Identification of the first low-molecular-weight inhibitors of matriptase-2

Article abstract of DOI:10.1021/jm100183e

As recently discovered, matriptase-2, a type II transmembrane serine protease, plays a crucial role in body iron homeostasis by down-regulating hepcidin expression, which results in increased iron levels. Thus, matriptase-2 represents a novel target for the development of enzyme inhibitors potentially useful for the treatment of systemic iron overload (hemochromatosis). A comparative three-dimensional model of the catalytic domain of matriptase-2 was generated and utilized for structure-based virtual screening in combination with similarity searching and knowledge-based compound design. Two N-protected dipeptide amides containing a 4-amidinobenzylamide as P1 residue (compounds 1 and 3) were identified as the first small molecule inhibitors of matriptase-2 with K_i values of 170 and 460 nM, respectively. An inhibitor of the closely related protease matriptase (compound 2, K_i = 220 nM), with more than 50-fold selectivity over matriptase-2, was also identified.

Full text of DOI:10.1021/jm100183e

J. Med. Chem. 2010, 53, 5523–5535 5523
DOI: 10.1021/jm100183e
Identification of the First Low-Molecular-Weight Inhibitors of Matriptase-2
†
‡
‡
†
,§
Mihiret Tekeste Sisay,‡,§ Torsten Steinmetzer, Marit Stirnberg, Eva Maurer, Maya Hammami, Jurgen Bajorath,* and
€
,‡
€
Michael Gutschow*
‡Pharmaceutical Institute, Pharmaceutical Chemistry I, University of Bonn, An der Immenburg 4, D-53121 Bonn, Germany, §Department of Life
Science Informatics, B-IT, Limes Program Unit Chemical Biology and Medicinal Chemistry, University of Bonn, Dahlmannstrasse 2,
D-53113 Bonn, Germany, and †Institute of Pharmaceutical Chemistry, Philipps University Marburg, Marbacher Weg 6, D-35032 Marburg,
Germany
Received February 11, 2010
As recently discovered, matriptase-2, a type II transmembrane serine protease, plays a crucial role in
body iron homeostasis by down-regulating hepcidin expression, which results in increased iron levels.
Thus, matriptase-2 represents a novel target for the development of enzyme inhibitors potentially useful
for the treatment of systemic iron overload (hemochromatosis). A comparative three-dimensional
model of the catalytic domain of matriptase-2 was generated and utilized for structure-based virtual
screening in combination with similarity searching and knowledge-based compound design. Two
N-protected dipeptide amides containing a 4-amidinobenzylamide as P1 residue (compounds 1 and
3) were identified as the first small molecule inhibitors of matriptase-2 with Ki values of 170 and 460 nM,
respectively. An inhibitor of the closely related protease matriptase (compound 2, Ki = 220 nM), with
more than 50-fold selectivity over matriptase-2, was also identified.
Introduction
phylogenetic analysis of the serine protease domain and the
stem region. These include the hepsin/TMPRSS subfamily,
the human airway trypsin-like protease (HAT) differentially
expressed in squamous cell carcinoma (DESC) subfamily, the
matriptase subfamily, and corin as the representative of a
further subfamily. The members of the matriptase subfamily
represent recently identified TTSPs with a unique stem com-
position and phylogenetically related serine protease do-
mains. Known members of the matriptase subfamily are
matriptase,9,10 matriptase-2,11 matriptase-3,12 and the mosaic
polyprotease, polyserase-1,13 as well as its shorter splice-
variant termed serase-1B.14
Matriptase (membrane-type serine protease 1, MT-SP1,
suppressor of tumorigenicity 14), the most studied represen-
tative of the matriptase subfamily, was originally identified
with a novel gelatinolytic activity in human breast cancer
cells.10 By activating its potential substrates, e.g. pro-single-
chain urokinase-type plasminogen activator (pro-uPA) and the
proform of hepatocyte growth factor (HGF/scatter factor),
matriptase seems to play a relevant role in extracellular matrix
degradation and cell scattering, whereas the cleavage of pro-
filaggrin is important for epithelial development.5,7,15,16 Ma-
triptase was shown to be overexpressed in a vast array of
human tumors of epithelial origin including breast, prostate,
and ovarian cancers and has been implicated in tumor growth
and metastasis in murine models of prostate cancer.7,9,10,17,18
Therefore, selective inhibition of matriptase has therapeutic
potential for the treatment of growth and metastasis of cancer.
On the other hand, matriptase-2 (TMPRSS6, transmem-
brane serine protease 6), was first identified in 2002by Velasco
and co-workers as a novel membrane-bound mosaic serine
protease predominantly expressed in the liver.11 The extra-
cellular stem region of matriptase-2 consists of a SEA (sea
urchin sperm protein, enteropeptidase, agrin) domain-like
The vast majority of known serine proteases are either
secreted or sequestered in the cytoplasmic storage organelles
awaiting signal-regulated release. Over the last years, a struc-
turally distinctnew classofserine proteaseshasbeen identified
that are transmembrane proteins containing an extracellular
trypsin-like serine protease domain.1 These enzymes are
ideally positioned to interact with other proteins and mediate
the proteolysis of various proteins on the cell surface, of the
extracellular matrix, and on adjacent cells. Therefore, locali-
zation to the cell surface gives these proteases an excellent
opportunity to regulate signal transduction between cells and
their extracellular environment. Even though the exact patho-
physiological roles of many membrane anchored serine pro-
teases remain to be elucidated, a number of these proteases are
indicated to be involved in different stages of cancer progres-
sion including growth, invasion, migration, and metastasis,1-7
and also in several important physiological functions such as
digestion, cardiac function and blood pressure regulation,
hearing, iron metabolism, and epithelial homeostasis.2,4,6,8
The family of type II transmembrane serine proteases
(TTSPsa) possess a short intracellular N-terminal tail, a
transmembrane domain, and a large extracellular portion
containing a variable stem region and a C-terminal serine
protease catalytic domain of the chymotrypsin fold.1,2,5,8 The
TTSPs can be divided into four subfamilies based on the
*To whom correspondence should be addressed. J.B.: phone, þ49
228 2699 306; fax, þ49 228 2699 341; E-mail, bajorath@bit.uni-bonn.de.
For M.G.: phone, þ49 228 732317; fax, þ49 228 732567; E-mail,
guetschow@uni-bonn.de.
aAbbreviations: TTSPs, type II transmembrane serine proteases;
PDB, Protein Data Bank; HEK, human embryonic kidney; MOE,
Molecular Operating Environment; Tc, Tanimoto coefficient, HepG2,
hepatocellular carcinoma, human; FBS, fetal bovine serum.
r
2010 American Chemical Society
Published on Web 07/14/2010
pubs.acs.org/jmc

5524 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
Results and Discussion
Modeling of the Catalytic Domain of Matriptase-2. Be-
cause an X-ray crystal structure of human matriptase-2 is
currently not available, we attempted to model the structure
of the catalytic domain of matriptase-2. The model was
generated using crystal structures of closely related enzymes
by homology modeling, a procedure by which the coordi-
nates of atoms of a target protein are predicted based on a
topological sequence alignment of the target and a template
protein(s) of known structure.36 In a previous study, as part
of the first characterization of the enzyme, a 3D model of
matriptase-2 was built by means of a semiautomated model-
ing server.11 However, because it was intended to utilize the
model for inhibitor design, we generated a computer model
of matriptase-2 with extensive manual intervention in order
to ensure high accuracy of the model in the active site region.
We carried out a sequence search in the Protein Data Bank
(PDB)37 using the amino acid sequence of the catalytic
domain of matriptase-2 as query. As expected, the catalytic
domain of matriptase showed the highest sequence identity
(45%, PDB ID 1EAX), followed by plasma kallikrein (42%,
2ANY), coagulation factor XIa (40%, 1ZSK), the catalytic
domain of hepsin (40%, 1Z8G), plasminogen (40%, 1QRZ),
R-tryptase (39%, 2F9N), and thrombin (37%, 1ETS).
Among available X-ray crystallographic structures of ma-
triptase, the one with highest resolution (1EAX, 1.30 A)38
was taken as a primary template. In addition, structure-
based multiple sequence alignment was carried out among
matriptase-2 and its close homologues (matriptase, plasma
kallikrein, hepsin, thrombin) as well as the prototype en-
zymes factor Xa and trypsin. This procedure provided addi-
tional information about conserved and variable regions of
matriptase-2 compared to the homologous enzymes. Gaps in
the initial alignment were manually adjusted prior to model-
ing. The optimal sequence alignment of the catalytic do-
mains of matriptase and matriptase-2 (Figure 2A) extracted
from the multiple alignment reflects the high degree of simi-
larity between the two enzymes. The major difference occurs
in the so-called “60 insertion loop”, where matriptase has a
nine-residue insertion (DDRGFRYSD), three amino acids
longer than that in matriptase-2 (EDSMAS).38 Two addi-
tional single residue deletions in matriptase-2 are located
within the “37 loop” and the “70-80 loop” of matriptase.38
On the basis of the final sequence alignment, we then
generated initial 3D models with three different modeling
tools (see Experimental Section). Comparative analysis of
these initial models revealed that notable differences between
these models were limited to the loop regions and regions of
insertions and deletions described above. To model the “60
insertion loop” and regions of insertions and deletions,
alternative loops were generated and evaluated. Finally, loop
conformations were selected that most closely followed the
spatial paths of the corresponding loops in the high-resolution
X-ray structure of matriptase. Particular attention was given to
the side chain rotamers of nonconserved residues within the
active site region. These side chain conformations were adjusted
manually by selecting low energy rotamers that matched the
side chain pose of the corresponding residues in matriptase as
much as possible. After controlled energy refinement (see
Experimental Section), the final model of the matriptase-2
catalytic domain displayed good stereochemical quality (with
89% of the residues falling in the most favored regions of the
Ramachandran plot) and sequence-structure compatibility
Figure 1. Modular structure of matriptase-2. Different domains
are labeled as follows: T, transmembrane domain; SEA, sea urchin
sperm protein, enteropeptidase, agrin domain; CUB, complement
factor C1s/C1r, urchin embryonic growth factor, bone mor-
phogenetic protein 1 domain; L, low density lipoprotein receptor
class A domain. HDS indicates the catalytic triad of the catalytic
domain.
