Discovery of an orally efficacious inhibitor of coagulation factor Xa which incorporates a neutral P1 ligand

Article abstract of DOI:10.1021/jm020384z

The discovery and SAR of ketopiperazino methylazaindole factor Xa inhibitors are described. Structure-activity data suggesting that this class of inhibitors does not bind in the canonical mode were confirmed by an X-ray crystal structure showing the neutral haloaromatic bound in the S₁ subsite. The most potent azaindole, 33 (RPR209685), is selective against related serine proteases and attains higher levels of exposure upon oral dosing than comparable benzamidines and benzamidine isosteres. Compound 33 was efficacious in the canine AV model of thrombosis.

Full text of DOI:10.1021/jm020384z

J . Med. Chem. 2003, 46, 681-684
681
Discover y of a n Or a lly Effica ciou s
In h ibitor of Coa gu la tion F a ctor Xa
Wh ich In cor p or a tes a Neu tr a l P 1 Liga n d
Yong Mi Choi-Sledeski,*,† Robert Kearney,†
Gregory Poli,† Henry Pauls,† Charles Gardner,†
Yong Gong,† Michael Becker,† Roderick Davis,†
Alfred Spada,† Guyan Liang,† Valeria Chu,†
Karen Brown,† Dennis Collussi,† Robert Leadley, J r.,†
Sam Rebello,† Phillip Moxey,† Suzanne Morgan,†
Ross Bentley,† Charles Kasiewski,†
Sebastien Maignan, J ean-Pierre Guilloteau, and
Vincent Mikol
F igu r e 1.
Department of Medicinal Chemistry and Department of
Biology, Aventis Pharmaceuticals, Route 202-206,
Bridgewater, New J ersey 08807-0800, and Department of
Structural Biology, Aventis Pharmaceuticals,
molecule. The discovery of ketopiperazines as an alter-
native scaffold required reoptimization of the “privi-
leged” P1 and P4 ligands (i.e., fragments known to
impart good potency) from the earlier pyrrolidinone
series. In this communication, we describe the evolution
of the azaindole pyrrolidinones represented by 1 to
ketopiperazines 3 and subsequent optimization to yield
orally efficacious inhibitors of fXa (Figure 1).
Although it is well precedented that arylamidines
generally lead to potent inhibitors, molecules containing
such highly basic functions are often poorly absorbed8b
and/or are associated with undesirable side effects.8b
During the course of this work, it became clear that not
any benzamidine replacement8a is sufficient to impart
good oral bioavailability. This was exemplified by ami-
noquinazoline 4,8a which is highly potent against fXa
but displays poor pharmacokinetic properties. In the
endeavor to improve PK parameters and retain potency,
X-ray structural work9 was essential in understanding
the binding modes and interpreting the SAR of the
ketopiperazine fXa inhibitors.
13, Quai J . Guesde, F-94403 Vitry/ Seine, France
Received September 5, 2002
Abstr a ct: The discovery and SAR of ketopiperazino methy-
lazaindole factor Xa inhibitors are described. Structure-
activity data suggesting that this class of inhibitors does not
bind in the canonical mode were confirmed by an X-ray crystal
structure showing the neutral haloaromatic bound in the S1
subsite. The most potent azaindole, 33 (RPR209685), is selec-
tive against related serine proteases and attains higher levels
of exposure upon oral dosing than comparable benzamidines
and benzamidine isosteres. Compound 33 was efficacious in
the canine AV model of thrombosis.
In tr od u ction . Thrombosis-related diseases, includ-
ing myocardial infarction, deep vein thrombosis, and
unstable angina, often have life-threatening conse-
quences, and thus, considerable research is directed
toward developing novel antithrombotic agents. Current
therapies, using heparin1 and warfarin,2 are based on
indirect inhibition of thrombin, the terminal enzyme in
the coagulation cascade. These therapies are limited by
side effects, mode of administration, stringent monitor-
ing of drug levels, slow onset of action, and excessive
bleeding. The trypsin-like serine protease factor Xa (fXa)
occupies a unique position at the final convergence point
of the intrinsic and extrinsic pathways,3 thus inhibiting
coagulation without directly affecting platelet function.
