Elucidation of the contribution of active site and exosite interactions to affinity and specificity of peptidylic serine protease inhibitors using non-natural arginine analogs

Article abstract of DOI:10.1124/mol.111.072280

There is increasing interest in developing peptides for pharmacological intervention with pathophysiological functions of serine proteases. From phage-displayed peptide libraries, we previously isolated peptidylic inhibitors of urokinase-type plasminogen

Full text of DOI:10.1124/mol.111.072280

Supplemental material to this article can be found at:
http://molpharm.aspetjournals.org/content/suppl/2011/06/30/mol.111.072280.DC1.html
0026-895X/11/8004-585–597$25.00
MOLECULAR PHARMACOLOGY
Copyright © 2011 The American Society for Pharmacology and Experimental Therapeutics
Vol. 80, No. 4
72280/3714931
Mol Pharmacol 80:585–597, 2011
Printed in U.S.A.
Elucidation of the Contribution of Active Site and Exosite
Interactions to Affinity and Specificity of Peptidylic Serine
□
S
Protease Inhibitors Using Non-Natural Arginine Analogs
Masood Hosseini, Longguang Jiang, Hans Peter Sørensen, Jan K. Jensen,
Anni Christensen, Sarah Fogh, Cai Yuan, Lisbeth M. Andersen, Mingdong Huang,
Peter A. Andreasen, and Knud J. Jensen
Department of Natural Sciences and Environment, Faculty of Life Sciences, University of Copenhagen, Copenhagen, Denmark
(M.H., K.J.J.); Department of Molecular Biology, Aarhus University, Aarhus, Denmark (M.H., H.P.S., J.K.J., A.C., S.F., L.M.A., P.A.A.);
and Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou, China (L.J., C.Y., M.H.)
Received March 19, 2011; accepted June 30, 2011
ABSTRACT
There is increasing interest in developing peptides for pharmacolog- Arg analogs and achieved considerable improvement in the
ical intervention with pathophysiological functions of serine pro- affinity of the peptides to their targets. Using chimeras of hu-
teases. From phage-displayed peptide libraries, we previously iso- man and murine urokinase-type plasminogen activator as well
lated peptidylic inhibitors of urokinase-type plasminogen activator, a as X-ray crystallography, we delineated the relative contribution
potential target for intervention with cancer invasion. The two pep- of the P1 residue and exosite interactions to the affinity and
tides, upain-1 (CSWRGLENHRMC) and mupain-1 (CPAYSRYLDC), specificity of the inhibitors for their target enzyme. The effect of
are competitive inhibitors of human and murine urokinase-type inserting a particular non-natural amino acid into the P1 posi-
plasminogen activator, respectively. Both have an Arg as the tion is determined by the fact that changes in interactions of the
P1 residue, inserting into the S1 pocket in the active site of the P1 residue in the S1 pocket lead to changed exosite interac-
enzymes, but their specificity depends to a large extent on tions and vice versa. These findings are of general interest
interactions outside the enzymes’ active sites, so-called exosite when the affinities and specificities of serine protease inhibitors
interactions. Here we describe upain-2 (CSWRGLENHAAC) to be used for pharmacological intervention are considered and
and the synthesis of a number of upain-2 and mupain-1 vari- could pave the way for potential drug candidates for the treat-
ants in which the P1 Arg was substituted with novel non-natural ment of cancer.
ties, comparable with those of monoclonal antibodies, but
Introduction
because of their smaller size are amenable to chemical mod-
In recent years, peptides have attracted renewed interest
ification. They have predictable absorption, distribution, me-
as drug candidates. Most often, they display high specifici-
tabolism, and excretion properties and display a general ab-
sence of drug-drug interactions. Although peptides are labile
This work was supported by the National Natural Science Foundation of
to proteolysis, stable forms can be developed by chemical
modification. Moreover, new delivery methods have enabled
the formulation of peptide preparations suitable for in vivo
use (for review, see Jorgensen and Nielsen, 2009).
One important class of potential therapeutic targets is the
trypsin family (clan SA) of serine proteases, which has many
pathophysiological functions. Therefore, the interest in ser-
ine protease inhibitors is increasing (for review, see Walker
and Lynas, 2001). Moreover, serine proteases are classic
China [Grants 30811130467, 30973567, 30770429]; the Danish National Re-
search Foundation [Grant 26-331-6]; the Lundbeck Foundation [Grant R19-
A2173]; the Danish Cancer Society [Grant DP 07043, DP 08001]; the Danish
Research Agency [Grant 272-06-0518]; and the Novo Nordisk Foundation
[Grant R114-A11382].
All authors are affiliated with the Danish-Chinese Centre for Proteases and
Cancer (http://www.proteasesandcancer.org).
Article, publication date, and citation information can be found at
http://molpharm.aspetjournals.org.
doi:10.1124/mol.111.072280.
□S The online version of this article (available at http://molpharm.
aspetjournals.org) contains supplemental material.
ABBREVIATIONS: uPA, urokinase-type plasminogen activator; Fmoc, 9H-fluoren-9-ylmethoxycarbonyl; Dap, L-2,3-diaminopropionic acid; Boc,
di-tert-butyl dicarbonate; HPLC, high-performance liquid chromatography; HBTU, N-[(1H-benzotriazol-1-yl)(dimethylamino)methylene]-N-meth-
ylmethanaminium hexafluorophosphate N-oxide; TFA, trifluoroacetic acid; HOBt, N-hydroxybenzotriazole; HOAt, 1-hydroxy-7-azabenzotriazole;
DIPEA, N,N-diisopropylethylamine; NMP, N-methyl-2-pyrrolidone; Alloc, allyloxycarbonyl; HRMS, high-resolution mass spectrometry; ES, elec-
trospray; Cbz, carboxybenzyl; Fmoc-OSu, 9-fluorenylmethoxycarbonyl-N-hydroxysuccinimide; MS, mass spectrometry; wt, wild-type; S-2444,
pyro-Glu-Gly-Arg-p-nitroanilide; PDB, Protein Data Bank.
585

