) >180* 
Murine plasma kallikrein S-2302 116±40 (3) >650* 
huPA S-2444 85±0.3 (3) >870* 
Human tPA S-2288 331±0.1 (3) >710* 
Human plasmin S-2403 106±1 (3) >620* 
Human thrombin S-2238 17.3±0.9 (3) >350* 
Human aPC S-2366 711±0.4 (3) >950* 
Human Factor VIIa SpetrozymefVIIa >500 (3)† >680* 
Human Factor Xa S-2765 101±0.5 (3) >680* 
Human plasma kallikrein S-2302 300±0.3 (3) >680* 
Bovine β-trypsin S-2222 60.0±0.2 (3) 20.7±0.5 (3)* 
Protease Substrate Km(μM) Ki(μM) 
muPA S-2444 1190±0.3 (3) 0.412±0.180 (7) 
Murine tPA S-2765 187±13 (3) >580* 
Murine plasmin S-2366 179±7 (3) >640* 
Murine thrombin S-2238 37±9 (3) >370* 
Murine β-trypsin S-2222 55±8 (3) 45±3 (3)* 
Murine aPC S-2366 584±39 (3) >730* 
Murine Factor IXa SpectrozymefIXa >1250† >675* 
Murine Factor Xa S-2765 78±23 (3) >180* 
Murine plasma kallikrein S-2302 116±40 (3) >650* 
huPA S-2444 85±0.3 (3) >870* 
Human tPA S-2288 331±0.1 (3) >710* 
Human plasmin S-2403 106±1 (3) >620* 
Human thrombin S-2238 17.3±0.9 (3) >350* 
Human aPC S-2366 711±0.4 (3) >950* 
Human Factor VIIa SpetrozymefVIIa >500 (3)† >680* 
Human Factor Xa S-2765 101±0.5 (3) >680* 
Human plasma kallikrein S-2302 300±0.3 (3) >680* 
Bovine β-trypsin S-2222 60.0±0.2 (3) 20.7±0.5 (3)* 
*

Significantly different from the corresponding value for wild-type mupain-1 (P<0.01).

Saturating levels of substrate could not be reached in the absence of tissue factor and thus the indicated values represent a lower limit for the Km value under the tested conditions

Effect of active site reagents on the mupain-1–muPA binding

PAB and amiloride contain an amidino or a guanidino group respectively that insert into the S1 pocket in the active site of muPA. To investigate whether mupain-1 binds muPA via interaction in the S1 pocket, the ability of PAB and amiloride to displace mupain-1-phage particles from murine uPA was determined by a phage-ELISA. The experiments showed that both PAB and amiloride are capable of displacing mupain-1 phage particles in a dose-dependent manner, suggesting overlapping binding sites between mupain-1 and the two chemical compounds (results not shown).

Mutagenesis of residues in the mupain-1 sequence

In order to evaluate the importance of individual residues in mupain-1 for the binding to muPA, all the residues (except Ala3) were independently substituted with an alanine residue, either as a chemical peptide or in the mupain-1–D1D2 background. A linear variant with the N- and C-terminal cysteine residues replaced with serine residues was also constructed.

The effect of these substitutions on the inhibitory properties of the mupain-1 sequence was measured, using plots of Vi/V0 against inhibitor concentration at a single substrate concentration near the Km value. Likewise, we measured the binding affinity by the use of SPR, using the steady-state binding levels at several different peptide concentrations, as the association and dissociation kinetics were too fast to allow determination of the rate constants (Table 4). The results obtained by the two types of analyses agreed well. The analyses demonstrated that substitution of Pro2, Tyr4, Ser5, Arg6 and Tyr7 resulted in a strongly decreased affinity for muPA, whereas substitution of Leu8 or Asp9 had little or no effect. Furthermore, the cysteine residues in positions 1 and 10, and therefore the cyclical nature of the peptide sequence, were found to be necessary for binding. Upain-1, a cyclic peptide that inhibits huPA competitively, inhibited muPA very poorly, with a Ki value 200-fold higher than that for inhibition by mupain-1.

Table 4
Alanine-scanning analysis of the importance of individual mupain-1 residues required for muPA inhibition

The Ki values for inhibition of muPA by the indicated peptides were determined by an assay with the chromogenic substrate S-2444, in HBS with 0.1% BSA at a temperature of 37 °C. The reported inhibition constants were determined from the combined data of the indicated number of independent experiments and global evaluation by eqn (4). Values are means±S.D for the indicated number of experiments (in parentheses). The binding of the indicated peptides to muPA was characterized by the use of SPR with a BIACORE T100 instrument equipped with a CM5 chip coupled with a monoclonal antibody against the A-chain of muPA. After coating with 250 RU of muPA, the peptide variants were passed over the chip in various concentrations between 0 μM and 5 μM at 25 °C. The KD values were determined from the concentration-dependence of the steady-state levels (eqn 5). Values are means±S.D. for the indicated number of independent determinations (in parentheses). n.d., not determined.

