Nonbenzamidine isoxazoline derivatives as factor Xa inhibitors

Article abstract of DOI:10.1016/S0960-894X(03)00080-5

Factor Xa (fXa) is an important serine protease in the blood coagulation cascade. Inhibition of fXa has emerged as an attractive target for potential therapeutic applications in the treatments of both arterial and venous thrombosis. Herein, we describe a series of non-benzamidine isoxazoline derivatives as fXa inhibitors. The chloroaniline group was found to be the most potent benzamidine mimic in this series. Chloroaniline 1 (ST368) has a K_i value of 1.5 nM against fXa and is highly selective for fXa relative to thrombin and trypsin.

Full text of DOI:10.1016/S0960-894X(03)00080-5

Bioorganic & Medicinal Chemistry Letters 13 (2003) 1023–1028
Nonbenzamidine Isoxazoline Derivatives as Factor Xa Inhibitors
Mimi L. Quan,* Christopher D. Ellis, Ming Y. He, Ann Y. Liauw,
Patrick Y. S. Lam, Karen A. Rossi, Robert M. Knabb, Joseph M. Luettgen,
Matthew R. Wright, Pancras C. Wong and Ruth R. Wexler
Bristol-Myers Squibb Co., PO Box 80500, Wilmington, DE 19880-0500, USA
Received 15 October 2002; revised 21 January 2003; accepted 22 January 2003
Abstract—Factor Xa (fXa) is an important serine protease in the blood coagulation cascade. Inhibition of fXa has emerged as an
attractive target for potential therapeutic applications in the treatments of both arterial and venous thrombosis. Herein, we describe
a series of non-benzamidine isoxazoline derivatives as fXa inhibitors. The chloroaniline group was found to be the most potent
benzamidine mimic in this series. Chloroaniline 1 (ST368) has a Ki value of 1.5 nM against fXa and is highly selective for fXa
relative to thrombin and trypsin.
# 2003 Elsevier Science Ltd. All rights reserved.
Introduction
compounds5À7 with improved inhibitory activity, effi-
cacy, and pharmacokinetics. These compounds are
potent fXa inhibitors and anticoagulants as shown by
SK549 and SM084.5À7 However, these fXa inhibitors
still have a relatively short half-life and low oral bio-
availability, in part due to the basicity of the benzamidine
group. Therefore, the benzamidine group was replaced
with less basic and non-basic benzamidine mimics to try
to enhance the pharmacokinetic profile and oral bio-
availability of these inhibitors.
Thrombotic disorders are a major cause of mortality in
Western society. Thrombosis is the result of a blood
clot, or a thrombus, in a blood vessel results in occlu-
sion of the blood flow. Thrombin, the final serine pro-
tease in the blood coagulation cascade, is responsible
for the conversion of fibrinogen to insoluble fibrin
eventually leading to blood clotting. Factor Xa (fXa), a
trypsin-like serine protease, is a critical enzyme which in
combination with fVa and ca2+ on a phospholipid sur-
face forms the prothrombinase complex that converts
biologically inactive prothrombin to thrombin. The
activation of thrombin by fXa is a highly amplified
process.1 As a result, inhibition of fXa may be more
effective than direct inhibition of thrombin in regulation
of haemostasis. In addition, inhibitors of fXa may have
a more favorable efficacy: safety ratio compared to
direct inhibitors of thrombin as demonstrated in animal
models of thrombosis.2 Accordingly, fXa has emerged
as an important target for the development of new
antithrombotic agents.3
Results and Discussion
A variety of benzamidine mimics were employed in an
isoxazoline and a pyrazole series, which afforded potent
and selective fXa inhibitors.8À10 Several structure-based
P1 libraries were designed and synthesized. Results from
these libraries indicated that chloroaniline was a rela-
tively potent and selective benzamidine mimic, approxi-
mately 15–100-fold less potent for fXa than the
corresponding benzamidine analogues.10 As shown in
Figure 1, chloroaniline 2 was determined to be
approximately 20-fold less potent compared to the ben-
zamidine analogue 3 while chloroaniline 4 was about
60-fold less active than the corresponding benzamidine
5. These results suggest that the structural activity rela-
tionships derived from substitution of the 5-position of
the isoxazoline may be different when a chloroaniline
group is employed as the P1 moiety. Therefore a series
Previous reports from our laboratory have described a
series of isoxazoline benzamidine derivatives as fXa
inhibitors.4À7 We have shown that the bisbenzamidine
compounds4 can be elaborated to monobasic
*Corresponding author. Tel.: +1-609-818-5301; fax: +1-609-818-
3331; e-mail: mimi.quan@bms.com
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00080-5

