Factor Xa inhibitors based on a 2-carboxyindole scaffold: SAR of neutral P1 substituents

Article abstract of DOI:10.1016/j.bmcl.2004.06.020

A series of novel, highly potent 2-carboxyindole-based factor Xa inhibitors is described. Structural requirements for neutral ligands, which bind in the S1 pocket of factor Xa were investigated with the 2-carboxyindole scaffold. This privileged fragment assembly approach yielded a set of equipotent, selective inhibitors with structurally diverse neutral P1 substituents.

Full text of DOI:10.1016/j.bmcl.2004.06.020

Bioorganic & Medicinal Chemistry Letters 14 (2004) 4191–4195
Factor Xa inhibitors based on a 2-carboxyindole scaffold: SAR of
neutral P1 substituents
ꢀ *
Marc Nazare, Melanie Essrich, David W. Will, Hans Matter, Kurt Ritter,
€
Matthias Urmann, Armin Bauer, Herman Schreuder, Angela Dudda, Jorg Czech,
Martin Lorenz, Volker Laux and Volkmar Wehner
Aventis Pharma Deutschland GmbH, D-65926 Frankfurt am Main, Germany
Received 27 April 2004; accepted 8 June 2004
Available online
Abstract—A series of novel, highly potent 2-carboxyindole-based factor Xa inhibitors is described. Structural requirements for
neutral ligands, which bind in the S1 pocket of factor Xa were investigated with the 2-carboxyindole scaffold. This privileged
fragment assembly approach yielded a set of equipotent, selective inhibitors with structurally diverse neutral P1 substituents.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
linked to the extremely poor bioavailability observed for
the inhibitors bearing this functionality.4 Here we report
the synthesis of novel, potent and selective fXa inhibi-
tors based on a 2-carboxyindole scaffold, which contain
neither a benzamidine nor a guanidine moiety, and
incorporate a variety of neutral P1 ligands directed
towards the fXa S1 pocket.
The currently available anticoagulants for the treatment
and prevention of thromboembolic diseases including
deep vein thrombosis, myocardial infarction and pul-
monary embolism are handicapped by several short-
comings, including slow onset of action, mode of
administration and severe, potentially life-threatening
side effects like excessive bleeding requiring stringent
monitoring of drug-levels.1 The serine protease factor
Xa (fXa) has emerged as an attractive target for the
therapy of thrombosis-related diseases since it is a key
enzyme in the activation cascade of the coagulation
system linking the extrinsic and intrinsic activation
pathways.2 It is anticipated that inhibition of factor Xa
should prevent thrombus formation without compro-
mising normal hemostasis and platelet function.3 An
orally available fXa inhibitor should therefore be a
superior antithrombotic agent overcoming the short-
comings associated with the current treatment. A major
obstacle to the development of an orally available drug
from the many potent and selective inhibitors available
is often the need to incorporate a highly basic benz-
amidine or guanidine serving as an arginine mimetic in
the S1 pocket of fXa. In general these moieties are
Figure 1 gives a schematic representation of the design
rationale for the 2-carboxyindole fXa inhibitors. In a
previous communication5 we described the optimization
of a benzoic acid scaffold incorporating a neutral P1
ligand, which interacts in the S1 pocket via the so-called
chloro binding mode.6 We succeeded in reaching low
nanomolar activity against fXa. However, concurrently
we were intrigued by the additional opportunities the
identification of alternative scaffolds with different
topology to the 1,3 connectivity of the P1 and P4 groups
of the benzoic acid series would present. We suspected
that modification of both the central scaffold and the
connectivity pattern might allow us to incorporate
different S1 and S4 directed ligands, and thus open
alternative routes to optimize physicochemical and
pharmacological profiles while maintaining fXa binding
affinity. In addition the recently described7 successful
use of ketopiperazines as scaffolds with a 1,4-connective
relationship of P1 and P4 substituents indicated that the
investigation of alternative connectivities could be a
promising approach. We decided, therefore, to investi-
gate N-alkylated 2-indolecarboxamides as nonchiral,
Keywords: Factor Xa inhibitors.