region, two CUB (complement factor C1s/C1r, urchin em-
bryonic growth factor, bone morphogenetic protein 1) do-
mains, and three repeats of LDLRA (low density lipoprotein
receptor class A) domains (Figure 1).1,2,11,19,20
In contrast to matriptase, recent studies have shown that
expression of matriptase-2 correlates with suppression of the
invasiveness and migration of breast and prostate cancer
cells.21,22 However, precise functions of matriptase-2 in cancer
remain to be further elucidated. Another interesting finding
that recently attracted much attention is the correlation
between mutations in the gene encoding matriptase-2 and
iron-refractory iron-deficiency anemia (IRIDA).23-26 Iron is
an essential trace element in mammalian metabolism, and due
to its generation of bioreactive superoxide anions and hydro-
xyl radicals, levels of plasma iron require tight regulation.26,27
Hepcidin, a small peptide hormone synthesized in the liver, is
the homeostatic regulator of plasma iron levels and iron tissue
distribution. It inhibits iron absorption from the intestine,
regulates iron recycling and release from iron stores, and
controls iron transport through the placenta. Hepcidin medi-
ates the internalization and degradation of the iron exporter
ferroportin, located on the surface of intestinal enterocytes,
macrophages, and hepatocytes, thereby inhibiting iron release
into the plasma.26 As such, control of hepcidin expression
represents
a critical checkpoint for maintaining iron
balance.28,29 Matriptase-2 suppresses hepcidin expression
through proteolytic processing of cell surface hemoju-
velin,23,25,26,30-32 a membrane-bound protein promoting hep-
cidin expression.28,33,34 Because of the involvement in such a
critical physiological process, matriptase-2 emerges as a po-
tentiallyimportant pharmaceuticaltarget. Therefore, selective
matriptase-2 inhibitors could be beneficial as pharmacologi-
cal tools to further investigate its exact role in regulating iron
homeostasis and might also be used for therapeutic interven-
tion of frequent iron disorders such as systemic iron overload
(hemochromatosis) or iron loading anemias where the level of
hepcidin is inappropriately low.
Although peptide-based matriptase-2 inhibitors, i.e., apro-
tinin and leupeptin, have previously been reported,11,35 small
molecule matriptase-2 inhibitors have not been developed so
far. Herein, we report on the generation of a model of the
catalytic domain of matriptase-2 and its application in virtual
screening and compound evaluation. The model was used to
interactively design four substrate-analogue inhibitors, two
amidino- and two chloro-substituted benzylamides. A virtual
screening campaign was carried out in the presence of these
compounds. On the basis of the screening calculations, the
designed benzamidines were assigned a high priority. We
synthesized these four substrate analogue inhibitors and in-
vestigated their inhibitory profile in in vitro assays, leading to
the identification of the first low-molecular-weight inhibitors
of matriptase-2.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5525
Figure 2. Model generation. (A) Structure-oriented sequence alignment of the catalytic domains of matriptase and matriptase-2 providing the
basis for modeling. Amino acid numbers for human matriptase-2 are given for the complete sequence of the enzyme.20 Numbering for human
matriptase is based on chymotrypsinogen numbering.38 Residues of the catalytic triad are shown in bold, and conserved residues are shaded in
gray. Secondary structure elements in the X-ray structure of matriptase are indicated with flat arrows (β-sheets) and cylinders (R-helices).38 (B)
Cartoon representation of the final model of the catalytic domain of matriptase-2 representing the chymotrypsin fold. The catalytic triad is
shown in stick presentation. The picture was generated with PyMol.66 (C) Residues in the active site region of matriptase-2 are shown in stick
representation and color-coded according to the pocket they form.
comparable to experimental structures (with a ProSA z-score of
8.13). Stereochemical and sequence-structure compatibility
quality control data for the model are provided in the Support-
ing Information (Figures S1-S3), confirming the overall good
quality of the model. However, it is important to note that the
major criterion for modeling accuracy in this case has been the
close resemblance of the matriptase and matriptase-2 active
sites. This has made it possible to apply a rather conservative
modeling strategy to generate the matriptase-2 active site
model. Because we intended to utilize the model for inhibitor
design, this has been the most important aspect of model
building and assessment.

5526 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
Model Analysis. Chymotrypsin-like serine proteases dis-
play a common fold (Figure 2B), and their distinct properties
arise from sequence differences in the substrate binding
region that modulate the shape and chemical nature of
binding pockets. The modeled active site structure of ma-
triptase-2 with the amino acid residues forming the corre-
sponding pockets is displayed in Figure 2C. At the center lies
the S1 specificity pocket. The backbone of Trp783 (chy-
motrypsinogen numbering 215), Gly784(216), Gln759(192),
and the Cys758(191)-Cys787(220) disulfide bridge form the
entrance of the pocket, Gly794(226), Tyr796(228), and
Val781(213) form the inner wall, and a critical, specificity
determining residue, Asp756(189), is located at the bottom of
the S1 pocket. Matriptase-2 has an alanine at the subsequent
position 757(190), similar to plasma kallikrein, factor Xa and
thrombin, whereas matriptase, trypsin, and uPA have a
serine residue at this position. This makes the bottom of
the S1 pocket of matriptase-2 slightly more hydrophobic
compared to matriptase. Binding of arginine to the S1 site is
stabilized by a direct salt bridge between its guanidinium
group and the carboxylate group of Asp756(189). However,
it is also known that binding of a P1-lysine residue involves
the formation of a hydrogen bond to the hydroxyl group of
serine and, therefore, less preferred for proteases with ala-
nine at the bottom of the S1 pocket.39 Accordingly, arginine
at the P1 position of peptidic substrates is preferred by
matriptase-2.11,35 The S2 binding site is bordered by His665-
(99) and His617(57) located above Asp668(102) and Ser782-
(214). His665(99) separates the small S2 site from the large
S3/S4 pocket. This pocket lies to the left of the S1 site, on top
of the Trp783(215) indol. The lower part of the S3/S4 extends
to Leu785(217), making this spot more hydrophobic com-
pared to matriptase, which has an aspartic acid instead of
leucine at the corresponding position. The upper part of the
S3/S4 pocket is formed by the backbone carbonyl oxygens of
Glu662(96), Asp663(97), and Ser664(98) (corresponding to
Asp, Phe, and Thr in matriptase, respectively), representing a
negatively charged environment. This part accepts a posi-
tively charged residue as has been shown in a substrate
mapping study.35 Different from matriptase, where the
benzyl group of Phe97 creates a solvent shield, these residues
are solvent exposed. The S10 pocket is located just above the
disulfide bridge Cys602(42)-Cys618(58) between the cata-
lytic His617(57) and the side chain of Ile601(41), whereas the
S30 site is adjacent to the S10 pocket and bordered by
Met624(60D). The S20 pocket lies in a cleft between Gln759-
(192) and Ile716(151). This pocket is smaller than that of
matriptase due to the substitution of a glycine by isoleucine
at position 716.
substitute for benzamidine.41-43 We considered introducing
the chlorine at this ring system in order to build in a favorable
lipophilic contact with Tyr796(228) inside the S1 pocket.
Having designed the S1 anchor points, we next focused on
the small S2 pocket in matriptase-2. A proline could be well
placed into this small pocket, which also provided a linker to
S3/S4 site. A prominent feature of the upper part of the S3/S4
site in matriptase-2 is the arrangement of backbone carbonyl
groups of residues Glu662(96), Asp663(97), and Ser664(98),
as discussed above. To complement the negatively charged
environment, an arginine residue could be well utilized, reach-
ing into the upper part of the pocket. However, D-configuration
of this arginine residue was essential to direct the side chain into
the upper part of the S3/S4 pocket. As a control, we also
designed test compounds with a D-cyclohexylalanine instead of
the D-arginine at this position. This hydrophobic moiety is
expected not to reach into the charged upper region of the
pocket, but to interact with the lower hydrophobic part of the
S3/S4 pocket. A common N-terminal benzylsulfonamide group
was chosen for our manually designed inhibitors as has been
incorporated in series of potent substrate analogue inhibitors of
related serine proteases.43-45
The design effort resulted in four N-protected dipeptide
amides as candidate compounds that are shown in Figure 3B.
These four compound suggestions originating from “know-
ledge-based” design were included into a virtual screen for
matriptase-2 inhibitors. A set of 15000 ZINC compounds
was preselected on the basis of remote chemical and approxi-
mate shape similarity to known matriptase inhibitors (see
Experimental Section) and subjected to the docking protocol
detailed in the Experimental Section. Importantly, test com-
pounds were flexibly docked, and no assumptions about
putative compound binding modes were made to restrict the
calculations. Compounds 1 and 3 were among the structures
we prioritized based on a detailed analysis of the docking
results and their relative ranking position (see Supporting
Information). We decided to synthesize and test the four
manually designed compounds. Future investigations will be
directed toward the evaluation of other highly ranked struc-
tures as potential inhibitors of matriptase-2 (see Supporting
Information).
Synthesis. The convergent synthesis of compound 1 is
shown in Scheme 1. The Boc-protected O-acetyl hydroxy-
amidine 5 was used as a precursor for the benzamidine moiety
of 1 (and also 3). Reductive cleavage followed by deprotec-
tion of the Boc group provided 4-amidinobenzylamine dihy-
drochloride (6). Nω-Nitro-D-arginine was N-protected with
benzyl sulfonylchloride, and the resulting derivative 7 was
coupled with proline methyl ester. Saponification and sub-
sequent catalytic hydrogenation to remove the nitro group
afforded the dipeptide 8. The final compound 1 was then
obtained by PyBOP-mediated amide coupling. To prepare
the dipeptide amide 2 (Scheme 2), the benzylamine derivative
Inhibitor Design and Virtual Screening Analysis. With the
aid of the modeled active site of matriptase-2, shown in
Figure 2C, we manually designed four substrate analogue
inhibitors. The structures of known potent inhibitors of
matriptase,40 the X-ray structure of a matriptase/inhibitor
complex (shown in Figure 3A), and comparison of active site
features, as discussed above, were taken into account to
support these modeling approach. Specifically, we first
focused our design efforts on the deep S1 pocket where the
critical negatively charged Asp756(189) residue must interact
with an inhibitor in a complementary fashion. Therefore, we
first placed a positively charged benzamidine group into this
pocket and, as an alternative, a benzene ring with a 2-amino-
methyl-5-chloro substitution pattern. This moiety has pre-
viously been observed in thrombin inhibitors as a suitable
9
46 was reacted in a similar manner with 8. Deprotection of
the Boc group was accomplished with TFA to produce 2.
