Development of potent and selective factor Xa inhibitors

Article abstract of DOI:10.1016/S0960-894X(01)00311-0

The development of potent and selective small molecule inhibitors of factor Xa is described.

Full text of DOI:10.1016/S0960-894X(01)00311-0

Bioorganic & Medicinal Chemistry Letters 11 (2001) 1797–1800
Development of Potent and Selective Factor Xa Inhibitors
Roopa Rai, Aleksandr Kolesnikov, Yong Li, Wendy B. Young,* Ellen Leahy,
Paul A. Sprengeler, Erik Verner, William D. Shrader, Jana Burgess-Henry, Joan C. Sangalang,
Darin Allen, Xi Chen, Bradley A. Katz, Christine Luong, Kyle Elrod and Lynne Cregar
Departments of Medicinal Chemistry, Structural Chemistry and Enzymology, Axys Pharmaceuticals, Inc., 385 Oyster Point Blvd.,
South San Francisco, CA 94080, USA
Received 5 March 2001; accepted 14 March 2001
Abstract—The development of potent and selective small molecule inhibitors of factor Xa is described. # 2001 Elsevier Science
Ltd. All rights reserved.
Chemistry: Results and Discussion
Introduction
2-(3-Bromo-2-hydroxy-5-methyl-phenyl)-1H-indole-5-
The coagulation cascade is a complex scheme of enzy-
matic reactions that ultimately generates fibrin, the
foundation of all blood clots. Disruption or mal-
functioning of this homeostatic process leads to throm-
boembolic disease. Much effort over the past few
decades has focused on direct inhibition of key enzymes
within this pathway to affect blood coagulation and to
potentially generate safer antithrombotic agents. Factor
Xa (fXa), a pivotal enzyme in blood coagulation, cata-
lyzes the conversion of prothrombin to thrombin (fIIa).
Positioned at the junction of the intrinsic and extrinsic
pathways, it is believed that intervention at this key
position may enable concomitant control over both
pathways which, in turn, may lead to a more effective
drug therapy.
carboxamidine (1) was identified in-house as a non-
selective serine protease inhibitor.10 This compound
demonstrated good inhibition for factor Xa (Ki0=100
nM), as well as submicromolar inhibition of thrombin
(fIIa), urokinase-type plasminogen activator (uPa),
trypsin, and low micromolar inhibition of plasmin.11
A crystal structure of 1 complexed with thrombin
revealed a surprising array of hydrogen bond interac-
tions between the ligand and the catalytic residues of the
protein.10 The key interactions in this array include
three extremely short, low barrier H-bonds between the
phenolate of 1, Ser195 hydroxyl, and a water bound in
the oxyanion hole as depicted in Figure 1a. Additional
ligand–protein hydrogen bonds occur between the
indole NH, 195 OH, the phenol O and the protonated
His57 NH. This binding motif appears to be general
since it is observed for many structurally related small
molecules in a number of different serine proteases.12
Utilizing this crystal structure, we constructed a model
of 1 in factor Xa as depicted in Figure 1b. The model
suggested that the core scaffold 1 could be further
functionalized to exploit entry into the S10, S1b, and
potentially, the S4 pockets of factor Xa for binding. Our
intent was to improve the potency of 1 for factor Xa
There are numerous reports detailing the design and
synthesis of peptide, peptidomimetic and small molecule
factor Xa inhibitors.1ꢀ5 Although there are no direct
factor Xa inhibitors currently being marketed as drugs,
there are several compounds in early phase clinical
trials.4 At Axys Pharmaceuticals, we have focused on
the generation of selective, small molecule inhibitors of
factor Xa that utilize unique binding modes.6ꢀ10 Herein,
we describe the structure-based design and development
of 1 to a selective subnanomolar inhibitor of factor Xa.
*Corresponding author. Fax: +1-650-829-1019; e-mail: wendy_young@
axyspharm.com
0960-894X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(01)00311-0