Sequence Peptide name Chemical peptide Ki (μM) Fusion protein Ki (μM) Chemical peptide KD (μM) 
CPAYSRYLDC Mupain-1 wild-type 0.41±0.18 (7) 5.20±0.10 (3) 0.33±0.11 (4)* 
CPLYNRMIGC Mupain-2 n.d. 13.2±0.4 (3) n.d. 
CAAYSRYLDC Mupain-1 P2A 13±0.4 (4)* n.d. 4.9±0.26 (2)* 
CPAASRYLDC Mupain-1 Y4A 67±1.9 (4)* n.d. 22.6±12.8 (4)* 
CPAYARYLDC Mupain-1 S5A 57±1.0 (4)* 480±13 (3)* 45.2±4.20 (2)* 
CPAYSAYLDC Mupain-1 R6A >830 (2)* 280±18 (3)* >300 (4)* 
CPAYSRALDC Mupain-1 Y7A 4.0±0.05 (4)* n.d. 2.5±0.5 (3) 
CPAYSRYADC Mupain-1 L8A 1.4±0.03 (4)* n.d. 0.83±0.15 (2) 
CPAYSRYLAMupain-1 D9A 0.89±0.02 (4) n.d. 0.57±0.06 (2) 
SPAYSRYLDS Linearized mupain-1 >430 n.d. >300* 
CSWRGLENHRMC Upain-1 n.d. 1200±93 n.d. 
Sequence Peptide name Chemical peptide Ki (μM) Fusion protein Ki (μM) Chemical peptide KD (μM) 
CPAYSRYLDC Mupain-1 wild-type 0.41±0.18 (7) 5.20±0.10 (3) 0.33±0.11 (4)* 
CPLYNRMIGC Mupain-2 n.d. 13.2±0.4 (3) n.d. 
CAAYSRYLDC Mupain-1 P2A 13±0.4 (4)* n.d. 4.9±0.26 (2)* 
CPAASRYLDC Mupain-1 Y4A 67±1.9 (4)* n.d. 22.6±12.8 (4)* 
CPAYARYLDC Mupain-1 S5A 57±1.0 (4)* 480±13 (3)* 45.2±4.20 (2)* 
CPAYSAYLDC Mupain-1 R6A >830 (2)* 280±18 (3)* >300 (4)* 
CPAYSRALDC Mupain-1 Y7A 4.0±0.05 (4)* n.d. 2.5±0.5 (3) 
CPAYSRYADC Mupain-1 L8A 1.4±0.03 (4)* n.d. 0.83±0.15 (2) 
CPAYSRYLAMupain-1 D9A 0.89±0.02 (4) n.d. 0.57±0.06 (2) 
SPAYSRYLDS Linearized mupain-1 >430 n.d. >300* 
CSWRGLENHRMC Upain-1 n.d. 1200±93 n.d. 
*

Significantly different from the corresponding value for wild-type mupain-1 (P<0.01).

Mapping of the binding site for mupain-1 on uPA by alanine-scanning mutagenesis

To describe the binding interface for mupain-1 on muPA, we used site-directed mutagenesis of muPA expressed recombinantly in HEK-293T cells. Principally, the alanine substitutions were chosen on the basis of a three-dimensional model of muPA aligned with the upain-1–uPA crystal structure (Figure 4) [20]. Residues in muPA closer than 5 Å (1 Å=0.1 nm) to, and pointing towards upain-1 in this model were decided to be of particular interest in relation to the binding of mupain-1 to muPA, assuming that mupain-1 and upain-1 have similar binding sites. In order to estimate the binding of the muPA variants relative to the binding of the wild-type, we used a phage-particle ELISA and a BIACORE set-up, in both cases employing directly conditioned medium from transfected HEK-293T cells. In the ELISA, the muPA variants from the conditioned medium were captured in Ni-NTA-coated wells and the relative binding of phage particles subsequently scored using a control in which the capture levels of the various variants were estimated with a monoclonal anti-muPA antibody (Figure 3A). In the BIACORE set-up, the muPA variants from the conditioned medium were captured on a monoclonal anti-muPA antibody which was immobilized on a CM5 chip, and the binding of mupain-1–D1D2 to captured muPA was recorded (Figure 3B).

Effect of alanine substitutions of specific residues in muPA on the binding of mupain-1

Figure 3
Effect of alanine substitutions of specific residues in muPA on the binding of mupain-1

(A) Phage-particle ELISA. Medium from HEK-293T cells transfected with the corresponding cDNAs were used directly in the assay. muPA was captured in Ni-NTA coated wells. After the removal of excess muPA, mupain-1 phage particles were allowed to bind and were subsequently detected by a HRP-conjugated anti-M13 antibody. Background signals from wells without muPA were subtracted before the signals were normalized to the actual capture level of muPA in the wells and the signal obtained with wild-type muPA. (B) BIACORE binding analysis. muPA variants from conditioned medium from HEK-293T cells were captured on a monoclonal anti-muPA antibody which was immobilized on a CM5 chip and the binding of mupain-1–D1D2 to muPA was monitored. Background signals from binding analysis without muPA has been subtracted and the data are normalized to the binding signal obtained for wild-type.