* Corresponding author. Tel.: +49-69-305-15150; fax: +49-69-331-399;
e-mail: marc.nazare@aventis.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.06.020

4192
M. Nazarꢀe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4191–4195
O
NH2
N
NHBoc
O
H
(a), (b)
N
NH
N
A:
1
O
N
O
P4
N
H
S
N
N
O
O
S
P4
N
Ki(fXa) = 18 nM
Ki(fXa) = 4 nM
Cl
Cl
Cl
P1
1,4-Connectivity
Neutral P1 ligand
Br
1,3-Connectivity
Neutral P1 ligand
P1
NH2
HN
Cl
O
(c)
N
B:
C:
N
2
Cl
O
N
N
H
N
Br
O
O
S
P1 ligand
O
O
O
O
(f), (g)
(e)
(d)
N
O
N
O
S
O
1,2-Connectivity
O
O
Cl
S
3
5
4
S
Amidino-P1 ligand
N
N
H
N
N
H
Cl
Cl
Cl
NH
NH2
Scheme 1. Reagents and conditions: (a) acetone, NaBH3CN, AcOH,
methanol, rt, 16 h; (b) HCl/MeOH (8 M), rt, 16 h; (c) BrCH2COBr,
pyridine, toluene, rt, 2 h; (d) KOtBu, then oxalic acid diethyl ester,
toluene, rt 3 h; (e) NH2OHꢀHCl, ethanol, reflux, 6 h; (f) NaBH4, eth-
anol, 0 °C, 3 h; (g) NBS, PPh3, CH2Cl2, rt, 1 h.
N+
P1
P4
Ki(fXa) = 19 nM
P1
P4
NH
NH
H2N
H2N
Ki(fXa) = 90 nM
Figure 1. Rationale for the design of 2-carboxyindole factor Xa
inhibitors with neutral P1 ligands.
indole-2-carboxylic acid ethyl ester (Scheme 2). N-aryl-
ated compounds 8–10 were synthesized according to
route 1 by Cu(OAc)2 mediated coupling of aryl boronic
acids to the indole nitrogen.9 The N-alkylated precur-
sors were synthesized under standard conditions using
the corresponding bromides or tosylates in the presence
of NaH in DMF. Saponification of the ester 6 was fol-
lowed by amide coupling of the acid with isopropyl-
rigid a-amino acid mimetics offering an alternative and
potentially superior trajectorial orientation for the P1/
P4 ligands. This scaffold had previously been reported
in combination with less favourable benzamidine S1
ligands.8 In addition, in terms of library generation the
chemistry for the decoration of 2-carboxyindoles has the
advantage of being very flexible, enabling the attach-
ment of the respective ligands first at the indole nitrogen
or alternatively at the carboxy group.
Here we describe 2-carboxyindole derivatives as fXa
inhibitors, with special emphasis on the SAR of the P1
ligand.
O
N
H
O
(a)
(b), (c)
O
Route 1
Route 2
2. Chemistry
Scheme 1 shows the synthesis of representative ligand
building blocks, which were subsequently attached to
the 2-carboxyindole scaffold. The P4 ligand 1 used
throughout was prepared from commercially available
N-Boc protected 4-aminopiperidine in two steps by
reductive amination with acetone followed by acid cat-
alyzed removal of the Boc group. Bromoacetamide 2
was prepared in one step by acylation of 2-amino-5-
chloro pyridine with bromoacetyl bromide. The biaryl
bromide 5 was synthesized starting from 1-(5-chloro-
thiophen-2-yl)-ethanone by condensation with oxalic
acid diethyl ester followed by regioselective formation of
the isoxazole ring by treatment with hydroxylamine.
Reduction of the ester in 4 and subsequent bromination
yielded bromide 5.
O
O
N
N
N
N
H
H
R
7
6
(a)
(b), (c)
O
N
R
N
H
N
8-29
Scheme 2. Reagents and conditions: (a) ArB(OH)2, Cu(OAc)2, pyri-
dine, NEt3, CH2Cl2, 50 °C, 5 days; or Ar–(CH2)–Hal, NaH, DMF,
80 °C, 1 h; or Ar–(CH2)2–OTos, NaH, DMF, rt, 16 h; or Ar–
NHCO(CH2)–Br, NaH, DMF, rt, 2 h; (b) LiOH, THF/H2O, 60 °C, 2 h;
(c) amine 1, TOTU, N-ethylmorpholine, CH2Cl2; or BOP–Cl, NEt3,
CH2Cl2.
