Investigation of mechanism-based thrombin inhibitors: Implications of a highly conserved water molecule for the binding of coumarins within the S pocket

Article abstract of DOI:10.1016/j.bmcl.2005.12.070

The synthesis of novel coumarins bearing on the lateral side chain in the 3-position an amine or a guanidine group is described. In vitro evaluation highlighted 14d which possesses a meta aniline side chain as a very potent THR inhibitor. Surprisingly, the introduction of a guanidine moiety always led to a decrease in THR inhibiting properties. We, thus, used docking experiments to rationalize the SAR in the series. This study showed the crucial role of a conserved water molecule in the specificity pocket of THR during docking simulation in order to explain the inactivity of guanidine derivatives.

Full text of DOI:10.1016/j.bmcl.2005.12.070

Bioorganic & Medicinal Chemistry Letters 16 (2006) 2017–2021
Investigation of mechanism-based thrombin inhibitors:
Implications of a highly conserved water molecule for the binding
of coumarins within the S pocket
a,
a
b
Raphael Frederick, Caroline Charlier,b, Severine Robert, Johan Wouters,
´ ´
´
¨
Bernard Masereela and Lionel Pocheta,*
aDepartment of Pharmacy, Drug Design and Discovery Center, University of Namur, FUNDP,
61, rue de Bruxelles, B-5000 Namur, Belgium
bDepartment of Chemistry, Drug Design and Discovery Center, University of Namur, FUNDP,
61, rue de Bruxelles, B-5000 Namur, Belgium
Received 4 November 2005; revised 19 December 2005; accepted 19 December 2005
Available online 18 January 2006
Abstract—The synthesis of novel coumarins bearing on the lateral side chain in the 3-position an amine or a guanidine group is
described. In vitro evaluation highlighted 14d which possesses a meta aniline side chain as a very potent THR inhibitor. Surprisingly,
the introduction of a guanidine moiety always led to a decrease in THR inhibiting properties. We, thus, used docking experiments to
rationalize the SAR in the series. This study showed the crucial role of a conserved water molecule in the specificity pocket of THR
during docking simulation in order to explain the inactivity of guanidine derivatives.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Thrombin (THR) is a trypsin-like serine protease which
plays a pivotal role in the process of haemostasis and
thrombosis. Besides exerting multifunctional activities
in the coagulation cascade, it is also one of the main
activators of platelet secretion and aggregation.1,2 So,
THR has earlier been recognized as a key target for
the development of new antithrombotics.
departure of the chlorine atom promotes the formation
of a highly reactive quinone methide which could be
alkylated and thus leads to the irreversible inactivation
of the enzyme.
In the present report, we investigated the synthesis and
structure–activity relationships (SAR) of novel couma-
rins bearing an amine or a guanidine on the ester lateral
side chain in the 3-position. These basic moieties could
advantageously interact with the negatively charged
Asp189 in the specificity (S) pocket. This strategy should
therefore lead to new potent THR inhibitors. Pyridinic
derivatives have been also investigated.
As part of a project aiming at the development of cou-
marins as selective serine protease inhibitors, we have
recently described a series of coumarins as THR and
factor Xa (FXa) inhibitors.3 They are characterized by
a chloromethyl moiety in the 6-position and a hydro-
phobic alkyl-, aryl-, heteroaryl-ester, amide or thioster
in the 3-position. These compounds were found to act
as mechanism-based inhibitors. The first step of their
inhibition mechanism consists in the nucleophilic attack
by the activated hydroxyl group of Ser195 on the lac-
tone moiety, leading to the acyl-enzyme. Then, the
The synthetic routes of the newly designed coumarins
are depicted on Schemes 1 and 2. The introduction of
the ester lateral side chains on the 3-position needs the
preparation of the protected guanidino (5a–d) and ami-
no alcohols (6a–f). Among the various methods devel-
oped to prepare Boc-protected guanidines,4,5 we chose
the N,N0-bis(tert-butoxycarbonyl)-N00-triflylguanidine 3
which allows to access the alcohols (5a–d) in one step,
starting from the commercially available amino deriva-
tives 4.6,7 The intermediate 3 was easily obtained from
the guanidinium hydrochloride 1, by reaction with di-
Keywords: Thrombin; Coumarins; Docking; Serine protease; Mecha-
nism-based inhibitor.
*
Corresponding author. Tel.: +32 81 72 42 92; fax: +32 81 72 42 99;
e-mail: lionel.pochet@fundp.ac.be
These authors contributed equally to this work.