D-Cyclohexylalanine was converted to the sulfonamide 10
(Scheme 3), and the following reaction steps yielded the
dipeptide 11 and the final dipeptide amide 3. The preparation
of compound 4 (Scheme 4) was essentially the same as for 2
but started with D-cyclohexylalanine.
Expression of Human Matriptase-2. To study the inhibi-
tory effect of compounds 1-4 on human recombinant
matriptase-2, the whole matriptase-2 construct was cloned
and expressed in human embryonic kidney (HEK) cells with

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5527
Figure 3. Compound design. (A) shows a crystallographic matriptase inhibitor complex (PDB ID 2GV6) that was used to deduce a likely
orientation of putative matriptase-2 inhibitors in the binding site of the homology model. Specificity pockets are labeled. (B) shows the four
designed candidate compounds (1-4).
a Myc-tag at the C-terminal end of the protein. As shown in
Figure 4A, an approximately 30 kDa C-terminal fragment of
matriptase-2 was detectable using anti-c-Myc antibody in
the conditioned medium of transfected HEK cells (HEK-
MT2). This form represents the catalytic domain released
from the cell surface after processing of the zymogen, as
described previously.32 No signal was detectable in HEK
cells expressing the empty vector (HEK-mock). The catalytic
domain of matriptase-2 was purified from the conditioned
medium of HEK-MT2 cells by immunoaffinity chromatog-
raphy. After isolation, the purity of matriptase-2 was
checked on SDS-Page (Figure 4B). One single band of
approximately 30 kDa was visible after purification repre-
senting the catalytic domain of matriptase-2.
Enzyme Inhibition Assays. The activity profile of matrip-
tase-2 in the conditioned medium and of the purified cata-
lytic domain of matriptase-2 was characterized using the
chromogenic substrate Boc-Gln-Ala-Arg-para-nitroanilide.
We obtained similar Km and Ki values (Table 1) from the
experiments with either conditioned medium or purified
enzyme. Moreover, no detectable activity was observed in
the conditioned medium of nontransfected HEK cells or
HEK cells expressing the empty vector (HEK-mock). Thus,
it is possible to use the conditioned medium as a source for
matriptase-2 activity. The Km values for matriptase-2 in the
conditioned medium (210 μM) and for purified matriptase-2
(159 μM) in the micromolar range were similar to human
recombinant matriptase (Table 1) and to the reported value
(257 μM)47 for the purified catalytic domain of murine
matriptase overexpressed in insect cells measured with the
corresponding fluorogenic substrate.
In control experiments, the inhibition of matriptase-2
activity in the conditioned medium and of purified matrip-
tase-2 by aprotinin and leupeptin was measured (Table 1).
Similar inhibition of both enzymatic activities was obtained,
further confirming that indeed the released matriptase-2 in
the conditioned medium accounts for the cleavage of the
chromogenic substrate. Aprotinin was more active than
leupeptin to inhibit matriptase-2 by at least 2 orders of
magnitude, as it has already been found before.11 Human
recombinant matriptase, included in these experiments, was
also inhibited by aprotinin in the low nanomolar range. It
has been reported that aprotinin, when used at a concentra-
tion of 300 nM, led to a nearly complete inhibition of both
matriptase and matriptase-2.35 In accordance to this pre-
vious report,35 a somewhat stronger inhibition of matriptase

5528 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
Scheme 1a
a Reagents and conditions: (a) (1) AcOH, H2, Pd/C; (2) HCl; (b) (1) H-Pro-OMe ꢀ HCl, DIPEA, DMF, PyBOP, (2) MeOH, 1 M NaOH (1:1), (3)
AcOH, H2, Pd/C; (c) DIPEA, DMF, PyBOP.
Scheme 2a
Scheme 4a
a Reagents and conditions: (a) (1) compound 9, DIPEA, DMF,
PyBOP, (2) TFA.
a Reagents and conditions: (a) (1) DIPEA, DMF, PyBOP, (2) TFA.
The inhibitory potency of the dipeptide amides 1-4
(Figure 3B) toward matriptase-2 was determined and com-
pared with their inhibitory activity against the catalytic
domain of the structurally related matriptase (Table 1).
Compound 1 containing a D-arginine and a benzamidine
moiety was the most potent inhibitor for both enzymes with a
3-fold higher potency for matriptase (Ki=55 nM) than for
matriptase-2 in the conditioned medium (Ki = 190 nM) and
purified matriptase-2 (Ki = 170 nM). Dipeptides with a
4-amidinobenzylamide group, such as 1, are known potent
inhibitors of thrombin and factor Xa.44,48,49 The D-arginine
and D-cyclohexylalanine derivatives 2 and 4, both lacking the
benzamidine moiety, had only marginal inhibitory activity
against matriptase-2 (Ki > 10 μM). Among the four dipeptide
amides, only 3 exhibited a higher potency toward matrip-
tase-2 in the conditioned medium (Ki = 290 nM) and the
purified matriptase-2 (Ki = 460 nM) than against matriptase
(Ki = 770 nM). As expected for hits arising from a virtual
screening approach without selectivity constraints, the inhibi-
tory activity of 1-4 is not limited to matriptases. The related
serine proteases thrombin and trypsin were also inhibited, and
the data are given in the Supporting Information (Table S2).
Scheme 3a
a Reagents and conditions: (a) (1) H-Pro-OMe ꢀ HCl, DIPEA,
DMF, HBTU, (2) MeOH, 1 M NaOH (1:1); (b) compound 6, DIPEA,
DMF, PyBOP.
than matriptase-2 by leupeptin (1.9 μM versus 4.1 μM) was
observed.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5529
Figure 4. Expression and purification of human matriptase-2. (A) Conditioned medium of HEK cells transfected with an expressing vector
harboring matriptase-2 cDNA (HEK-MT2) and without matriptase-2 cDNA (HEK-mock) was characterized by Western blot analysis using
anti-c-Myc antibody. (B) The catalytic domain of matriptase-2 was isolated by affinity chromatography using immobilized anti-c-Myc
antibody. Lane 1: flow-through; lanes 2-4: wash fractions; lanes 5-7: elution fractions; M: molecular mass markers.
Table 1. Kinetic Parameters for Inhibition of Human Matriptase-2 and Human Matriptase by the Compounds 1-4
Ki ( SEM (μM)a
human matriptase-2
compd
in conditioned medium of HEK-MT2 cells
purified enzyme
0.17 ( 0.02b
>10c
recombinant human matriptase
1
2
3
4
0.19 ( 0.01
>10c
0.055 ( 0.003
0.22 ( 0.01
0.77 ( 0.15
2.1 ( 0.3
0.0098 ( 0.002
1.9 ( 0.4
0.29 ( 0.02
>10c
0.46 ( 0.06
>10c
aprotinind
leupeptind
0.0050 ( 0.0005
2.4 ( 0.2
0.013 ( 0.002
4.1 ( 0.7
a IC50 values were determined from duplicate measurements with at least five different inhibitor concentrations. Ki values were calculated using the
equation Ki = IC50/(1 þ [S]/Km). Km values obtained for Boc-Gln-Ala-Arg-para-nitroanilide were 210 ( 7 μM for matriptase-2 in conditioned medium
of HEK-MT2 cells, 159 ( 21 μM for purified matriptase-2, and 381 ( 33 μM for recombinant matriptase. b Triplicate measurement with five different
inhibitor concentrations. c Duplicate measurement with three different inhibitor concentrations. d IC50 values. Reactions were followed over 20 min to
approach steady-state. Steady-state velocities were plotted against inhibitor concentrations to obtain IC50 values.
Structure-Activity Relationships. The highly prioritized
compounds 1 and 3 were found to be inhibitors of matrip-
tase-2 but not compounds 2 and 4. Figure 5A shows the final
docking pose of the active compound 1. Because of accuracy
limitations, it is generally not possible to predict detailed
intermolecular interactions from docked poses, and we do
not attempt to do so. However, likely structural character-
istics of matriptase-2 inhibition can be explored on the basis
of the model. Most probably, the S1 specificity pocket
accommodates the benzamidine moiety. Then the amidine
group is in contact distance to the side chain of Asp756(189)
at the bottom of the S1 pocket. Furthermore, in the docked
pose, the guanidino group of arginine occupies the upper
part of the S3/S4 pocket and is in contact distance to the
carbonyl groups of Glu662(96), Asp663(97), and Ser664(98).
In addition, the proline binds to the S2 pocket. The inhibitor
forms a short antiparallel β-sheet to the backbone of Ser782-
(214) and Gly784(216) as has been experimentally shown in
the binding of similar inhibitors to related enzymes.44,45 The
position of the benzylsulfonamide moiety of the inhibitor
could not be deduced with a high level of confidence. It
resides either at the lower part of the S3/S4 pocket or a
shallow hydrophobic subsite behind the S1 binding site,
packing against the Cys758(191)-Cys787(220) disulfide
bridge. The latter orientation would be supported by the
X-ray crystal structure of the complex of factor Xa with
a structurally related benzamidine inhibitor where the ben-
zylsulfonamide moiety was located in a corresponding
subsite.44 The binding of compound 3 likely resembles that
of compound 1 except that the cyclohexyl group would be
positioned within the S3/S4 pocket for favorable interactions
with Trp783(215) and Leu785(217) (see Supporting Infor-
mation, Figure S4).
Compound 1 is also a potent inhibitor of matriptase.