1024
M. L. Quan et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1023–1028
Figure 1. Comparison of benzamidine and chloroaniline in the P1 position.
of compounds was prepared with chloroaniline in the P1
position and various substituents at the 5-position of
the isoxazoline.
Ethyl substitution at the terminal urea nitrogen
decreased fXa affinity significantly (9). Replacement of
the terminal urea NH2 group with either a methyl group
(10) or hydrogen (11) gave compounds 4–6 times less
active. The sulfamide analogue 12 was 2Â less potent.
The fXa activity was regained when the terminal NH2 of
12 was replaced with a methyl moiety (13). Other sulfo-
namides (14–19) also showed low nanomolar affinity for
fXa, although no significant improvement over 13 was
observed (Table 1). As previously observed, all of the
chloroanilines demonstrated very good selectivity for
fXa relative to thrombin and trypsin.10 This selectivity
can be explained by the smaller S1 pocket in trypsin
active site relative to Xa, thereby causing unfavorable
interaction by the change from the smaller hydrophobic
residue alanine to the somewhat larger and more
hydrophilic serine at position 190.
Five-membered ring heterocycles at the 5-position of the
isoxazoline were initially evaluated because analogues
such as 5 showed good fXa potency in the benzamidine
series.5À6 Tetrazole 4, pyrazole 6, and 1,2,4-triazole 7
were found to have similar fXa affinity in the range of
10–35 nM (Table 1). During the synthesis of 4, a more
potent by-product (fXa Ki of 2.5 nM) was isolated and
was identified to be urea 8. To follow up on this lead, the
series of compounds shown in Table 1 were prepared.
Table 1. Isoxazoline derivatives with various substitutions at R5
Several sets of enantiomers were obtained by chiral HPLC
separation as shown in Table 2. In each case, the negative
Table 2. Comparison of the enantiomers
Compd
R
fXa Ki
(nM)
Thrombin
Ki (nM)
Trypsin
Ki (nM)
5 (Æ)
4 (Æ)
0.17
10
1100
>21,000
31
>1600
6 (Æ)
7 (Æ)
14
34
>21,000
>21,000
>1600
>4200
Compd
R
fXa Ki
(nM)
Thrombin
Ki (nM)
Trypsin
Ki (nM)
20 (À)
21 (+)
22 (À)
23 (+)
7.1
320
19,000
>21,000
>21,000
>21,000
>1600
>1600
>1600
>1600
8 (Æ)
NHCONH2
NHCONHEt
NHCOMe
NHCOH
NHSO2NH2
NHSO2Me
NHSO2-nPr
NHSO2CH2CF3
NHSO2Ph
2.5
69
11
15
5.5
1.9
3.7
3.3
1.7
1.2
3.0
2.2
9300
>6300
>21,000
14,000
5900
1300
18,000
6600
>21,000
>21,000
17,000
6200
>1600
>4200
>1600
>8000
>8000
>8000
>8000
>8000
>8000
>8000
>8000
>4200
9 (Æ)
10 (Æ)
11 (Æ)
12 (Æ)
13 (Æ)
14 (Æ)
15 (Æ)
16 (Æ)
17 (Æ)
18 (Æ)
19 (Æ)
15
1100
24 (À)
25 (+)
1 (À)
NHCONH2
NHCONH2
NHSO2Me
NHSO2Me
2.5
69
1.5
800
>21,000
800
>1600
>6000
>8000
>4200
NHSO2-thiophen-3-yl
NHSO2-3-pyridyl
NHSO2CH2Ph
26 (+)
240
>21,000
fXa, thrombin, and trypsin Ki values were obtained using human
purified enzymes. See ref 11 for assay conditions.
fXa, thrombin, and trypsin Ki values were obtained using human
purified enzymes. See ref 11 for assay conditions.