586
Hosseini et al.
subjects for studies of catalytic and inhibitory mechanisms
Upain-1 inhibits human uPA with a Ki of approximately 40
(for review, see Branden and Tooze, 1991). However, the ␮M at 37°C. Mupain-1 has a considerably higher, although
human genome contains approximately 200 serine protease still not quite satisfactory, affinity to murine uPA, with a Ki
genes (for review, see Puente et al., 2008). Although different of approximately 0.4 ␮M at 37°C. In an effort to engineer the
serine proteases have different substrate specificities, the peptides to higher affinity while maintaining specificity and
catalytic site architectures are highly similar. Achieving suf- to understand enzyme-inhibitor interactions, we have first
ficient specificity is, therefore, a challenge of utmost impor- made two amino acid substitutions in upain-1 to obtain
tance when serine protease inhibitors are developed. Devel- upain-2 and, second, substituted the P1 L-Arg residues in
opment of small molecule inhibitors has proved to be a mupain-1 and upain-2 with a variety of novel non-natural
daunting task (for review, see Rockway et al., 2002). To arginine analogs. By testing different P1 residues in uPA
develop inhibitors with better specificity, a number of pepti- variants with different S1 pockets and different surface
dylic serine protease inhibitors have recently been isolated loops, we have evaluated the importance of S1 pocket inter-
by screening phage-displayed peptide libraries, including in- actions versus exosite interactions for the affinity and spec-
hibitors of chymotrypsin (Krook et al., 1997–1998), factor ificity of the peptides.
VIIa (Dennis et al., 2000), prostate-specific antigen (Wu et
al., 2000), kallikrein-2 (Hekim et al., 2006; Heinis et al.,
Materials and Methods
2009), mannose-binding lectin-associated serine protease-1
and -2 (Kocsis et al., 2010), furins (Hajdin et al., 2010), and
isoforms of trypsin (Wu et al., 2010).
One example of a serine protease that is a potential drug
target is urokinase-type plasminogen activator (uPA), which
General. Chemicals were purchased from Sigma-Aldrich (St. Louis,
MO), Iris Biotech GmbH (Marktredwitz, Germany) and Rapp Polymere
GmbH (Tu¨bingen, Germany) and used without further purification
unless otherwise stated. Fmoc-HArg(Pbf)-OH, Fmoc-L-Dap-N-amidine(N,
catalyzes the conversion of plasminogen to the active pro- NЈ-di-Boc)-OH, Fmoc-L-NorArg(N,NЈ-di-Boc)-OH, Fmoc-citrulline-OH,
Fmoc-L-4-guanidino-phenylalanine(N,NЈ-di-Boc)-OH, Fmoc-L-2,
3-diaminopropionic acid-OH, Fmoc-L-bishomoarginine(Pmc)-OH,
Fmoc-D-Arg(2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl)-OH,
Fmoc-L-Ala-4-piperidyl(Alloc)-OH, N-␣-(Fmoc)-3-(3-pyridyl)-L-alanine,
and Fmoc-L-canavanine(Boc)-OH were commercially available. 4-(2,3-
Bis(tert-butoxycarbonyl)guanidino)benzoic acid was synthesized ac-
cording to a procedure published previously (Kayser et al., 2007). NMR
was performed using a Bruker Avance 300 spectrometer. Analytical
HPLC was performed on a UltiMate 3000 (Dionex, Sunnyvale, CA),
using a Gemini 110 Å C18 column (3 ␮m, 4.6 ϫ 50 mm; Phenomenex,
tease plasmin. Plasmin in turn catalyzes directly the degra-
dation of extracellular matrix proteins. Up-regulation of uPA
is implicated in tissue remodeling in several pathological
conditions. In particular, uPA is central to the invasive ca-
pacity of malignant tumors (for review, see Andreasen et al.,
2000). We previously isolated two peptides, one (upain-1,
CSWRGLENHRMC) being an inhibitor of human uPA (Han-
sen et al., 2005) and the other (mupain-1, CPAYSRYLDC)
being an inhibitor of murine uPA (Andersen et al., 2008).
Both are constrained in a circular form by a disulfide bridge. Torrance, CA) with a flow rate of 1.0 ml/min and a linear gradient going
from 95% H2O-5% acetonitrile with 0.1% HCOOH to 100% acetonitrile
with 0.1% HCOOH over 10 min. Preparative HPLC was performed
using a UltiMate 3000, equipped with a Gemini-NX C18 110 Å column
(Phenomenex) running at a flow rate of 10.0 ml/min and a linear
gradient going from 95% H2O/5% acetonitrile with 0.1% TFA to 100%
acetonitrile with 0.1% TFA over 30 min. High-resolution mass spectra
were obtained on a Micromass LCT high-resolution time-of-flight in-
strument by direct injection. Ionization was performed in positive elec-
trospray mode.
Both are highly specific for the target enzyme against which
they were selected. Thus, upain-1 binds approximately 500-
fold better to human uPA than to murine uPA, whereas
mupain-1 binds more than 2000-fold better to murine uPA
than to human uPA. Both bind very selectively to uPA com-
pared with other serine proteases from the same species
(Hansen et al., 2005; Andersen et al., 2008). Analysis of the
inhibitory mechanism showed that both upain-1 and mu-
pain-1 are competitive inhibitors of their target enzymes
General Methods for Solid-Phase Peptide Synthesis. Pep-
(Hansen et al., 2005; Andersen et al., 2008). Site-directed tide synthesis was performed using N␣-Fmoc-protected amino acids
and a HBTU/HOBt activation protocol using a TentaGel resin with
Rink amide linker (0.23 mmol/g), HBTU (3.8 Eq), HOBt/HOAt (4:1,
4 Eq), Fmoc-protected amino acids (4 Eq), and DIPEA (7.2 Eq) in
NMP. Preactivation was for 5 min, and single couplings were for 90
min. Fmoc deprotection was performed using piperidine-NMP (1:4)
for 2 ϩ 15 min. Peptides with Alloc-protected amino acids were
deprotected to a free amine by treating the fully protected and
N-acetylated peptides with a mixture of Pd(PPh3)4 (0.05 Eq) and
Me2NH ꢀ BH3 (0.2 Eq) in degassed CH2Cl2 (30 min) and washed with
NMP (5ϫ). The peptides were then treated with N,NЈ-di-Boc-1H-
mutagenesis (Hansen et al., 2005; Andersen et al., 2008) and
X-ray crystal structure analysis (Zhao et al., 2007) showed
that Arg4 of upain-1 and Arg6 of mupain-1 are the P11
residues of these peptides, i.e., are inserted into the enzymes’
S1 or specificity pocket, the primary interaction site between
substrate and enzyme. In the case of upain-1, X-ray crystal
structure analysis showed that the side-chain of the peptide’s
Glu7 blocks the enzyme’s oxyanion hole (Zhao et al., 2007).
Upain-1 and mupain-1 thus directly occupy the structures of
the active sites, which are essential for the catalytic machin- pyrazole-1-carboxamidine (5 Eq) in NMP overnight. After peptide
assembly, the resins were washed extensively with NMP and
CH2Cl2, before peptide release with TFA-H2O-triethylsilane (95:2.5:
2.5). Peptide release normally proceeded for 2 h before the TFA-
peptide mixture was collected by filtration. The resin was addition-
ally washed with TFA (two times), and the TFA mixtures were
pooled. TFA was removed under a stream of nitrogen, and the pep-
tide was precipitated with diethyl ether. The synthesized peptides
were dissolved in a minimum amount of H2O-acetonitrile (2:1) before
being purified by preparative HPLC. The purified peptides were
dissolved in H2O-acetonitrile (2:1) to a final concentration of 1 mM.
The solution was brought to pH 7.5 to 8 with NH3 in methanol. The
ery for peptide bond hydrolysis (for review, see Hedstrom,
2002). However, the great specificity of both peptides is
achieved by a large interaction area, involving not only the
active sites but also surface-exposed loops specific for indi-
vidual proteases (i.e., exosite interactions).
1 P1, P2, P3, P4, etc. and P1Ј, P2Ј, P3Ј, etc. are of the nomenclature of
Schechter and Berger (Abramowitz et al., 1967) and denote those residues on
the amino-terminal and carboxyl sides of the scissile bond (P1–P1Ј), respec-
tively. The corresponding binding sites in the active sites of the enzymes are
referred to as S1, S2, S3, S4, etc., and SЈ1, SЈ2, SЈ3, etc.

