Modulating the selectivity of matriptase-2 inhibitors with unnatural amino acids

Article abstract of DOI:10.1016/j.ejmech.2017.02.006

Matriptase-2, a type II transmembrane serine protease (TTSP), is expressed in the liver and regulates iron homeostasis via the cleavage of hemojuvelin. Matriptase-2 emerges as an attractive target for the treatment of conditions associated with iron overload, such as hemochromatosis or beta-thalassemia. Starting from the crystal structure of its closest homolog matriptase, we constructed a homology model of matriptase-2 in order to further optimize the selectivity of serine trap peptidomimetic inhibitors for matriptase-2 vs matriptase. Careful modifications of the P4, P3 and P2 positions with the help of unnatural amino acids led to a thorough understanding of Structure-Activity Relationship and a >60-fold increase in selectivity for matriptase-2 vs matriptase. Additionally, the introduction of unnatural amino acids led to significant increases in plasma stability. Such compounds represent useful pharmacological tools to test matriptase-2 inhibition in a context of iron overload.

Full text of DOI:10.1016/j.ejmech.2017.02.006

Accepted Manuscript
Modulating the selectivity of Matriptase-2 inhibitors with unnatural amino acids
Catherine St-Georges, Antoine Désilets, François Béliveau, Mariana Ghinet,
Sébastien P. Dion, Éloic Colombo, Pierre-Luc Boudreault, Rafael J. Najmanovich,
Richard Leduc, Éric Marsault
PII:
S0223-5234(17)30075-2
10.1016/j.ejmech.2017.02.006
EJMECH 9202
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 2 December 2016
Revised Date: 19 January 2017
Accepted Date: 5 February 2017
Please cite this article as: C. St-Georges, A. Désilets, F. Béliveau, M. Ghinet, S.P. Dion, Éloic Colombo,
P.-L. Boudreault, R.J. Najmanovich, R. Leduc, E. Marsault, Modulating the selectivity of Matriptase-2
inhibitors with unnatural amino acids, European Journal of Medicinal Chemistry (2017), doi: 10.1016/
j.ejmech.2017.02.006.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Modulating the Selectivity of Matriptase-2 Inhibitors
with Unnatural Amino Acids
Catherine St-Georges,†,‡ Antoine Désilets, †,‡ François Béliveau, †,‡ Mariana Ghinet, †,‡ Sébastien
P. Dion, †,‡ Éloic Colombo, †,‡ Pierre-Luc Boudreault, †,‡ Rafael J. Najmanovich, §,‡ Richard
Leduc, †,‡,* and Éric Marsault†,‡,*
†Department of Pharmacology-Physiology, §Department of Biochemistry, ‡Institut de
Pharmacologie de Sherbrooke, Université de Sherbrooke, 3001 12e Avenue Nord, Sherbrooke,
Québec, J1H 5N4, Canada
KEYWORDS. Type 2 transmembrane serine proteases, matriptase-2, matriptase, slow tight
binding inhibitors, selectivity, peptidomimetics, ketobenzothiazole, serine trap.
1

ACCEPTED MANUSCRIPT
ABSTRACT. Matriptase-2, a type II transmembrane serine protease (TTSP), is expressed in the
liver and regulates iron homeostasis via the cleavage of hemojuvelin. Matriptase-2 emerges as an
attractive target for the treatment of conditions associated with iron overload, such as
hemochromatosis or beta-thalassemia. Starting from the crystal structure of its closest homolog
matriptase, we constructed a homology model of matriptase-2 in order to further optimize the
selectivity of serine trap peptidomimetic inhibitors for matriptase-2 vs matriptase. Careful
modifications of the P4, P3 and P2 positions with the help of unnatural amino acids led to a
thorough understanding of Structure-Activity Relationship and a >60-fold increase in selectivity
for matriptase-2 vs matriptase. Additionally, the introduction of unnatural amino acids led to
significant increases in plasma stability. Such compounds represent useful pharmacological tools
to test matriptase-2 inhibition in a context of iron overload.
2

ACCEPTED MANUSCRIPT
INTRODUCTION. Serine proteases are involved in numerous biological processes [1]. Type
II transmembrane serine proteases (TTSPs) are a subfamily of serine proteases characterized by
an N-terminal cytoplasmic tail followed by a transmembrane domain, a stem region and an
extracellular C-terminal serine protease domain that contains the canonical catalytic triad [2]
composed of histidine (His57), aspartic acid (Asp102) and serine (Ser195) [1]. TTSPs are divided
in four subfamilies: HAT/DESC (human airway trypsin-like protease/differentially expressed in
squamous cell carcinoma), hepsin/TMPRSS (transmembrane protease, serine), matriptase and
corin [3].
In physiological conditions, the expression of matriptase-2 (encoded by the TMPRSS6 gene) is
restricted to the liver, where it is found on the surface of hepatocytes with its substrate
hemojuvelin (HJV) [4,5]. The cleavage of HJV by matriptase-2 has been shown to play a
significant role in the regulation of iron homeostasis in mice and human [6–8]. Intact HJV acts as
a co-receptor for BMP/SMAD signaling, which increases transcription of hepcidin, the major
regulator of iron levels in circulation [7,9–13]. Hepcidin then binds to the iron transporter
ferroportin, causes its degradation, which leads to a reduction of iron transport to the
bloodstream [14]. Matriptase-2 cleaves HJV at the plasma membrane of hepatocytes, acting de
facto as a negative regulator of hepcidin expression [4]. Consistently, loss-of-function mutations
of matriptase-2 in both mice and humans results in iron-refractory iron deficiency anemia, a form
of anemia which does not benefit from oral iron supplementation [15–17]. Iron overload diseases
such as hemochromatosis and β-thalassemia are two important inherited blood disorders [18].
Currently, only two treatments are available: phlebotomy (bloodletting) for hemochromatosis,
and iron chelators for β-thalassemia. However, these treatments are invasive or associated with
significant side-effects, respectively [19]. Therefore, regulation of the hepcidin-activating
3