As part of the prothrombinase complex, factor Xa is the
singular enzyme responsible for thrombin activation;
thus, fXa has been identified as an attractive target for
antithrombotic therapy.4
Ch em istr y. A representative synthesis of the azain-
dole inhibitor is described in Scheme 1. Alkylation of
Sch em e 1a
In a previous communication, we described the opti-
mization of azaindole P1 and heteroaromatic P4 groups
in the pyrrolidinone series leading to double-digit na-
nomolar azaindole inhibitors of factor Xa.5 Although
these compounds, 1, exhibited good affinity for fXa in
vitro, only modest oral bioavailability was obtained in
our canine model as determined by an ex vivo anti-fXa
bioassay.6 Concurrently, a new series7 of factor Xa
inhibitors (e.g., 2) was identified that seemed more
tolerant of changes in the putative P1 region of the
a
* To whom correspondence should be addressed. Phone: (908) 231-
2429. Fax: (908) 231-2429. E-mail: yong-mi.choi@aventis.com.
† Department of Medicinal Chemistry and Department of Biology.
Department of Structural Biology.
(a)NaH, THF/DMF, propargylbromide, 0 °C;(b)(i)Pd(PPh3)2Cl2,
Et3N, CuI, DMF, 100 °C, (ii) DBU, 50 °C; (c) Pd black, H2O,
HCO2H, MeOH; (d) ArSO2Cl, Et3N, CH3CN; (e) 30% TFA, CH2Cl2.
10.1021/jm020384z CCC: $25.00 © 2003 American Chemical Society
Published on Web 02/04/2003

682 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 5
Letters
Sch em e 2a
Ta ble 2. 5-Azaindole Modifications
Ki (nM)
fXa thrombin trypsin (% absorbed)a metabolisma,b
CaCo-2
R1
a
(a) nBuLi, THF, (EtO)2POCl; (b) nBuLi, THF; (c) nBu4NI,
3 H
20 Me
4
3
2900 >2900 6.0 (mod)
>3900 >2900 3.0 (mod)
high (64%)
high (60%)
high (99%)
ND
acetone; (d) PPh3, SO2Cl2, DCM.
21 -CH2CO2Me 18 >3900 >2900 47.6 (high)
22 -CH2CO2H 154 >3900 >2900 ND
Ta ble 1. In Vitro Activity of 5,6-Fused Biaryl Benzamidine
Replacements
23 -CH2CONH2
24 -CH2CH2OH
8
3
>3900 >2900 7.0 (mod)
>3900 >2900 15.3 (mod)
mod (40%)
mod (47%)
a
b
mod ) moderate. Rat liver microsomes.
derivative 14 was 11-fold less active, while the 4-aza-
indole 13, paralleling the pyrrolidinone result,5 was
dramatically less active. The activities of these ana-
logues appear to loosely correlate with the basicity of
the azarene group except in the case of the 4-amino-5-
azaindole analogue 17. This derivative is 1 log more
basic than the simple 5-azaindole and was designed to
interact with D189 in the S1 active site in a bidentate
fashion; thus, it was surprising that the potency dropped
255-fold versus the parent compound 3. A significant
loss in activity was also observed with the oxo analogue
18, whereas the neutral N-oxide 19 was a modest
inhibitor with a Ki of 38 nM.
Table 2 summarizes attempts to optimize the series
by substitution at the N-1 position of the 5-azaindole
ring. This position offers a chemical handle to append
moieties that could modify physicochemical properties,
improve potency by additional interaction with the en-
zyme, and improve pharmacokinetic properties by block-
ing a possible site of metabolism. Two in vitro measures
of metabolism and permeability, i.e., microsomal stabil-
ity and CaCo-2 cell monolayer absorption,11 respectively,
were monitored as a function of N-1 substitution.
Incorporation of a negative charge, e.g., compound 22,
had a detrimental effect on inhibitor potency, although
in general, substitution at the N-1 position had little
effect on Ki. Note that carboxamide 23 and hydroxyethyl
24 analogues, both capable of H-bonding interactions,
were no more potent than the parent compound 3. These
results are difficult to reconcile with a binding mode
that places the azaindole system in the congested S1
specificity pocket and suggest that the appended chains
are not in intimate contact with the enzyme. Perme-
ability was significantly improved in only one case, ester
21, which was rapidly degraded presumably to acid 22.