Exosite Interactions of Peptidylic Serine Protease Inhibitors
587
peptides were oxidized to form disulfide bridges, either by addi- reomers (Hartung et al., 2005). 1H (300 MHz, CDCl3) ␦ ppm 8.02 (s,
tion of 1.2 Eq of H2O2 (30–60 min) or by bubbling atmospheric air 1H), 7.76 (d, J ϭ 7.5 Hz, 1H), 7.60 (d, J ϭ 6.3 Hz, 1H), 7.40 (t, J ϭ 7.3
through the solution (1–3 days). The use of H2O2 is only applicable Hz, 1H), 7.35 to 7.24 (m, 1H), 5.93 (dd, J ϭ 16.3, 9.6 Hz, 1H), 5.34 to
in peptides without a methionine (upain-2 and mupain-1 vari- 5.03 (m, 1H), 4.71 to 4.32 (m, 3H), 4.23 (t, J ϭ 7.0 Hz, 1H), 4.03 to
ants). The oxidization was stopped with the addition of acetic acid 3.75 (m, 1H), 3.12 to 2.46 (m, 1H), 1.97 to 1.04 (m, 4H). 13C (75 MHz,
(0.1 ml) followed by evaporation and HPLC purification.
CDCl3) ␦ ppm 176.09, 156.52, 156.24, 155.50, 143.95, 143.74, 141.37,
Synthesis of Fmoc-Dap(4-(2,3-bis(Boc)guanidino)benzoate))-OH 133.00, 127.80, 127.16, 126.13, 125.22, 120.06, 117.60, 116.49, 67.63,
(Protected Form of 11). Benzotriazol-1-yloxy-tripyrrolidinophospho- 67.20, 66.38, 51.89, 51.64, 49.78, 49.03, 47.20, 44.72, 36.04, 32.41,
nium hexafluorophosphate (1040 mg, 2.0 mmol) and DIPEA (0.35 ml, 2.0
30.62, 29.87, 24.43. MS (ES) calculated for C27H31N2O6 (M ϩ H),
mmol) were added to solution of 4-(2,3-bis(tert-butoxycarbonyl) 479.2. Found, 479.1.
a
guanidino)benzoic acid (800 mg, 2.1 mmol) in CH2Cl2 (5 ml). The mixture
was stirred for 10 min at room temperature and transferred to a solution
Synthesis of Fmoc-3-(N,Nꢁ-di-Boc-guanidine)-Tyr-OH (Pro-
tected Form of 14). A solution of H-Tyr(3-NO2)-OH (1000 mg, 4.42
of Fmoc-Dap-OH (326 mg, 1.0 mmol) in CH2Cl2 (6 ml). DIPEA (0.35 ml, 2.0 mmol) in dioxane-H2O (2:1, 20 ml), NaHCO3 (1280 mg, 15.2 mmol),
mmol) was added, and the mixture was stirred for 18 h at room tempera- and Fmoc-OSu (1710 mg, 5.1 mmol) in a round-bottomed flask was
ture. The mixture was evaporated and purified by column chromatography stirred overnight at room temperature. After the solution was de-
with ethyl acetate-acetic acid (99:1) to afford the product as a yellow oil (421
gassed, Pd/C was added, and a hydrogen balloon was connected. The
mg, 61%). 1H NMR (300 MHz, CDCl3) ␦ ppm 7.78–7.48 (m, 10H), 7.38–7.20 mixture was stirred for 2 h, filtered (celite), and evaporated. To this
(m, 4H), 4.45–4.05 (m, 3H), 3.88–3.63 (m, 2H), 1.50–1.62 (bs, 9H). 13C (75 solid was added CH2Cl2 (5 ml), which formed a suspension, and
MHz, CDCl3) ␦ ppm 172.2, 168.6, 156.9, 153.6, 143.7, 143.6, 141.2, 141.1, N,NЈ-di-Boc-1H-pyrazole-1-carboxamidine (1530 mg, 4.94 mmol) was
139.8, 129.4, 128.7, 128.2, 127.6, 127.0, 125.1, 122.2, 120.9, 119.8, 67.3, added and the mixture was left stirring overnight. The mixture was
55.1, 46.9, 28.0. HRMS (ES) exact mass calculated for C26H26N5O5 (M ϩ evaporated with celite and purified by column chromatography with
H Ϫ 2Boc), 488.1934. Found, 488.1938.
ethyl acetate-heptane-acetic acid (1:1:0.01) providing 1980 mg (68%).
Synthesis of Fmoc-4-(N,Nꢁ-di-Cbz-guanidine)-cyclohexyla- 1H NMR (300 MHz, CDCl3) ␦ ppm 10.1 (bs, 4H), 7.82 to 7.13 (m,
lanine-OH (Protected Form of 15). To a degassed (argon) solution 11H), 7.04 to 6.77 (m, 3H), 5.48 (d, J ϭ 7.7 Hz, 1H), 4.53 to 4.12 (m,
of Boc-Phe(4-NO2)-OH (1 g, 3.22 mmol) in acetic acid (3 ml) and 4H), 3.32 to 2.95 (m, 2H), 1.61 to 1.40 (bs, 18H).13C (75 MHz, CDCl3)
methanol (3 ml) was added PtO2 (500 mg), and H2 was bubbled ␦ ppm 174.9, 160.9, 155.8, 153.1, 153.0, 148.4, 143.8, 143.7, 141.3,
through the mixture until completion of the reaction as determined 141.2, 132.9, 129.0, 128.7, 128.2, 127.7, 127.1, 125.3, 125.2, 125.1,
by HPLC (9 h). The mixture was coevaporated several times with 124.7, 124.6, 121.0, 119.9, 84.7, 80.9, 67.1, 54.9, 47.1, 36.7, 28.1, 21.4.
toluene. The green solid was dissolved in dimethylformamide (5 ml), HRMS (ES) exact mass calculated for C25H25N4O5 (M ϩ H-2Boc),
and to this mixture was added N,NЈ-bis(benzyloxycarbonyl)-1H- 461.1825. Found, 461.1831.
pyrazole-1-carboxamidine (1250 mg, 3.30 mmol) and Et3N (0.45 ml,
Synthesis of Fmoc-L-phenylglycine-3-(N,Nꢁ-di-Boc-guanidine)-
3.23 mmol) and left stirring overnight. The mixture was evaporated, OH (Protected Form of 13). L-Phenylglycine (5 g, 33 mmol) was
and ethyl acetate (30 ml) was added and extracted with aqueous dissolved in 97% H2SO4 (12 ml) and cooled to 0°C. To this solution
NaHCO3 (10%, three 20-ml extractions). The aqueous solution was was added 37% HNO3 (1.5 ml, 36 mmol) dropwise. The mixture was
acidified to pH 2 to 3 with aqueous HCl (37%) and extracted with stirred at 0°C for 1 h and neutralized with 25% NH4OH/H2O (20 ml).
CH2Cl2 (three 30-ml extractions) and evaporated. TFA-CH2Cl2 (1:1, When the mixture reached pH 7, a yellow precipitate formed, which
10 ml) was added, and the mixture was stirred for 15 min and then was filtered off and washed with ice-cold deionized H2O. The precip-
evaporated to an oil and left on high vacuum overnight. The solid itate was recrystallized with H2O to afford 3-nitrophenylglycine as a
was dissolved in H2O-dioxane (1:2, 60 ml). To this was added Fmoc- white solid (Sassatelli et al., 2006). The solids were dissolved in
OSu (1.20 g, 3.56 mmol) and NaHCO3 (550 mg, 6.55 mmol), and the H2O-dioxane (2:3, 100 ml). Fmoc-OSu (7.0 g, 21 mmol) and Na2CO3
mixture was left stirring overnight. The mixture was evaporated and (3.6 g, 34 mmol) were added. The mixture was stirred at room
redissolved in acetic acid (5 ml), and celite (3 g) was added. The temperature (18 h) and acidified to pH 2 by 37% HCl. To this was
slurry was evaporated and purified by silica gel column chromatog- added H2O (150 ml), and the mixture was extracted with ethyl
raphy with ethyl acetate-heptane-acetic acid (50:50:1) providing 813 acetate (100 ml). The organic phase was washed with H2O (100 ml)
mg (35%). The product is presumably a cis/trans isomer (Rao et al., and brine (100 ml) and evaporated. Acetic acid (50 ml) and Pd/C (300
1991). 1H NMR (300 MHz, CDCl3) ␦ ppm 7.55 to 7.63 (d, J ϭ 7.3 Hz, mg) was added, and the mixture was left stirring under 1 atm H2 for
2H), 7.52 to 7.61 (m, 2H), 7.32 to 7.45 (m, 10H), 5.16 to 5–05 (m, 5H), 12 h. The mixture was filtered with celite and evaporated to a yellow
4.31 to 4.05 (m, 4H), 2.59 (s, 0.2H), 2.30 (s, 0.8H), 2.05 to 1.02 (m,
8H). 13C (75 MHz, CDCl3) ␦ ppm 176.6, 176.5, 163.8, 156.1, 155.2, added N,NЈ-di-Boc-1H-pyrazole-1-carboxamidine (400 mg, 1.3 mmol)
155.1, 154.0, 153.8, 148.5, 143.9, 143.8, 143.6, 141.3, 137.8, 136.8, and triethylamine (0.35 ml, 2.5 mmol), and the mixture was stirred
136.8, 134.6, 134.5, 134.5, 125.0, 120.0, 68.4, 68.1, 67.1, 51.7, 49.5, for 3 h at room temperature. The mixture was evaporated and
47.1, 46.3, 38.6, 32.4, 32.2, 30.5, 29.2, 29.1, 28.1, 26.9, 21.4. HRMS purified by column chromatography with heptane-ethyl acetate-ace-
(ES) exact mass calculated for C41H43N4O8 (M ϩ H): 719.3081. tic acid (40:40:1) to give the product (620 mg, 85%) as a yellow foam.
Found: 719.3044.
1H NMR (300 MHz, CDCl3) ␦ ppm 1.34 to 1.63 (m, 18 H) 4.07 to 4.55
oil (8.72 g). To 450 mg of this oil in dimethylformamide (10 ml) was
Synthesis of Fmoc-L-3-(N-amidino-3-piperidyl)alanine-OH (m, 3 H) 4.94 to 5.42 (m, 1 H) 6.00 to 6.43 (m, 1 H) 7.03 to 8.10 (m, 15
(Precursor for 17). Fmoc-L-3Pal-OH (700 mg, 1.8 mmol) was dis- H). 13C (75 MHz, CDCl3) ␦ ppm 28.1, 31.2, 47.2, 58.0, 67.2, 105.6,
solved in acetic acid (10 ml) in a round-bottomed flask, and PtO2 (500 119.9, 122.1, 123.3, 123.7, 124.3, 125.2, 127.1, 127.7, 128.3, 129.1,
mg) was added. The solution was degassed, the flask was connected 129.5, 133.0, 136.5, 137.9, 138.3, 141.3, 143.9, 153.2, 154.2, 155.5,
to a hydrogen balloon, and the mixture was left stirring overnight.
160.4, 173.1 HRMS (ES) exact mass calculated for C34H39N4O8,
The mixture was filtered and evaporated to a clear oil. The oil was 631.2768. Found, 631.2767.
further evaporated with toluene (three 30-ml washes) to remove
Synthesis of (S)-2-Acetamido-6-guanidinohexanamide. To a
acetic acid residues. The oil was dissolved in acetonitrile-H2O (2:1, 50 solution of Fmoc-Lys(Alloc)-OH (181 mg, 0.4 mmol) in NMP (2 ml)
ml) and DIPEA (0.15 ml, 0.9 mmol), and Alloc2O (285 mg, 2.0 mmol) was added HBTU (140 mg, 0.37 mmol) and DIPEA (0.12 ml, 0.72
was added. The mixture was stirred at room temperature until the
mmol), and the mixture was added to TentaGel resin (417 mg, 0.1
completion of the reaction as determined by HPLC (1 h). The mixture mmol). The mixture was carefully shaken (2 h) and washed several
was evaporated and purified by column chromatography with ethyl times with NMP. The resin was treated with piperidine-NMP (1:4,
acetate-heptane-acetic acid (1:1:0.02) to afford the product as a white 2 ϩ 15 min), washed several times with NMP and CH2Cl2, and
foam (650 mg, 75%). The product is presumably a mixture of diaste- treated with Ac2O-CH2Cl2 (1:2, two 7-min treatments). To the resin