M. L. Quan et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1023–1028
1025
Table 3. Compounds with nonbenzamidine mimics at P1
enantiomer was found to be more potent. This is in agree-
ment with what was observed earlier for the benzamidine
series. Methylsulfonamide ST368 (1) was the most potent
fXa inhibitor prepared in this series (fXa Ki=1.5 nM).
The chloroaniline moiety in 13 was also replaced by
other benzamidine mimics10 as shown Table 3. Amino-
benzisoxazole 27 and p-methoxyphenyl 28 have fXa Ki
values of 4.6 and 10 nM, about two- and five-fold less
potent than the corresponding chloroaniline 13, respec-
tively. These two compounds are also highly selective
for fXa related to other serine protease.
Compd
P1
fXa Ki
(nM)
Thrombin
Ki (nM)
Trypsin
Ki (nM)
1.9
1300
>8000
13
Computer modeling studies using Insight II (V.98.0,
MSI, San Diego, CA) and the CFF98 Forcefield to
compare SM084 and 20 (Fig. 2) have shown that 20
penetrates deeper down into the S1 specificity pocket in
order for the aniline group to interact with Asp189. The
aniline moiety is capable of interacting with Asp189 in a
similar way as achieved for SM084; but the NH2 of 20 is
not able to form the additional hydrogen bond with
Gly218 of the enzyme backbone as observed with
SM084. However, the chlorine is capable of forming a
lipophilic interaction with Val213, which may have
resulted in the enhanced fXa affinity. A nitrogen in the
tetrazole group of 20 could potentially form a hydrogen
bond with either Arg143 or Gln192. Comparing 24 and 1
with 20, it appears that both the urea CO and the
27
28
4.6
5200
>5200
>4200
10
>6300
fXa, thrombin, and trypsin Ki values were obtained using human
purified enzymes. See ref 11 for assay conditions.
Figure 2. Modeling of SM084 and 20 in the factor Xa active site.

1026
M. L. Quan et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1023–1028
Table 4. PKprofiles of selective compounds in dogs
methylsulfonamide SO ought to be capable of forming a
hydrogen bond with Arg143 or Gln192. Additionally, the
NH group is within binding distance of Gly218. The
enhanced Xa affinity may be due to the urea and the
sulfonamide being better hydrogen bond acceptors than
the tetrazole. In addition, the methyl, phenyl, and thio-
phenyl groups have the potential for a favorable lipo-
philic interaction with the Cys 220/Cys191 disulfide bond.
Compd
Dose (mg/kg)
Clearance (L/h/kg)
t1/2 (h)
SM084
24
ST368 (1)
5.0
0.5
0.5
1.32
0.43
0.53
1.1
2.32
7.25
PKData for SM084 were obtained by intravenous dosing in a discrete
study. Data for other compounds were obtained from intraveneous
cassette dosing studies N-in-1 iv studies.
Computer modeling also shows that the amino group of
aminobenzisoxazole can form a hydrogen bond with
Gly218 and Asp189 and the other nitrogen can interact
with Asp189 through a water molecule. Compared to the
chloroaniline, the aminobenzisoxazole group shows simi-
lar interactions with the enzyme and therefore is antici-
pated to have similar affinity. Although the methoxy
group of the p-methoxyphenyl can interact with the
enzyme in a similar way to the chloro group of the
chloroaniline, it lacks the potential interaction with
Table 5. Antithrombotic efficacy of selective compounds
Compd
Human fXa
Ki (nM)
Rabbit fXa
Ki (nM)
ID50
(mmol/kg/h)
SM084
24
ST368 (1)
0.11
2.5
1.5
0.11
4.7
2.3
0.032
5.5
8.0
ID50 values were determined in the rabbit Arterio-Venous shunt
model.7
Scheme 1. Synthesis of chloroaniline derivatives. (a) NH2OH/EtOH/pyridine; (b) NaN3 or tetrazole/DMSO; (c) bleach/CH2Cl2; (d) LiOH/THF; (e)
SOCl, 35/Et3N/CH2Cl2; (f) TFA, SnCl2/EtOAc; (g) P(OEt)3/THF, H+; (h) RCOCl or RSO2Cl/Et3N; (i) TFA, SnCl2/EtOAc; (j) ClSO2NCO/formic
acid/Et3N; (k) TFA, SnCl2/EtOAc.