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.12.070

´ ´
R. Frederick et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2017–2021
2018
NTf
NHBoc
NH2
NH2
NH
Cl
a
b
BocHN
H2N
BocHN
NHBoc
1
3
2
X
HO
X NH
3, c
a
b
c
d
e
f
2-ethyl
3-propyl
p-phenyl
m-phenyl
p-cyclohexyl
4-piperidine
NHBoc
BocN
5a-d
HO
X
NH2
4
HO
X
NHBoc
d
6a-f
Scheme 1. Reagents and conditions: (a) Boc2O, NaOH, 0 ꢁC; (b) Tf2O, Et3N, ꢀ78 ꢁC; (c) CH2Cl2, rt; (d) Boc2O, dioxane, rt, 2 h.
O
O
O
O
O
O
HO
OC2H5
a
c, d
H
b
HO
H
Cl
Cl
O
OH
OH
O
7
8
9
10
O
O
NH
NH2
O
O
NBoc
N NHBoc
X
X
Cl
O
N
H
Cl
O
f
H
X
a
b
c
d
e
f
2-ethyl
3-propyl
p-phenyl
m-phenyl
p-cyclohexyl
4-piperidine
O
O
11 a-d
e
13 a-d
O
O
O
X
f
X
Cl
O
NH2
Cl
O
NHBoc
O
O
O
14 a-f
12 a-f
Scheme 2. Reagents and conditions: (a) HCOH, HCl, 80 ꢁC, 20 min; (b) diethyl malonate, piperidine, AcOH, EtOH, reflux, 17 h; (c) HCl (3 N),
EtOH, reflux, 3 h; (d) SOCl2, reflux, 3 h; (e) 5a–d or 6a–f, dioxane, pyridine, rt, 2 h; (f) TFA, CH2Cl2 (1:1), rt, 10 min.
tert-butyl-dicarbonate and sodium hydroxide at 0 ꢁC
followed by reaction at ꢀ78 ꢁC with triflic anhydride
and Et3N (Scheme 1).8 The Boc-protected amino alco-
hols (6a–f) were synthesized by reaction of the hydrox-
yl-amino (4a–f) with di-tert-butyl-dicarbonate.
However, the introduction of an amine moiety on linear
(14a–b) or cyclic (14e–f) aliphatic side chains always
leads to a decrease in THR inhibitory potency. Com-
pound 14d, which possesses a meta-aminophenyl as side
chain in the 3-position, is the best compound in this ser-
ies with an IC50 value of 1.98 lM. For comparison pur-
pose, we determined their kinetic parameters ki and KI
(Table 2).12 14d possesses a ki/KI ratio equal to
3570 Mꢀ1 sꢀ1 and thus appears to be about 10-fold less
potent than 16, the best THR inhibitor of our previous
study.3 Moreover, the introduction of basic moieties
such as guanidine, which proved to be very efficient on
other series,13 leads to an important decrease in THR
inhibitory potency.
Then, the suitable N-protected alcohol was reacted with
the acyl halide (10), obtained by a previously described
procedure,9 to afford (11a–d) and (12a–f) (Scheme 2).8
Finally, deprotection of derivatives (11a–d) and (12a–
d) led to the targeted guanidines (13a–d) and amines
(14a–f).8
Pyridinic compounds 15a–e were obtained according to
our previously published procedure.9,10
In the pyridine series (15a–e), only derivatives possessing
the pyridine nitrogen in the 20-position (15a–b) inhibit
THR and particularly when a chlorine atom is located
in the 30-position.
Table 1 summarizes the THR inhibitory potency of the
newly synthesized compounds.11 Surprisingly, guanidine
derivatives (13a–d) are found to be almost inactive. Only
13a and 13c, bearing, respectively, a 2-guanidino-ethyl
and a 4-guanidino-phenyl as side chain in the 3-position,
possess a weak THR inhibitory potency. On the con-
trary, in the amino series, potent THR inactivators are
obtained when the amine group is introduced on a phen-
yl ring (14c–d), particularly in the meta-position (14d).