The synthesis of the fully decorated indole scaffold with
the various putative P1 ligands was carried out by two
optional routes starting from commercially available

M. Nazarꢀe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4191–4195
4193
Table 1. 2-Carboxyindole-based factor Xa inhibitors incorporating
simple P1 ligands
piperidine 1 using either TOTU or BOP–Cl in CH2Cl2 to
yield the final derivatives 8–29. Route 2 essentially
postponed the diversifying indole N-alkylation step to
the end of the sequence and was successfully applied for
the synthesis of compounds 26–29 (Table 3). This syn-
thetic route proved to be superior for the preparation of
compounds incorporating the N-acetamide linker,
which turned out to be rather unstable under the basic
or acidic conditions required for ester cleavage. In
addition the use of route 2 prevented cyclization to the
six membered N-aryl-indolo-1,3-ketopiperazine, which
occurred to a considerable extend under the acyl acti-
vation conditions used in route 1.
H
N
N
N
O
R1
Compds
R1
Ki (fXa) nM
8
2100
Cl
OMe
9
2480
Cl
10
11
9010
2500
3. Results and discussion
Cl
OMe
The use of a scaffold providing new trajectories into the
S1 and S4 pocket required a review of privileged P4
ligands (i.e., fragments known to impart good potency)
under the stipulation that the neutral P1 ligand should
be attached at the indole nitrogen. A solution phase
library of 140 compounds containing 14 prototypic
conformationally rather mobile P1 ligands and 10
potential P4 ligands at the indole scaffold (results not
shown) revealed N-isopropyl-4-aminopiperidine as a
highly potent P4 ligand. This group was selected for the
present SAR studies on the P1 substituents and was
incorporated in all the inhibitors described in this
communication.
12
89
Cl
13
14
15
654
73
Cl
Cl
474
Cl
OMe
16
>10,000
Table 1 shows the SAR for a series of simple, mono-
cyclic P1 groups, with varying lengths of spacer. Clearly
the directly arylated inhibitors 8–10 were too short for
close interaction with Tyr 228 at the base of the S1
pocket,6 although some nuances in activity were evident.
In contrast to the equipotent para-chlorophenyl com-
pound 8 and the meta-methoxyphenyl derivative 9, the
meta-chlorophenyl compound 10 was much less active.
A different pattern of activity was observed for com-
pounds 11–13 with a methylene spacer connecting the
P1 ligand with the scaffold. para-Chlorobenzyl com-
pound 11 was the least potent in this series. meta-
Chlorobenzyl 13, showed intermediate potency and,
remarkably, the meta-methoxybenzyl compound 12, had
a Ki value under 100 nM,10 indicating a strong and
specific interaction in the S1 pocket. Substituted phen-
ethyl ligands, which we previously used successfully in
combination with a benzoic acid scaffold,5 were incor-
porated in compounds 14–16. The para-chlorophenyl-
ethyl moiety in compound 14 exhibited the most
promising activity as P1 ligand in this context. The
dichloro derivative 15 was approximately 6-fold less
potent than 14 and the meta-methoxyphenylethyl sub-
stitution (16) was even less potent.
(Table 2). A chlorobenzothiophene P1 ligand in 17 did
indeed improve potency compared to 14. Surprisingly
extension to a biarylic system resulted in a rather dra-
matic fall in activity (18–21), indicating the sensitivity of
the interaction between the chloroaryl fragment and Tyr
228 to changes in the trajectory and spacing of the P1
ligand. Moreover, a more detailed analysis revealed the
critical role played by the first heterocyclic ring and its
polarity distribution. Isoxazole 18 was only half as ac-
tive as the reversed derivative 19. Compounds 20 and 21
incorporating a oxadiazole and thiazole ring, respec-
tively, were even less active suggesting that distinct polar
interactions were required at one side of the ring
whereas such interactions at the other side of the ring
were deleterious. This generally disappointing loss in
activity on rigidifying the ethyl spacer of the para-
chlorophenyl ring could be more than completely
reversed in the biaryl series by replacement of the para-
chlorophenyl moiety by chlorothiophene (22–25). For
example, exchange of the para-chlorophenyl group in 18
for a chlorothienyl group in 22 resulted in a dramatic
270-fold increase in activity.