M. L. Quan et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1023–1028
1027
Scheme 2. Synthesis of aminobenzisoxazole SV27. (a) NH2OH/EtOH/pyridine; (b) bleach/CH2Cl2; (c) P(OEt)3/THF, H+, CH3SO2Cl; (d) LiOH/
THF; (e) (COCl)2, 35/Et3N/CH2Cl2; (f) (CH3)2CNOH/KO-t-Bu, EtOH/H+, TFA.
Asp189. Hence, the 5-fold decrease in fXa affinity is not
surprising.
the azide intermediate 36. Reduction of the azide group
with P(OEt)3 followed by acid hydrolysis afforded
amine 37. Acylation of 37 with acetyl chloride gave 38a.
Reaction with the requisite sulfonyl chloride produced
the sulfonamide 38b. Deprotection of the sulfonamide
with TFA followed by reduction provided compounds
10 and 13–19. Reaction of amine 37 with ClSO2NCO/
formic acid/Et3N afforded 39 and 40. These two com-
pounds were separated and then carried on the desired
targets individually by deprotection followed by reduc-
tion of the nitro group with SnCl2.
To determine the pharmacokinetic profile of these com-
pounds relative to the corresponding benzamidine,
Compounds 24 and ST368 (1) were studied in dogs
dosed at 0.5 mg/kg intravenously (Table 4). ST368
showed an improved clearance value (0.5 L/kg/h vs 1.32
L/kg/h) and prolonged half-life (7.25 h vs 1.1 h) com-
pared with SM084. These two compounds (24 and 1)
were evaluated for antithrombotic efficacy in a rabbit
arterio-venous shunt thrombosis model7 and the results
relative to SM084 are shown in Table 5. The ID50 values
were determined to be 5.5 and 8.0 mmol/kg/h, respec-
tively, compared to 0.032 mmol/kg/h for SM084. In the
in vitro clotting assay using human plasma, compounds
24 and ST368 double the APTT at 11 and 4.6 mM,
respectively.
The synthesis of the aminobenzisoxazole 27 is shown in
Scheme 2. 3-Cyano-4-fluorobenzaldehyde was con-
verted to oxime 42 with hydroxylamine. Reaction of 42
with bleach followed by [2+3] cycloaddition with 43 gave
isoxazoline 44. Reduction of the azide group with P(OEt)3
followed by sulfonylation with methanesulfonyl chloride
produced sulfonamide 45. The ester of 45 was hydrolyzed
to the corresponding acid 46, which was then coupled
with biarylamine 35 to afford amide 47. Reaction with
acetone oxime followed by heating in EtOH/HCl gener-
ated the aminobenzisoxazole. Lastly, removal of the
t-butyl group with TFA gave the desired product 27.
Compound 28 with a methoxyphenyl P1 was prepared
similarly using p-methoxybenzaldehyde as the starting
material.
Although we have improved the pharmacokinetic pro-
file in terms of clearance and half-life with ST368, the
A-V shunt and APTT results indicate that ST368 is not
potent enough for further preclinical progression. The
low efficacy of ST368 may be due to high protein bind-
ing. Therefore, compounds discussed in this paper were
not evaluated in oral bioavailability studies.
Synthesis
Conclusions
The tetrazole analogues 4 and 20 were prepared utilizing
the same methods previously described for the synthesis
of SM084,6 but with the use of 4-chloro-3-nitro-
benzaldehyde as the starting material (Scheme 1). The
urea 8 was isolated as a by product from the last two
steps of the tetrazole formation reactions (TFA, SnCl2).
The triazole and pyrazole analogues were prepared in a
similar manner as the preparation of the tetrazoles. The
other chloroaniline derivatives were synthesized from
We have designed and synthesized a series of isoxazo-
line compounds employing P1 benzamidine. The chloro-
aniline moiety was found to be the best benzamidine
replacement in this series. The best compound ST368
(1), while a 14-fold less potent than the corresponding
benzamidine SM084, was found to have both better
selectivity profile for fXa relative to other serine pro-
teases and an improved pharmacokinetic profile in