When modeled into the binding site of matriptase, it became
apparent that the accommodation of the benzylsulfonamide
moiety would represent the major difference. In matriptase,
this group is oriented toward the lower part of the S3/S4
pocket but could not extend into the shallow hydrophobic
subsite behind the S1 pocket because matriptase has a tyro-
sine residue at position 146 whose side chain would block
access of an inhibitor to this subsite (see Supporting Infor-
mation, Figure S5). The guanidino moiety of 1 extends
toward the upper part of the S3/S4 binding pocket forming
hydrogen bonding to the Phe97 carbonyl oxygen and
cation-π interaction with Phe99 and Trp215, as has been
similarly described for other inhibitors of matriptase.50
The replacement of D-arginine in 1 by D-cyclohexylalanine
in 3 resulted in a stronger decrease in potency against
matriptase compared to matriptase-2. This finding might
be rationalized by the reduced hydrophobic character of the
lower part of the S3/S4 pocket in matriptase compared to
matriptase-2. In matriptase-2, the cyclohexyl ring of 3 might
favorably interact with Leu-785, but in matriptase, this
interaction would be absent and the cyclohexyl ring extends
toward the upper part of the S3/S4 pocket (see Supporting

5530 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
rationalized in structural terms. Additional inhibitors with
structural variations at the P1 position will be required to better
understand selectivity determinants in this case. From the
X-ray structure of a related N-(3-chlorobenzyl)prolinamide
inhibitor in complex with thrombin (PDB ID: 2ZC9),42 it can
be inferred that compound 2 might bind similarly to matriptase
as the benzamidine-based inhibitor 1. However, while the
amidine moiety of 1 would interact with Asp189 of matriptase,
the chlorobenzylamine group of 2 would be buried inside the S1
pocket and the chloro atom would point toward the aromatic
ring of Tyr228. Figure 5B shows a model representing this
putative binding mode.
It is also important to note that the docking strategy we
applied was not per se validated by reproducing the binding
mode of matriptase inhibitors. Rather, the experimental
confirmation of inhibitory activity of two of the compounds
that we designed and that were highly ranked in our docking
calculations supported the approach. On this basis, we also
proposed the putative approximate binding mode of these
inhibitors, consistent with our design strategy, but also being
fully aware of the intrinsic limitations of such predictions,
even if they are experimentally supported.
Conclusion
In conclusion, we have identified the first low-molecular-
weight inhibitors of matriptase-2. Future investigations will
be directed toward structural modifications and the evalua-
tion of their (cyto)toxicity profile. The most potent matrip-
tase-2 inhibitor of this study, compound 1, was, however, not
selective for matriptase-2 over matriptase. Substitution of the
guanidine moiety by cyclohexyl slightly reduced the potency
against matriptase-2 (1 versus 3). This effect, however, is more
pronounced in the case of matriptase than matriptase-2,
which might be due to the difference in hydrophobic character
in the lower part of the S3/S4 pocket and the loss of the
cation-π interaction in matriptase. Replacement of the benz-
amidine by the chlorobenzylamine moiety (1 and 3 versus 2
and 4) resulted in a complete loss of inhibitory activity toward
matriptase-2. Compound 2 was identified as an inhibitor with
strong preference for matriptase over matriptase-2. Such
selectivity might be crucial in the development of matriptase
inhibitors as anticancer agents because simultaneous inhibi-
tion of matriptase-2 is likely to have undesirable effects on
body iron metabolism. However, the N-protected dipeptide
amides 1 and 3 described herein will be used as leads to
develop inhibitors with selectivity toward matriptase-2. The
benzamidine substructure present in 1 and 3 is an established
pharmacophore in drug development.45,51,52 Therefore, to
improve potency and achieve selectivity toward matriptase-2,
future structural modifications might be made while keeping
the P1 benzamidine in the molecules. On the other hand,
further exploring the P1 site in these compounds is thought to
lead to a better understanding of selectivity determinants in
matriptases. The selective inhibition of matriptase-2 may
serve as a new potential strategy in the treatment of primary
hemochromatosis and iron loading anemias.
Figure 5. Modeled complexes. (A) Model of the matriptase-
2/compound 1 complex. (B) Model of the matriptase/compound 2
complex. Shown are manually adjusted docking poses of these
compounds. Specificity pockets are labeled.
Information, Figure S6). In addition, in the case of matrip-
tase, the loss of the cation-π interaction contributes to the
decreased activity (1 versus 3).
In compounds 2 and 4, the benzamidine moiety is replaced
by a para-chlorobenzylamine group. These compounds were
inactive against matriptase-2. However, both compounds, in
particular 2, inhibited matriptase. Previously reported X-ray
crystal structures of structurally similar inhibitors bound to
thrombin41-43 showed that a chlorobenzene group occupied
the S1 pocket with the chloro atom forming a van der Waals
contact with Tyr228, which is conserved in matriptase and
matriptase-2. Rittle et al. investigated thrombin inhibitors
with a chlorobenzene group and introduced an aminomethyl
residue in para-position to the chlorine, as present in 2 and 4,
which further improved the inhibitory activity. In the X-ray
structure, the amino group forms a salt bridge with Glu-192
and a hydrogen bond to Gly-216 at the entrance of the S1
pocket of thrombin.43 The major difference between matrip-
tase and matriptase-2 inside the S1 pocket is the presence of
serine at position 190 in matriptase instead of alanine in
matriptase-2. Assuming a binding mode of compounds 2 and
4 in matriptase similar to the one seen in thrombin,43 which also
has alanine at position 190, we would expect the benzamidine-
chlorobenzylamine replacement to be tolerated by matriptase-2
rather than matriptase. However, compound 2was identified as
a selective inhibitor of matriptase, which can currently not be
Experimental Section
General Synthetic Methods and Materials. All reagents in-
cluding amino acids, amino acid derivatives, coupling reagents,
and reagents for synthesis were obtained from Novabiochem,
Iris, Fluka, Aldrich, or Acros. 4-Amidinobenzylamine dihy-
drochloride (6) was prepared as described.53 Analytical HPLC
experiments were performed on a Shimadzu LC-10A system

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5531
(column: Nucleodur C18, 5 μm, 100 A, 4.6 mm ꢀ 250 mm,
were dissolved in DMF (2 mL) at 0 °C and treated with PyBOP
(104 mg, 0.20 mmol). The mixture was stirred at 0 °C for 15 min
and 4 h at room temperature. The solvent was removed in vacuo,
and the product was purified by preparative HPLC. Product-
containing fractions were combined, evaporated, and lyophilized
to give compound 1 (95 mg, 60%) as a white powder; mp 161-
163 °C. HPLC: 16.8 min. MS calcd 556.25, found 557.2 (M þ H)þ.
13CNMR(D2O) δ24.14, 24.28 (Cγ-Arg, Cγ-Pro), 28.01, 29.39 (Cβ-
Arg, Cβ-Pro), 40.36, 42.80 (Cδ-Arg, NH-CH2-Carom), 47.73 (Cδ-
Pro), 54.95 (CR-Arg), 59.17, 61.17 (CR-Pro, CH2SO2), 126.21,
127.66, 127.87, 128.43, 128.85, 129.04, 130.83, 144.57 (Carom),
156.78 (NH-CdNH), 165.98 (Carom-CdNH), 172.13, 174.39
(CdO); ca. 116 (F3C), the TFA CO signal was not detected.
Benzylsulfonyl-D-arginyl-proline-(2-aminomethyl-5-chlorobenzyl)-
€
Machery-Nagel, Duren, Germany) with a linear gradient of
acetonitrile (10-80% in 70 min, detection at 220 nm) containing
0.1% TFA at a flow rate of 1 mL/min. The final inhibitors were
purified to more than 95% purity (detection at 220 nm) by
preparative HPLC (pumps Varian PrepStar Model 218) using
a C8 column (Nucleodur, 5 μm, 100 A, 32 mm ꢀ 250 mm,
€
Macherey-Nagel, Duren, Germany) and a linear gradient of
acetonitrile containing 0.1% TFA at a flow rate of 20 mL/min.
All inhibitors were finally obtained as TFA salts after lyophili-
zation. The molecular mass of the synthesized compounds was
determined using a QTrap 2000 ESI spectrometer (Applied
Biosystems) or an AutoSpek spectrometer (Micromass). The
1H and 13C NMR spectra were recorded on a ECX-400 (Jeol
Inc., USA) at 400 and 100 MHz, respectively, and are referenced
to internal solvent signals. All tested compounds possessed a
purity of at least 95%.
Benzylsulfonyl-Nω-nitro-D-arginine (7). Nω-Nitro-D-arginine
(4.0 g, 18.25 mmol) was suspended in H2O (250 mL) and DIPEA
(2.36 g, 3.18 mL, 18.25 mmol). The suspension was treated with
3 equiv of benzylsulfonyl chloride (10.44 g, 54.76 mmol) solved
in acetonitrile (50 mL) and 6 equiv of DIPEA (14.15 g, 19.05 mL,
109.5 mmol) in several portions at 0 °C over a period of 1 h and
was stirred additional 12 h at room temperature. The acetoni-
trile was removed in vacuo, and the basic aqueous phase was
extracted twice with ethyl acetate. The product containing the
aqueous phase was acidified by addition of 5% KHSO4 to pH
2-3, partially evaporated, and extracted 3ꢀ with ethyl acetate.
The organic phase was washed 3ꢀ with brine and dried over
Na2SO4. During evaporation of the solvent, the product starts
to crystallize, and compound 7 was obtained (1.9 g, 28%) as
white crystals; mp 165-167 °C. MS calcd 373.11, found 374.2
(M þ H)þ. HPLC: 20.25 min. 13C NMR (DMSO-d6) δ 24.67
(Cγ-NO2-Arg), 29.22 (Cβ-NO2-Arg), ca. 40.1 (Cδ-NO2-Arg),
55.45 (CR-NO2-Arg), 58.56 (CH2SO2), 127.97, 128.23, 130.23,
130.88 (Carom), 159.30 (CdNH), 173.53 (CdO).
amide Bis(trifluoroacetate) (2). Bzls-D-Arg-Pro-OH TFA (8) (800
3
mg, 1.48 mmol), 2-(tert-butoxycarbonylaminomethyl)-5-chloro-
benzylamine (9)46 (400 mg, 1.48 mmol), and 2 equiv of DIPEA
(383 mg, 516 μL, 2.96 mmol) were solved in DMF (15 mL) and
treated with PyBOP (770 mg, 1.48 mmol) at 0 °C. The mixture was
stirred 15 min at 0 °C and then overnight at room temperature. The
solvent was evaporated in vacuo, and the remaining residue (HPLC:
38.81 min) was dissolved in TFA (10 mL). The mixture was stirred at
room temperature for 1.5 h, the solvent was removed in vacuo, and
the product was purified by preparative HPLC. The product-
containing fractions were combined, evaporated, and lyophilized
to yield compound 2 (0.64 g, 0.79 mmol) as a white powder; mp
123-125 °C. HPLC: 22.02 min. MS calcd 577.22, found 578.1 (M þ
H)þ. 13C NMR (D2O) δ 24.17, 24.21 (Cγ-Arg, Cγ-Pro), 28.00, 29.32
(Cβ-Arg, Cβ-Pro), 39.23, 40.01, 40.40 (Cδ-Arg, NH-CH2-Carom
,
NH2-CH2), 47.71 (Cδ-Pro), 54.93 (CR-Arg,), 59.16, 61.01 (CR-Pro,
CH2SO2), 128.30, 128.47, 128.74, 128.90, 128.95, 129.10, 130.86,
131.25, 134.94, 138.16 (Carom), 156.75 (CdNH), 172.42, 174.07
1
2
(CdO), 116.40 (q, J = 292 Hz, F3C), 163.00 (q, J = 36 Hz,
F3CCO2H).