Figure 3
Effect of alanine substitutions of specific residues in muPA on the binding of mupain-1

(A) Phage-particle ELISA. Medium from HEK-293T cells transfected with the corresponding cDNAs were used directly in the assay. muPA was captured in Ni-NTA coated wells. After the removal of excess muPA, mupain-1 phage particles were allowed to bind and were subsequently detected by a HRP-conjugated anti-M13 antibody. Background signals from wells without muPA were subtracted before the signals were normalized to the actual capture level of muPA in the wells and the signal obtained with wild-type muPA. (B) BIACORE binding analysis. muPA variants from conditioned medium from HEK-293T cells were captured on a monoclonal anti-muPA antibody which was immobilized on a CM5 chip and the binding of mupain-1–D1D2 to muPA was monitored. Background signals from binding analysis without muPA has been subtracted and the data are normalized to the binding signal obtained for wild-type.

Three-dimensional model of muPA and the binding site for mupain-1

Figure 4
Three-dimensional model of muPA and the binding site for mupain-1

Shown are a ribbon (A) and surface (B) representation of a three-dimensional model of muPA that was generated using SWISSmodel [30]. It is based on a sequence alignment of the protease domain of huPA and muPA and the X-ray crystal structure in Zhao et al. [20]. The orientation of the structures in (A) and (B) is the same. The identification of mupain-1 binding residues in muPA was based on mupain-1 phage-particle ELISA and SPR analysis using mupain-1–D1D2. (A) The active site of muPA and its surroundings. Residues of the catalytic triad are labelled by identity as well as surface-exposed loops which are coloured green. The side chains of the residues involved in binding of mupain-1 are depicted in red, whereas those having no or only a weak effect on binding of mupain-1 when changed individually to alanine residues are depicted in blue. Amino acids are numbered according to chymotrypsin. (B) Mupain-1 binding residues in muPA, as based on phage-particle ELISA and SPR binding analysis. Alanine substitutions of the residues coloured red resulted in pronounced loss of binding of mupain-1, except for K192A which, for unknown reasons, could not be expressed. However, the implication of Lys192 in the binding of mupain-1 is supported by the mutagenesis of huPA. Alanine substitutions of the residues coloured blue showed no or only a weak reduction in the binding of mupain-1.

Figure 4
Three-dimensional model of muPA and the binding site for mupain-1

Shown are a ribbon (A) and surface (B) representation of a three-dimensional model of muPA that was generated using SWISSmodel [30]. It is based on a sequence alignment of the protease domain of huPA and muPA and the X-ray crystal structure in Zhao et al. [20]. The orientation of the structures in (A) and (B) is the same. The identification of mupain-1 binding residues in muPA was based on mupain-1 phage-particle ELISA and SPR analysis using mupain-1–D1D2. (A) The active site of muPA and its surroundings. Residues of the catalytic triad are labelled by identity as well as surface-exposed loops which are coloured green. The side chains of the residues involved in binding of mupain-1 are depicted in red, whereas those having no or only a weak effect on binding of mupain-1 when changed individually to alanine residues are depicted in blue. Amino acids are numbered according to chymotrypsin. (B) Mupain-1 binding residues in muPA, as based on phage-particle ELISA and SPR binding analysis. Alanine substitutions of the residues coloured red resulted in pronounced loss of binding of mupain-1, except for K192A which, for unknown reasons, could not be expressed. However, the implication of Lys192 in the binding of mupain-1 is supported by the mutagenesis of huPA. Alanine substitutions of the residues coloured blue showed no or only a weak reduction in the binding of mupain-1.

A more than 5-fold reduction in mupain-1 binding was found for the following mutants: K41A, H57A, Y99A, D189A and S195A, whereas the substitutions Q35A, Q60A, K143A, E146A and R217A had no or a very moderate influence on the binding of mupain-1. In addition, the mutant K15A, introduced to ensure that the protein remained in the inactive pro-enzyme form, reduced the binding of mupain-1 to unmeasurably low levels (Figure 3). For unknown reasons, the mutant K192A could not be expressed, and the importance of this residue in the binding of mupain-1 has therefore not been elucidated.