The results in Table 1 established that the structural
requirements for the P1 ligand in the most potent
inhibitor (14) were a terminal chloro substituent com-
bined with an extended spacer. We next turned our
attention to the optimization of this rather conforma-
tionally mobile P1 ligand, combining further extension
of the linker length with the required rigidification
Apparently this was due to the slight trajectorial change
in the presentation of the chlorine (essential for binding)
on shifting from para-chlorophenyl to its chlorothienyl
analogue. This gain in activity can be attributed to an
optimal positioning of the chlorothiophene enabling a
close contact of the chlorine atom with the centroid of

4194
M. Nazarꢀe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4191–4195
Table 2. 2-Carboxyindole-based factor Xa inhibitors incorporating
extended and rigidified P1 ligands
Table 3. 2-Carboxyindole-based factor Xa inhibitors incorporating
acetamide rigidified P1 ligands
H
H
N
N
N
N
N
R1
N
R1
O
O
Compds
R1
Ki (fXa) nM
Compds
R1
Ki (fXa) nM
H
Cl
14
73
N
26
15
S
Cl
Cl
O
H
N
17
18
19
20
21
22
23
40
810
400
2230
2730
3
S
Cl
27
28
29
3
287
1
O
N
O
O
Cl
O
H
N
N
Cl
Cl
O
Cl
N
N
H
N
N
N
Cl
Cl
O
S
O
N
O
N
S
intermediate length to the flexible ethyl spacer in com-
pound 14, and the biaryl derivatives 18–25. In this series
the acetamide-linked chlorothiophene in 26 proved still
to be considerably active, but in contrast to the biaryl
series the activity of the para-chlorophenyl analogue 27
surpassed that of all other para-chlorophenyl derivatives
synthesized. Altering the trajectory of the chlorine atom
in the meta-chlorophenyl derivative 28 resulted in a
considerable drop in activity. However, potency could
be increased by introducing a nitrogen resulting in the
most potent inhibitor (29) in this series (Ki ¼ 1 nM).
This outcome again reflects that in conjunction with the
chloro binding mode, polar interactions along the S1
pocket of the fXa enzyme can influence the affinity of
inhibitors with neutral P1 ligands.
Cl
Cl
N
3
S
N
24
25
4
S
N
S
S
Cl
Cl
57
S
the aromatic ring of Tyr 228 located at the base of the S1
pocket, as described previously.6 Although the impact of
this interaction masked the more subtle differences in
interaction between the first isoxazole ring in compound
22 and the reversed isoxazole 23 or the thiadiazole
derivative 24 the presence of a less polar thiazole ring in
compound 25 was still deleterious, resulting in a 19-fold
loss in activity.
Table 4 shows the activity of key compounds (17, 22, 27,
29) against related proteases and confirms the high
selectivity of the compounds in this series. These com-
pounds were evaluated (Table 5) for their anticoagulant
The final type of rigidified P1 ligand investigated was the
acetamides (26–29) shown in Table 3. These are of
Table 4. Selectivity profile for selected 2-carboxyindole-based factor Xa inhibitors
Compds
fXa Ki/lM
Thrombin Ki/lM
Trypsin Ki/lM
Kallikrein Ki/lM
t-PAa Ki/lM
17
22
27
29
0.040
0.003
0.003
0.001
9.04
2.76
>10
>10
>100
>100
>100
>100
>100
15.4
>100
>100
>100
>100
>100
>100
a Human tissue plasminogen activator.
Table 5. Anti-fXa, anticoagulant activity, predicted absorption and metabolic stability of selected compounds
dPTa/lM
APTTb/lM
PTb/lM
S9c %
Papp nm/s
d
Compds
fXa Ki/lM
17
22
27
29
0.040
0.003
0.003
0.001
0.897
0.426
0.027
0.028
17.04
5.16
1.81
1.08
14.13
5.41
0.42
0.35
98
94
67
95
1.5
12.4
72.9
96.5
a Concentration required for 50% inhibition of dilute prothrombin time.
b Concentration required to double plasma clotting time of APTT and PT, respectively.
c Percent compound remaining after 2 h incubation with human S9 liver fraction.
d Apparent permeability coefficient in Caco-2 cells.

M. Nazarꢀe et al. / Bioorg. Med. Chem. Lett. 14 (2004) 4191–4195
4195
activity in the standard coagulation assay for activated
References and notes
partial thromboplastin time (APTT) and prothrombin
time (PT), as well as the more sensitive dilute PT (dPT)
with diluted thromboplastin reagent.10 Surprisingly,
compounds 27 and 29, which in the enzyme assay are
approximately equipotent to 22 are around 16 times
more potent than this compound in the more stringent
and predictive dPT antithrombotic assay, as well as
being more potent in the APTT and PT assays (Table
5).10 This lack of clear correlation between enzyme
inhibition and potency in coagulation assays has previ-
ously been reported.5
1. (a) Hirsh, J. N. Eng. J. Med. 1991, 324, 156; (b) Freedman,
M. D. Am. J. Ther. 1996, 3, 771.
2. Leadley, R. J., Jr. Curr. Top. Med. Chem. 2001, 1, 151.
3. Kaiser, B. Drugs Future 1998, 23, 423.
4. (a) Pinto, D. J. P.; Orwat, M. J.; Wang, S.; Fevig, J. M.;
Quan, M. L.; Amparo, E.; Cacciola, J.; Rossi, K. A.;
Alexander, R. S.; Smallwood, A. M.; Luettgen, J. M.;
Liang, L.; Aungst, B. J.; Wright, M. R.; Knabb, R. M.;
Wong, P. C.; Wexler, R. R.; Lam, P. Y. S. J. Med. Chem.