The rate of microsomal metabolic turnover for carbox-
imide 23 and hydroxyethyl 24 analogues was lower than
for the parent 3, but these modifications had only a
moderate effect on reducing metabolism. In general, this
approach maintained in vitro potency, had modest
effects on absorption, but could not concurrently reduce
microsomal metabolism.
the commercially available CBZ-ketopiperazine intro-
duces the alkyne 5 functionality that undergoes the key
Pd-catalyzed crossing-coupling reaction with (3-iodo-
pyridin-4-yl)carbamic acid tert-butyl ester 6, followed by
base-induced cyclization to yield the protected 5-aza-
indole 7 in a one-pot process. Sequential deprotection
of CBZ under transfer hydrogenation conditions, forma-
tion of the sulfonamide with the corresponding sulfonyl
chloride, and then acid-catalyzed removal of the BOC
group yield the desired compound 3. The requisite
alkenylsulfonyl chloride was prepared by treatment of
the Wittig reagent 8 with aldehyde 9 to afford the ethyl
sulfonate ester 10, which was subsequently converted
to the tetrabutylammonium salt (Scheme 2). Treatment
of the salt 11 with sulfuryl chloride and triphenylphos-
phine yields 12.
In Vitr o Op tim iza tion . Using our preferred 6-chlo-
robenzothiophene sulfonamide group as the invariant,
we examined a number of 5,6-fused biaryls (Table 1) as
replacements for the undesired benzamidine. In contrast
to the pyrrolidinone scaffold, in which the 6-azaindole
1a yielded the more potent analogue,5 the 5-azaindole
compound 3 in the ketopiperazine series afforded the
more potent inhibitor with a Ki of 4 nM. The 6-azaindole
The aryl sulfonamide portion of the molecule was
subsequently investigated using privileged ligands12
(Table 3), which had imparted good activity in our
pyrrolidinone series of inhibitors. The pairs 26, 27 and
28, 29 exemplify the necessity of the halogen atom for
good potency. A 240-fold drop in potency was observed
by removing the halogen atom, 26, and an even greater
loss was observed by placing the chlorine at the 3-posi-

Letters
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 5 683
Ta ble 3. Optimization of Sulfonamides
tion, 25. On the basis of the crystal structure of the
6-azaindole pyrrolidine 1a with human fXa13 and as-
suming that the aryl sulfonamide was engaged in
lipophilic interactions in the S4 site, weakly basic
azarenes (28-31) were examined as a means to improve
potency by accessing the cation hole region of the
enzyme. Compound 30 was designed to retain the
empirically observed key chloro interaction while posi-
tioning the nitrogen toward the side chain of E97. To
our surprise, this compound showed poor fXa inhibitory
activity. Regioisomer 28 maintained good activity; how-
ever, the presence of nitrogen did not enhance its
potency against fXa. Divergent SAR trends between the
pyrrolidinone and ketopiperazine scaffolds were par-
ticularly apparent with the thienopyridine sulfonamide
containing analogues (e.g., 29 and 31). The thienopyri-
dine group, which ordinarily imparted good anti-fXa
activity in combination with benzamidine12 or its iso-
steric replacement14 in the pyrrolidinone series, was a
poor inhibitor in the latter series. X-ray crystallographic
data clarified the reasons we were unable to improve
inhibitory activity using this approach (vide infra).
In a survey of chlorine-containing aryl sulfonamides,
it became clear that the trajectory of the halogenated
aryl was critical for optimal fXa binding. The chloro-
styryl derivative 32, which can be viewed as a ring-
opened derivative of compound 3, was 12-fold less
potent. The activity was regained by substituting the
phenyl ring with thiophene 33, resulting in the most
potent inhibitor identified in this series (fXa, Ki ) 1 nM).
None of the other entries in Table 3 resulted in potency
enhancements. In particular, inhibitor 35, the satu-
rated, more conformationally mobile derivative of 33,
was 2 orders of magnitude less active. The aryl sulfon-
amide SAR within the ketopiperazine series underscores
the importance of the aryl halogen interaction with the
enzyme for optimum potency. The SAR trends diverge
from those established with canonical pyrrolidinone
inhibitors, suggesting an alternative binding mode for
this series.
group was making hydrophobic contact in the S4 region.