Changing the susceptibility of huPA to mupain-1

By combining the biochemical data and the structural information given by a sequence alignment of the protease domain of huPA and muPA, we identified two amino acids (Lys41 and Tyr99) in muPA which are highly important for the binding of mupain-1 and which are not conserved beween huPA and muPA. Even though the importance of Lys192 of muPA could not be tested by alanine substitution, it may, according to a three-dimensional model of muPA, be located in the binding area of mupain-1, and this residue is also not conserved between huPA and muPA. In an attempt to confer mupain-1 sensitivity to huPA, we decided to graft the residues present at positions 41, 99 and 192 in muPA on to huPA. In order to evaluate the importance of individual amino acids on the species specificity, we produced seven huPA mutants which combined the mutations at the three positions in all possible ways, i.e. three single mutants, three double mutants and one triple mutant. The Km and Vmax values for S-2444 hydrolysis by the different huPA variants were determined using the conditioned medium from HEK-293T cells transfected with the corresponding cDNAs. The huPA variants were tested for binding to mupain-1 in SPR binding analysis. We found that binding between mupain-1 peptide and huPA wild-type, huPA V41K, huPA Q192K and huPA V41K/Q192K was below measurable levels with a 50 μM concentration of mupain-1. However, the huPA variants H99Y, V41K/H99Y, H99Y/Q192K and V41K/H99Y/Q192K showed clear binding to mupain-1 in the SPR binding analysis and the KD values for the binding between mupain-1 and these variants are listed in Table 5. Substituting the histidine residue at position 99 with a tyrosine residue greatly enhanced the binding of mupain-1 to huPA, resulting in a KD of 12.0 μM. A tyrosine residue at position 99 combined with a lysine residue at position 192 did not further enhance the binding to mupain-1. However, combining the tyrosine residue at position 99 with a lysine residue at position 41 enhanced the binding to mupain-1 2-fold compared with the single substitution at position 99. The affinity between mupain-1 and huPA is further enhanced by grafting all three amino acids at positions 41, 99 and 192 in muPA on to huPA; we thereby obtained a KD value of 4.2 μM, only approx. 10-fold higher than the corresponding value for the binding of mupain-1 to muPA (Table 5). The Ki values for inhibition of the different huPA variants by mupain-1 were determined using the conditioned medium and a single substrate concentration near the Km value. When no inhibition was observed even at the highest concentration tested (100 μM), a Ki value was estimated on the basis of eqn (4), assuming a lower limit for detection of approx. 10% inhibition. In excellent agreement with the KD values, the single substitution at position 99 resulted in a Ki of 15.3 μM, whereas combining this substitution with a substitution at position 41 further decreased the Ki to 6.9 μM. The largest effect on Ki was again observed when grafting all three amino acids from muPA on to huPA, resulting in a Ki of 3.6 μM, which again is approx. 10-fold higher compared with the same value determined for muPA. The Km value for hydrolysis of S-2444 was also markedly influenced by substitution of amino acids on the beforementioned positions in huPA. The Km value for the triple mutant of huPA was comparable with the corresponding value obtained for muPA wild-type (Table 5).

Table 5
Effects of grafting amino acid residues from muPA on to huPA on the Km values for hydrolysis of S-2444 and Ki values for inhibition by mupain-1

Kinetic parameters were determined in HBS with 0.1% BSA at pH 7.4 and at 37 °C with conditioned medium from HEK-293T cells transfected with the corresponding huPA cDNAs. The reported inhibition constants were determined from the combined data of three independent experiments and global evaluation by eqn (4). Numbers in parentheses indicate the number of experiments. n.b., no binding of mupain-1 at a concentration of 50 μM

huPA variant Km(mM) Vmax(10−7 M/s) Ki (μM) KD(μM) 
Wild-type 0.31±0.03 (3) 2.0±0.6 (3) >550 n.b. 
V41K 0.47±0.07 (3) 2.6±0.3 (3) >630 n.b. 
H99Y 0.61±0.11 (3) 2.3±0.3 (3) 15.3±2.0 (3) 12.0±4.7 (3) 
Q192K 0.76±0.14 (3) 4.0±0.6 (3) >600 n.b. 
V41K/H99Y 1.04±0.21 (3) 4.0±0.6 (3) 6.9±1.0 (3) 6.0±2.3 (3) 
V41K/Q192K 1.20±0.27 (3) 3.7±0.9 (3) >710 n.b. 
H99Y/Q102K 1.80±0.28 (3) 4.0±0.6 (3) 9.4±1.3 (3) 12.2±5.7 (3) 
V41K/H99Y/Q192K 3.35±0.78 (3) 3.2±0.6 (3) 3.6±0.8 (3) 4.2±1.3 (3) 
huPA variant Km(mM) Vmax(10−7 M/s) Ki (μM) KD(μM) 
Wild-type 0.31±0.03 (3) 2.0±0.6 (3) >550 n.b. 
V41K 0.47±0.07 (3) 2.6±0.3 (3) >630 n.b. 
H99Y 0.61±0.11 (3) 2.3±0.3 (3) 15.3±2.0 (3) 12.0±4.7 (3) 
Q192K 0.76±0.14 (3) 4.0±0.6 (3) >600 n.b. 
V41K/H99Y 1.04±0.21 (3) 4.0±0.6 (3) 6.9±1.0 (3) 6.0±2.3 (3) 
V41K/Q192K 1.20±0.27 (3) 3.7±0.9 (3) >710 n.b. 
H99Y/Q102K 1.80±0.28 (3) 4.0±0.6 (3) 9.4±1.3 (3) 12.2±5.7 (3) 
V41K/H99Y/Q192K 3.35±0.78 (3) 3.2±0.6 (3) 3.6±0.8 (3) 4.2±1.3 (3) 

The stability of mupain-1 in serum

The ability of certain amounts of mupain-1 to inhibit muPA changed little, if at all, whether it was assayed directly or after having been incubated for 20 h with serum (Figure 5). The change observed corresponded maximally to an approx. 50% loss of inhibitory active peptide during the incubation.