588
Hosseini et al.
was added degassed CH2Cl2 and Pd(PPh3)4 (30 mg, 0.026 mmol) and active form and could thus be used directly for enzymatic assays.
BH3 ꢀ HN(CH3)2 complex (25 mg, 0.45 mmol). The mixture was Three sets of mutant cDNAs were cloned into pcDNA3.1 and used to
shaken for 30 min and washed several times with NMP. To the resin obtain the corresponding proteins. First, murinized human uPA
was added N,NЈ-di-Boc-1H-pyrazole-1-carboxamidine (310 mg, 1.0 variants, harboring the point mutations H99Y,2 Q192K, or H99Y-
mmol) in NMP (2 ml), and the mixture was shaken overnight at room
Q192K, were those described previously (Andersen et al., 2008).
temperature. The resin was treated with TFA-triethylsilane-H2O Second, humanized murine uPA variants, in which the point muta-
(95:2.5:2.5, 2 h). The mixture was evaporated under a stream of N2, tion K192Q was introduced into murine uPA or in which the human
37, 60, and 99 loops were grafted unto murine uPA, were constructed
by standard cloning procedures and customized DNA synthesis
(GenScript, Piscataway, NJ). More specifically, the murine uPA vari-
ant with human 37, 60, and 99 loops harbors the mutations Q35R,
K36R, N37H, K37aR, Q60aD, L60bY, N63D, and Y99H. Third, cDNA
corresponding to chimeras of human and mouse uPA, in which po-
sitions 22 to 120 or 121 to 250 of human uPA replaced the corre-
sponding residues in a murine uPA background, were synthesized by
GenScript. Conditioned media from the HEK293T transfectants
were used directly for the analysis of kinetic parameters and inhibi-
tion constants for the various uPA variants. In a few cases (wt
human uPA, wt murine uPA, murine uPA with human residues 16 to
120, and murine uPA with human residues 121 to 250), it was shown
that the same kinetic parameters and inhibition constants were
obtained with conditioned media and purified preparations.
The cloning strategy and the production of the recombinant uPA
catalytic domain (residues 159–411) to be used for crystallization
were described previously (Zhao et al., 2007). In brief, the recombi-
nant uPA catalytic domain was secreted from a stable Pichia pastoris
strain (X-33) after induction by methanol and captured by a cation
exchange column. The protein was further purified on a gel filtration
column (Superdex 75 HR 10/30 column; GE Healthcare, Chalfont St.
Giles, Buckinghamshire, UK) equilibrated with 20 mM sodium phos-
phate, pH 6.5, 150 mM NaCl. The protein was eluted as a single peak
under these conditions, with a retention time of approximately 13.6
ml. The recombinant uPA catalytic domain expressed this way is an
active protease with an activity comparable with that of full-length
two-chain uPA (Zhao et al., 2007). The protein was dialyzed in 20
mM potassium phosphate, pH 6.5, overnight and concentrated to 10
mg/ml, using stirred ultrafiltration cells (Millipore Amicon Biosepa-
rations, model 5124; Thermo Fisher Scientific, Waltham, MA) before
protein crystallization.
and the product was purified by preparative HPLC to provide 14 mg
(64%). 1H NMR (300 MHz, D2O) ␦ 4.25 (dd, J ϭ 8.9, 5.4 Hz, 1H), 3.20
(t, J ϭ 6.9 Hz, 2H), 1.92 to 1.69 (m, 2H), 1.61 (dd, J ϭ 14.4, 6.9 Hz,
2H), 1.52 to 1.35 (m, 2H). 13C NMR (75 MHz, D2O) ␦ ppm 156.7, 53.4,
40.8, 30.4, 27.3, 22.1, 21.6. HRMS (ES) exact mass calculated for
C9H20N5O2 (M ϩ H), 230.1617. Found, 230.1609.
Synthesis of (S)-2-Acetamido-3-(4-guanidinophenyl)propana-
mide. HBTU (140 mg, 0.37 mmol) and DIPEA (0.12 ml, 0.72 mmol)
were added to a solution of Fmoc-4-guanidino-phenylalalnine(N,NЈ-di-
Boc)-OH (258 mg, 0.4 mmol) in NMP (2 ml), and the mixture was added
to Tentagel resin (417 mg, 0.1 mmol). The mixture was carefully shaken
(2 h) and washed several times with NMP. To the resin was added
piperidine-NMP (1:4, 2 ϩ 15 min), and the mixture was washed several
times with NMP and CH2Cl2. Subsequently the resin was treated with
Ac2O-CH2Cl2 (1:2, two 7-min treatments) and washed several times
with CH2Cl2. The resin was then treated with TFA-triethylsilane-H2O
(95:2.5:2.5, 2 h). The mixture was evaporated under a N2 flow, and the
product was purified by preparative HPLC (11 mg, 44%). 1H NMR (300
MHz, D2O) ␦ 7.57 to 7.25 (m, 4H), 4.66 to 4.54 (m, 1H), 3.25 (dd, J ϭ
14.0, 6.0 Hz, 1H), 3.03 (dd, J ϭ 13.9, 9.3 Hz, 1H), 1.96 (s, 3H). 13C NMR
(75 MHz, D2O) ␦ 137.5, 136.1, 132.5, 130.3, 130.1, 128.2, 125.5, 122.6,
118.7, 118.2, 117.8, 114.0, 54.1, 36.1, 21.1. HRMS (ES) exact mass
calculated for C12H18N5O2 (M ϩ H), 264.1461. Found, 264.1476.
Synthesis of (S)-2-Acetamido-3-(1-carbamimidoylpiperidin-
4-yl)propanamide. To a solution of Fmoc-L-Ala(4-Pip(Alloc))-OH
(191 mg, 0.4 mmol) in NMP (2 ml) was added HBTU (140 mg, 0.37
mmol) and DIPEA (0.12 ml, 0.72 mmol), and the mixture was added
to a TentaGel resin (417 mg, 0.1 mmol). The mixture was carefully
shaken (2 h) and washed several times with NMP. The resin was
treated with piperidine-NMP (1:4, 2 ϩ 15 min), and the mixture
washed several times with NMP and CH2Cl2. Then the resin was
treated with Ac2O-CH2Cl2 (1:2, two 7-min treatments) and washed
several times with CH2Cl2. To the resin was added degassed CH2Cl2,
BH3 ꢀ HN(CH3)2 complex (25 mg, 0.45 mmol), and Pd(PPh3)4 (30 mg,
0.026 mmol). The mixture was shaken for 30 min and washed several
times with NMP. N,NЈ-Di-Boc-1H-pyrazole-1-carboxamidine (310
mg, 1.0 mmol) in NMP (2 ml) was added, and the mixture was
shaken overnight at room temperature. The resin was treated with
TFA-triethylsilane-H2O (95:2.5:2.5, 2 h). The mixture was evapo-
rated under an N2 flow, and the product was purified by preparative
HPLC (10 mg, 39%). 1H NMR (300 MHz, D2O) ␦ 4.22 (t, J ϭ 7.3 Hz,
1H), 3.70 (d, J ϭ 13.8 Hz, 2H), 2.95 (dd, J ϭ 23.7, 10.9 Hz, 2H), 1.94
(s, 3H), 1.76 to 1.52 (m, 5H), 1.29 to 1.02 (m, 2H). 13C NMR (75 MHz,
D2O) ␦ ppm 177.3, 174.4, 155.6, 51.1, 45.8, 45.7, 37.0, 31.4, 30.9, 29.7,
21.6. HRMS (ES) exact mass calculated for C11H22N5O2 (M ϩ H),
256.1774. Found, 256.1782.
Determination of KM Values for uPA-Catalyzed Hydrolysis
of S-2444. To determine the KM and maximal velocity (Vmax) values
for the uPA-catalyzed hydrolysis of the chromogenic substrate
S-2444, a 200-␮l 2-fold dilution series of the substrate S-2444 (0–4
mM) in a buffer of 10 mM HEPES, pH 7.4, 140 mM NaCl (HEPES-
buffered saline), with 0.1% bovine serum albumin, was incubated 2
min at 37°C before the addition of a fixed concentration of uPA (4 nM
final concentration). The initial reaction velocities (Vi), monitored as
the changes in absorbance at 405 nm, were plotted against the initial
substrate concentration ([S]) and nonlinear regression analysis was
used to determine the Km according to eq. 1:
Vi ϭ Vmax[S]/([S] ϩ KM)
(1)
The Km values for all the uPA variants used in the present study
are listed in Supplemental Table 1.
Recombinant uPA Expression and Purification. Two-chain
human uPA, originating from urine, was purchased from Wakamoto
Pharmaceutical Company (Tokyo, Japan) or Prospec (Rehovot, Is-
rael). Human uPA concentrations were determined from the absor-
Determination of Ki Values for the Inhibition of uPA Vari-
ants by Upain-2 and Mupain-1 Variants. The concentrations of
the peptide variants were determined by measurements of OD280
and the use of sequence-derived extinction coefficients provided by
the Protparam tool provided by the ExPASy server (located at http://
www.expasy.org). For routine determination of Ki values for the
inhibition of the various uPA variants under steady-state inhibition
conditions, a fixed concentration of purified uPA or uPA-containing
conditioned media from transfected cells (2 nM uPA as the final
concentration) was preincubated in 200 ␮l of HEPES-buffered saline
bance at 280 nm, using an extinction coefficient of 1.36 ml ꢀ mgϪ1
ꢀ
cmϪ1 and Mr of 54,000. Two-chain murine uPA was purchased from
Molecular Innovations (Novi, MI). Wild-type and mutant recombi-
nant human and murine uPA were expressed in and in some cases
purified from human embryonic kidney 293T cells transfected with
the corresponding cDNAs in pcDNA3.1 (Petersen et al., 2001; An-
dersen et al., 2008).
When the cells were cultured without addition of protease inhib-
itors able to inhibit the conversion of single-chain pro-uPA to active
two-chain uPA, such as aprotinin, approximately 50% of the uPA in
the conditioned medium and the purified preparations was in the
2 Amino acid numbering used in the text and figures refers to the standard
chymotrypsinogen template numbering.