To our surprise, the chlorothiophene of 33 was bound
in the S1 pocket; closest contact to D189 was ca. 3 Å.
The chlorine atom was located in an area normally
occupied by a water molecule in canonical benzamidine/
factor Xa co-complexes.13 The chlorine atom is engaged
in direct contact with the centroid of the aromatic ring
of Y228 located on the back wall of the S1 binding
pocket. Reminiscent of pyrrolidinone binding, the car-
bonyl oxygen of piperazine forms an H bond with the
NH of G218. The azaindole group fills the S4 site with
the indole NH directed toward the solvent, while the
pyridine nitrogen forms an H bond with the structural
water molecule bridging I175 and T28 at the entrance
of the hydrophobic pocket. This binding mode explains
the preference for the 5-azaindole in this series and the
tolerance to substitution at N-1.
Structural evidence for the halogen-S1 tyrosine in-
teraction, although unprecedented in factor Xa, has been
observed in thrombin/inhibitor X-ray structures.15 Com-
pound 33 represents a parallel example in fXa inhibition
wherein a basic group thought to be necessary to form
the crucial salt bridge with D189 can be eliminated and
replaced by a neutral group capable of contributing
enough favorable binding energy to maintain good
affinity for the enzyme. The SAR necessitating the
presence of halogen suggests that ketopiperazines are
binding in the “reversed” mode in all cases, unlike the
pyrrolidinone series where multiple binding modes are
possible and divergent SAR trends are observed among
the isomeric azaindoles.
Biologica l Activity. Key compounds in this series
showing excellent selectivity (>1000-fold) against re-
lated serine proteases were evaluated for anticoagulant
properties in different species; APTT prolongation16 in
plasma was in the low micromolar range (Table 4). An
ex vivo anti-fXa bioassay6 was used to measure and
compare bioavailability, half-life, and plasma levels of
the inhibitors (Table 5). In contrast to the amino-
quinazoline inhibitor 4, which exhibits negligible plasma
levels when administered orally at 10 mg/kg in dogs,
the analogous azaindole 3 achieves plasma levels of 660
nM at the same dose. The N-alkylated azaindole 24
exhibits plasma levels similar to 3 but has a superior
half-life (2 h). At half the dose, 33 reaches micromolar
plasma levels with a concomitant increase in half-life;
X-r a y Cr ysta llogr a p h y. The X-ray crystal structure
of 33 in human fXa, obtained at a resolution of 2.1 Å,9
clarified the stringent requirements for good potency
that was observed within the ketopiperazine series. Our
working hypothesis had assumed that the azaindole was
located in the S1 site and the halogenated aromatic

684 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 5
Letters
(3) Furie, B.; Furie, B. C. Molecular Basis of Blood Coagulation. In
Hematology: Basic Principles and Practices; Hoffman, R., Benz,
E. J ., Shattil, S. J ., Furie, B., Cohen, H. J ., Silberstein, L. E.,
Eds.; Churchill Livingstone; New York, 1995; pp 1566-1587.
(4) Adang, A. E. P.; Rewinkel, J . B. M. A New Generation of Orally
Active Antithrombotics: Comparing Strategies in the GPIIb/IIIa,
Thrombin and Factor Xa Areas. Drugs Future 2000, 25, 369-
383.
(5) Gong, Y.; Becker, M.; Choi-Sledeski, Y. M.; Davis, R. S.; Salvino,
J . M.; Chu, V.; Brown, K.; Pauls, H. Solid-Phase Parallel
Synthesis of Azarene Pyrrolidinones as Factor Xa Inhibitors.
Bioorg. Med. Chem. Lett. 2000, 10, 1033-1036.
(6) Leadley, R. J ., J r.; Morgan, S.; Bentley, R.; Bostwick, J .;
Kasiewski, C.; Chu, V.; Brown, K.; Ewing, W. R.; Pauls, H.;
Spada, A.; Perrone, M.; Dunwiddie, C. Pharmacodynamic Activ-
ity and Antithrombotic Efficacy of RPR120844, a Novel Inhibitor
of Coagulation Factor Xa. J . Cardiovasc. Pharmacol. 1999, 34,
791-799.