2001, 44, 566; (b) Pauls, H. W.; Ewing, W. R. Curr. Top.
Med. Chem. 2001, 1, 83.
ꢀ
5. Nazare, M.; Matter, H.; Klingler, O.; Al-Obeidi, F.;
Schreuder, H.; Zoller, G.; Czech, J.; Lorenz, M.; Dudda,
A.; Peyman, A.; Nestler, H. P.; Urmann, M.; Bauer, A.;
Laux, V.; Wehner, V.; Will, D. W. Bioorg. Med. Chem.
Lett. 2004, 14, 2801.
The same compounds were also tested in human S9 liver
fractions and Caco-2 cell monolayers as in vitro pre-
dictive assays for metabolism and permeation, respec-
tively. In these assays (Table 5) the compounds proved
to be metabolically stable and, with the exception of 17,
showed excellent predicted absorption properties.11
6. (a) Adler, M.; Kochanny, M. J.; Ye, B.; Rumennik, G.;
Light, D. R.; Biancalana, S.; Whitlow, M. Biochemistry
2002, 41, 15514; (b) Maignan, S.; Guilloteau, J.-P.; Choi-
Sledeski, Y. M.; Becker, M. R.; Ewing, W. R.; Pauls, H.
W.; Spada, A. P.; Mikol, V. J. Med. Chem. 2003, 46, 685.
7. Choi-Sledeski, Y. M.; Kearney, R.; Poli, G.; Pauls, H.;
Gardner, C.; Gong, Y.; Becker, M.; Davis, R.; Spada, A.;
Liang, G.; Chu, V.; Brown, K.; Collussi, D.; Leadley, R.,
Jr.; Rebello, S.; Moxey, P.; Morgan, S.; Bentley, R.;
Kasiewski, C.; Maignan, S.; Guilloteau, J.-P.; Mikol, V.
J. Med. Chem. 2003, 46, 681.
8. Matter, H.; Defossa, E.; Heinelt, U.; Blohm, P.-M.;
Schneider, D.; Mueller, A.; Herok, S.; Schreuder, H.;
Liesum, A.; Brachvogel, V.; Loenze, P.; Walser, A.; Al-
Obeidi, F.; Wildgoose, P. J. Med. Chem. 2002, 45, 2749.
9. Mederski, W. K. R.; Lefort, M.; Germann, M.; Kux, D.
Tetrahedron 1999, 55, 12757.
In conclusion, we have developed a new class of potent
nonguanidino/amidino fXa inhibitors based on a simple
2-carboxyindole scaffold. We have investigated the
structural requirements for neutral P1 ligands in com-
bination with this scaffold, and used these to develop a
series of selective, structurally diverse, equipotent
inhibitors. The compounds are predicted to be meta-
bolically stable and well absorbed. They are capable of
reducing the dilute prothrombin time, a recognized
surrogate parameter for in vivo antithrombotic activity
in the clinic, and are therefore promising drug sub-
stances for the treatment of thrombotic diseases.
10. In vitro assays for determination of fXa and thrombin
activity were performed as described by Ostrem, J. A.; Al-
Obeidi, F.; Safar, P.; Safarova, A.; Stringer, S. K.; Patek,
M.; Cross, M. T.; Spoonamore, J.; LoCascio, J. C.;
Kasireddy, P.; Thorpe, D. S.; Sepetov, N.; Lebl, M.;
Wildgoose, P.; Strop, P. Biochemistry 1998, 37, 1053.
11. (a) Artursson, P.; Palm, K.; Luthman, K. Adv. Drug
Delivery Rev. 2001, 46, 27; (b) Fichert, T.; Yazdanian, M.;
Proudfoot, J. R. Bioorg. Med. Chem. Lett. 2003, 13, 719.
Acknowledgements
We would like to thank Astrid Sihorsch, Marga
€
Kammerer-Dienst, Lynn Bayer, Dirk Timme, Norbert
Laub and Stefan Engel for their excellent technical
assistance.