Exosite Interactions of Peptidylic Serine Protease Inhibitors
589
with 0.1% bovine serum albumin at 37°C, with various concentra- analogs. Concerning upain-1, the novel P1 residues were
tions of upain-2 or mupain-1 variant peptides (0–400 ␮M) for 15 min
before the addition of the chromogenic substrate S-2444 in concen-
trations approximately equal to the KM value for each particular
variant. The initial reaction velocities were monitored at an absor-
bance of 405 nm. The inhibition constants (Ki) were subsequently
determined from the nonlinear regression analyses of plots for Vi/Vo
versus [I]0 using eq. 2, derived under an assumption of competitive
inhibition:
substituted into the modified sequence CSWRGLENHAAC
(upain-2).3 This modification was chosen because the cycliza-
tion step could be performed much faster when Arg10 of
upain-1 (CSWRGLENHRMC) was replaced with Ala and,
secondly, H2O2 could be used as an oxidant when Met11 was
absent. As predicted from our previous report (Hansen et al.,
2005), the modified sequence bound to human uPA with a Ki
indistinguishable from that of the original peptide: 35 Ϯ 8
␮M (n ϭ 5) for the modified sequence versus 30 Ϯ 1 ␮M (n ϭ
3) for the original sequence. When planning which arginine
analogs to test as the P1 residue in upain-2 and mupain-1, we
considered varying several parameters, all expected to affect
the interaction of the P1 residue with the S1 pocket (Table 1).
Vi/V0 ϭ ͑Ki ϫ ͑KM ϩ ͓S͔ ͒/͑͑Ki ϫ ͓S͔ ͒ ϩ ͑KM ϫ ͑Ki ϩ [I]0))) (2)
0
0
where Vi and V0 are the reaction velocities in the presence and
absence of inhibitor, respectively; [S]0 and [I]0 are the substrate and
inhibitor concentrations, respectively, and KM is the KM for the uPA
catalyzed hydrolysis of S-2444. In eq. 2, it is assumed that [S]free
Ϸ
[S]0 and [I]free Ϸ [I]0. These conditions were fulfilled because less
than 10% of the substrate was converted to product in the assays and
because the assay typically contained a final concentration of uPA
variant of 2 nM and inhibitor concentrations in the micromolar
range.
In cases for which we observed no measurable inhibition (i.e.,
Ͻ10%) at 400 ␮M, the maximal inhibitor concentration used, the
accuracy of the assay allowed us to conclude that the Ki value was
above 1000 ␮M (indicated as “Ͼ1000 ␮M” in the tables). In other
cases in which we did observe a measurable inhibition at 400 ␮M, we
were still unable to accurately determine Ki values more than 500
␮M (indicated as “Ͼ500 ␮M” in the tables).
The validity of performing the Ki determinations with uPA-con-
taining conditioned media from transfected cells was verified by
controls in which the determinations were performed with condi-
tioned media as well as with purified preparations. These controls
were performed with wt murine uPA, wt human uPA, human uPA
Q192K, murine uPA, and human residues 16 to 120, obtaining in-
distinguishable values with the types of samples.
X-Ray Crystal Structure Analysis. The crystallization trials
were performed using the method of sitting drop vapor diffusion. The
crystals of the uPA catalytic domain were obtained by equilibrating
against a reservoir solution containing 2.0 M ammonium sulfate, 50
mM sodium citrate, pH 4.6, and 5% PEG 400 at room temperature.
The crystals appeared in approximately 3 days. For the uPA-upain-
2-13 complex, the crystals of uPA were then soaked for 2 weeks in
crystallization mother liquid containing 1 mM upain-2-13. Before
X-ray data collection, the crystals were soaked for 5 min in a cryo-
protectant solution (2.0 M ammonium sulfate, 5% polyethylene gly-
col 400, 100 mM Tris-HCl, pH 7.4, and 20% glycerol).
1. The chirality of the P1 residue was tested by substituting
L-Arg (1) with D-Arg (3).
2. The length of the aliphatic side chain was tested by sub-
stituting arginine with homologs containing one to five
methylene groups between the ␣-carbon and the gua-
nidino group (4–7). Besides determining how far the gua-
nidino moiety can insert into the S1 pocket, the spacer
length could also have an effect on the binding entropy
and the strength of hydrophobic interactions.
3. The exact nature of the basic group could be tested by
replacing the guanidino group of L-Arg with a urea moiety
(pKa ϭ 12), an amino group (pKa ϭ 10), or a guanidino-oxo
group (pKa ϭ 7) (Boyar and Marsh, 1982).
4. The structure of the carbon skeleton between the ␣-carbon
and the basic group was varied by introducing aliphatic or
aromatic rings (11–14). Aromatic rings would be spatially
more demanding and restrict the conformational orienta-
tions and potentially mimic already reported small mole-
cule inhibitors. They would also result in a lower pKa
value for the guanidino moiety. Cyclic aliphatic groups
would restrict the conformational orientations (15–17).
Although a variety of arginine analogs are commercially
available, more arginine analogs had to be synthesized de novo
(Fig. 1) to complete the planned scan of the P1 position (Table 1).
We here designed and synthesized the following novel arginine
analogs: 4-guanidino-L-cyclohexylalanine (15), 3-guanidino-L-
tyrosine (14), 3-guanidino-L-phenylglycine (13), (S)-2-amino-3-
(4-guanidinobenzamido)propanoic acid (11), and L-3-(N-
amidino-3-piperidyl)alanine (17).
With the de novo synthesized arginine analogs and the
commercially available analogs at hand for insertion into the
P1 position, linear forms of the peptides were assembled by
solid-phase peptide synthesis. The peptides were obtained as
the N-terminal acetylated and C-terminal amidated forms
and were cyclized by disulfide bridge formation.
Effects of P1 Substitutions on the Inhibition of Hu-
man uPA by Upain-2 and of Murine uPA by Mupain-1.
We investigated the effect of substituting the different ana-
logs into the P1 position of upain-2 or mupain-1 on their
ability to inhibit human or murine uPA (Table 1). The bind-
ing of the peptide to their targets was critically dependent on
the side chain of the P1 residue, because substitution of P1
Arg with Ala (2) abolished all inhibitory activity.
X-ray diffraction data for the crystals were collected at the BL17U1
beamline (Shanghai Synchrotron Radiation Facility, Shanghai, China).
All diffraction data were indexed and integrated with the HKL2000
program package (Otwinowski and Minor, 1997).
The crystal structure of the uPA-upain-2-13 complex was solved
by molecular replacement with the CCP4 program package (Collab-
orative Computational Project, Number 4, 1994), using the structure
of the uPA catalytic domain (PDB file 2NWN) (Zhao et al., 2007) as
the search model. The electron density for the peptide was clearly
visible in the uPA active site and was modeled on the basis of the
Fo Ϫ Fc difference map. The structure was refined with the CCP4
program package and manually adjusted with the molecular graph-
ics program Coot (Emsley and Cowtan, 2004) iteratively until the
convergence of the refinement. Solvent molecules were added using
a Fo Ϫ Fc Fourier difference map at 2.5 ␴ in the final refinement step.
The final structures were analyzed by software PyMOL (DeLano,
2002).
Results
3 Each of the amino acids tested as the P1 residue was given a number n
between 1 and 19 (Table 1). The corresponding variants of upain-2 and mu-
Non-Natural Arginine Analogs. The P1 Arg residue in
mupain-1 was systematically substituted with a series of Arg pain-1 were named upain-2-n and mupain-1-n, respectively.

590
Hosseini et al.
Fig. 1. Synthesis of novel arginine analogs for use as P1 residues in upain-2 and mupain-1. Chemical synthesis of the protected forms of new Arg
analogs 11, 13, 14, and 15 as well as the protected form of the precursor for 17. After incorporation of 17 into peptides, the Alloc protection was removed
and the liberated amine guanidylated on-resin. Similar chemical processes were used for the construction of peptides containing homoarginine.