ACCEPTED MANUSCRIPT
pathway constitutes a promising approach to treat both diseases. Indeed, several studies in mice
have shown that genetic reduction of TMPRSS6 levels ameliorates both hemochromatosis and β-
thalassemia phenotypes, making matriptase-2 a potential therapeutic target for these iron
overload disorders [20,21].
TTSPs possess different functions but similar substrate specificities [22]. Matriptase,
structurally the most similar TTSP to matriptase-2, is one of the most studied TTSPs [23,24]. It
is highly expressed in most epithelial tissues, where it exerts its main function in maintaining
epithelial integrity [22]. The over-expression of matriptase has been linked to the progression of
breast carcinoma [25] and other epithelial cancers [26]. Matriptase has also been linked to the
replication of the influenza virus [27,28], and the progression of osteoarthritis [29]. A previous
study demonstrated that mutation G827R in the catalytic domain of matriptase is associated with
the development of autosomal recessive ichthyosis diseases with hypotrichosis in humans [30],
whereas genetic knockout of matriptase in mice gives rise to animals presenting severe barrier
defects that die shortly after birth. In this context, selectivity for matriptase-2 vs matriptase
appears important if one ultimately wishes to tackle iron overload disorders using matriptase-2
inhibitors.
In the past few years, several groups reported different classes of matriptase-2 inhibitors using
various scaffolds [31–36]. Most studies however did not report selectivity for matriptase-2 vs
matriptase [33,34], or reported non-selective inhibitors [31,32,36,37]. To the best of our
knowledge, the only selective matriptase-2 inhibitors were reported by Gitlin et al [35]. These
belong to macrocyclic analogues of the Sunflower Trypsin Inhibitor 1, a bicyclic
tetradecapeptide with molecular weight >1,500 Da. On the other hand, our group recently
reported and used a new series of slow, tight-binding matriptase inhibitors designed based on the
4

ACCEPTED MANUSCRIPT
N-terminal portion of the auto-catalytic activation sequence of matriptase (Arg-Gln-Ala-Arg),
flanked by a ketobenzothiazole serine trap on the scissile bond [27,38,39]. Subsequent in silico
screening with all combinations of the 20 natural amino acids on the P4, P3 and P2 positions led
to the identification of the Tyr-Tyr-Val-Arg-Ketobenzothiazole (Y-Y-V-R-Kbt, compound 1,
Figure 1) sequence as a potent inhibitor of matriptase-2 [40]. Herein, we report a series of 29
potent peptidomimetic inhibitors of matriptase-2, in which positions P4, P3 and P2 were varied
using unnatural amino acids. Inhibition constants (Ki) of the compounds were measured against
matriptase-2 and its close homolog matriptase, with the goal to improve both potency and
selectivity. The most potent and selective compounds were further evaluated to determine their
mode of action and dissociation half-time.
RESULTS AND DISCUSSION.
Homology modelling and docking. A homology model of matriptase-2 was built using
MODELER [41] and GROMACS [42], starting from the published crystal structure of
matriptase (PDB code 3NCL), as described previously [40]. Using this model, five residues only
were found to differ between the two enzymes in the otherwise identical substrate binding
pockets of matriptase-2 vs matriptase: Ala757 vs Ser800 (S1), His665 vs Phe708 (S2), Glu712 vs
Tyr755 (S3), Asp663 vs Phe706 and Leu785 vs Asp828 (S4) [40]. The molecules synthesized below
were docked into the above homology model and the crystal structure of matriptase to rationalize
their mode of interaction with both enzymes (details of docking procedures can be found in the
Supporting Information section).
5

ACCEPTED MANUSCRIPT
Structure-Activity Relationships. We previously reported a peptidomimetic inhibitor with a
Ki of 2.6 ± 0.1 nM for matriptase-2 and 12-fold selectivity vs matriptase (sequence Y-Y-V-R,
analogue 1, Table 1, Figure 1). Using 1 as a reference, the P4, P3 and P2 positions (Schechter
and Berger nomenclature, Figure 1) [43] were modified with unnatural and natural amino acids
to explore structure-activity relationship (SAR), improve or conserve potency for matriptase-2,
and improve selectivity vs matriptase. It has been shown by others and ourselves that arginine in
P1 is the favored amino acid for matriptase and matriptase-2 [38,44,45], and for this reason the
P1 was kept invariant at this stage. Likewise, in the present article the ketobenzothiazole was not
modified because it provides excellent inhibition. Inhibition constants (Ki) for both recombinant
proteases were determined as described in the experimental section, and selectivity ratios are
reported as the ratio of Ki (matriptase) vs Ki (matriptase-2).
P4 modifications. Table 1 reports the influence of structural variations in position P4 of the Y-
Y-V-R-Kbt scaffold (analogue 1) on inhibition and selectivity.
Firstly, removal of the N-terminal amino group was attempted in order to improve the
proteolytic stability of the final compounds. Desamino-compounds are anticipated to be more
biologically stable because the corresponding aminated products may be substrates of
endogenous aminopeptidases. Indeed, whereas reference compound 1 was completely degraded
after 24h in mouse plasma, its desamino analogue 2 showed 40% presence of the parent
compound in the same conditions (Figure 2) [46]. Moreover, (H)Y-Y-V-R (analogue 2)
possessed similar inhibition of matriptase-2 (Ki 2 nM) and a 17-fold selectivity compared to
reference compound 1 (Ki 2.6 nM, 12-fold selectivity). The influence of lipophilicity and steric
bulk on position P4 was next explored. Thus, the Tyr residue was replaced by desamino-
6

ACCEPTED MANUSCRIPT
adamantane (analogue 3) and desamino-Leu (analogue 4). Results showed a small decrease in
potency of both analogues against matriptase-2 (Ki 6.0 nM), which translated in a reduced
selectivity. Hence, aromatic amino acids such as tyrosine are slightly favored compared to
lipophilic amino acids on matriptase-2 at position P4, whereas both residues were equally
tolerated by matriptase. To ascertain this further, Tyr was replaced by Phe (analogue 5), leading
to a similar inhibition compared to Y-Y-V-R-Kbt on matriptase-2 (Ki 1.74 nM). Concomitantly,
a decrease in potency against matriptase (Ki 43 nM) doubled selectivity compared to 1 (25-fold
vs 12-fold).
P3 modifications. The S3 subpocket was then probed with different types of amino acids
(Table 2). Replacement of Tyr by Phe in position P3 (analogue 6) decreased potency on both
proteases (Ki 9 nM and 63 nM for matriptase-2 and matriptase, respectively), demonstrating the
importance of the hydroxyl group at position P3 on both enzymes. Analogue 7 with chloro
substituent in position 3 of Phe displayed a potency similar to the reference compound on both
protease (Ki 5 nM and 34 nM for matriptase-2 and matriptase respectively). To further explore
S3 subpocket specificity, changes were implemented on the P3 position of analogue 5.
Introduction of a 4-thiazolyl group in P3 (analogue 8) had a minor effect on inhibition of both
proteases (Ki 1.2 nM and 12 nM for matriptase-2 and matriptase, respectively). On the other
hand, introduction of a benzyl group on the phenol moiety (analogue 9) decreased potency for
matriptase-2 (Ki 3.54 nM vs 1.74 nM for 5) and matriptase (Ki 71 nM vs 43 nM for 5), reducing
selectivity to 20. Finally, elongation of the side chain by one carbon atom using
homophenylalanine (hPhe, analogue 10), preserved potency for matriptase-2 (Ki 2.1 nM) while
decreasing potency for matriptase (Ki 65 nM), further improving selectivity to 32-fold in favor of
7