(7) Ewing, W. R.; Becker, M. R.; Li, A.; Davis, R. S.; J iang, J . Z.;
Zulli, A.; Choi-Sledeski, Y. M.; Pauls, H. W.; Myers, M. R.; Spada,
A. P.; Maignan, S.; Guilloteau, J .-P.; Mikol, V.; Brown, K. D.;
Colussi, D.; Chu, V.; Leadley, R. J .; Cheney, D.; Mason, J . Design
and SAR of Factor Xa Inhibitors Employing Pyrrolidinone and
Piperazinone Scaffolds. Abstracts of Papers, 219th National
Meeting of the American Chemical Society, San Francisco, CA,
March 26-30, 2000; American Chemical Society: Washington,
DC, 2000; MEDI-161.
Ta ble 4. Selectivity Profile and APTTs for Selected fXa
Inhibitorsa
Ki(APC)/ Ki(plasmin)/ Ki(tPA)/ 2(APTT),b 2(APTT),b
Ki(fXa) Ki(fXa)
Ki(fXa)
Ki(fXa) human (µM) dog (µM)
3
24
33
4.0
3.0
1.1
>4600
>6100
>17000
>1800
>2400
>17000
>2200
>2900
>8000
2.0
1.7
0.59
5.0
ND
2.2
a
Source of enzyme: human fXa, human APC, human plasmin,
b
recombinant tPA. 2 × APTT is defined as the concentration of
inhibitor required to double the activated partial thromboplastin
time.
Ta ble 5. Pharmacokinetic Parameters after Oral Dosing in
Dogs
Ki(fXa)
dose
F
T1/2
Cmax
(nM)
compd
(nM)
(mpk, po)
(%)
(min)
4
3
24
33
0.8
4
3
10
10
10
5
<5%
99
11
ND
36
139
52
10
658
783
1
97
1638
the estimated bioavailability for this inhibitor is 97%.
These results represent a substantial improvement over
previous inhibitors in terms of their pharmacokinetic
properties.
(8) (a) Pauls, H. W.; Ewing, W. R. The Design of Competitive, Small-
Molecule Inhibitors of Coagulation Factor Xa. Curr. Top. Med.
Chem. 2001, 1 (2), 83-100. (b) The above issue is dedicated
exclusively to fXa and contains reviews from various industrial
research groups.
(9) Maignan, S.; Guilloteau, J .-P.; Choi-Sledeski, Y. M.; Becker, M.
R.; Ewing, W. R.; Pauls, H. W.; Spada, A. P.; Mikol, V. Molecular
Structures of Human Factor Xa Complexed with Ketopiperazine
Inhibitors: Preference for a Neutral Group in the S1 Pocket. J .
Med. Chem. 2003, 46, 685-690.
On the basis of the favorable in vitro and PK profile
of inhibitor 33, efficacy studies were conducted in a
canine arteriovenous thrombosis model.17 Upon oral
dosing (20 mg/kg), the time-to-occlusion was prolonged
1.9-fold (145 vs 77 min) on the venous side and 1.8-fold
(147 vs 80 min for vehicle) on the arterial side. A
concomitant reduction in venous thrombus weight (34%)
and a smaller reduction in arterial thrombus weight
(13%) were observed.
(10) Greenhill, J . V. Comprehensive Heterocyclic Chemistry; Katritz-
ky, A. R., Rees, C. W., Eds.; Pergamon Press: New York, 1984;
Vol. 4, pp 498-502.
(11) Walter, E.; Kissel, T.; Reers, M.; Dickneite, G.; Hoffmann, D.;
Stuber, W. Pharm. Res. 1995, 12, 360. In our assay, an
absorption of <2% is considered poorly absorbed, 2-20% is
moderate, and >20% is highly absorbed.
(12) Becker, M. R.; Ewing, W. R.; Davis, R. S.; Pauls, H. W.; Ly, C.;
Li, A.; Mason, H. J .; Choi-Sledeski, Y. M.; Spada, A. P.; Chu,
V.; Brown, K. D.; Colussi, D.; Leadley, R. J .; Bentley, R.;
Bostwick, J .; Kasiewski, C.; Morgan, S. Synthesis, SAR and in
Vivo Activity of Novel Thienopyridine Sulfonamide Pyrrolidi-
nones as Factor Xa Inhibitors. Bioorg. Med. Chem. Lett. 1999,
9, 2753-2758.