Exosite Interactions of Peptidylic Serine Protease Inhibitors
591
Substitution of P1 Arg in upain-2 with a number of L-Arg
analogs with linear aliphatic side chains (3–10) in almost all
cases resulted in a strong loss of activity. Neither a change in
chirality, variation in the chain length, nor exchange of the
guanidino group with a urea moiety, an amino group, or a
guanidino-oxo group was accepted. The only exception was
substitution with L-homo-arginine (6), which resulted in a
slight improvement of affinity. Moreover, substitution with
aromatic (11–14) or aliphatic ring structures (15–17) always
resulted in a strong loss of inhibitory activity.
sis and X-ray crystal structure determination (Hansen et
al., 2005; Zhao et al., 2007; Andersen et al., 2008), the
exosite interactions are very different for upain-2 and
mupain-1. The exosite interactions may restrict the way
the P1 residue inserts into the S1 pocket and vice versa,
thus determining how a certain P1 substitution affects the
affinity.
To distinguish between these two possibilities, we con-
structed chimeras in which exosite and S1 pocket residues
Similar to upain-2, substitution of the P1 Arg in mupain-1 were exchanged between human and murine uPA. The choice
with L-Arg analogs containing linear aliphatic side chains of exosites was based on previous results from site-directed
(3–10) in all cases resulted in loss of inhibitory activity, with mutagenesis and X-ray crystal structure analysis (Fig. 2). We
an increase in Ki of between 5 and Ͼ2000-fold. The response previously demonstrated that “murinization” of human uPA
to substitution with L-Arg analogs with aromatic ring struc- in position 99 (i.e., generation of human uPA H99Y) made
tures (11–14) was more varied, because the substitution with human uPA sensitive to inhibition by mupain-1, however
the Phe analog 12 resulted in a slight increase in affinity, with a Ki value approximately 20-fold higher than the Ki
whereas other substitutions resulted in a strong loss of af- value for inhibition of murine uPA (Andersen et al., 2008).
finity. Moreover, inhibition of murine uPA could still be We therefore used human uPA H99Y to study the importance
achieved with L-Arg analogs with aliphatic ring structures of exosite interactions for mupain-1 binding. We also previ-
(15–17). In fact, substitution with the amidino compound 16 ously demonstrated the importance of the so-called 37 and 60
resulted in a 10-fold increase in affinity (Table 1). Thus, in loops and of position 99 of human uPA for binding of upain-1
contrast to upain-2, with mupain-1 we managed to achieve a (Hansen et al., 2005; Zhao et al., 2007)(Fig. 2). We now
considerable increase in affinity to the target enzyme by constructed and expressed a “humanized” murine uPA, with
suitable substitutions.
a human 37-loop (murine uPA Q35R, K36R, N37H, and
The effects of substituting the same amino acid into the P1 K37aR), a human 60-loop (murine uPA Q60aD, L60bY, and
position was in several cases strikingly different for upain-2 N63D), and the human 99-residue grafted onto murine uPA
targeting human uPA compared with mupain-1 targeting mu- (murine uPA Y99H). To evaluate the contribution from the
rine uPA. Substitution with L-homo-arginine (6) increased the S1 pocket, we constructed a human uPA variant that was
affinity of upain-2 to human uPA but decreased the affinity of murinized in position 192 (human uPA Q192K) and a variant
mupain-1 to murine uPA 5-fold; substitution with the Phe de- of murine uPA that was humanized in position 192 (murine
rivative 12 or the amidino compound 16, respectively, de- uPA K192Q). Position 192 is the only position with a noncon-
creased the affinity of upain-2 to human uPA, but increased the served amino acid among the positions with residues lining
affinity of mupain-1 to murine uPA.
the S1 pocket (Fig. 2).
We confirmed the competitive nature of the inhibition with
As a more radical way of exchanging residues, we con-
selected peptides (i.e., mupain-1-12 and mupain-1-16). The structed variants of murine uPA in which either the N-ter-
Ki values determined for inhibition of murine uPA, calcu- minal ␤-barrel (residues 16–120) or the C-terminal ␤-barrel
lated with eq. 2, were found to be indistinguishable at sub- (residues 120–250) of the catalytic domain was replaced by
strate concentrations ranging from 0.2 to 3 times the Km the corresponding sequences of human uPA. All identified
value.
exosite interactions between upain-1 and human uPA involv-
An important finding was that the amidino compound 16, ing nonconserved residues are present in the N-terminal
L-homo-arginine (6), and the Phe derivative 12, as free N-ter- ␤-barrel. Likewise, all identified exosite interactions between
minal acetylated and C-terminal amidated amino acids, were upain-1 and human uPA involving nonconserved residues
unable to inhibit human or murine uPA measurably in con- are present in this ␤-barrel. In contrast, the S1 pocket is
centrations up to 400 ␮M (data not shown).
totally embedded in the C-terminal ␤-barrel (Fig. 2).
Effects of P1 Substitutions on the Inhibition of Chi-
We then tested the susceptibility of the constructed chime-
meras between Human and Murine uPa by Mupain-1 ras to the original upain-2 and mupain-1 peptides and to
and Upain-2. We considered two different hypotheses for selected P1-substituted variants of upain-2 and mupain-1.
the different effects of P1 substitutions on upain-2 inhibition The variants selected were variants with the amidino com-
of human uPA and on mupain-1 inhibition of murine uPA:
pound 16, L-homo-arginine (6), or the Phe derivative 12 as
the P1 residue, because of the differential effect these sub-
1. The different responses to the substitutions in human stitutions have on the affinity of upain-2 to human uPA and
uPA on one side and murine uPA on the other could be due of mupain-1 to murine uPA (Table 2).
to the new amino acids in the P1 positions fitting differ-
ently into the S1 pockets of human and murine uPA, upain-2 or mupain-1 is determined by the exosites, not posi-
respectively. tion 192. In almost all cases, substitution with L-homo-argi-
By inspecting the data, it is seen that the susceptibility to
2. The different responses to the P1 substitutions on inhibi- nine (6) either increased or left unchanged the affinity of
tion of human and murine uPA could be related to the fact upain-2 to the uPA variants, but decreased the affinity of
that the effect of a certain P1 substitution is strongly mupain-1, whereas the reverse was true for substitutions
affected by the exosite interactions. As expected from the with amino acids 12 and 16. Strikingly, the latter P1 substi-
different sequences of the two peptides outside the P1 tutions even resulted in a measurable affinity of mupain-1 to
position and as demonstrated by site-directed mutagene- wt human uPA and human uPA Q192K, which were not

592
Hosseini et al.
TABLE 1
Ki values for inhibition of human uPA by upain-2 with P1 substitutions or inhibition of murine uPA by or mupain-1 variants with P1 substitutions
Ki values for the inhibition of S-2444 hydrolysis by human uPA by the indicated peptides were determined at 37°C. The table shows mean Ϯ S.D., with the number of
determinations indicated in parentheses.
Ki for Inhibition
P1 Residue
Chemical Structure of P1 Residue
Human uPA: Upain-2 Variants
Murine uPA: Mupain-1 Variants
␮M
L-Arginine (1)
L-Alanine (2)
D-Arginine (3)
35 Ϯ 8 (5)
Ͼ1000**
Ͼ1000**
0.55 Ϯ 0.08 (8)
Ͼ1000***
Ͼ1000***
L-2,3-Diaminopropionic acid-amidine (4)
Ͼ1000**
Ͼ1000**
Ͼ1000***
Ͼ1000***
L-Nor-arginine (5)
L-Homo-arginine (6)
23 Ϯ 4 (3)*
2.88 Ϯ 0.58 (5)***
L-Bis-homo-arginine (7)
Ͼ1000**
N.D.
L-Citrulline (8)
L-Ornithine (9)
Ͼ1000**
Ͼ1000**
Ͼ1000***
Ͼ500***

Exosite Interactions of Peptidylic Serine Protease Inhibitors
593
TABLE 1—Continued
Ki for Inhibition
P1 Residue
Chemical Structure of P1 Residue
Human uPA: Upain-2 Variants
Murine uPA: Mupain-1 Variants
␮M
Canavanine (10)
Ͼ1000**
169 Ϯ 9 (3)***
L-4-Guanidino-benzoic acid-2,3-
diaminopropionic acid (11)
Ͼ1000**
Ͼ1000***
L-4-Guanidino-phenylalanine (12)
L-3-Guanidino-phenylglycine (13)
Ͼ500**
0.29 Ϯ 0.02 (5)***
a
316 Ϯ 10 (3)**
343 Ϯ 26 (3)***,
L-3-Guanidino-tyrosine (14)
Ͼ1000**
396 Ϯ 35 (3)***
L-4-Guanidino-cyclohexyl-alanine (15)
L-3-(N-Amidino-4-piperidyl)alanine (16)
Ͼ1000**
Ͼ1000**
0.92 Ϯ 0.25 (3)***
0.045 Ϯ 0.010 (4)***
L-3-(N-Amidino-3-piperidyl)alanine (17)
N.D.
3.00 Ϯ 0.34 (3)***
N.D., not determined.
a Two peptides could be isolated with the same mass, probably stemming from epimerization of the sensitive benzylic carbon. The other peptide had no measurable activity.
* Significantly different from the Ki values for inhibition of human uPA by upain-2 (P Ͻ 0.05).
** Significantly different from the Ki values for inhibition of human uPA by upain-2 (P Ͻ 0.005).
*** Significantly different from the Ki values for inhibition of murine uPA by mupain-1 (P Ͻ 0.005).

594
Hosseini et al.
Fig. 2. Residues in human or murine uPA, which were implicated in
binding of upain-1 and mupain-1, respectively. A, alignment of the cata-
lytic domains of human and murine uPA. The residue numbering indi-
cated is by chymotrypsin template numbering (Spraggon et al., 1995).
The N-terminal ␤-barrel is highlighted in green and the C-terminal
␤-barrel in wheat. The residues in the sequence of human uPA that are
highlighted in light green were implicated in upain-1 binding by site-
directed mutagenesis (Hansen et al., 2005): Arg35, Arg36, Arg37a, Tyr40,
Phe59, Asp60a, Tyr60b, Tyr64, Tyr94, His99, Tyr151, and Trp186. The
residues in the sequence of murine uPA that are highlighted in dark red
were implicated in mupain-1 binding by site-directed mutagenesis (An-
dersen et al., 2008): Lys41, His57, Tyr99, Lys143, Asp189, and Ser195.
B, overview of the structure of the uPA-upain-1 complex. uPA is depicted
in surface mode and upain-1 in sticks with CPK colors. The N-terminal
␤-barrel of uPA is colored green; the C-terminal ␤-barrel of uPA is colored
wheat. N-terminal ␤-barrel exosite residues implicated in upain-1 bind-
ing by site-directed mutagenesis are indicated by light green. The lining
of the S1 specificity pocket is colored red. Gln192 is colored firebrick. The
figure was constructed from PDB file 2NWN (Zhao et al., 2007).
␮
measurably inhibited by the original mupain-1. Thus, the
directions of the changes in affinities to the chimeras after
substitutions of the P1 residues of upain-2 and mupain-1
were the same as those observed with wt human and murine
uPA, although the fold changes were not exactly the same. In
contrast, mutagenesis of position 192 did not change the
direction of the changes. These results favor hypothesis 2
over hypothesis 1.
There were only a few exceptions to this generalization.
Thus, substitution of the P1 residue of upain-2 with the
amidino compound 16 (upain-2-16) unexpectedly resulted in
a measurable inhibition of murine uPA, although the affinity
remained low. In addition, substitution of the P1 residue of
mupain-1 with L-homo-arginine (6) (mupain-1-6) resulted in
a slightly increased affinity to murine uPA with human 37
and 60 loops and position 99, whereas P1 substitution of
upain-2 with L-homo-arginine (6) seemed to slightly decrease
the affinity to the same uPA variant.
X-Ray Crystal Structure Analysis of a Complex be-
tween a Upain-2 Variant and Human uPA. This analysis