ACCEPTED MANUSCRIPT
matriptase-2. As represented in Figure 3, the selectivity of analogue 10 may be explained by the
replacement of Glu712 in pocket S3 of matriptase-2 (cyan) by Tyr755 in matriptase (pink), which
creates a steric clash that presumably decreases Ki.
P2 modifications. Subsequently, modifications to the P2 position were implemented (Table
3). Replacement of Val by the sterically demanding diphenylalanine (Dip, analogue 11) led to a
significant decrease in potency for both proteases (Ki 340 nM and 1380 nM for matriptase-2 vs
matriptase). This presumably creates a clash with His665 in the S2 subpocket (Figure 4),
consistently with the preference of matriptase-2 for small aliphatic residues like Ala or Val in
position P2 discussed by Wysocka et al [45]. Replacement of Val by His led to a decrease in
potency for matriptase-2 (analogue 12, Ki 56 nM) while preserving potency for matriptase (Ki 48
nM), reversing selectivity in this series. Knowing that matriptase-2 possesses a His665 residue in
the S2 subpocket instead of the Phe708 residue for matriptase, this replacement presumably
provokes an electrostatic repulsion between the two imidazole residues. Similar effects were
observed with heterocycles such as thiazolyl or thienyl group (analogues 14 and 17), decreasing
selectivity toward matriptase-2. Likewise, replacement of Phe by 2-naphthylalanine or Trp
(analogues 16 and 18) reduced selectivity for matriptase-2. It also decreased with the addition of
one or two hydroxyl groups (Tyr, analogue 15, (2,3-OH)Phe), analogue 13) on the aromatic ring
of Phe. Altogether, aromatic modifications in position P2 failed to increase selectivity for
matriptase-2 compared to reference compound 1.
The Val residue in position P2 was then replaced by branched aliphatic or polar amino acids,
starting from analogue 8 (H)F-hF-V-R-Kbt (Table 4). Replacement of Val by D-Val (analogue
19) decreased potency for both proteases (Ki 7 nM and 240 nM for matriptase-2 vs matriptase,
8

ACCEPTED MANUSCRIPT
respectively). Replacement by Leu (analogue 20) had no effect on potency or selectivity
compared to Val. Interestingly, introduction of Ile (analogue 21) resulted in an important loss of
potency for matriptase (Ki 230 nM) and a slight loss for matriptase-2 (Ki 4 nM), improving
selectivity to 58-fold in favor of matriptase-2. Fine-tuning of side chain stereochemistry via
replacement of
24) at the P2 position affected potency diversely on both proteases.
displayed a similar potency on both proteases compared to the reference (Ki 2.6 nM and 51 nM,
for matriptase-2 and matriptase, respectively). Stereochemistry inversion ( -allo-Ile, analogue
24) led to an important decrease in potency for both proteases compared to the -analogue (Ki
690 nM and Ki 12000 nM for matriptase-2 and matriptase, respectively). -amino acids are not
L-Ile for
D
-Ile (analogue 22),
L-allo-Ile (analogue 23) and
D-allo-Ile (analogue
L-allo-Ile (analogue 23)
D
L
D
well tolerated in position S2 of matriptase-2, and even less in matriptase (analogue 19 or 24).
Linear side chains such as aminobutyric acid (Abu, analogue 25) and Nva (analogue 26)
displayed similar potency for matriptase-2 and matriptase compared to analogue 1. The bulkier
tert-butyl glycine (analogue 27) in position P2 was associated with a drastic decrease in
inhibition for both proteases (Ki 560 nM and Ki 15000 nM for matriptase-2 and matriptase,
respectively), delineating the steric limits of the S2 subpocket. Introduction of directional H-
bond donors in that subpocket via L-threonine (analogue 28) or L-allo-threonine (analogue 29)
was well tolerated, leading to the most selective inhibitor in this series with analogue 29 (Ki 3.1
for matriptase-2, 64-fold selective vs matriptase). Selectivity may be explained by hydrogen
bond interaction between water, the hydroxyl group of L
-allo-threonine (analogue 29) and His665
in the S2 supbocket of matriptase-2 (Figure 5). Replacement of this residue by Phe708 in
matriptase does not allow this positive interaction, translating into an increased selectivity for
matriptase-2. Altogether, these results suggest that the binding pocket tolerates a certain degree
9

ACCEPTED MANUSCRIPT
of steric hindrance, and testifies to its asymmetry. It also demonstrates the preference of
matriptase-2 for branched, lipophilic residues.
Plasma stability. In order to assess the impact of unnatural amino acids in positions P4, P3
and P2 on stability of inhibitor 2, inhibitors 8 and 29 were incubated in mouse plasma. As
demonstrated in Figure 2, both compounds displayed significantly increased plasma stability
compared to reference inhibitor 1 and its desamino congener 2, with >75% parent compound
remaining after 24h incubation for both inhibitors, as opposed to 40% for 2 and 0% for 1.
Kinetic analysis. The kinetic parameters of the most potent compounds were determined
using progress curve analysis. All compounds in this study harbor a ketobenzothiazole serine
trap as the warhead, which is mechanistically expected to form a reversible, covalent bond in the
form of a hemiketal with the catalytic serine residue of the protease [47]. Indeed, results from our
previous studies using R-Q-A-R-Kbt showed that the compound acts as a reversible, slow tight-
binding inhibitor of matriptase [38,39]. Similarly to hepatitis C serine protease inhibitors
telaprevir and boceprevir which both possess a ketoamide serine trap [48], the proposed mode of
action involves a two-step mechanism characterized by the initial rapid formation of a non-
covalent enzyme-inhibitor complex, followed by the slower formation of a high affinity, tightly
bound covalent complex.
Simple steady-state analysis of substrate cleavage rate by the protease in the presence of
inhibitors provides overall Ki but gives no detailed information on mechanism of inhibition or
target dissociation half-time. These parameters were determined via progress curves analysis
with the most potent matriptase-2 inhibitors (analogues 2, 8 and 25) (Table 5). Initially, one- vs
10