With a view to helping in the understanding of these sur-
prising results, we explored the binding mode of repre-
sentative compounds within the THR active site by
means of molecular modelling. Derivatives 13c, 13d,
14c and 14d were docked into the THR cavity (from

´ ´
R. Frederick et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2017–2021
2019
Table 1. Inhibitory effect on THR and physical and synthetic data for compounds 13a–d, 14a–f and 15
O
R
Cl
O
O
O
b
Compound
R
Yielda (%)
67
mp (ꢁC)
IC50 (lM)
H2N
13a
141–143
107 (23–495)
>300
NH
N
H
H2N
13b
13c
13d
78
15
11
144–145
157–161
137–140
NH
N
H
NH
151 (69–330)
NH2
HN
NH2
NH
HN
>300
NH2
14a
14b
72
86
149–151
127–129
284 (137–589)
>300
NH2
14c
14d
22
15
280–288
335c
48 (25–93)
NH2
NH2
1.98 (0.76–5.17)
14e
14f
76
45
215–218
210–213
>300
>300
NH2
NH
15a
15b
15c
15d
15e
—
—
—
—
—
—
—
—
—
—
0.87 (0.30–2.01)
N
N
Cl
1.39 (0.64–3.01)
CH3
N
>300
nid
Cl
Cl
N
Br
nid
N
a Overall yield on steps e, f (Scheme 2).
b Calculated by nonlinear regression from dose–response curves. Values in parentheses are 95% confidence intervals.
c Decomposition temperature.
d ni, no inactivator at maximum solubility.

´ ´
R. Frederick et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2017–2021
2020
Table 2. Comparison of the kinetic parameters ki and KI for
derivatives 14d and 16
formed in the presence of H2O395. This latter was al-
lowed to switch on and off during the simulation, that
is, GOLD predicted if the molecule was displaced or
not by the inhibitors.
O
R
Cl
O
O
O
As can be seen in Figure 1, all the complexes involve the
conserved water molecule, underlining the importance
to consider it in the docking of coumarins. In the amino
series (Fig. 1a), both compounds (14c, d) adopt a similar
orientation within the THR active site. The phenyl ring
of coumarin favourably interacts with the hydrophobic
proximal (P) pocket, composed of Tyr60A and Trp60D.
The chloromethyl moiety fills in part the distal binding
(D) pocket, interacting through hydrophobic contact
with Trp215. The main difference in the binding of the
two compounds arises from the specificity (S) pocket.
While both aniline groups are stabilized through H-
bonding with Asp189, an additional H-bond with
Gly219 or with H2O395 (Fig. 1a; represented in cyan)
is observed for the meta (14d) and para (14c) derivatives,
respectively. In both complexes, the lactone carbonyl
moiety is located next to the catalytic Ser195, allowing
the formation of the acyl-enzyme. Indeed, the distance
Compound
R
THR ki/KI (Mꢀ1 sꢀ1
)
14d
30-Aminophenyl
3570 655
(0.0020 sꢀ1, 0.548 lM)
37,000 5150
(0.016 sꢀ1, 0.424 lM)
16a
20,50-Dichlorophenyl
a Compound 30 in Ref. 3.
PDB entry 1A4W) using the automated GOLD
program.14 To take protein flexibility into account, the
selected enzyme–inhibitor complexes were further
refined by molecular mechanics using the DISCOVER3
module15 (CVFF forcefield) from INSIGHTII.16
Initially, the docking study was conducted on the THR
active site, in the absence of water molecules. This
unfortunately did not allow to explain the observed bio-
logical trends, that is, the inactivity of guanidine deriva-
tives (13c, d) versus the activity of amine compounds
(14c, d). According to a recent paper17 that highlighted
the importance of water in the binding properties of
inhibitors, especially in the case of FXa, another tryp-
sin-like serine protease structurally related to THR, we
thought about the possible role played by water in the
binding of these coumarins to THR. Therefore, we com-
pared the X-ray structures of several THR-inhibitor
complexes (PDB entries: 1A4W, 1BA8, 1OYT, 1QHR,
1SB1, 1EB1, 8KME and 1D9I). Their superimposition
revealed a highly conserved water molecule (H2O395),
located just above the aromatic ring of Tyr228 (distance
˚
of 4.2 and 4.0 A between the lactone carbonyl C atom
of 14c and 14d, respectively, and the hydroxyl O atom
of Ser195, activated via H-bonding with the carbonyl
O atom of exocyclic ester, is consistent with the nucleo-
philic attack of this residue onto the lactone, as de-
scribed in the inhibition mechanism.3,18 As both
compounds are equally stabilized within the THR active
site, and exhibit a proper orientation with regard to
Ser195 (for acyl-enzyme formation), no direct conclu-
sion can be derived from these results that would explain
the small difference in potency of 14c and 14d.