Stability of mupain-1 in mouse serum or fetal bovine serum

Figure 5
Stability of mupain-1 in mouse serum or fetal bovine serum

Mupain-1 (400 μM) was incubated with 77% mouse serum (A) or 77% fetal bovine serum (B) for 0, 1, 2 or 20 h. The serum was then acid-treated to inactivate the α2-macroglobulin. Aliquots of incubation mixtures, corresponding to different initial concentrations of mupain-1 added to the serum, were then mixed with 4 nM muPA. The remaining activity of muPA was determined by an S-2444 assay. A background achieved with serum without mupain-1 has been subtracted from all samples. In all cases serum constitutes 2.5% of the total reaction volume in the assay.

Figure 5
Stability of mupain-1 in mouse serum or fetal bovine serum

Mupain-1 (400 μM) was incubated with 77% mouse serum (A) or 77% fetal bovine serum (B) for 0, 1, 2 or 20 h. The serum was then acid-treated to inactivate the α2-macroglobulin. Aliquots of incubation mixtures, corresponding to different initial concentrations of mupain-1 added to the serum, were then mixed with 4 nM muPA. The remaining activity of muPA was determined by an S-2444 assay. A background achieved with serum without mupain-1 has been subtracted from all samples. In all cases serum constitutes 2.5% of the total reaction volume in the assay.

DISCUSSION

In the present study, we describe the isolation of mupain-1, a muPA-binding peptide from a phage-displayed library of disulfide-bridged constrained peptides. The peptide, of the format CX8C, is a competitive inhibitor of muPA. It displayed a Ki for inhibition and a KD for binding of approx. 400 nM. It is highly specific for muPA. The affinity for almost all tested serine proteases were several-hundred-fold higher than for muPA. The one exception was β-trypsin, which was inhibited with a 50-fold lower efficiency. Strikingly, mupain-1 did not measurably bind to huPA, having a Ki for inhibition by mupain-1 more than 2500-fold higher than muPA. Mupain-1 expressed in fusion with the N-terminal domains, D1 and D2, of the phage coat protein g3p had a markedly lower affinity than the synthetic peptide. This observation is remarkable, since a previous study with a peptide of similar format, binding to huPA, showed higher affinity for its target when fused to D1D2 [19].

We undertook a partial mapping of the mupain-1 binding site of muPA by alanine substitution mutagenesis and in this way identified five residues of importance for binding. Three of these, His57, Asp189 and Ser195, are part of the actual active site. His57 and Ser195 are parts of the catalytic triad and Asp189 is localized at the bottom of the specificity pocket. The importance of these residues is in agreement with the fact that mupain-1 is a competitive inhibitor. In addition, Lys41 and Tyr99 were also shown to be important for binding, showing that mupain-1 also targets areas flanking the actual active site. When planning the alanine-substitution mutagenesis, we also suspected Lys192 of being involved in mupain-1 binding, but were unable to express this variant.

The high specificity of mupain-1 for muPA can be explained, at least in part, by the binding to these residues, as different residues are present in these positions in the other serine proteases tested (Table 6). In particular, there is a striking specificity of mupain-1 for muPA over huPA. Likewise, we previously isolated a peptide sequence, upain-1 (CSWRGLENHRMC), a competitive inhibitor of huPA, but with very low affinity for muPA [19]. In both cases, the species specificity is likely to rely on binding to loops outside the actual active site, such as the 37, 60 and 99 loops. But it should be noted that differences between muPA and huPA must also exist within the actual active site. Although the activation loop in both man and mouse plasminogen is CPGRVVGGC, the Km value for S-2444 hydrolysis by muPA and huPA is approx. 10-fold higher for muPA than for huPA (Table 4). In order to provide further information about the basis for this specificity, we grafted the residues present in positions 41, 99 and 192 of muPA on to huPA. The triple mutant of huPA, having its natural residues replaced by the counterparts from muPA, had a Km value for S-2444 hydrolysis comparable with that of muPA wild-type. Moreover, the binding affinity for mupain-1 increased more than 400-fold by the triple substitution, and the Ki was only 10-fold higher than that for wild-type muPA. This increase in affinity for mupain-1 was mainly contributed by the H99Y substitution, which causes an at least 100-fold increase in affinity. Residue number 99 is localized in the 99-loop which lines the S2 pocket. This position has previously been implicated in determining the substrate specificity of factor VIIa [27]. His99 also played a role in the recognition of upain-1 by uPA.

Table 6
Amino acid residue variation among related serine proteases in the position critical for binding to muPA

If no residue is indicated, the protease does not have that position filled. Residue numbers are according to the chymotrypsin template numbering. The alignment is that given by Mackman et al. [10] or a sequence-based alignment, performed with ClustalW (www.ebi.ac.uk/Tools/clustalw/) in the case of trypsin, aPC and the murine proteases.