Benzylsulfonyl-D-cyclohexylalanine (10). D-Cyclohexylala-
nine (2.0 g; 11.68 mmol) was suspended in acetonitrile (90
mL), H2O (60 mL), and DIPEA (2.03 mL, 1.51 g, 11.68 mmol).
The suspension was treated with benzylsulfonyl chloride (6.68 g,
35.05 mmol) and DIPEA (10.17 mL, 7.55 g, 58.4 mmol) in small
portions at 0 °C over a period of 1 h and stirred additional 4 h at
room temperature. The solvent was removed in vacuo, and the
residue was dissolved in a mixture of ethyl acetate and 5%
KHSO4. The organic phase was washed 3ꢀ with 5% KHSO4
and brine, dried over Na2SO4, and evaporated to give 10 (3.1 g,
82%) as a white solid; mp 110-112 °C. MS calcd 325.13, found
348.1 (M þ Na)þ. HPLC: 45.62 min. 13C NMR (CDCl3) δ 25.85,
26.00, 26.26, 32.26, 33.28, 33.47 (Ccyclohexyl), 40.84 (Cβ-Cha),
54.10 (CR-Cha), 60.15 (CH2SO2), 128.64, 128.79, 128.88, 130.77
(Carom), 177.93 (CdO).
Benzylsulfonyl-D-argininyl-proline Trifluoroacetate (8). Bzls-
D-Arg(NO2)-OH (1.0 g (2.68 mmol), 1.15 equiv of H-Pro-
OMe HCl (0.51 g, 3.08 mmol), and 2.5 equiv of DIPEA (0.87
3
g, 1.17 mL, 6.7 mmol) were dissolved in DMF (15 mL) at 0 °C
and treated with 1.05 equiv of PyBOP (1.46 g, 2.81 mmol). The
mixture was stirred at 0 °C for 15 min and then at room
temperature overnight. The solvent was removed in vacuo,
and the remaining oil was dissolved in ethyl acetate and washed
3ꢀ with 5% KHSO4, saturated NaHCO3 solution, and brine,
respectively. The organic layer was dried over Na2SO4 and
evaporated. The remaining residue (1.3 g crude brown oil,
HPLC: 29.05 min) was dissolved in methanol (8 mL) and 1 M
NaOH (8 mL). The mixture was stirred for 3 h at room
temperature and diluted with ethyl acetate and 5% KHSO4.
The organic phase was washed 3ꢀ with brine, dried over
Na2SO4, and evaporated. The remaining oil (0.982 g, HPLC
22.67 min, contains approximately 7% of Bzls-D-Arg(NO2)-OH
as impurity) was dissolved in 90% acetic acid (30 mL) and
hydrogenated in presence of 10% Pd/C (100 mg) at 35 °C
overnight. The catalyst was removed by filtration, the solvent
was evaporated in vacuo, and the product was purified by
preparative HPLC. The product-containing fractions were
combined, evaporated, and lyophilized to obtain compound 8
(0.55 g, 38%) as a white TFA salt; mp < 90 °C (dec). MS calcd
425.17, found 426.2 (M þ H)þ. HPLC: 18.27 min. 13C NMR
(DMSO-d6) δ 24.75, 25.14 (Cγ-Arg, Cγ-Pro), 29.11, 29.29 (Cβ-
Arg, Cβ-Pro), 40.74 (Cδ-Arg), 47.02 (Cδ-Pro), 54.47 (CR-Arg),
59.01, 59.20 (CR-Pro, CH2SO2), 128.24, 128.61, 130.64, 131.39
(Carom), 157.23 (CdNH), 170.26, 173.78 (CdO), 116.98 (q, 1J=
296 Hz, F3C), 159.11 (q, 2J = 33 Hz, F3CCO2H).
Benzylsulfonyl-D-cyclohexylalanyl-proline (11). Bzls-d-Cha-
OH (10) (1.34 g, 4.1 mmol), 1.1 equiv of H-Pro-OMe HCl
3
(0.75 g, 4.5 mmol), and 2.5 equiv of DIPEA (1.324 g, 1.78 mL,
10.25 mmol) were solved in DMF (20 mL) and treated with
HBTU (1.55 g, 4.1 mmol) at 0 °C. The mixture was stirred at 0 °C
for 30 min and at room temperature overnight. The solvent was
removed in vacuo, and the remaining oil was dissolved in ethyl
acetate and washed 3ꢀ with 5% KHSO4, saturated NaHCO3
solution, and brine, respectively. The organic phase was dried
over Na2SO4 and evaporated. The remaining ester (2.2 g of
amorphous orange solid, HPLC: 53.41 min) was dissolved in
methanol (15 mL) and 1 M NaOH (15 mL). The mixture was
stirred for 4 h at room temperature and diluted with ethyl
acetate and 5% KHSO4. The organic phase was washed 1ꢀ
with 5% KHSO4 and 3ꢀ with brine, dried over Na2SO4, and
evaporated to give compound 11 (1.3 g, 75%) as a white solid; mp
172-176 °C. MS calcd 422.19, found 423.2 (M þ H)þ. HPLC:
45.95 min. 13C NMR (CDCl3) δ 24.55, 25.79, 26.06, 26.34, 28.28,
32.13, 33.05, 33.99 (Ccyclohexyl, Cβ-Pro, Cγ-Pro), 41.29 (Cβ-Cha),
47.05 (Cδ-Pro), 53.01 (CR-Cha), 59.68, 59.78 (CR-Pro, CH2SO2),
128.39, 128.69, 129.17, 130.93 (Carom), 172.73, 173.66 (CdO).
Benzylsulfonyl-D-argininyl-proline-(4-amidinobenzyl)amide Bis-
(trifluoroacetate) (1). Bzls-D-Arg-Pro-OH TFA (8) (108 mg, 0.20
3
mmol), 1.1 equiv of 4-amidinobenzylamine 2HCl (6)53 (49 mg,
3
0.22 mmol), and 2.5 equiv of DIPEA (64 mg, 87 μL, 0.50 mmol)

5532 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
Benzylsulfonyl-D-cyclohexylalanyl-proline-(4-amidinobenzyl)-
amide Trifluoroacetate (3). First, 250 mg (0.592 mmol) of Bzls-
D-Cha-Pro-OH (11) (250 mg, 0.59 mmol), 1.1 equiv of 4-amidi-
compounds and using broadened “rule of five” screen (mole-
cular weight: 200-600; logP: -2-6; donors: 1-10; acceptors:
1-10; rotatable bonds: 0-18), reducing its size to 3684443
unique compounds. Then, eight known inhibitors of matriptase
(compounds 2, 8, 18, 20, 29, 31, 56, and 59 from ref 40) were
taken as a reference set for nearest neighbor and centroid
similarity searching using MACCS structural keys61 as a finger-
print and the Tanimoto coefficient (Tc)62 as the similarity
measure. Compounds falling into the Tc interval 0.6-0.8 were
preselected in order to retain compounds with some structural
resemblance to matriptase inhibitors but omit analogues or
other notably similar compounds; a total of 206644 compounds
from the nearest neighbor search and 203039 from centroid
search was obtained. These preselections were reranked using
the Molprint2D63 fingerprint, which is a higher-resolution
similarity search tool than MACCS keys, and the top 50000
compounds from each list were taken. Considering compounds
occurring in both lists, a total of 67346 unique compounds
remained. These database compounds were ranked on the basis
of an approximate shape matching procedure using MOE
relative to the known matriptase inhibitors, and the top 15000
compounds were then considered for docking and the four
manually designed inhibitors were added to this compound
pool.
The homology model of matriptase-2 and the matriptase
X-ray crystal structure with PDB ID 1EAX38 were used for
docking. As docking programs, DOCK664 and FlexX65 were
applied for flexible ligand docking. For FlexX, the active site
was prepared by taking all residues into account within 6.5 A
around a crystallographic inhibitor of matriptase (PDB ID
2GV6)40 after superposition of the homology models and the
X-ray structure. For DOCK6, a 10 A radius was applied to
generate and select spheres representing the active site. In both
cases, docking parameters were adjusted by redocking the
inhibitor of matriptase and reproducing its crystallographic
pose.
Initial shape- and energy-based docking calculations were
performed with DOCK6, and the final docking and ranking was
performed with the FlexX.65 A total of 2167 compounds pro-
duced a DOCK6 energy score of less than -20 kcal/mol. These
compounds were redocked using FlexX. Finally, 10 poses of the
30 top-scoring database compounds (see Supporting Infor-
mation) were visually inspected and 13 compounds were selected
as candidates. The selection was mainly based on (i) active site
shape/chemical complementarity, (ii) occupation of the S1
specificity pocket by a basic group, (iii) omission of compounds
with solvent-exposed hydrophobic groups, and (iv) assessment
of conformational strains.
Cell Culture. Human embryonic kidney (HEK) 293 cells,
which do not express endogenous matriptase-2 and HepG2
(hepatocellular carcinoma, human) cells expressing matrip-
tase-2, were cultured in Dulbecco’s Modified Eagle’s medium
(DMEM), supplemented with penicillin (100 U/mL), strepto-
mycin (100 μg/mL), glutamine (0.2 M), and fetal bovine serum
(10%; FBS; all substances from Invitrogen, Karlsruhe,
Germany) under a humidified atmosphere of 5% CO2.
nobenzylamine 2HCl (6)53 (144 mg, 0.65 mmol), and 3 equiv
3
of DIPEA (309 μL, 229 mg, 1.77 mmol) were solved in DMF
(10 mL) and treated with PyBOP (307 mg, 0.59 mmol) at 0 °C.