Exosite Interactions of Peptidylic Serine Protease Inhibitors
595
␧
was followed by X-ray crystal structure analysis. However, these structures. In addition, in this complex, the N atom is
we were restricted to analyzing complexes between human hydrogen-bonded to Gly219 (Fig. 3B). In the uPA-upain-1
␩
uPA and upain-2 variants, because it has not yet been pos- complex, only one of the N atoms of the guanidino group
␩
sible to crystallize complexes between mupain-1 and murine hydrogen bonds with Asp189, whereas the other N hydrogen
uPA. Among the complexes between human uPA and upain-2 bonds with Gly219. The contact surface areas in the S1
variants, we were now able to determine the crystal structure pocket in the upain-1 complex and the uPA-upain-2-13 com-
of human uPA in complex with an upain-2 analog (upain-2- plex are quite similar to each other (150 versus 154 Å2).
13) to a high resolution (1.32 Å) (Fig. 4A). Peptide upain-2-13 Thus, the P1-S1 interaction in the uPA-upain-2-13 complex
has a 10-fold lower affinity to human uPA than upain-2 is similar to that of the uPA-upain-1 complex; however, the
(Table 1).
guanidino group in the uPA-upain-2-13 complex probably
Structural analysis revealed that the uPA moiety of the has a more favorable hydrogen bonding pattern than that in
uPA-upain-2-13 complex has a conformation almost identical the uPA-upain-1 complex.
to that of uPA in complex with the original upain-1 (Zhao et
In contrast to this conservation of P1-S1 interaction, the
al., 2007) (root mean square deviation of 0.35 Å). Further- P3-P1Ј segments are quite different in the uPA-upain-2-13
more, the guanidino group of 13 forms five hydrogen bonds complex and uPA-upain-1 complex. In the structure of uPA-
with uPA residues in the S1 pocket. This hydrogen bonding upain-2-13, the peptide is moved toward the solvent and
scheme is similar, although not identical, to that of the P1 away from uPA by approximately 1.7 Å compared with
L-Arg in the complex between uPA and upain-1 (Zhao et al., upain-1 (Fig. 3C). As a result, upain-2-13 has lost an intra-
2007) (Fig. 3). In the uPA-upain-2-13 complex, the guanidino molecular hydrogen bond (Trp3 to Ala6), which exists in
group aligns well with the carboxyl group of Asp189, allowing upain-1. In addition, Trp3 in the uPA-upain-2-13 complex
the formation of two bifurcated hydrogen bonds between makes less contact with uPA compared with that in the
uPA-upain-1 complex (64 versus 82 Å2), suggesting a poten-
tially nonoptimal interaction of the inhibitor with uPA. How-
ever, the contacts between upain-2-13 and the exosites in the
N-terminal ␤-barrel remain unchanged. Thus, the anchoring
of this peptide in the S1 pocket and to the N-terminal ␤-bar-
rel exosites leads to a change of the P3-P1Ј segment and
weakening of the interactions of Trp3 with uPA. This result
is in agreement with hypothesis 2.
Of importance, both the uPA-upain-1 complex and the
uPA-upain-2-13 complex display decisive differences from
the recently determined Michaelis complex between S195A
uPA and its natural proteinaceous serpin inhibitor PAI-1
(Lin et al., 2011). In the uPA-PAI-1 Michaelis complex, the
P2-P1Ј sequence inserts into the active site of uPA in a
substrate-like fashion. Although the hydrogen bonding
scheme of the P1 residue in the uPA-PAI-1 Michaelis com-
plex is similar to that in the uPA-upain-1 complex and the
uPA-upain-2-13 complex, the localizations of the P2 and P1Ј
residues are decisively different. The difference is caused by
the inability of the bulky Trp residue in the P2 position of the
inhibitory peptides to insert into the S2 pocket of uPa (Fig.
3B) and explains why upain-1 and upain-2-13 are unable to
align into the active site in a substrate-like fashion, whereas
PAI-1 is.
Discussion
We here report data elucidating the relative importance of
active site and exosite interactions for the binding of pepti-
Fig. 3. X-ray crystal structure analysis of the complex between human
dylic inhibitors to uPA. The data presented show that exosite
interactions strongly influence the effect of a certain P1 sub-
stitution on the Ki values. From the data, it seems reasonable
to conclude that there is a mutual dependence between the
strength of the exosite interactions and the exact fit of the P1
residue into the S1 pocket, thus determining whether a cer-
tain P1 substitution affects the affinity positively or nega-
tively. The determination of the Ki values was supplemented
by X-ray crystal structure analysis of a upain-2 variant in
complex with human uPA, demonstrating that the key to the
effect of the P1 substitutions in upain-2 is the communication
between the P1-S1 interactions and the interactions of the
uPA and upain-2-13. A, overview of the complex between human uPA and
upain-2-13. The uPA structure is shown as ribbons and upain-2-13 is
indicated in sticks with CPK colors. The electron density map (␴-
weighted 2Fo-Fc omit map contoured at 1.0 ␴) of upain-2-13 is shown in
green. B, close-ups of the interactions of the P1 residues of various
inhibitors within the S1 pocket of uPA. Top, uPA-upain-2-13 complex;
middle, uPA-upain-1 complex (PDB file 2NWN) (Zhao et al., 2007); bot-
tom, PAI-1-uPA Michaelis complex (PDB file 3PB1) (Lin et al., 2011).
Besides the P1 residues of the inhibitors, the figure also shows the P2
residues (Trp in upain-2-13 and upain-1 and Ala in PAI-1) and the P1Ј
residues (Gly in upain-2-13 and upain-1 and Met in PAI-1) of the inhib-
itors and Asp189, Ser190, and Gly219 of uPA. C, comparison of the
structures of upain-2-13 and upain-1 in their complexes with uPA. The
figure was constructed by aligning the uPA moieties of the complexes.
Upain-2-13 is shown in sticks with CPK colors and upain-1 is in green.

596
Hosseini et al.
monoclonal antibodies (Danø et al., 1980; Kaltoft et al., 1982;
Petersen et al., 2001; Lund et al., 2008; Blouse et al., 2009)
are available but lack some of the advantageous properties of
peptides. Some aryl-amidine-based, small molecule inhibi-
tors had a strong preference (up to 203-fold) for human uPA
over murine uPA (Klinghofer et al., 2001). Whereas naph-
thamidine targeted only the S1 pocket and was equally po-
tent with human and murine uPA, the specificity for human
uPA increased with increasing number and increasing size of
the side chains. X-ray crystal structure analysis demon-
strated that the most specific inhibitors interacted with
Asp60a in human uPA, but no mutagenesis was performed to
test this suggestion. Nevertheless, the observation is in
agreement with our finding that specificity depends on exo-
site interactions (Fig. 2) (Hansen et al., 2005; Zhao et al.,
2007; Andersen et al., 2008).
The strong species specificities of upain-2 and mupain-1
depend on the interaction of the peptides with species-spe-
cific surface loops outside the active site (Fig. 2). Human or
murine uPA can be made susceptible to mupain-1 or upain-2
by grafting mouse-specific residues onto human uPA or hu-
man-specific residues onto murine uPA (Table 2). In this
context, one also notes that substitution of the P1 L-Arg
residue with the amidino compound 16, L-homo-arginine (6),
Fig. 4. The relationship between the Ki values for mupain-1 substituted
with non-natural amino acids and the Ki values for mupain-1. The figure
is based on the Ki values presented in Table 2 for wt human and murine or the Phe derivative 12 may deteriorate the species speci-
uPA and human-murine uPA chimeras. Data points corresponding to Ki
values more than 500 ␮M were omitted.
ficity. This could be due to a larger fraction of the binding
energy depending on interactions within the S1 pocket or
other target-independent contributions. Several examples il-
Trp3 of upain-2. Taken together, the data are in agreement with lustrate this point. Thus, the original mupain-1 is a Ͼ2000-
the above formulated hypothesis 2 rather than hypothesis 1.
fold better inhibitor of murine uPA than of human uPA, but
Searching for quantitative patterns in the changes in Ki mupain-1-12 is only a 300-fold better inhibitor of murine uPA
values after substitution of the P1 L-Arg with non-natural than of human uPA. Nevertheless, mupain-1-16, the best
amino acids, we plotted the Ki values for inhibition of the mupain-1 variant developed until now, still displays an ap-
various uPA variants by the various non-natural mupain-1 proximately 2000-fold selectivity for murine uPA over human
variants versus the corresponding values for the original uPA. This high specificity could indicate that mupain-1-16
mupain-1 in a double logarithmic plot (Fig. 4). A similar plot also will be specific among serine proteases in general, which,
could not be constructed for upain-2, because the Ki values combined with a favorable Ki (45 nM), makes mupain-1-16 a
were in too many cases above the level allowing an accurate candidate for use in in vivo experiments.
determination. With the mupain-1 variants, we observed a
linear relationship in such a plot (Fig. 4). Although there is
no obvious molecular explanation for the linearity of the plot,
it does allow us to draw an interesting inference about the
effect of the substitutions. With both L-homo-arginine (6) and
the Phe analog 12, the slopes of the lines were less than 1
(0.62 and 0.66, respectively), indicating that P1 substitutions
with these amino acids generally caused more increase or
less decrease in affinity, the lower the affinity of the original
mupain-1 peptide was to the target enzyme. However, with
the amidino compound 16, the slope was 0.89, showing that
the fold change in Ki was almost independent of the target.
On the basis of this observation, one may suggest that the
effect of the substitutions also partly depends on a target-
independent element. This dependence cannot be explained
by either hypothesis 1 or hypothesis 2 but may rather be
related to a destabilization of the unbound state of the peptide.
Comparison of the roles of murine and human uPA in
xenografted tumors may help to delineate the relative impor-
tance of tumor and host uPA in tumor growth and spread.
Availability of species-specific inhibitors is therefore impor-
tant. Although a number of inhibitors of human uPA have
been developed, the species specificity has been studied in
only a few cases. Species-specific inhibitory polyclonal and
Authorship Contributions
Participated in research design: Sørensen, K. J. Jensen, Hosseini,
Huang, and Andreasen.
Conducted experiments: Yuan, Sørensen, Hosseini, Jiang, Chris-
tensen, and J. K. Jensen.
Contributed new reagents or analytic tools: Sørensen, Andersen,
Hosseini, and Fogh.
Performed data analysis: K. J. Jensen, Hosseini, Huang, Andrea-
sen, Sørensen, Christensen, and J. K. Jensen.
Wrote or contributed to the writing of the manuscript: K. J. Jensen,
Hosseini, Huang, Andreasen, Sørensen, and J. K. Jensen.
References
Abramowitz N, Schechter I, and Berger A (1967) On the size of the active site in
proteases. II. Carboxypeptidase-A. Biochem Biophys Res Commun 29:862–867.
Andersen LM, Wind T, Hansen HD, and Andreasen PA (2008) A cyclic peptidylic
inhibitor of murine urokinase-type plasminogen activator: changing species spec-
ificity by substitution of a single residue. Biochem J 412:447–457.
Andreasen PA, Egelund R, and Petersen HH (2000) The plasminogen activation
system in tumor growth, invasion, and metastasis. Cell Mol Life Sci 57:25–40.
Blouse GE, Dupont DM, Schar CR, Jensen JK, Minor KH, Anagli JY, Gårdsvoll H,
Ploug M, Peterson CB, and Andreasen PA (2009) Interactions of plasminogen
activator inhibitor-1 with vitronectin involve an extensive binding surface and
induce mutual conformational rearrangements. Biochemistry 48:1723–1735.
Boyar A and Marsh RE (1982) L-Canavanine, a paradigm for the structures of
substituted guanidines. J Am Chem Soc 104:1995–1998.
Branden C and Tooze J (1991) Introduction to Protein Structure. Garland Publishing,
New York.