ACCEPTED MANUSCRIPT
two-step mechanism was confirmed (Figure 6A), then Ki, Ki* and target dissociation half-time
were determined. Time-course of product formation plotted as a function of time were fitted to
both models using Dynafit software (Figure 6B) [48]. Data fitting quality was evaluated by
calculation of residuals (difference between measured and fitted data) for each model, and best
fit was chosen as the mechanism of action to calculate kinetic parameters. To account for
intrinsic loss of enzyme activity, an enzyme inactivation constant (kiE) was added to the models,
which improved data fitting. Examples of progress curves are shown on Figure 6B for analogue
2, following the two-step inhibition model (left panel). Very small residuals for reaction time
were calculated, indicating very good curve fitting (lower panel).
Using this approach, compounds 2, 8 and 25 were shown to mechanistically act as slow, tight-
binders that followed the two-step mechanism on matriptase-2. Indeed, following the residual
protease activity over time in the presence of inhibitor, a diminution of velocity in time (curve)
was observed, characteristic of the slow nature of the high-affinity, covalent bond formation
(Figure 6B). Furthermore, previous results obtained by our group with matriptase and a
ketobenzothiazole serine trap support the transient nature of the covalent bond since dilution of a
high concentration enzyme-inhibitor complex results in the slow recovery of residual activity
over time [38]. Importantly, formation of a transient covalent bond between the catalytic serine
residue of the protease and the ketobenzothiazole serine trap was reported by others for thrombin
[47]. All these observations strongly suggest the actual formation of a covalent reversible bond
between ketobenzothiazole harboring compounds 2, 8 and 25 and matriptase-2.
The calculated Ki (in case of two-step inhibitors, the dissociation constant representing
the higher affinity complex, Ki*) gave similar results compared to the steady-state analysis (Table
5). Analogue 8 is the most potent matriptase-2 inhibitor, with a sub-nanomolar Ki* of 0.4 nM and
11

ACCEPTED MANUSCRIPT
a dissociation half-time of 0.8h (Table 5). The same analogue is also very potent for matriptase
with a calculated Ki* in the low nanomolar range (4 nM) and high dissociation half-time (1.2h).
From our experience with slow, tight binding inhibitors possessing Ki in the low nM range,
progress curve analysis over a long period yields more reliable results because it accounts for the
slow association and dissociation rate of the tightly bound complex. For analogues 2 and 25,
potency was still very good for matriptase-2, with Ki*s of 1.2 nM and 1.5 nM and dissociation
half-time of 0.4h and 0.3h, respectively. Interestingly, these two compounds do not follow the
two-step model of inhibition on matriptase, suggesting that the covalent, high affinity complex
may not be formed. This is reflected by Ki > 20nM and a dissociation half-time in the order of
seconds (Table 5). The selectivity calculated with Ki obtained from progress curves analysis with
these three compounds gave similar results to that calculated using steady-state cleavage rate.
These results confirm that the best analogues are potent, tightly bound matriptase-2 inhibitors
with long target dissociation half-time and mechanistically suggests the formation of a covalent-
reversible bond between inhibitor and the protease.
CONCLUSION. In summary, we present herein the SAR analysis of slow, tight binding
inhibitors of matriptase-2, a potential therapeutic target for the treatment of iron overload
disorders. Careful modifications to the P4, P3 and P2 positions with unnatural amino acids
increased selectivity for matriptase-2 vs matriptase, leading to a more thorough understanding of
the structural tenants of selectivity between the two enzymes. Two low-nM, selective inhibitors,
(H)F-hF-I-R-Kbt (21) and (H)F-hF-(allo)T-R-Kbt (29) possessing >55-fold selectivity in favour
of matriptase-2 were identified. Molecular modeling was instrumental to better rationalize
selectivity results, and kinetic parameters support a two-step inhibition consistent with the
12

ACCEPTED MANUSCRIPT
proposed mechanism. The use of a serine trap contributed to high affinity and long target
dissociation half-time, two characteristics that could be advantageous in a therapeutic context
[49]. Finally, replacement of residues in positions P4, P3 and P2 with unnatural amino acids
significantly increased plasma stability. This is the first report of semi-peptidic inhibitors
selective and potent for matriptase-2. Subsequent steps include to assess the modulation of
secreted hepcidin level by the best inhibitors in cultured human hepatocytes, and confirmation
of effect in vivo using murine models of iron overload, with measurements of circulating iron
and hepcidin levels. These will be reported in due course.
EXPERIMENTAL SECTION.
1. Synthetic procedures
General methods: Amino acids and coupling reagents were obtained from Chem-Impex
International (USA) and used as received. All other reagents and solvents were purchased from
Sigma-Aldrich (Canada) or Fisher Scientific (USA). Tetrahydrofuran (THF) was dried over
sodium benzophenone ketyl; DCM on over P2O5; Methanol over magnesium. Celite (AW
Standard Super-Cel® NF) was obtained from Sigma-Aldrich (Canada). Thin layer
chromatography was carried out on glass plates covered with silica gel (250 µm) 60 F-254
(Silicycle). Flash chromatography was carried out with Silicaflash® P60 (40-63 µm, Silicyle).
Chlorotrityl Chloride (CTC) Resin was obtained from Matrix Innovation and usually used with a
loading of 1.2 mmol/g. Reactions on resin were conducted in 60-mL polypropylene cartridges
(obtained from Applied Separations, Allentown, PA) and Teflon stopcocks. Reactors were gently
rocked on an orbital shaker at 172 rpm during solid phase chemistry. The resin was washed with
13

ACCEPTED MANUSCRIPT
the indicated solvent for 2-5 min with 10 mL solvent per gram of resin. Purity was analyzed on a
Waters UPLC H-Class with UV detection PDA equipped with an Acquity UPLC CSH C18
1.7µm 2.1 x 50 mm2 column. MS spectra were recorded on a Waters SQD 2 detector
(electrospray) instrument with a linear gradient of 5-95% CH3CN and H2O containing 0.1%
formic acid. Final products were purified to > 95% purity (UPLC-UV) using a Waters
Preparative LC (Sample Manager 2767 (Fraction collector), Binary gradient module 2545, with
two 515 HPLC pump and a System Fluidics Organizer (SFO), Photodiode Array Detector 2998:
column X Select CSH Prep C18 5 µm OBD 19 x 250 mm2 column, buffer: A: 0.1% HCOOH in
H2O, B: 0.1% HCOOH in ACN, Flow 20 mL/min). The gradient was 10-60% of acetonitrile at a
flow rate of 20 mL/min. All inhibitors were obtained as formic salts or TFA salts after
lyophilisation.
Scheme 1. General Solid-Phase method for the synthesis of tripeptide 1-29.
14

ACCEPTED MANUSCRIPT
The solid phase methodology was reported in Duchêne D. et al., J. Med. Chem, 2014, 57 (23),
10198-10204.
Scheme 2. Synthesis of tetrapeptidic analogues 1-29.
15