(13) Maignan, S.; Guilloteau, J .-P.; Pouzieux, S.; Choi-Sledeski, Y.
M.; Becker, M. R.; Klein, S. I.; Ewing, W. R.; Pauls, H. W.; Spada,
A. P.; Mikol, V. Crystal Structures of Human Factor Xa Com-
plexed with Potent Inhibitors. J . Med. Chem. 2000, 43, 3226-
3232.
(14) Choi-Sledeski, Y. M.; Becker, M. R.; Green, D. M.; Davis, R.;
Ewing, W. R.; Mason, H. J .; Ly, C.; Spada, A.; Liang, G.; Cheney,
D.; Barton, J .; Chu, V.; Brown, K.; Colussi, D.; Bentley, R.;
Leadley, R.; Dunwiddie, C.; Pauls, H. W. Aminoisoquinolines:
Design and Synthesis of an Orally Active Benzamidine Isostere
for the Inhibition of Factor Xa. Bioorg. Med. Chem, Lett. 1999,
9, 2539-2544.
(15) Tucker, T. J .; Brady, S. F.; Lumma, W. C.; Lewis, S. D.; Gardell,
S. J .; Naylor-Olsen, A. M.; Yan, Y.; Sisko, J . T.; Stauffer, K. J .;
Lucas, B. J .; Lynch, J . J .; Cook, J . J .; Stranieri, M. T.; Holahan,
M. A.; Lyle, E. A.; Baskin, E. P.; Chen, I.-W.; Dancheck, K. B.;
Krueger, J . A.; Cooper, C. M.; Vacca, J . P. Design and Synthesis
of a Series of Potent and Orally Bioavailable Noncovalent
Thrombin Inhibitors That Utilize Nonbasic Groups in the P1
Position. J . Med. Chem. 1998, 41, 3210-3219.
Con clu sion . In summary, a series of ketopiperazine
inhibitors with an azaindole P4 group and nonbasic
chloroaryl groups have been shown to be potent and
selective inhibitors of factor Xa. The X-ray crystal
structure of 33 and fXa reveal that the key interaction
in the S1 pocket does not involve interaction with D189
as is typical for basic inhibitors. The critical interaction
between chlorine and the aromatic ring of Y228 in the
S1 pocket may be a general phenomenon, which may
be exploited with other serine proteases containing this
residue. Consequently, additional opportunities to ob-
tain orally active inhibitors will be possible with the
elimination of a highly basic charged group in the
molecule. The properties of our inhibitors are such that
good exposure is obtained upon oral dosing, and in the
case of 33 (RPR209685), antithrombotic efficacy is
observed. The arylsulfonamido ketopiperazine azain-
doles represent an important milestone in the develop-
ment of orally efficacious factor Xa inhibitors.
(16) Chu, V.; Brown, K.; Colussi, D.; Choi, Y. M.; Green, D.; Pauls,
H. W.; Spada, A.; Perrone, M.; Leadley, R. J .; Dunwiddie, C. In
Vitro Characterization of a Novel Factor Xa Inhibitor, RPR
130737. Thromb. Res. 2000, 99 (1), 71-82.
Su p p or tin g In for m a tion Ava ila ble: Experimental de-
tails for 6, 33, 3, 20, 17, and 19. This material is available
free of charge via the Internet at http://pubs.acs.org.
(17) Rebello, S. S.; Bentley, R. G.; Morgan, S. R.; Kasiewski, C. J .;
Chu, V.; Perrone, M. H.; Leadley, R. J . Antithrombotic Efficacy
of a Novel Factor Xa Inhibitor, FXV673, in a Canine Model of
Coronary Artery Thrombolysis. Br. J . Pharmacol. 2001, 133 (7),
1190-1198.
Refer en ces
(1) Hirsh, J . Heparin. N. Engl. J . Med. 1991, 324, 156-174.
(2) Freedman, M. D. Warfarin and Other “Anti”-Vitamin K Anti-
coagulants. Pharmacodynamics and Clinical Use. Am J . Ther.
1996, 3, 771-783.
J M020384Z