  Amino acid residue 
Proteinase Position… 41 99 192 
huPA Val His Gln 
muPA Lys Tyr Lys 
Human tPA Leu Tyr Gln 
Murine tPA Leu Tyr Gln 
Human plasmin Phe – Gln 
Murine plasmin Phe – Gln 
Human thrombin Leu Leu Glu 
Murine thrombin Leu Leu Glu 
Human Factor VIIa Leu Thr Lys 
Murine Factor VIIa Leu Iso Lys 
Human Factor Xa Phe Tyr Gln 
Murine Factor Xa Phe Tyr Gln 
Human trypsin Phe Leu Gln 
Murine trypsin Phe Leu Gln 
Bovine trypsin Phe Leu Gln 
Human plasma kallikrein Leu Gly Lys 
Murine plasma kallikrein Leu Gly Lys 
  Amino acid residue 
Proteinase Position… 41 99 192 
huPA Val His Gln 
muPA Lys Tyr Lys 
Human tPA Leu Tyr Gln 
Murine tPA Leu Tyr Gln 
Human plasmin Phe – Gln 
Murine plasmin Phe – Gln 
Human thrombin Leu Leu Glu 
Murine thrombin Leu Leu Glu 
Human Factor VIIa Leu Thr Lys 
Murine Factor VIIa Leu Iso Lys 
Human Factor Xa Phe Tyr Gln 
Murine Factor Xa Phe Tyr Gln 
Human trypsin Phe Leu Gln 
Murine trypsin Phe Leu Gln 
Bovine trypsin Phe Leu Gln 
Human plasma kallikrein Leu Gly Lys 
Murine plasma kallikrein Leu Gly Lys 

We undertook a comprehensive alanine-scanning mutagenesis of mupain-1 and found that the cyclic structure as well as residues Pro2, Tyr4, Ser5, Arg6 and Tyr7 were of decisive importance for the affinity to muPA, whereas Leu8 and Asp9 could be substituted with an alanine residue without affecting the affinity to muPA. Since Arg6 is the only arginine residue, it must be the P1 residue and insert into the specificity pocket. Accordingly, Ser5 will have to be the P2 residue and Tyr4 the P3 residue, whereas Tyr7 must be the P1′ residue. Leu8 and Asp9 are likely to point away from the muPA surface.

Peptidylic inhibitors selected from phage-displayed libraries provide generally applicable tools for studying protease specificity and inhibitory and catalytic mechanisms (for a review, see [28]). Why is mupain-1 an inhibitor, and not a substrate? We previously asked the same question in the case of upain-1, the inhibitor of huPA. The three-dimensional structure of the upain-1–uPA complex [20] showed that the bulky tryptophan residue in the P2 position is too large to be accommodated in the narrow P2 binding site of uPA. The proper alignment of the scissile bond into the active site is therefore not possible. Instead, the side chain of the glutamate residue in position 7 of upain-1 bends into the oxyanion hole and further prevents hydrolysis. A similar mechanism may be predicted in the case of mupain-1, which has a serine residue in the P2 position, also likely to be too large for the P2 binding site of muPA.

It seems that peptide-based drugs are now realistic alternatives to other biopharmaceuticals such as antibodies. Most of the past limitations to peptides have been removed by new technologies. Phage-display technologies can provide novel peptides that bind protein targets with high affinity and specificity (for a review, see [29]). Peptides have specificities comparable with those of monoclonal antibodies and protein protease inhibitors, but at the same time a size eventually allowing for chemical synthesis and modification. Clearance rates may be reduced by derivatization with poly(ethylene glycol). In the case of mupain-1, we have demonstrated that the peptide is stable in serum. After appropriate affinity maturation and perhaps incorporation of unnatural amino acids, mupain-1 may become a useful high-affinity inhibitor of muPA in murine models of tumour invasion and metastasis.

The technical assistance of Anni Christensen is gratefully acknowledged. This work was supported by grants from the Danish Cancer Society, the Danish Cancer Research Foundation, the Danish Research Agency, the Carlsberg Foundation, the Novo-Nordisk Foundation and European Union FP6 contract LSHC-CT-2003 h503297 (the Cancer Degradome) to P. A. A.