Exosite Interactions of Peptidylic Serine Protease Inhibitors
597
Collaborative Computational Project, Number 4 (1994) The CCP4 suite: programs
for protein crystallography. Acta Crystallogr D Biol Crystallogr 50:760–763.
Danø K, Nielsen LS, Møller V, and Engelhart M (1980) Inhibition of a plasminogen
activator from oncogenic virus-transformed mouse cells by rabbit antibodies
against the enzyme. Biochim Biophys Acta 630:146–151.
DeLano (2002) The PyMOL Molecular Graphics System, San Carlos, CA.
Dennis MS, Eigenbrot C, Skelton NJ, Ultsch MH, Santell L, Dwyer MA, O’Connell
MP, and Lazarus RA (2000) Peptide exosite inhibitors of factor VIIa as anticoag-
ulants. Nature 404:465–470.
Emsley P and Cowtan K (2004) Coot: model-building tools for molecular graphics.
Acta Crystallogr D Biol Crystallogr 60:2126–2132.
Hajdin K, D’Alessandro V, Niggli FK, Scha¨fer BW, and Bernasconi M (2010) Furin
targeted drug delivery for treatment of rhabdomyosarcoma in a mouse model.
PLoS One 5:e10445.
Krook M, Lindbladh C, Eriksen JA, and Mosbach K (1997–1998) Selection of a cyclic
nonapeptide inhibitor to ␣-chymotrypsin using a phage display peptide library.
Mol Divers 3:149–159.
Lin Z, Jiang L, Yuan C, Jensen JK, Zhang X, Luo Z, Furie BC, Furie B, Andreasen
PA, and Huang M (2011) Structural basis for recognition of urokinase-type plas-
minogen activator by plasminogen activator inhibitor-1. J Biol Chem 286:7027–
7032.
Lund IK, Jo¨gi A, Rønø B, Rasch MG, Lund LR, Almholt K, Gårdsvoll H, Behrendt N,
Rømer J, and Høyer-Hansen G (2008) Antibody-mediated targeting of the uroki-
nase-type plasminogen activator proteolytic function neutralizes fibrinolysis in
vivo. J Biol Chem 283:32506–32515.
Otwinowski Z and Minor W (1997) Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol 276:307–326.
Petersen HH, Hansen M, Schousboe SL, and Andreasen PA (2001) 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 268:4430–4439.
Hansen M, Wind T, Blouse GE, Christensen A, Petersen HH, Kjelgaard S, Mathia-
sen L, Holtet TL, and Andreasen PA (2005) A urokinase-type plasminogen acti-
vator-inhibiting cyclic peptide with an unusual P2 residue and an extended pro-
tease binding surface demonstrates new modalities for enzyme inhibition. J Biol
Chem 280:38424–38437.
`
Puente XS, Ordo´n˜ez GR, and Lo´pez-Otin (2008) The Cancer Degradome: Proteases
and Cancer Biology. Springer, New York.
Rao PN, Peterson DM, Acosta CK, Bahr ML, and Kim HK (1991) Synthesis of
cis-4-aminocyclohexyl-D-alanine and trans-4-aminocyclohexyl-D-alanine deriva-
tives and determination of their stereochemistry. Org Prep Proced Int 23:103–110.
Rockway TW, Nienaber V, and Giranda VL (2002) Inhibitors of the protease domain
of urokinase-type plasminogen activator. Curr Pharm Des 8:2541–2558.
Sassatelli M, Debiton E, Aboab B, Prudhomme M, and Moreau P (2006) Synthesis
and antiproliferative activities of indolin-2-one derivatives bearing amino acid
moieties. Eur J Med Chem 41:709–716.
Spraggon G, Phillips C, Nowak UK, Ponting CP, Saunders D, Dobson CM, Stuart DI,
and Jones EY (1995) The crystal structure of the catalytic domain of human
urokinase-type plasminogen-activator. Structure 3:681–691.
Walker B and Lynas JF (2001) Strategies for the inhibition of serine proteases. Cell
Mol Life Sci 58:596–624.
Wu P, Leinonen J, Koivunen E, Lankinen H, and Stenman UH (2000) Identification
of novel prostate-specific antigen-binding peptides modulating its enzyme activity.
Eur J Biochem 267:6212–6220.
Wu P, Weisell J, Pakkala M, Pera¨kyla¨ M, Zhu L, Koistinen R, Koivunen E, Stenman
UH, Na¨rva¨nen A, and Koistinen H (2010) Identification of novel peptide inhibitors
for human trypsins. Biol Chem 391:283–293.
Hartung R, Hitzel-Zerrahn F, Mu¨ller T, and Pietsch J (2005) inventors; Degussa AG,
assignee. Process for the hydrogenation of aromatic compounds. World patent
WO/2005/014526. 2005 Feb 17.
Hedstrom L (2002) Serine protease mechanism and specificity. Chem Rev 102:4501–
4524.
Heinis C, Rutherford T, Freund S, and Winter G (2009) Phage-encoded combinato-
rial chemical libraries based on bicyclic peptides. Nat Chem Biol 5:502–507.
Hekim C, Leinonen J, Na¨rva¨nen A, Koistinen H, Zhu L, Koivunen E, Va¨isa¨nen V,
and Stenman UH (2006) Novel peptide inhibitors of human kallikrein 2. J Biol
Chem 281:12555–12560.
Jorgensen L and Nielsen HM (2009) Delivery Technologies for Biopharmaceuticals Peptides,
Proteins, Nucleic Acids and Vaccines. John Wiley & Sons, Chichester, UK.
Kaltoft K, Nielsen LS, Zeuthen J, and Danø K (1982) Monoclonal antibody that
specifically inhibits a human Mr 52,000 plasminogen-activating enzyme. Proc Natl
Acad Sci USA 79:3720–3723.
Kayser KJ, Glenn MP, Sebti SM, Cheng JQ, and Hamilton AD (2007) Modifications
of the GSK3␤ substrate sequence to produce substrate-mimetic inhibitors of Akt as
potential anti-cancer therapeutics. Bioorg Med Chem Lett 17:2068–2073.
Klinghofer V, Stewart K, McGonigal T, Smith R, Sarthy A, Nienaber V, Butler C,
Dorwin S, Richardson P, Weitzberg M, et al. (2001) Species specificity of amidine-
based urokinase inhibitors. Biochemistry 40:9125–9131.
Zhao G, Yuan C, Wind T, Huang Z, Andreasen PA, and Huang M (2007) Structural basis of
specificity of a peptidyl urokinase inhibitor, upain-1. J Struct Biol 160:1–10.
Kocsis A, Ke´kesi KA, Sza´sz R, Ve´gh BM, Balczer J, Dobo´ J, Za´vodszky P, Ga´l P, and
Pa´l G (2010) Selective inhibition of the lectin pathway of complement with phage
display selected peptides against mannose-binding lectin-associated serine pro-
tease (MASP)-1 and -2: significant contribution of MASP-1 to lectin pathway
activation. J Immunol 185:4169–4178.
Address correspondence to: Dr. Knud J. Jensen, IGM, Faculty of Life
Sciences, University of Copenhagen, Thorvaldsensvej 40, 4 sal, Copenhagen,
Denmark. E-mail: kjj@life.ku.dk