In the guanidine series (Fig. 1b), 13c and 13d adopt two
distinct binding modes within the THR active site. Both
guanidinophenyls fill the S pocket and are stabilized by
several H-bonds involving Gly219, Asp189 and H2O395,
˚
between the O atom and the phenyl centroid of ꢁ3.5 A)
in the S pocket. New docking studies were thus per-
Figure 1. Superimposition of the docking simulation for (a) 14c (cyan) and 14d (orange), (b) 13c (cyan) and 13d (orange) and (c) 16 (cyan) and 15a
(orange). For (a) and (b), the conserved water molecule and H-bonds are represented in the colour of the inhibitor with which they interact. For (c)
the red sphere represents the position of the conserved water molecule before its ejection of the active site during the docking simulation.

´ ´
R. Frederick et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2017–2021
2021
3. Frederick, R.; Robert, S.; Charlier, C.; de Ruyck, J.;
Wouters, J.; Pirotte, B.; Masereel, B.; Pochet, L. J. Med.
Chem. 2005, 48, 7592.
4. Bernatowicz, M. S.; Wu, Y.; Matsueda, G. R. Tetrahedron
Lett. 1993, 34, 3389.
5. Kim, K. S.; Qian, L. Tetrahedron Lett. 1993, 34, 7677.
6. Feichtinger, K.; Sings, H.; Baker, T.; Matthews, K.;
Goodman, M. J. Org. Chem. 1998, 63, 8432.
7. Feichtinger, K.; Zapf, C.; Sings, H.; Goodman, M. J. Org.
Chem. 1998, 63, 3804.
whereas the coumarinic nucleus presents different orien-
tations. In 13d (Fig. 1b, coloured in orange), the two
carbonyl moieties are in the same direction (syn-confor-
mation) and the coumarin ring and chloromethyl group
interact favourably within the P and D pockets, respec-
tively. In 13c (Fig. 1b, coloured in cyan) in contrast, the
carbonyl moieties lie in opposition (anti-conformation)
and the coumarin ring interacts solely with the D pock-
et, and especially Trp215.
8. Experimental details: (a) To a solution of 10 (2.27 mmols)
generated in situ according to known procedure, in dry
dioxane (5 mL), is added the alcohol 5a–d, 6a–f (1.1 equiv)
and dry pyridine (1.1 equiv) and the mixture was stirred
vigorously during 90 min. Then the solvents were
removed, the residue redissolved in chloroform
(200 mL), washed (3·) with 0.1 N HCl and brine and
then dried over MgSO4 and evaporated. This allowed to
obtain 11a–d, 12a–f which were recrystallized either in hot
AcOEt or acetonitrile. Deprotection was then realized by
dissolving 11a–d, 12a–f in a 50:50 (v/v) mixture of
methylene chloride/trifluoroacetic acid to generate 13a–d,
14a–f. Compound 14d: overall yield: 15%; mp: 280–
288 ꢁC; 1H NMR (DMSO-d6, 400 MHz) d 4.88 (2H, s)
6.40 (1H, d, J = 8 Hz) 6.67 (1H, s) 6.73 (1H, d, J = 8 Hz)
7.24 (1H, t, J = 8 Hz) 7.51 (1H, d, J = 9.2 Hz) 7.85 (dd,
1H, J = 9.2 Hz, J = 2 Hz) 8.05 (1H, d, J = 2 Hz) 9.02 (1H,
s); ESI-MS m/z (MH)+ = 330.1 (MMH)+ = 659.0. Com-
pound 13a: 67%; mp: 141–143 ꢁC ;1H NMR (DMSO-d6,
400 MHz) d 3.54 (m, 2H), 4.34 (t, 2H, J = 5.6 Hz), 4.87 (s,
2H), 7.27–7.31 (m, 4H) 7.49 (d, 1H, J = 8.4 Hz), 7.83 (m,
2H), 7.96 (d, 1H, J = 2.4 Hz), 8.81 (s, 1H); ESI-MS m/z
(MH)+ = 324.1. (b) To a solution of N,N0-bis(tert-butox-
In comparison to amino derivatives, the bulky guanidine
forces the coumarin ring to move away from the catalyt-
ic Ser195, increasing the distance between the lactone
carbonyl C atom and the hydroxyl O atom of Ser195
(distance of 8.01 and 6.43 A for 13c and 13d, respective-
ly). This clearly accounts for the inactivity of guanidine
versus amine derivatives.