ACCEPTED MANUSCRIPT
Reagents and conditions: (a) HN(Me)OMe, HATU, DIPEA, THF, r.t., 92%; (b) Benzothiazole,
n-BuLi, THF, -78°C, 48%; (c) NaBH4, MeOH, -20°C, 74% (d) HCl/Dioxane, 1h, 99%; (e)
HATU, DIPEA, DMF, r.t., 60%, (f) Dess-Martin periodinane, DCM, quant., (g) TFA:H2O
(95:5), r.t. (h) reverses phase prep arative HPLC.
Warhead synthesis: Compound 32-36 was prepared as described in Duchêne et al., J. Med.
Chem, 2014, 57 (23), 10198-10204 with minor modification.
16

ACCEPTED MANUSCRIPT
Scheme 3. Synthesis of (H)Tyr(OtBu)-OH (40).
Reagents and conditions: (a) Acetyl chloride, MeOH, 0°C, quant.; (b) Isobutylene, sulfuric
acid, DCM, -78°C, 40%; (c) LiOH, Dioxane:H2O, MeOH, 90%.
Methyl 3-(4-hydroxyphenyl)propanoate (38): To a solution of 3-(4-hydroxyphenyl)propanoic
acid (2.50g, 15.04 mmol) in methanol (40 mL) were added dropwise 5.35 mL acetyl chloride
(5.0eq, 75.20 mmol) at 0°C. The reaction mixture was stirred over 1.5h at 0°C. After completion
of the reaction, the mixture was concentrated in vacuo and ethyl acetate was added. The organic
layer was washed with NaHCO3 (3 x 50 mL) and concentrated in vacuo to give methyl 3-(4-
hydroxyphenyl)propanoate 38 (2.70 g, 100%) as a colorless oil. UPLC-MS: 1.62 min, 100%. 1H
NMR (300 MHz, CD3OD) δ (ppm) 7.08-6.94 (m, 2H), 6.80-6.64 (m, 2H), 3.69-3.49 (m, 3H),
2.87-2.70 (m, 2H), 2.60-2.45 (m, 2H). 13C NMR (75 MHz, CD3OD) δ (ppm) 174.4, 155.8, 131.8,
129.3, 115.3, 51.1, 36.1, 30.2. MS (ES) m/z 181.7 (M+H)+. HRMS calculated for C10H12O3:
203.0679; found: 203.0679 (M+Na)+.
Methyl 3-(4-(tert-butoxy)phenyl)propanoate (39): To a solution of methyl 3-(4-
hydroxyphenyl)propanoate (4.50g, 24.99 mmol) in dry DCM (5 mL) in a sealed tube were added
a few drops of sulfuric acid at -78°C. Isobutylene (approximately 60 mL) was condensed at the
same temperature, the sealed tube was closed and the reaction mixture was stirred overnight at
room temperature. The sealed tube was opened and the mixture was allowed to warm up slowly
17

ACCEPTED MANUSCRIPT
to ambient temperature. The mixture was concentrated in vacuo and purified by flash
chromatography (Et2O/Hexane 10:90 to 30:80) to give methyl 3-(4-(tert-
1
butoxy)phenyl)propanoate 39 (2.36g, 40%) as a white solid. UPLC-MS: 1.62 min, 98.09%. H
NMR (300 MHz, CDCl3) δ (ppm) 7.12-7.06 (m, 2H), 6.93-6.87 (m, 2H), 3.67 (s, 3H), 2.92 (t,
13
2H), 2.62 (t, 2H), 1.33 (s, 9H). C NMR (75 MHz, CDCl3) δ (ppm) 173.4, 153.7, 135.3, 128.7,
124.2, 78.2, 51.6, 35.9, 30.3, 28.8. MS (ES) m/z 237.7 (M+H)+. HRMS: calculated for C14H20O3:
259.1305; found: 259.1310 (M+Na)+.
3-(4-(tert-butoxy)phenyl)propanoic
acid (40):
To
a
solution
of
3-(4-(tert-
butoxy)phenyl)propanoate (3.22g, 13.64 mmol) in a mixture of dioxane (40 mL), water (10 mL)
and methanol (3 mL), was added lithium hydroxide (816 mg, 34.09 mmol). The reaction mixture
was stirred overnight at room temperature. After completion of the reaction, the aqueous phase
was acidified to pH 6 and ethyl acetate was added to the mixture. The organic layer was washed
with HCl 0.1M (3x 50 mL), brine (1 x 50 mL) and concentrated in vacuo to give 3-(4-(tert-
1
butoxy)phenyl)propanoic acid 40 (2.72g, 90%) as a yellow oil. UPLC-MS: 1.46 min. 100%. H
NMR (300 MHz, CDCl3) δ (ppm) 7.15-7.07 (m, 2H), 6.96-6.89 (m, 2H), 2.93 (t, 2H), 2.67, (t,
13
2H), 1.34 (s, 9H). C NMR (75 MHz, CDCl3) δ (ppm) 179.1, 153.7, 135.0, 128.6, 124.2, 78.3,
35.7, 29.9, 28.8. HRMS: calculated for C13H18O3: 245.1148; found: 245.1146 (M+Na)+.
Scheme 4. Synthesis of Fmoc-(S)-(2,3-OH)Phe-OH (45).
18

ACCEPTED MANUSCRIPT
Reagents and conditions: (a) Fmoc-OSu, NaHCO3, dioxane:H2O, 12h, 75%; (b) Benzyl
bromide, K2CO3, DMF, 96%; (c) 2,2-Dimethoxypropane, APTS, benzene, reflux, 80%; (d) H2,
Pd/C 10%, THF, quant.
N-(9-Fluorenylmethoxycarbonyl)-3,4-dihydroxy-L-phenylalanine (42):[50]
Compound 42 was prepared as described in Song et al, J Med Chem, 2006, 49(5), 1585-1596.
After completion of the reaction, the aqueous layer was cooled on an ice bath and acidified with
6 N HCl to pH 3. This aqueous solution was extracted with ethyl acetate. The combined organic
layers were washed with brine and dried over anhydrous Na2SO4. Solvent was removed in vacuo.
The crude was loaded onto a silica gel column and eluted with DCM/Hexane 90:10 to
DCM/MeOH 96:4 to get N-Fmoc-L-Dopa-OH 42 as a white solid (4.8g, 75%). UPLC-MS: 1.75
1
min, 95.88%. H NMR (400 MHz, CD3OD) δ (ppm) 7.74 (d, 2H, J=8.0Hz), 7.56 (d, 2H,
J=8.0Hz), 7.35 (t, 2H), 7.30-7.22 (m, 2H), 6.75-6.72 (m, 1H), 6.70 (d, 1H, J=8.0Hz), 6.60-6.53
(m, 1H), 4.42-4.34 (m, 1H), 4.32-4.07 (m, 3H), 3.06 (dd, 1H, J=8.0, 4.0Hz), 2.86-2.77 (m, 1H).
13C NMR (101 MHz, CD3OD) δ (ppm) 176.1, 159.0, 146.9, 145.8, 143.1, 130.6, 129.4, 128.8,
127.0, 126.9, 126.7, 126.6, 122.4, 121.5, 118.8, 117.9, 116.9, 69.1, 68.7, 58.2, 57.6, 50.3, 50.2,
50.1, 49.9, 49.7, 49.4, 49.2, 49.0, 48.9, 39.1, 38.7. (ES) m/z 420.8 (M+H)+. HRMS: calculated
for C24H21NO6: 442.1261; found: 442.1267 (M+Na)+.
19