Abbreviations

     
  • aPC

    activated protein C

  •  
  • HBS

    Hepes-buffered saline

  •  
  • HEK

    human embryonic kidney

  •  
  • His6

    hexahistidine

  •  
  • HRP

    horseradish peroxidase

  •  
  • huPA

    human urokinase-type plasminogen activator

  •  
  • muPA

    murine urokinase-type plasminogen activator

  •  
  • Ni-NTA

    Ni2+-nitrilotriacetate

  •  
  • PAB

    p-aminobenzamidine

  •  
  • RU

    response unit

  •  
  • S-2222

    benzyl-IEGR-p-nitroaniline

  •  
  • S-2238

    H-D-Pro-piperidine-Arg-p-nitroaniline

  •  
  • S-2288

    H-D-IPR-p-nitroaniline

  •  
  • S-2302

    H-D-PFR-p-nitroanline

  •  
  • S-2366

    pyro-EPR-p-nitroaniline

  •  
  • S-2403

    pyro-EFK-p-nitroaniline

  •  
  • S-2444

    pyro-EGR-p-nitroaniline

  •  
  • S-2765

    benzyloxycarbonyl-D-RGR-p-nitroaniline

  •  
  • SpectrozymeFVIIa

    methansulfonyl-D-cyclohexylalanyl-butyl-Arg-p-nitroaniline

  •  
  • SPR

    surface plasmon resonance

  •  
  • tPA

    tissue plasminogen activator

  •  
  • uPA

    urokinase-type plasminogen activator

  •  
  • uPAR

    uPA receptor

  •  
  • VLK-AMC

    H-D-VLK-7-amido-4-methylcoumarin

References

References
1
Hedstrom
L.
Serine protease mechanism and specificity
Chem. Rev.
2002
, vol. 
102
 (pg. 
4501
-
4524
)
2
Barrett
A. J.
Rawlings
N. D.
Woessner
J. F.
Handbook of Proteolytic Enzymes
1998
London
Academic Press
3
Dano
K.
Moller
V.
Ossowski
L.
Nielsen
L. S.
Purification and characterization of a plasminogen activator from mouse cells transformed by an oncogenic virus
Biochim. Biophys. Acta
1980
, vol. 
613
 (pg. 
542
-
555
)
4
Dano
K.
Andreasen
P. A.
Grondahl-Hansen
J.
Kristensen
P.
Nielsen
L. S.
Skriver
L.
Plasminogen activators, tissue degradation, and cancer
Adv. Cancer Res.
1985
, vol. 
44
 (pg. 
139
-
266
)
5
Andreasen
P. A.
Egelund
R.
Petersen
H. H.
The plasminogen activation system in tumor growth, invasion, and metastasis
Cell. Mol. Life Sci.
2000
, vol. 
57
 (pg. 
25
-
40
)
6
Andreasen
P. A.
Kjoller
L.
Christensen
L.
Duffy
M. J.
The urokinase-type plasminogen activator system in cancer metastasis: a review
Int. J. Cancer
1997
, vol. 
72
 (pg. 
1
-
22
)
7
Kaltoft
K.
Nielsen
L. S.
Zeuthen
J.
Dano
K.
Monoclonal antibody that specifically inhibits a human Mr 52,000 plasminogen-activating enzyme
Proc. Natl. Acad. Sci. U.S.A.
1982
, vol. 
79
 (pg. 
3720
-
3723
)
8
Petersen
H. H.
Hansen
M.
Schousboe
S. L.
Andreasen
P. A.
Localization of epitopes for monoclonal antibodies to urokinase-type plasminogen activator: relationship between epitope localization and effects of antibodies on molecular interactions of the enzyme
Eur. J. Biochem.
2001
, vol. 
268
 (pg. 
4430
-
4439
)
9
Laboissiere
M. C.
Young
M. M.
Pinho
R. G.
Todd
S.
Fletterick
R. J.
Kuntz
I.
Craik
C. S.
Computer-assisted mutagenesis of ecotin to engineer its secondary binding site for urokinase inhibition
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
26623
-
26631
)
10
Mackman
R. L.
Katz
B. A.
Breitenbucher
J. G.
Hui
H. C.
Verner
E.
Luong
C.
Liu
L.
Sprengeler
P. A.
Exploiting subsite S1 of trypsin-like serine proteases for selectivity: potent and selective inhibitors of urokinase-type plasminogen activator
J. Med. Chem.
2001
, vol. 
44
 (pg. 
3856
-
3871
)
11
Schweinitz
A.
Steinmetzer
T.
Banke
I. J.
Arlt
M. J.
Sturzebecher
A.
Schuster
O.
Geissler
A.
Giersiefen
H.
Zeslawska
E.
Jacob
U.
, et al. 
Design of novel and selective inhibitors of urokinase-type plasminogen activator with improved pharmacokinetic properties for use as antimetastatic agents
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
33613
-
33622
)
12
Spencer
J. R.
McGee
D.
Allen
D.
Katz
B. A.
Luong
C.
Sendzik
M.
Squires
N.
Mackman
R. L.
4-Aminoarylguanidine and 4-aminobenzamidine derivatives as potent and selective urokinase-type plasminogen activator inhibitors
Bioorg. Med. Chem. Lett.
2002
, vol. 
12
 (pg. 
2023
-
2026
)
13
Sperl
S.
Jacob
U.