1798
R. Rai et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1797–1800
while also enhancing its selectivity versus key anti-tar-
gets such as thrombin, uPa, plasmin, and trypsin.
methylene tether of compound 6 is not of sufficient
length to allow the phenyl ring to access the S4 pocket.
It is more likely that the aromatic ring is bound at the
S1b site above the 191–220 disulfide and the selectivity
observed is due to differences between the enzymes at
this site. Our modeling indicates that the three carbon
analogue 8 may bind outside of the S4 pocket, poten-
tially explaining its poor inhibition. Clearly, a range of
longer and more complex tethers could be generated at
this site to access the factor Xa S4 ‘cationic binding
pocket’,5 although, at this juncture, we avoided this
strategy opting for lower molecular weight solutions.
Initial chemistry efforts focused on preparing C30 sub-
stituted analogues of the original bromo derivative 1 to
probe the S10 pocket. Accordingly, analogues 2–5 were
generated as depicted in Table 1. Replacement of the
bromine of 1 with a hydrogen, chlorine, methyl or
phenyl gave analogues that were all less efficient at
inhibiting factor Xa, although the chlorine analogue
was nearly equipotent. The bromine at C30 may be
optimal for several reasons. First, there are favorable
interactions with the residues (including His57) of the
relatively small S10 pocket of fXa and secondly, the
bromine shields the network of short hydrogen bonds
from water which increases the enthalpic contribution.12
Thirdly, the electronic properties of the bromine may
strengthen the short hydrogen bond network.
In order to improve the selectivity versus thrombin, we
next exploited sequence differences between thrombin
and factor Xa at residue 192. Thrombin has a glutamic
acid at this position and factor Xa has a glutamine. An
anionic moiety in the vicinity should repel the thrombin
Glu192 while potentially H-bonding to the factor Xa
Gln192. Other factor Xa inhibitors have taken advan-
tage of this difference as well.5 Our model suggests that
substitution at C50, the position para to the phenol,
might achieve the desired interaction with amino acid
192 in factor Xa and thrombin.
Most reported factor Xa inhibitors exploit the S4
pocket to gain binding affinity. In an attempt to
improve the potency of 1 and enhance its overall selec-
tivity profile, we sought to probe the S4 pocket as well.
Modeling suggested that substituents at either the
indole C3 or the C50-position of the phenolic ring might
bind in this site. A stepwise extension of a benzene
moiety from the indole C3 position was accomplished to
give 6–8. Of the three carbon homologues generated,
compound 6 offered the best binding, demonstrating a
5-fold enhancement in potency for factor Xa as com-
pared to 1. Additionally, the selectivity versus uPa was
improved 15-fold. The model suggested that the one
A series of derivatives that systematically extend a car-
boxylic acid group from the aryl ring (9–13) and incor-
porate varying charged moieties (14–16) at the C50
position of 6 is listed in Table 2. Almost all these ana-
logues demonstrated increased inhibition for factor Xa
(as compared to 6) although a range in selectivity versus
thrombin was clearly noted. As demonstrated by ana-
logues 10 and 11, a one carbon spacer between the aryl
ring and acid moiety caused the largest decrease in
thrombin inhibition. Such a decrease in thrombin inhi-
bition combined with an increase in fXa inhibition pro-
vided improved overall selectivity (i.e., 333-fold in the
case of 10). The incorporation of a phosphoric acid
group, as in 14, also demonstrated increased inhibition
of factor Xa with 660-fold selectivity versus thrombin.
As tetrazoles are known isosteres of carboxylic acids,
Table 1. Modifications at C30 and C3
Compd
R1
R2
Ki (mM)
fXa
fIIa
uPa Plasmin Trypsin
1
2
3
4
Br
H
Cl
CH3
H
H
H
H
0.10
4.3
0.13
1.2
0.36
23.0
0.46
6.9
0.11
12.4
0.19
2.3
1.2
21.0
1.0
0.64
8.2
0.7
5.0
10.0
5
6
H
0.64
0.02
4.0
0.11
1.6
1.5
4.6
1.8
2.4
Br
Br
Br
0.26
7
8
0.35
7.6
0.71
6.6
2.3
8.6
8.8
32
3.4
Figure 1. (a) H-bonding array displayed by 1 with the catalytic resi-
dues of thrombin in the crystal strucutre; (b) model of 1 in factor Xa
built from the thrombin crystal structure showing the S1, S10 and S4
pockets.
42