The mixture was stirred 15 min at 0 °C and then 4 h at room
temperature. The solvent was evaporated in vacuo, and the
product was purified by preparative HPLC. The product-con-
taining fractions were combined, evaporated, and lyophilized to
yield compound 3 (0.31 g, 77%) as a white TFA salt; mp
139-141 °C. HPLC: 36.67 min. MS calcd 553.27, found 554.2
(M þ H)þ. 13C NMR (DMSO-d6) δ (mixture of conformers,
major signals) 23.63, 24.37, 25.55, 25.86, 25.94, 26.04, 29.42,
31.32, 32.99, 33.39 (Ccyclohexyl, Cβ-Pro, Cγ-Pro), 41.67, 44.86,
45.83, 45.87, 46.82 (Cβ-Cha, Cδ-Pro, NH-CH2-Carom), 52.06
(CR-Cha), 58.39, 59.95 (CR-Pro, CH2SO2), 126.23, 127.19,
127.67, 127.97, 128.01, 130.40, 130.93, 145.85 (Carom), 165.20
(CdNH), 170.68, 171.94 (CdO), 116.65 (q, 1J = 297 Hz, F3C),
158.53 (q, 2J = 33 Hz, F3CCO2H).
Benzylsulfonyl-D-cyclohexylalanyl-proline-(2-aminomethyl-5-
chlorobenzyl)amide Trifluoroacetate (4). Bzls-D-Cha-Pro-OH
(11) (200 mg, 0.473 mmol), 2-(tert-butoxycarbonylamino-
methyl)-5-chlorobenzylamine (9)46 (128 mg, 0.473 mmol), and
2.5 equiv (206 μL, 153 mg, 1.18 mmol) of DIPEA were solved in
DMF (10 mL) and treated with PyBOP (246 mg, 0.473 mmol) at
0 °C. The mixture was stirred 15 min at 0 °C and 4 h at room
temperature. The solvent was evaporated in vacuo, and the
remaining residue (HPLC: 63.98 min) was dissolved in TFA
(10 mL). The mixture was stirred at room temperature for 1.5 h,
the solvent was removed in vacuo, and the product was purified
by preparative HPLC. The product-containing fractions were
combined, evaporated, and lyophilized to give compound 4
(0.24 g, 74%) as a white TFA; mp 118-120 °C. HPLC: 42.17
min. MS calcd 574.24, found 575.1 (M þ H)þ. 13C NMR (DMSO-
d6) δ (mixture of conformers, major signals) 24.38, 25.53, 25.79,
25.86, 25.94, 26.04, 29.40, 31.27, 32.97, 33.38 (Ccyclohexyl, Cβ-Pro,
Cγ-Pro), 38.40, 44.80, 45.83, 45.86, 46.84 (Cβ-Cha, Cδ-Pro, NH-
CH2-Carom, NH2-CH2), 52.09 (CR-Cha), 58.41, 59.94 (CR-Pro,
CH2SO2), 126.99, 127.69, 128.01, 128.06, 130.31, 130.46, 130.88,
131.07, 133.29, 139.99 (Carom), 170.67, 172.00 (CdO), 116.93 (q,
1J = 298 Hz, F3C), 158.08 (q, 2J = 33 Hz, F3CCO2H).
Template Search, Sequence Alignment, and Model Building.
The amino acid sequence of the catalytic domain of matriptase-2
was retrieved from the Swiss-Prot sequence database (http://
expasy.org/sprot/), and a template search was performed
against sequences of proteins from the PDB37 using the Basic
Local Alignment Search Tool (BLAST) algorithm.54 Initial
pairwise sequence alignments were performed using the
ClustalW55 sequence alignment tool. A structure-based multiple
sequence alignment was generated using the protein alignment
MOE-Align function of the Molecular Operating Environment
(MOE).56 Three homology modeling software tools, each with
different core modeling methods, MOE, Modeller,36 and the
SwissModel modeling server,57 were used to generate the initial
models. All energy minimizations, visualizations, and manual
modeling were performed using the MOE graphical environ-
ment. To model the “60 insertion loop” and regions of insertions
and deletions, alternative loops were generated and evaluated
with MOE and the DOPE loop optimization protocol of
Modeller. The best intermediate model was selected and sub-
jected to controlled energy minimization while tethering the
protein backbone and side chain atoms conserved in matriptase
and matriptase-2. The stereochemical quality of intermediate
and final refined models was examined using the MOE protein
quality evaluation function, an implementation of PRO-
CHECK.58 Side chain packing quality and sequence-structure
compatibility scores were calculated with ProSA-web.59
Cloning and Transfection. Total mRNA was isolated from
HepG2 cells expressing endogenous matriptase-2 with TRI-
ZOL-Reagent (Invitrogen, Karlsruhe, Germany) and tran-
scribed into cDNA using Omniscript RT kit (Qiagen, Hilden,
Germany) and Oligo dT primer (Invitrogen, Karlsruhe,
Germany) according to the instruction manuals. For construc-
tion of the expression vector pcDNA4-matriptase-2-Myc-His
A, TMPRSS6 cDNA was obtained by PCR using primers
matriptase-2-fw (50-GCTCCCGGTACCATGCCCGTGGCC-
GAGGCCC-30) and matriptase-2-rv (50-GCTCCCCTCGAG-
GGTCACCACTTGCTGGATCC-30; Acc65I and XhoI recog-
nition sites in bold, TMPRSS6 cDNA underlined). After its
cloning into the Acc65I/XhoI restriction sites of plasmid
pcDNA4-Myc-His A (Invitrogen, Karlsruhe, Germany), the
Virtual Screening. The ZINC60 database containing a total of
5627809 compounds was filtered to remove toxic and reactive

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5533
TMPRSS6 DNA fragment was confirmed by sequencing. The
resulting expression vector pcDNA4-matriptase-2-Myc-His A
encodes the matriptase-2 protein tagged at the C-terminus with
a c-Myc epitope (Myc-tag) and six histidine residues (His-tag).
HEK cells were transfected with pcDNA4-matriptase-2-
Myc-His A plasmid by lipofection with Lipofectamine TM
2000 (Invitrogen, Karlsruhe, Germany) according to the man-
ufacturer’s instructions. Forty-eight hours after transfection,
HEK cells were selected with Zeocin (100 μg/mL; Cayla SAS,
Toulouse, France). Single cell clones stably expressing the
human matriptase-2-Myc-His protein were raised.
Purification and Detection of Matriptase-2. For the purifica-
tion of the catalytic domain of matriptase-2 from the condi-
tioned medium of transfected HEK-MT2 cells, cells were seeded
into 75 cm2 flasks (5 ꢀ 105 cells per cm2) and cultured for one day
in full medium. After two washing steps with PBS media, it was
replaced by FBS-free Opti-MEM. Cells were cultured for
further two days to increase the protein concentration. Condi-
tioned medium was then collected and concentrated using
Amicon Ultra-15 centrifugal filters (3000 MWCO, Millipore,
Billerica, MA). To verify expression of matriptase-2, 10 μg of
total protein was separated by 10% SDS-PAGE under reducing
conditions, analyzed by Western blotting with monoclonal
mouse anti-c-myc antibody (clone 9E10; Sigma-Aldrich, Saint
Louis, MO) and alkaline phosphatase-conjugated goat anti-
mouse IgG (Merck, Darmstadt, Germany). Activity of alkaline
phosphatase was visualized with the chromogenic substrate
5-bromo-4-chloro-3-indolyl phosphate (BCIP). The catalytic
domain of matriptase-2 was isolated from the concentrated
conditioned medium by immunoprecipition using ProFound
c-Myc Tag IP/Co-IP Kit (Pierce Biotechnology, Rockford, IL)
as described by the manufacturer.
of the Deutsche Forschungsgemeinschaft (DFG). We thank
StephanieHautmann, Tobias Bald, Stefan Frank, and Maxim
Frizler for assistance.
Supporting Information Available: Structures of the 30 top-
ranked compounds from the virtual screening experiments.
Model evaluation report (Ramachandran diagram, ProSA,
and ANOLEA plots). Models of the complex of matriptase-2
with compound 3, and the complexes of matriptase with com-
pound 1 and 3, respectively. Data for inhibition of compounds
1-4 toward human thrombin and bovine trypsin. 13C NMR
spectra of compounds 1-4, 7, 8, 10, and 11. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Netzel-Arnett, S.; Hooper, J. D.; Szabo, R.; Madison, E. L.;
Quigley, J. P.; Bugge, T. H.; Antalis, T. M. Membrane anchored
serine proteases: a rapidly expanding group of cell surface proteo-
lytic enzymes with potential roles in cancer. Cancer Metastasis Rev.
2003, 22, 237–258.
(2) Szabo, R.; Wu, Q.; Dickson, R. B.; Netzel-Arnett, S.; Antalis,
T. M.; Bugge, T. H. Type II transmembrane serine proteases.
Thromb. Haemostasis 2003, 90, 185–193.
(3) Noel, A.; Maillard, C.; Rocks, N.; Jost, M.; Chabottaux, V.;
Sounni, N. E.; Maquoi, E.; Cataldo, D.; Foidart, J. M. Membrane
associated proteases and their inhibitors in tumour angiogenesis.
J. Clin. Pathol. 2004, 57, 577–584.
(4) Bugge, T. H.; Antalis, T. M.; Wu, Q. Type II transmembrane serine
proteases. J. Biol. Chem. 2009, 284, 23177–23181.
(5) Szabo, R.; Bugge, T. H. Type II transmembrane serine proteases
in development and disease. Int. J. Biochem. Cell Biol. 2008, 40,
1297–1316.
€
(6) Choi, S. Y.; Bertram, S.; Glowacka, I.; Park, Y. W.; Pohlmann, S.
Type II transmembrane serine proteases in cancer and viral infec-
tions. Trends Mol. Med. 2009, 15, 303–312.
(7) Lee, M. S. Matrix-degrading type II transmembrane serine pro-
tease matriptase: its role in cancer development and malignancy.
J. Cancer Mol. 2006, 2, 183–190.
Enzymatic Assay. The activity of matriptase-2 in the condi-
tioned medium of HEK-MT2 cells, of the purified catalytic
domain of matriptase-2 and of recombinant matriptase
€
(catalytic domain; Enzo Life Sciences, Lorrach, Germany) was
assayed in Tris saline buffer (50 mM Tris, 150 mM NaCl, pH
8.0) at 37 °C by monitoring the release of para-nitroaniline from
the chromogenic substrate Boc-Gln-Ala-Arg-para-nitroanilide
(Bachem, Bubendorf, Switzerland) at 405 nm using a Cary 100
UV-vis spectrophotometer (Varian, Darmstadt, Germany).