Arroyo de Prada
N.
Sturzebecher
J.
Wilhelm
O. G.
Bode
W.
Magdolen
V.
Huber
R.
Moroder
L.
(4-Aminomethyl)phenylguanidine derivatives as nonpeptidic highly selective inhibitors of human urokinase
Proc. Natl. Acad. Sci. U.S.A.
2000
, vol. 
97
 (pg. 
5113
-
5118
)
14
Wendt
M. D.
Rockway
T. W.
Geyer
A.
McClellan
W.
Weitzberg
M.
Zhao
X.
Mantei
R.
Nienaber
V. L.
Stewart
K.
Klinghofer
V.
Giranda
V. L.
Identification of novel binding interactions in the development of potent, selective 2-naphthamidine inhibitors of urokinase. Synthesis, structural analysis, and SAR of N-phenyl amide 6-substitution
J. Med. Chem.
2004
, vol. 
47
 (pg. 
303
-
324
)
15
Zeslawska
E.
Jacob
U.
Schweinitz
A.
Coombs
G.
Bode
W.
Madison
E.
Crystals of urokinase type plasminogen activator complexes reveal the binding mode of peptidomimetic inhibitors
J. Mol. Biol.
2003
, vol. 
328
 (pg. 
109
-
118
)
16
Frese
K. K.
Tuveson
D. A.
Maximizing mouse cancer models
Nat. Rev. Cancer
2007
, vol. 
7
 (pg. 
645
-
658
)
17
Dano
K.
Nielsen
L. S.
Moller
V.
Engelhart
M.
Inhibition of a plasminogen activator from oncogenic virus-transformed mouse cells by rabbit antibodies against the enzyme
Biochim. Biophys. Acta
1980
, vol. 
630
 (pg. 
146
-
151
)
18
Klinghofer
V.
Stewart
K.
McGonigal
T.
Smith
R.
Sarthy
A.
Nienaber
V.
Butler
C.
Dorwin
S.
Richardson
P.
Weitzberg
M.
, et al. 
Species specificity of amidine-based urokinase inhibitors
Biochemistry
2001
, vol. 
40
 (pg. 
9125
-
9131
)
19
Hansen
M.
Wind
T.
Blouse
G. E.
Christensen
A.
Petersen
H. H.
Kjelgaard
S.
Mathiasen
L.
Holtet
T. L.
Andreasen
P. A.
A urokinase-type plasminogen activator-inhibiting cyclic peptide with an unusual P2 residue and an extended protease binding surface demonstrates new modalities for enzyme inhibition
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
38424
-
38437
)
20
Zhao
G.
Yuan
C.
Wind
T.
Huang
Z.
Andreasen
P. A.
Huang
M.
Structural basis of specificity of a peptidyl urokinase inhibitor, upain-1
J. Struct. Biol.
2007
, vol. 
160
 (pg. 
1
-
10
)
21
Kvassman
J. O.
Verhamme
I.
Shore
J. D.
Inhibitory mechanism of serpins: loop insertion forces acylation of plasminogen activator by plasminogen activator inhibitor-1
Biochemistry
1998
, vol. 
37
 (pg. 
15491
-
15502
)
22
Olson
S. T.
Bock
P. E.
Kvassman
J.
Shore
J. D.
Lawrence
D. A.
Ginsburg
D.
Bjork
I.
Role of the catalytic serine in the interactions of serine proteinases with protein inhibitors of the serpin family. Contribution of a covalent interaction to the binding energy of serpin–proteinase complexes
J. Biol. Chem.
1995
, vol. 
270
 (pg. 
30007
-
30017
)
23
Koivunen
E.
Wang
B.
Dickinson
C. D.
Ruoslahti
E.
Peptides in cell adhesion research
Methods Enzymol.
1994
, vol. 
245
 (pg. 
346
-
369
)
24
Scott
J. K.
Smith
G. P.
Searching for peptide ligands with an epitope library
Science
1990
, vol. 
249
 (pg. 
386
-
390
)
25
Jensen
J. K.
Wind
T.
Andreasen
P. A.
The vitronectin binding area of plasminogen activator inhibitor-1, mapped by mutagenesis and protection against an inactivating organochemical ligand
FEBS Lett.
2002
, vol. 
521
 (pg. 
91
-
94
)
26
Wind
T.
Jensen
J. K.
Dupont
D. M.
Kulig
P.
Andreasen
P. A.
Mutational analysis of plasminogen activator inhibitor-1
Eur. J. Biochem.
2003
, vol. 
270
 (pg. 
1680
-
1688
)
27
Larsen
K. S.
Ostergaard
H.
Bjelke
J. R.
Olsen
O. H.
Rasmussen
H. B.
Christensen
L.
Kragelund
B. B.
Stennicke
H. R.
Engineering the substrate and inhibitor specificities of human coagulation Factor VIIa
Biochem. J.
2007
, vol. 
405
 (pg. 
429
-
438
)
28
Diamond
S.
Methods for mapping protease specificity
Curr. Opin. Chem. Biol.
2007
, vol. 
11
 (pg. 
46
-
51
)
29
Ladner
R. C.
Sato
A. K.
Gorzelany
J.
de Souza
M.
Phage display-derived peptides as therapeutic alternatives to antibodies
Drug Discovery Today
2004
, vol. 
9
 (pg. 
525
-
529
)
30
Schwede
T.
Kopp
J.
Guex
N.
Peitsch
M. C.
SWISS-MODEL: an automated protein homology-modeling server
Nucleic Acids Res.
2003
, vol. 
31
 (pg. 
3381
-
3385
)