R. Rai et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1797–1800
1799
compounds 15 and 16 were generated to study this
alteration. These analogues showed little to no
improvement in potency or selectivity as compared to
the nonionic analogue 6. Notably, the incorporation of
anionic moieties had little effect on the inhibition of
plasmin and trypsin, although it did enhance uPa
activity.
This result reinforces the notion that Glu192 of throm-
bin is responsible for the selectivity since the acids,
esters, and amides can interact with Gln192 of factor Xa
but only the esters and amides are well tolerated by
Glu192 of thrombin. Masking the anionic charge
removes the repulsive interaction with Glu192 of
thrombin and reduces the selectivity for the target
enzyme.
The in vitro anticoagulant effect (activated partial
thromboplastin times, aPTTs)11 of 6 and 9–16 are listed
in Table 2. For compounds with similar inhibition for
factor Xa, the range in efficacies can be attributed to the
ability of the compound to bind to plasma proteins and/
or the anti-target selectivity profile.
As phenols are candidates for in vivo metabolism (glu-
curonidation and sulfation) and amidines are expected
to limit absorption, we investigated the contribution of
these moieties on our elaborated lead molecule 12. We
hypothesized that there may be enough other positive
interactions between the inhibitor and protein to allow
removal of the hydroxyl or amidine. However, both
masking the hydroxy as a methyl ether (20) and remov-
ing it completely (21) reduced inhibition for factor Xa
90-fold and 2000-fold, respectively. Although the methoxy
analogue 20 still maintained appreciable factor Xa
potency, the selectivity versus almost all other anti-targets
A crystal structure of 10 was obtained in thrombin. The
structure reveals that the amidine moiety makes a hydro-
gen-bonded salt bridge with Asp189 and the array of
short hydrogen bonds with the catalytic residues is intact.
Interestingly, the benzyl group displays discrete disorder;
the major conformer is bound at the S1b disulfide bridge
site rather than at S4 which is consistent with our mod-
eling prediction. Also, there is no density for the C50
carboxylic acid consistent with it being disordered to
avoid repulsive interactions with Glu192 (Fig. 2).
In order to clearly discern the role of the carboxylic acid
in insulting thrombin, a series of acid analogues of 10
was generated (Table 3). The acid was masked as an
ester (17), secondary amide (18) and a primary amide
(19). Not surprisingly, in each case, the inhibition for
factor Xa was left intact while the inhibition for throm-
bin increased, thus decreasing the effective selectivity.
Table 2. Modifications at C50
Figure 2. Crystal structure of 10 in thrombin (2.0 A resolution). Note:
the C3 benzyl group is discretely disordered (thick green bonds repre-
sent 77% occupancy site, thin light green bonds display 1% occu-
pancy) and the C50 substituent is modeled as a methyl group due to
lack of density.
Compd
R
Ki (mM)
2xaPTT
(mM)
fXa fIIa uPa Plasmin Trypsin
Table 3. Modifications at C5, C20, and C50
6
–CH3
0.02
0.26 1.6
4.6
1.5
1.8
2.4
3.9
9
0.004 0.33 0.32
0.003 1.0 0.43
0.44
1.2
0.55
0.38
10
11
12
13
14
15
0.01
0.75 0.35
3.0
1.7
1.5
1.2
4.2
1.3
0.6
0.61
3.8
1.3
0.93
0.52
0.75
0.42
0.70
Compd
x
R1
R2
R3
Ki (mM)
0.0005 0.12 0.28
0.0007 0.04 0.27
0.0005 0.33 0.40
fXa
fIIa
10
17
18
19
12
20
21
22
23
1
1
1
1
2
2
2
2
2
Am
Am
Am
Am
Am
Am
Am
H
CONH2
OH
OH
OH
OH
OH
OMe
H
CO2H
CO2Me
CONMe2
CONH2
CO2H
CO2H
CO2H
CO2H
CO2H
0.003
0.003
0.002
0.002
0.0005
0.045
1.0
1.0
0.08
0.043
0.025
0.12
2.5
0.01
0.02
0.35 0.55
0.44 0.35
12
16
2.2
0.84
0.59
OH
OH
300
8.4
700
78