ACCEPTED MANUSCRIPT
Fmoc-L-Dopa-OBn (43): To Fmoc-L-Dopa-OH (10.6g, 25.29 mmoles) in DMF (300 mL) was
added 3.67g of potassium carbonate (1.05 eq, 26.55 mmol) at 0°C. The reaction was stirred for
10 minutes and 3.30 mL of benzyl bromide (1.1 eq, 27.82 mmol) was added dropwise. The
reaction mixture was stirred overnight at room temperature. After completion of the reaction,
ethyl acetate was added and the organic layer was washed with water (5 x 100 mL). The organic
layer was concentrated in vacuo and purified by flash chromatography (AcOEt/Hexane 20:80 to
60:40) to give N-Fmoc-L-Dopa-OBn 43 as a colorless oil (11g, 96%). UPLC-MS: 1.72 min,
100%. 1H NMR (400 MHz, CD3OD) δ (ppm) 7.75 (d, 2H, J=8.0Hz), 7.57 (d, 2H, J=8.0Hz), 7.35
(t, 2H), 7.31-7.19 (m, 7H), 6.71-6.67 (m, 1H), 6.65 (s, 1H), 6.54-6.48 (m, 1H), 5.09 (s, 2H),
4.45-4.38 (m, 1H), 4.28-4.22 (m, 2H), 4.16-4.09 (m, 1H), 2.98 (dd, 1H, J=8.0, 4.0Hz), 2.88-2.78
(m, 1H). 13C NMR (101 MHz, CD3OD) δ (ppm) 172.2, 157.0, 144.9, 143.9, 143.8, 143.7, 141.1,
135.7, 128.2, 128.1, 127.8, 127.4, 126.8, 124.9, 120.3, 119.5, 115.9, 115.0, 66.7, 66.5, 56.5, 56.0,
37.0, 36.7. (ES) m/z 510.9 (M+H)+. HRMS calculated for C31H27NO6: 532.1746; found:
532.1731 (M+Na)+.
(S)-benzyl
2-((((9H-fluoren-9-yl)methoxy)carbonyl)amino)-3-(2,2-dimethylbenzo[d]
[1,3]dioxol-5-yl)propanoate (44): To N-Fmoc-L-Dopa-OBn (2g, 3.93 mmol) in benzene (100
mL) was added 4.82 mL of 2,2-dimethoxypropane (10eq, 39.28 mmol) and 68 mg of p-
toluenesulfonic acid (0.1eq, 0.39 mmol). The reaction mixture was reflux with a Dean-Stark
overnight. After completion of the reaction, ethyl acetate was added and the organic layer was
washed with a saturated solution of NaHCO3 (3 x 70 mL). The organic layer was concentrated in
vacuo and purified by flash chromatography (AcOEt/Hexane 0:100 to 30:70) to give 44 (1.70g,
1
79%) as a white solid. UPLC-MS: 1.97 min, 100%. H NMR (400 MHz, CDCl3) δ (ppm) 7.82-
7.74 (d, 2H, J=8.0Hz), 7.63-7.55 (t, 2H), 7.47-7.29 (m, 9H), 6.60 (d, 1H, J=8.0Hz), 6.48 (s, 1H),
20

ACCEPTED MANUSCRIPT
6.42 (d, 1H, J=8.0Hz), 5.34 (d, 1H, J=8.0Hz), 5.19 (s, 2H), 4.75-4.65 (m, 1H), 4.49-4.42 (m,
13
1H), 4.41-4.34 (m, 1H), 4.24 (t, 1H), 3.05 (d, 2H, J=8.0Hz), 1.67 (s, 6H). C NMR (101 MHz,
CDCl3) δ (ppm) 171.4, 155.6, 147.7, 147.3, 146.7, 146.6, 143.9, 143.8, 141.3, 135.1, 128.7,
128.6, 128.5, 128.3, 127.7, 127.1, 126.0, 125.5, 125.2, 125.1, 124.9, 121.9, 120.0, 119.7, 119.3,
118.0, 110.3, 109.7, 109.5, 108.2, 67.4, 67.3, 67.2, 67.1, 55.6, 55.0, 54.9, 54.4, 47.2, 47.1, 38.3,
38.0, 35.8, 25.9, 25.8. (ES) m/z 580.9 (M+H)+. HRMS: calculated for C34H31NO6: 572.2032;
found: 572.2044 (M+Na)+.
(S)-2-((((9H-fluoren-9-yl)methoxy)carbonyl)amino)-3-(2,2-dimethylbenzo[d] [1,3]dioxol-5-
yl)propanoic acid (45): A solution of compound 44 (1.6g, 2.91 mmoles) and 10% Pd/C in
anhydrous THF (25 mL) was stirred under an atmospheric pressure of hydrogen at room
temperature for 14h. The catalyst was filtered off and the solvent evaporated to give compound
1
45 (1.3g, 100%). UPLC-MS: 1.72 min, 97.22%. H NMR (400 MHz, CDCl3) δ (ppm) 7.77 (d,
2H, J=8.0Hz), 7.57 (t, 2H), 7.41 (t, 2H), 7.36-7.29 (m, 2H), 6.69-6.62 (m, 1H), 6.60-6.52 (m,
2H), 5.24 (d, 1H, J=8,0Hz), 4.66 (quad, 1H), 4.52-4.42 (m, 1H), 4.42-4.34 (m, 1H), 4.27-4.16
(m, 1H), 3.16-3.01 (m, 2H), 1.66 (s, 6H). 13C NMR (101 MHz, CDCl3) δ (ppm) 175.6, 155.8,
147.7, 146.7, 143.8, 143.7, 141.3, 128.3, 127.8, 127.1, 125.1, 121.9, 120.0, 119.4, 118.1, 109.4,
108.3, 67.1, 54.6, 47.2, 37.4, 25.9, 25.8. (ES) m/z 460.8 (M+H)+. HRMS: calculated for
C27H25NO6: 482.1574; found: 482.1584 (M+Na)+.
21