Km values were determined with eight different substrate con-
centrations in duplicate experiments. Inhibition assays were
performed in duplicate or triplicate measurements with three
(2 and 4 at matriptase-2) or at least five (other experiments)
different inhibitor concentrations. IC50 values were obtained by
nonlinear regression according to equation v = v0/(1 þ [I]/IC50).
Then 10 mM inhibitor stock solutions of 1-4 and leupeptin
(Calbiochem. Darmstadt, Germany) and a 100 mM stock
solution of Boc-Gln-Ala-Arg-para-nitroanilide were prepared
in DMSO, and a 1 mM stock solution of aprotinin (Carl Roth,
Karlsruhe, Germany) in H2O. The final concentration of the
substrate was 400 μM and of DMSO was 1.5%. Into a cuvette
containing 979 μL prewarmed assay buffer, 11 μL of an inhibitor
solution and 4 μL of a substrate solution were added and
thoroughly mixed. The reaction was initiated by adding 6 μL
of an enzyme solution (5 μg/6 μL total protein of the conditioned
medium of HEK-MT2 cells; 28 ng/6 μL purified catalytic
domain of matriptase-2; 3 ng/6 μL of matriptase) and was
followed over 20 min.
(8) Hooper, J. D.; Clements, J. A.; Quigley, J. P.; Antalis, T. M. Type II
transmembrane serine proteases. Insights into an emerging class of
cell surface proteolytic enzymes. J. Biol. Chem. 2001, 276, 857–860.
(9) Lin, C. Y.; Anders, J; Johnson, M.; Sang, Q. A.; Dickson, R. B.
Molecular cloning of cDNA for matriptase, a matrix-degrading
serine protease with trypsin-like activity. J. Biol. Chem. 1999, 274,
18231–18236.
(10) Shi, Y. E.; Torri, J.; Yieh, L.; Wellstein, A.; Lippman, M. E.;
Dickson, R. B. Identification and characterization of a novel
matrix-degrading protease from hormone-dependent human
breartst cancer cells. Cancer Res. 1993, 53, 1409–1415.
(11) Velasco, G.; Santiago, C.; Victor, Q.; Luis, M. S.; Carlos, L.
Matriptase-2, a membrane-bound mosaic serine proteinase pre-
dominantly expressed in human liver and showing degrading
activity against extracellular matrix proteins. J. Biol. Chem. 2002,
277, 37637–37646.
(12) Szabo, R.; Netzel-Arnett, S.; Hobson, J. P.; Antalis, T. M.; Bugge,
T. H. Matriptase-3 is
a novel phylogenetically preserved
membrane-anchored serine protease with broad serpin reactivity.
Biochem. J. 2005, 390, 231–242.
(13) Cal, S.; Quesada, V.; Garabaya, C.; Lopez-Otin, C. Polyserase-I, a
human polyprotease with the ability to generate independent serine
protease domains from a single translation product. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 9185–9190.
(14) Okumura, Y.;Hayama, M.;Takahashi, E.; Fujiuchi, M.;Shimabukuro,
A.; Yano, M.; Kido, H. Serase-1B, a new splice variant of polyserase-
1/TMPRSS9, activates urokinase-type plasminogen activator and the
proteolytic activation is negatively regulated by glycosaminoglycans.
Biochem. J. 2006, 400, 551–561.
(15) Lee, S. L.; Dickson, R. B.; Lin, C. Y. Activation of hepatocyte
growth factor and urokinase plasminogen activator by matriptase,
an epithelial membrane serine protease. J. Biol. Chem. 2000, 275,
36720–36725.
(16) List, K.; Szabo, R.; Wertz, P. W.; Segre, J.; Haudenschild, C. C.;
Kim, S. Y.; Bugge, T. H. Loss of proteolytically processed filaggrin
caused by epidermal deletion of Matriptase/MT-SP1. J. Cell Biol.
2003, 163, 901–910.
Human thrombin (Calbiochem. Darmstadt, Germany) and
trypsin from bovine pancreas (Sigma, Steinheim, Germany)
were assayed with the chromogenic substrates H-D-Phe-L-Pip-
Arg-para-nitroanilide (Hyphen BioMed, Neuville-sur-Oise,
France) and Suc-Ala-Ala-Pro-Arg-para-nitroanilide (Bachem,
Bubendorf, Switzerland), respectively. Details are given in the
Supporting Information (Table S2).
(17) Oberst, M. D.; Johnson, M. D.; Dickson, R. B.; Lin, C. Y.; Singh,
B.; Stewart, M.; Williams, A.; Nafussi, A.; Smyth, J. F.; Gabra, H.;
Sellar, G. C. Expression of the serine protease matriptase and its
Acknowledgment. This work was supported by the
Graduiertenkolleg 677 and the Sonderforschungsbereich 645

5534 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15
Sisay et al.
inhibitor HAI-1 in epithelial ovarian cancer: correlation with
clinical outcome and tumor clinicopathological parameters. Clin.
Cancer Res. 2002, 8, 1101–1107.
a matrix-degrading transmembrane serine proteinase. J. Biol. Chem.
2002, 277, 2160–2168.
(39) Sichler, K.; Hopfner, K. -P.; Kopetzki, E.; Huber, R.; Bode, W.;
Brandstetter, H. The influence of residue 190 in the S1 site of
trypsin-like serine proteases on substrate selectivity is universally
conserved. FEBS Lett. 2002, 530, 220–224.
(18) Uhland, K. Matriptase and its putative role in cancer. Cell. Mol.
Life Sci. 2006, 63, 2968–2978.
(19) Park, T. J.; Lee, Y. J.; Kim, H. J.; Park, H. G.; Park, W. J. Cloning
and characterization of TMPRSS6, a novel type 2 transmembrane
serine protease. Mol. Cells 2005, 19, 223–227.
(20) Ramsay, A. J.; Reid, J. C.; Velasco, G.; Quigley, J. P.; Hooper, J. D.
The type II transmembrane serine protease matriptase-2;identi-
fication, structural features, enzymology, expression pattern and
potential roles. Front. Biosci. 2008, 13, 569–579.
(21) Parr, C.; Sanders, A. J.; Davies, G.; Martin, T.; Lane, J.; Mason,
M. D.; Mansel, R. E.; Jiang, W. G. Matriptase-2 inhibits breast
growth and invasion and correlates with favorable prognosis for
breast cancer patients. Clin. Cancer. Res. 2007, 13, 3568–3576.
(22) Sanders, A. J.; Parr, C.; Martin, T. A.; Lane, J.; Mason, M. D.;
Jiang, W. G. Genetic upregulation of matriptase-2 reduces the
aggressiveness of prostate cancer cells in vitro and in vivo and
affects FAK and paxillin localisation. J. Cell. Physiol. 2008, 216,
780–789.
(23) Du, X.; She, E.; Gelbart, T.; Truksa, J.; Lee, P.; Xia, Y.; Khovananth,
K.; Mudd, S.; Mann, N.; Moresco, E. M. Y.; Beutler, E.; Beutler, B.
The serine protease TMPRSS6 is required to sense iron deficiency.
Science 2008, 320, 1088–1092.
(24) Finberg, K. E.; Heeney, M. M.; Campagna, D. R.; Ays-inok, Y.;
Pearson, H. A.; Hartman, K. R.; Mayo, M. M.; Samuel, S. M.;
Strouse, J. J.; Markianos, K.; Andrews, N. C.; Fleming, M. D.
Mutations in TMPRSS6 cause iron-refractory iron deficiency
anaemia (IRIDA). Nat. Genet. 2008, 40, 569–571.
(25) Melis, M. A.; Cau, M.; Congiu, R.; Sole, G.; Barella, S.; Cao, A.;
Westerman, M.; Cazzola, M.; Galanello, R. A mutation in the
TMPRSS6 gene, encoding a transmembrane serine protease that
suppresses hepcidin production, in familial iron deficiency anemia
refractory to oral iron. Haematologica 2008, 93, 1473–1479.
(26) Ramsay, A. J.; Hooper, J. D.; Folgueras, A. R.; Velasco, G.;
Lopez-Otin, C. Matriptase-2 (TMPRSS6): a proteolytic regulator
of iron homeostasis. Haematologica 2009, 94, 840–849.
(27) De Domenico, I.; McVey Ward, D.; Kaplan, J. Regulation of iron
acquisition and storage: consequences for iron-linked disorders.
Nat. Rev. Mol. Cell. Biol. 2008, 9, 72–81.
(28) Niederkofler, V.; Salie, R.; Arber, S. Hemojuvelin is essential for
dietary iron sensing, and its mutation leads to severe iron overload.
J. Clin. Invest. 2005, 115, 2180–2186.
€
(40) Steinmetzer, T.; Schweinitz, A.; Sturzebecher, A.; Donnecke, D.;
€
Uhland, K.; Schuster, O.; Steinmetzer, P.; Muller, F.; Friedrich, R.;
Than, M. E.; Bode, W.; Sturzebecher, J. Secondary amides of
€
sulfonylated 3-amidinophenylalanine. New potent and selective
inhibitors of matriptase. J. Med. Chem. 2006, 49, 4116–4126.
(41) Stauffer, K. J.; Williams, P. D.; Selnick, H. G.; Nantermet, P. G.;
Newton, C. L.; Homnick, C. F.; Zrada, M. M.; Lewis, S. D.; Lucas,
B. J.; Krueger, J. A.; Pietrak, B. L.; Lyle, E. A.; Singh, R.; Miller-
Stein, C.; White, R. B.; Wong, B.; Wallace, A. A.; Sitko, G. R.;
Cook, J. J.; Holahan, M. A.; Stranieri-Michener, M.; Leonard,
Y. M.; Lynch, J. J., Jr.; McMasters, D. R.; Yan, Y. 9-Hydroxyaza-
fluorenes and their use in thrombin inhibitors. J. Med. Chem. 2005,
48, 2282–2293.
(42) Baum, B.; Mohamed, M.; Zayed, M.; Gerlach, C.; Heine, A.;
Hangauer, D.; Klebe, G. More than a simple lipophilic contact: a
detailed thermodynamic analysis of nonbasic residues in the S1
pocket of thrombin. J. Mol. Biol. 2009, 390, 56–69.