1800
R. Rai et al. / Bioorg. Med. Chem. Lett. 11 (2001) 1797–1800
Acknowledgements
The authors wish to thank Michael Venuti and Jochen
Knolle for helpful scientific discussions, Liling Fang for
analytical support and Barbara Nielsen for aid in pre-
paring this manuscript.
Scheme 1. Synthesis of 10. Reagents and conditions: (a) PhCHO,
NaOH, THF; (b) H2 Pd/C; (c) NBS, DMF; (d) 26, triethylamine,
MeOH, reflux; (e) PPA, 130 ꢁC.
was significantly diminished. Additionally, removing the
amidine completely (22) caused a 600,000-fold decrease
in potency, while replacement with an amide (23) was
somewhat less detrimental. Recently, we have generated
a series of analogues of 10 in which the 5-amidinoindole
moiety has been replaced with a less-basic heterocycle.
Such analogues show only a 50-fold loss in factor Xa
potency and improved anti-target selectivity. This work
will be reported shortly.
References and Notes
1. Zhu, B.-Y.; Scarborough, R. M. Curr. Opin. Cardiol. Pulm.
Renal Invest. Drugs 1999, 1, 63.
2. Fevig, J. M.; Wexler, R. R. Annu. Rep. Med. Chem. 1999,
34, 81.
3. Ewing, W. R.; Pauls, H. W.; Spada, A. P. Drugs Future
1999, 24, 771.
4. Zhu, B.-Y.; Scarborough, R. M. Annu. Rep. Med. Chem.
2000, 35, 83.
5. Rai, R.; Sprengeler, P. A.; Elrod, K. C.; Young, W. B.
Curr. Med. Chem. 2001, 8, 101.
6. Allen, D. A.; Hataye, J. M.; Hruzewicz, W. N.; Kolesnikov,
A.; Mackman, R. L.; Rai, R.; Spencer, J. R.; Verner, E. J.;
Young, W. B. PCT International Application WO 00/35886,
2000.
7. Day, R. F.; Young, W. B. PCT International Application
WO 99/26933, 1999.
8. Fatheree, P. R.; Jenkins, T. E.; Li, Y.; Linsell, M. S.; Rai,
R.; Shrader, W. D.; Trapp, S. G.; Young, W. B. PCT Inter-
national Application WO 99/26932, 1999.
9. Fatheree, P. R.; Jenkins, T. E.; Linsell, M. S.; Trapp, S. G.;
Verner, E. J.; Young, W. B. US6150379, 2000.
10. Verner, E. J.; Katz, B. A.; Spencer, J.; Allen, D.; Hataye,
J.; Hruzewicz, W.; Hui, H. C.; Kolesnikov, A.; Li, Y.; Luong,
C.; Martelli, A.; Kesavan, R.; Rai, R.; She, M.; Shrader, W.;
Sprengeler, P. A.; Trapp, S.; Wang, J.; Young, W. B.; Mack-
man, R. L. J. Med. Chem. 2001, in press.
11. Inhibition assays for fXa, fIIa, plasmin and trypsin were
performed as described (Cregar, L.; Elrod, K. C.; Putnam, D.;
Moore, W. R. Arch. Biochem. Biophys. 1999, 366, 125) with
the pH adjusted to 7.4. Coagulation experiments were carried
out with human plasma and a coagulation meter (Electra 800)
in accordance with the manufacturer’s instructions.
12. Katz, B. A.; Elrod, K.; Luong, C.; Rice, M.; Mackman,
R. L.; Sprengeler, P. A.; Spencer, J.; Hataye, J.; Janc, J.; Link,
J.; Litvak, J.; Rai, R.; Rice, K.; Sideris, S.; Verner, E.; Young,
W. J. Mol. Biol. 2001, 307, 1451.
The synthesis of the lead inhibitor 10 is illustrated in
Scheme 1. All other analogues depicted were synthe-
sized in an analogous manner, although the requisite
acetophenones were utilized.6 All compounds described
were generated by utilizing a Fischer–Indole cyclization
as the terminal step. The synthesis of 10 begins with 3-
acetyl-4-hydroxy-phenylacetic acid (24) as previously
outlined.6 Condensation of 24 with benzaldehyde, fol-
lowed by catalytic hydrogenation, and subsequent
treatment with N-bromosuccinimide gave 25. Ketone 25
was then treated with hydrazine 26 and the resultant
hydrazone was heated in polyphosphoric acid at 125–
130 ꢁC for less than 1 h to give 10.
Conclusion
Utilizing structure-based design, a nonselective screen-
ing lead was optimized to produce a series of selective,
subnanomolar inhibitors of factor Xa. Interestingly, this
series demonstrates a S1–S10 binding mode that is
unique from most other reported factor Xa inhibitors
which generally bind from S1 to S4. Pharmacokinetics,
in vivo animal efficacy, and non-amidino P1 analogue
development will be the subject of a future publication.