1028
M. L. Quan et al. / Bioorg. Med. Chem. Lett. 13 (2003) 1023–1028
terms of clearance and half-life. Efforts to further opti-
mize the pharmacokinetic profile of our fXa inhibitors
will be reported in due course.
8. Pinto, D. J.; Orwat, M. J.; Wang, S.; Fevig, J. M.; Quan,
M. L.; Amparo, E.; Cacciola, J.; Rossi, K. A.; Alexander,
R. S.; Wong, P. C.; Knabb, R. M.; Luettgen, J. M.; Aungst,
B. J.; Li, L.; Wright, M.; Jona, J. A.; Wexler, R. R.; Lam,
P. Y. S. J. Med. Chem. 2001, 44, 566.
9. Quan, M. L.; Wexler, R. R. Curr. Top. Med. Chem. 2001, 1,
137.
Acknowledgements
10. Lam, P. Y. S; Li, R.; Clark, C. G.; Adams, J. J.;
Galemmo, R. A.; Fevig, J. M.; Pinto, D. J.; Quan, M. L.; He,
M.; Han, Q.; Alexander, R. S.; Rossi, R. A.; Wong, P. C.;
Luettgen, J. M.; Aungst, B. J.; Wright, M. R.; Bai, S. A.;
Knabb, R. M.; Wexler, R. R. 220th ACS National Meeting,
Washington, DC, 2000, MEDI 329.
We wish to thanks Al Mical and his group for the chiral
separations, and Earl Crain and Carol Watson for con-
ducting the rabbit A-V shunt experiments.
References and Notes
11. Enzyme affinity assays. FXa, thrombin, and trypsin Ki
values were obtained from human purified enzymes. All
assays are run in micro-titer plates using a total volume of
250 mL in 0.1 M Sodium Phosphate Buffer containing 0.2 M
NaCl and 0.5% polyethylene glycol 6000 at pH 7.0. The
compounds are run at 10 mM, 3.16, 1.0, 0.316, 0.1, 0.0316,
0.01, and 0.00316 mM. Plates are read for 30 min at 405 nm.
Rates are determined for the controls (no inhibitor) and for
the inhibitors. % Enzyme activity is determined from these
rates and used in the following formula to determine Ki:
1. Elodi, S.; Varadi, K. Thromb. Res. 1979, 15, 617.
2. Harker, L. A.; Hanson, S. R.; Kelly, A. B. Thromb. Hae-
mostasis 1997, 78, 736.
3. (a) Betz, A. Exp. Opin. Ther. Patents 2001, 11, 1007. (b)
Rai, R.; Sprengeler, P. A.; Elrod, K. C.; Young, W. B. Current
Medicinal Chemistry 2001, 8, 101.
4. Quan, M. L.; Pruitt, J. R.; Ellis, C. D.; Liauw, A. Y.;
Galemmo, R. A., Jr.; Stouten, P. F. W.; Wityak, J.; Knabb,
R. M.; Thoolen, M. J.; Wong, P. C.; Wexler, R. R. Bioorg.
Med. Chem. Lett. 1997, 7, 2813.
5. Quan, M. L.; Liauw, A. Y.; Ellis, C. D.; Pruitt, J. R.; Car-
ini, D. J.; Bostrom, L. L.; Huang, P. P.; Harrison, K.; Knabb,
R. M.; Thoolen, M. J.; Wong, P. C.; Wexler, R. R. J. Med.
Chem. 1999, 42, 2752.
Ki=1000ÂInhibitor
Concentration/(((Km+S)-SÂACT)/
(ACTÂKm)-1); where S=Substrate Concentration; ACT=%
Enzyme Activity for inhibitor. All compounds were tested in
duplicate studies and were compared with the same internal
standards. These assays are described in detail in refs 12 and
13.
6. Quan, M. L.; Ellis, C. D.; Liauw, A. Y.; Alexander, R. S.;
Knabb, R. M.; Lam, G.; Wright, M. R.; Wong, P. C.; Wexler,
R. R. J. Med. Chem. 1999, 42, 2760.
12. Knabb, R. M.; Kettner, C. A.; Timmermans,
P. B. M. W.M.; Reilly, T. M. Thromb. Haemostasis 1992, 67,
56.
7. Wong, P. C.; Quan, M. L.; Crain, E. J.; Watson, C. A.; Wex-
ler, R. R.; Knabb, R. M. J. Pharmacol. Exp. Ther. 2000, 292, 351.
13. Kettner, C. A.; Mersinger, L. J.; Knabb, R. M. J. Biol.
Chem. 1990, 265, 18289.