˚
For comparison purpose, we performed the docking of
15a and 16 allowing GOLD to predict if H2O395 was dis-
placed or not (Fig. 1c). It appeared that the binding of 15a
and 16 produces the removal of water and occupation of
its binding site by the chlorine atom in the meta-position.
This observation supports our previous results which de-
scribed a lipophilic contact between this chlorine atom
and the aromatic ring of Tyr228 using GOLD with the
THR active site deprived of water.3 Moreover, water dis-
placement allows 15a and 16 to adopt a more favourable
binding mode in which the distance between the lactone
carbonyl C atom and the hydroxyl O atom of Ser195 is
ycarbonyl)-N00-triflylguanidine
3 (2.5 mmol) (prepared
˚
3.2A. This could explain the difference observed in the
rate of THR inactivation of 16 versus 14d (ki = 0.016 sꢀ1
and 0.0020 sꢀ1 for 16 and 14d, respectively).
using the procedure described in Refs. 6 and 7) in dry
methylene chloride (10 mL) is added Et3N (1.1 equiv)
under an argon atmosphere. To this stirred solution are
then added the amino derivatives 4 at room temperature.
After 2 h, the mixture is diluted with methylene chloride
(10 mL), washed twice with a saturated solution of
NaHSO4, once with brine, dried over MgSO4 and evapo-
rated. This allowed to obtain the guanidine derivatives
5a–d which are purified by liquid chromatography on silica
gel (eluent: CH2Cl2 100%). Compound 5a: yield: 85%; mp:
In conclusion, we described the synthesis of novel 3,6-di-
substituted coumarins which possess an amine or a gua-
nidine moiety in the 3-position. The best derivative (14d)
possesses a meta-aminophenyl side chain and displays a
high THR inhibitory potency. Nevertheless, surprisingly
the introduction of a guanidine moiety always led to a
decrease in THR inhibiting properties. Rationalization
of structure–activity relationships (SAR) in the series
using molecular modelling revealed the need to consider
a highly conserved water molecule (H2O395) within the
S cavity of THR to properly model the binding of cou-
marins bearing a basic moiety in the 3-position. This
water molecule is of prime importance when using dock-
ing simulation to rationalize SAR, particularly in a ser-
ies of mechanism-based inhibitors.
1
98–100 ꢁC; H NMR (CDCl3, 400 MHz) d 1.50 (s, 18H),
3.24 (m, 2H), 4.40 (t, 2H), 8.50 (s, 1H), 11.27 (s, 1H).
9. Pochet, L.; Doucet, C.; Schynts, M.; Thierry, N.; Bogg-
etto, N.; Pirotte, B.; Jiang, K. Y.; Masereel, B.; de Tullio,
P.; Delarge, J.; Reboud-Ravaux, M. J. Med. Chem. 1996,
39, 2579.
10. Doucet, C.; Pochet, L.; Thierry, N.; Pirotte, B.; Delarge,
J.; Reboud-Ravaux, M. J. Med. Chem. 1999, 42, 4161.
11. Inhibition assay was adapted from the previously
described procedure3 with an incubation time of 10 min
for the enzyme/inhibitor reaction.
12. For experimental details, see Ref. 3.
13. Vacca, J. P. Curr. Opin. Chem. Biol. 2000, 4, 394.
14. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor,
R. J. Mol. Biol. 1997, 267, 727.
15. Discover 3, version 2.98; Accelrys Inc.: San Diego,
CA, 1998.;
Acknowledgments
´ ´
´
R. Frederick and C. Charlier thank the ‘Fonds Special
de Recherche’ (FUNDP, Namur, Belgium) and ‘Fonds
National de la Recherche Scientifique’ (FNRS, Brux-
elles, Belgium), respectively, for financial support.
16. Insight II, version 2000; Accelrys Inc.: San Diego,
CA, 2000.;
17. Verdonk, M. L.; Chessari, G.; Cole, J. C.; Hartshorn, M.
J.; Murray, C. W.; Nissink, J. W.; Taylor, R. D.; Taylor,
R. J. Med. Chem. 2005, 48, 6504.
References and notes
´ ´
18. Pochet, L.; Dieu, M.; Frederick, R.; Murray, A. M.;
Kempen, I.; Pirotte, B.; Masereel, B. Tetrahedron 2003,
59, 4557.
1. Mann, K. G. Thromb. Haemost. 1999, 82, 165.
2. Dahlback, B. Lancet 2000, 355, 1627.