ACCEPTED MANUSCRIPT
Scheme 5. Synthesis of 3-(naphtalen-2-yl)propanoic acid (46).
Reagents and conditions: (a) 2-naphtaldehyde, pyridine: piperidine, quant.; (b) H2, Pd/C,
AcOEt, MeOH, quant.
Compound 47-48 was synthesized as reported in Kita et al, J. Am. Chem. Soc., 2001, 123: 3214-
3222.
2. Biology
Plasma stability
27 µL of mouse sodium heparin plasma (Cedarlane CLC8001H) and 6 µL of analogues in
DMSO at a concentration of 5 mM (final concentration 0.3 mM) were incubated at 37 °C for 0.5,
1, 2, 6 and 24h. The proteolytic degradation was stopped by adding 70 µL of methanol. After
vortexing and centrifugation at 13000 rpm for 20 min at 4 °C, the supernatant was filtered on a 4
mm nylon 0.2 µm syringe filter and analyzed by UPLC-MS.
Purification of active human serine proteases
Matriptase and Matriptase-2 were expressed and purified as previously described [44,51]. Both
proteases were active-site titrated with the burst titrant 4-methylumbelliferyl-p-guanidino
benzoate (MUGB).
Ki determination using steady-state velocities
Enzymatic assays and Ki determination were performed at room temperature in an assay buffer
containing 50 mM Tris-HCl, 150 mM NaCl and 500 µg/mL BSA at pH 7.4, as described
previously.[38] To determine which method to use for the evaluation of inhibition potency for
22

ACCEPTED MANUSCRIPT
compounds 1-29 towards proteases, 0.25 nM protease was added to a reaction buffer containing
0 nM, 2.5 nM or 1 mM of inhibitors and 200µM of a fluorogenic substrate (Boc-Gln-Ala-Arg-
AMC). Proteolytic activity was monitored by measuring the release of fluorescence (excitation;
360 nm, emission; 441 nm) in a FLX800 TBE microplate reader (Bio-Tek Instruments,
Winooski, VT, USA).
If inhibition occurs only at I/E > 10, data generated from plots of enzyme velocity as a function
of substrate concentration at several inhibitor concentrations were fitted by nonlinear regression
to equations describing different models of reversible inhibition (competitive, uncompetitive,
non-competitive and mixed model). The preferred model was used for Ki determination.
If substantial inhibition occurred using a ratio I/E ≤ 10, compounds were treated as tight-binding
inhibitors. Plots of enzyme velocity as a function of inhibitor concentrations were fitted by
nonlinear regression analysis to the Morrison equation for Ki determination of tight-binding
inhibitors [52].
All assays were performed at least three times in duplicates, and data were presented as mean ±
standard error of the mean (SEM). Nonlinear regression and statistical analysis were performed
using GraphPad Prism version 6.02 for Windows (GraphPad Software, San Diego, CA, USA).
Ki determination using progress curve analysis
Matriptase and matriptase-2 cleavage of Boc-Gln-Ala-Arg-AMC was monitored (excitation; 360
nm, emission; 460 nm) 1200 min using a FLX-800 TBE microplate reader (Bio-Tek Instruments,
Winooski, VT, USA).
Equations representing one- and two-step mechanisms of reaction were used to fit the data from
the progress curves obtained in the presence of different inhibitor concentrations. Data fitting
23

ACCEPTED MANUSCRIPT
was performed using Dynafit version 4.07.066 [53]. The mechanisms and kinetic parameters are
shown in Figure 6.
When rapid equilibrium was assumed, the ON rates for ES and EI formation (k1 and k3) were
fixed at 100 µM-1 s-1, and k2 at 8,400 s-1 for matriptase, 11,660 s-1 for matriptase-2 to satisfy
experimental Km value (84µM and 116,6 µM respectively). Calculated Kcat were fixed to 9.52 s-1
for matriptase and 6.53 s-1 for matriptase-2.
Enzyme inactivation rate (kiE) and enzyme concentration ([E]) were determined by curve fitting
in the absence of inhibitor when fixing substrate concentration ([S]). Determined values were
used as fixed values to determine k3, k4, k5 and k6 values with kinetics in presence of inhibitors.
Inhibitor concentrations ([I]) were fitted except for the lowest concentration that was fixed.
For the two-step model, the inhibition constants were calculated as: Ki = k4/k3, Ki* = Ki k6/(k5+k6),
kon = k5 and koff = k4 k6 / (k4 + k5 + k6) The dissociation half-life of the enzyme-inhibitor complex
was calculated as t1/2 = 0.693 / koff. For the one-step model, kon and koff were equal to k3 and k4,
respectively.
3. Molecular modelling
Preparation of the protein target
The structure of Matriptase (PDB code 3NCL) was obtained from the RSCB Protein Data Bank
(www.pdb.org). Because there is no reported crystallographic structure for matriptase-2, a
homology model was built using MODELER [41] and GROMACS [42] as described by this
team [40]. Water molecules were removed from the structure.
Preparation of Ligand Structures
After drawing the structure, all protein-ligand complexes were prepared using the Protonate 3D
tool then the partial charges were calculated and the ligands were energy-minimized using the
24

ACCEPTED MANUSCRIPT
Molecular Operating Environment software (MOE 2013.08) from the Chemical Computing
Group [54].
Docking parameters
Molecules were docked in each protein with the software MOE2013.08. All atoms of Mat and
Mat-2 were fixed and the ligands were allowed to be flexible. The ketone carbon atom making
the reversible covalent bond with the protein was fixed at a distance of 3.0 ± 0.1 Å of the
catalytic serine to constrain the position of the ketobenzothiazole group within the binding site.
The guanidine of the Arg in P1 was also fixed via two key interactions in the binding site.
Conformational search using LowModeMD was made with AMBER10:EHT as a molecular
mechanics forcefield with default parameters (Rejection limit: 100, RMS Gradient: 0.05,
Conformation Limit: 10000 and Iteration Limit: 10000). At the end of docking, the low energy
conformations of the inhibitors-protein complex were analysed for their binding interactions.
25