(43) Rittle, K. E.; Barrow, J. C.; Cutrona, K. J.; Glass, K. L.; Krueger,
J. A.; Kuo, L. C.; Lewis, S. D.; Lucas, B. J.; McMasters, D. R.;
Morrissette, M. M.; Nantermet, P. G.; Newton, C. L.; Sanders,
W. M.; Yan, Y.; Vacca, J. P.; Selnick, H. G. Unexpected enhance-
ment of thrombin inhibitor potency with o-aminoalkylbenzyla-
mides in the P1 position. Bioorg. Med. Chem. Lett. 2003, 20, 3477–
3482.
€
(44) Schweinitz, A.; Sturzebecher, A.; Sturzebecher, U.; Schuster, O.;
Sturzebecher, J.; Steinmetzer, T. New substrateanalogue inhibitors
€
€
of factor Xa containing 4-amidinobenzylamide as P1 residue: Part
1. Med. Chem. 2006, 2, 349–361.
(45) Schweinitz, A.;Steinmetzer, T.;Banke, I. J.;Arlt, M. J. E.;Sturzebecher,
€
A.; Schuster, O.; Geissler, A.; Giersiefen, H.; Zeslawska, E.; Jacob, U.;
€
€
Kruger, A.; Sturzebecher, J. Design of novel and selective inhibitors of
urokinase-type plasminogen activator with improved pharmacokinetic
properties for use as antimetastatic agents. J. Biol. Chem. 2004, 279,
33613–33622.
(46) Nelson, T. D.; LeBlond, C. R.; Frantz, D. E.; Matty, L.; Mitten,
J. V.; Weaver, D. G.; Moore, J. C.; Kim, J. M.; Boyd, R.; Kim,
P. -Y.; Gbewonyo, K.; Brower, M.; Sturr, M.; McLaughlin, K.;
McMasters, D. R.; Kress, M. H.; McNamara, J. M.; Dolling, U. H.
Stereoselective synthesis of a potent thrombin inhibitor by a novel
P2-P3 lactone ring opening. J. Org. Chem. 2004, 69, 3620–3627.
(47) Cho, E. G.; Kim, C.; Kim, M. G.; Kim, S. R.; Seong, I. S.; Chung,
C.; Schwartz, R. H.; Park, D. N-Terminal processing is essential for
release of epithin, a mouse type II transmembrane serine protease.
J. Biol. Chem. 2001, 276, 44581–44589.
(29) Nemeth, E.; Tuttle, M. S.; Powelson, J.; Vaughn, M. B.; Donovan,
A.; Ward, D. M.; Ganz, T.; Kaplan, J. Hepcidin regulates cellular
iron efflux by binding to ferroportin and inducing its internaliza-
tion. Science 2004, 306, 2090–2093.
(30) Guillem, F.; Lawson, S.; Kannengiesser, C.; Westerman, M.;
Beaumont, C.; Grandchamp, B. Two nonsense mutations in the
TMPRSS6 gene in a patient with microcytic anemia and iron
deficiency. Blood 2008, 112, 2089–2091.
€
(48) Hellstern, P.; Sturzebecher, U.; Wuchold, B.; Haubelt, H.; Seyfert,
€
U. T.; Bauer, M.; Vogt, A.; Sturzebecher, J. Preservation of in vitro
function of platelets stored in the presence of a synthetic dual
inhibitor of factor Xa and thrombin. J. Thromb. Haemostasis 2007,
5, 2119–2126.
ꢀ
(31) Folgueras, A. R.; de Lara, F. M.; Pendas, A. M.; Garabaya, C.;
ꢀ
Rodriguez, F.; Astudillo, A.; Bernal, T.; Cabanillas, R.; Lopez-Otın,
´
C.; Velasco, G. Membrane-bound serine protease matriptase-2
(Tmprss6) is an essential regulator of iron homeostasis. Blood 2008,
112, 2539–2545.
€
(49) Sturzebecher, A.; Donnecke, D.; Schweinitz, A.; Schuster, O.;
Steinmetzer, P.; Sturzebecher, U.; Kotthaus, J.; Clement, B.;
€
€
€
Sturzebecher, J.; Steinmetzer, T. Highly potent and selective sub-
(32) Silvestri, L.; Pagani, A.; Nai, A.; De Domenico, I.; Kaplan, J.;
Camaschella, C. The serine protease matriptase-2 (TMPRSS6)
inhibits hepcidin activation by cleaving membrane hemojuvelin.
Cell Metab. 2008, 8, 502–511.
strate analogue factor Xa inhibitors containing D-homophenylala-
nine analogues as P3 residue: part 2. ChemMedChem 2007, 2, 1043–
1053.
€
(50) Steinmetzer, T.; Donnecke, D.; Korsonewski, M.; Neuwirth, C.;
(33) Papanikolaou, G.; Samuels, M. E.; Ludwig, E. H.; MacDonald,
Steinmetzer, P.; Schulze, A.; Saupe, S. M.; Schweinitz, A. Modi-
fication of the N-terminal sulfonyl residue in 3-amidinophenylala-
nine-based matriptase inhibitors. Bioorg. Med. Chem. Lett. 2009,
19, 67–73.
ꢀ
M. L.; Franchini, P. L.; Dube, M. P.; Andres, L.; MacFarlane, J.;
Sakellaropoulos, N.; Politou, M.; Nemeth, E.; Thompson, J.;
Risler, J. K.; Zaborowska, C.; Babakaiff, R.; Radomski, C. C.;
Pape, T. D.; Davidas, O.; Christakis, J.; Brissot, P.; Lockitch, G.;
Ganz, T.; Hayden, M. R.; Goldberg, Y. P. Mutations in HFE2
cause iron overload in chromosome 1q-linked juvenile hemochro-
matosis. Nat. Genet. 2004, 36, 77–82.
€
(51) Steinmetzer, T.; Sturzebecher, J. Progress in the development of
synthetic thrombin inhibitors as new orally active anticoagulants.
Curr. Med. Chem. 2004, 11, 2297–2321.
(52) Soeiro, M. N.; de Castro, S. L.; de Souza, E. M.; Batista, D. G.;
Silva, C. F.; Boykin, D. W. Diamidine activity against trypano-
somes: the state of the art. Curr. Mol. Pharmacol. 2008, 1, 151–161.
(53) Becker, G. L.; Sielaff, F.; Than, M. E.; Lindberg, I.; Routhier, S.;
Day, R.; Lu, Y.; Garten, W.; Steinmetzer, T. Potent inhibitors of
furin and furin-like proprotein convertases containing decarboxy-
lated P1 arginine mimetics. J. Med. Chem. 2010, 53, 1067–1075.
(34) Malyszko, J. Hemojuvelin: the hepcidin story continues. Kidney
Blood Press Res. 2009, 32, 71–76.
ꢀ
ꢀ
(35) Beliveau, F.; Desilets, A.; Leduc, R. Probing the substrate specifi-
cities of matriptase, matriptase-2, hepsin and DESC1 with intern-
ally quenched fluorescent peptides. FEBS J. 2009, 276, 2213–2226.
(36) Sali, A.; Blundell, T. L. Comparative protein modeling by satisfac-
tion of spatial restraints. J. Mol. Biol. 1993, 234, 779–815.
€
(37) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.;
Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data
Bank. Nucleic Acids Res. 2000, 28, 235–242.
(38) Friedrich, R.; Fuentes-Prior, P.; Ong, E.; Coombs, G.; Hunter, M.;
Oehler, R.; Pierson, D.; Gonzalez, R.; Huber, R.; Bode, W.;
Madison, E. L. Catalytic domain structures of MT-SP1/matriptase,
(54) Altschul, S. F.; Madden, T. L.; Schaffer, A. A.; Zhang, J.; Zhang,
Z.; Miller, W.; Lipman, D. J. Gapped BLAST and PSI-BLAST: a
new generation of protein database search programs. Nucleic Acids
Res. 1997, 25, 3389–3402.
(55) Thompson, J. D.; Higgins, D. G.; Gibson, T. J. ClustalW: improv-
ing the sensitivity of progressive multiple sequence alignments

Article
through sequence weighting, position specific gap penalties
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 15 5535
(61) Durant, J. L.; Leland, B. A.; Henry, D. R.; Nourse, J. G. Reopti-
mization of MDL keys for use in drug discovery. J. Chem. Inf.
Comput. Sci. 2002, 42, 1273–1280.
and weight matrix choice. Nucleic Acids Res. 1994, 22, 4673–
4680.
(56) Molecular Operating Environment (MOE 2005.06); C.C.G. Inc.:
Montreal, Canada, 2005.
(57) Schwede, T.; Kopp, J.; Guex, N.; Peitsch, M. C. SWISS-MODEL:
an automated protein homology-modeling server. Nucleic Acids
Res. 2003, 31, 3381–3385.
(58) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton,
J. M. PROCHECK: a program to check the stereochemical quality
of protein structures. J. Appl. Crystallogr. 1993, 26, 283–291.
(59) Wiederstein, M.; Sippl, M. J. ProSA-web: interactive web service
for the recognition of errors in three-dimensional structures of
proteins. Nucleic Acids Res. 2007, 35, W407–W410.
(60) Irwin, J. J.; Shoichet, B. K. ZINC;a free database of commercially
available compounds for virtual screening. J. Chem. Model. 2005,
45, 177–182.
(62) Willett, P. Searching techniques for databases of two- and
three-dimensional chemical structures. J. Med. Chem. 2005, 48,
4183–4199.
(63) Bender, A.; Mussa, H. Y.; Glen, R. C.; Reiling, S. Molecular
similarity searching using atom environments, information-based
feature selection, and a naıve Bayesian classifier. J. Chem. Inf.
Comput. Sci. 2004, 44, 170–178.
(64) Meng, E. C.; Shoichet, B. K.; Kuntz, I. D. Automated docking with
grid-based energy evaluation. J. Comput. Chem. 1992, 13, 505–524.
(65) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A fast flexible
docking method using an incremental construction algorithm.
J. Mol. Biol. 1996, 261, 470–489.
(66) DeLano, W. L. The PyMOL molecular graphics system; DeLano
Scientific: San Carlos, CA, 2002; http://www.pymol.org.