ACCEPTED MANUSCRIPT
ASSOCIATED CONTENTS
Supporting Information
Characterization data for all compounds described in this article can be found at http://
Authors will release the atomic coordinates docking models and experimental data upon article
publication.
AUTHOR INFORMATION
Corresponding Authors
E-mail: Richard.Leduc@USherbrooke.ca
Phone: +819-821-8000 ext 75413
E-mail: Eric.Marsault@usherbrooke.ca.
Phone: +819-821-8000 ext 72433
Fax: 1 819-564-5400
ACKNOWLEDGMENT
The Canadian Institutes for Health Research, the Natural Sciences & Engineering Research
Council of Canada and the Canadian Foundation for Innovation are acknowledged for financial
support. EM, RJN and RL are members of the Québec Network for Drug Discovery
(www.rqrm.ca); EM and RJN are members of Proteo.
26

ACCEPTED MANUSCRIPT
ABBREVIATIONS.
AA, Amino Acid, ACN, Acetonitrile; APTS, p-Toluenesulfonic acid; Boc, tert-
Butyloxycarbonyl; DCM, Dichloromethane; DMF, Dimethylformamide; DIPEA, N,N-
Diisopropylethylamine;
Fmoc,
Fluorenylmethyloxycarbonyl;
HATU,
1-
(Bis(dimethylamino)methylene)-1H-1,2,3-triazolo(4,5-b)pyridinium 3-oxidhexafluorophosphate;
HFIP, Hexanfluoropropan-2-ol; iPrOH, Isopropanol; Kbt, Ketobenzothiazole; Pbf, 2,2,4,6,7-
Pentamethyldihydrobenzofuran-5-sulfonyl chloride; TFA, trifluoroacetic acid; THF,
Tetrahydrofuran; TLC, thin layer chromatography; TTSP, Type II transmembrane serine
proteases.
REFERENCES
[1]
[2]
[3]
[4]
L. Hedstrom, Serine Protease Mechanism and Specificity, Chem. Rev. 102 (2002) 4501–
4524.
T.M. Antalis, T.H. Bugge, Q. Wu, Membrane-anchored serine proteases in health and
disease., Prog. Mol. Biol. Transl. Sci. 99 (2011) 1–50.
T.H. Bugge, T.M. Antalis, Q. Wu, Type II transmembrane serine proteases., J. Biol.
Chem. 284 (2009) 23177–23181.
L. Silvestri, A. Pagani, A. Nai, I. De Domenico, J. Kaplan, C. Camaschella, The serine
protease matriptase-2 (TMPRSS6) inhibits hepcidin activation by cleaving membrane
hemojuvelin., Cell Metab. 8 (2008) 502–511.
[5]
A.J. Ramsay, J.C. Reid, G. Velasco, J.P. Quigley, J.D. Hooper, The type II
transmembrane serine protease matriptase-2--identification, structural features,
27

ACCEPTED MANUSCRIPT
enzymology, expression pattern and potential roles, Front. Biosci. 13 (2008) 569–579.
[6]
G. Papanikolaou, M.E. Samuels, E.H. Ludwig, M.L.E. MacDonald, P.L. Franchini, M.-P.
Dubé, L. Andres, J. MacFarlane, N. Sakellaropoulos, M. Politou, E. Nemeth, J.
Thompson, J.K. Risler, C. Zaborowska, R. Babakaiff, C.C. Radomski, T.D. Pape, O.
Davidas, J. Christakis, P. Brissot, G. Lockitch, T. Ganz, M.R. Hayden, Y.P. Goldberg,
Mutations in HFE2 cause iron overload in chromosome 1q–linked juvenile
hemochromatosis, Nat. Genet. 36 (2004) 77–82.
[7]
[8]
[9]
V. Niederkofler, R. Salie, S. Arber, Hemojuvelin is essential for dietary iron sensing, and
its mutation leads to severe iron overload., J. Clin. Invest. 115 (2005) 2180–2186.
F.W. Huang, J.L. Pinkus, G.S. Pinkus, M.D. Flemming, N.C. Andrews, A mouse model of
juvenile hemochromatosis, J. Clin. Invest. 115 (2005) 2187–2191.
B. Andriopoulos, E. Corradini, Y. Xia, S.A. Faasse, S. Chen, L. Grgurevic, M.D. Knutson,
A. Pietrangelo, S. Vukicevic, H.Y. Lin, J.L. Babitt, BMP6 is a key endogenous regulator
of hepcidin expression and iron metabolism., Nat. Genet. 41 (2009) 482–487.
[10] P. Celec, Hemojuvelin: a supposed role in iron metabolism one year after its discovery., J.
Mol. Med. (Berl). 83 (2005) 521–525.
[11] D. Meynard, L. Kautz, V. Darnaud, F. Canonne-Hergaux, H. Coppin, M.-P. Roth, Lack of
the bone morphogenetic protein BMP6 induces massive iron overload., Nat. Genet. 41
(2009) 478–481.
[12] J.J. Sekelsky, S.J. Newfeld, L.A. Raftery, E.H. Chartoff, W.M. Gelbart, Genetic
characterization and cloning of mothers against dpp, a gene required for decapentaplegic
28

ACCEPTED MANUSCRIPT
function in Drosophila melanogaster., Genetics. 139 (1995) 1347–1358.
[13] C. Savage, P. Das, A.L. Finelli, S.R. Townsend, C.Y. Sun, S.E. Baird, R.W. Padgett,
Caenorhabditis elegans genes sma-2, sma-3, and sma-4 define a conserved family of
transforming growth factor beta pathway components., Proc. Natl. Acad. Sci. U. S. A. 93
(1996) 790–794.
[14] G. Ramey, J.-C. Deschemin, B. Durel, F. Canonne-Hergaux, G. Nicolas, S. Vaulont,
Hepcidin targets ferroportin for degradation in hepatocytes., Haematologica. 95 (2010)
501–504.
[15] M.A. Melis, M. Cau, R. Congiu, G. Sole, S. Barella, A. Cao, M. Westerman, M. Cazzola,
R. Galanello, 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. 93 (2008) 1473–1479.
[16] K.E. Finberg, M.M. Heeney, D.R. Campagna, Y. Aydınok, H.A. Pearson, K.R. Hartman,
M.M. Mayo, S.M. Samuel, J.J. Strouse, K. Markianos, N.C. Andrews, M.D. Fleming,
Mutations in TMPRSS6 cause iron-refractory iron deficiency anemia (IRIDA), Nat.
Genet. 40 (2008) 569–571.
[17] A.R. Folgueras, F.M. De Lara, A.M. Pendás, C. Garabaya, F. Rodríguez, A. Astudillo, T.
Bernal, R. Cabanillas, C. López-Otín, G. Velasco, Membrane-bound serine protease
matriptase-2 (Tmprss6) is an essential regulator of iron homeostasis, Blood. 112 (2008)
2539–2545.
[18] A.A. Enein, N.A. El Dessouky, K.S. Mohamed, S.K.A. Botros, M.F. Abd El Gawad, M.
29