5-Amidinobenzo[b]thiophenes as dual inhibitors of factors IXa and Xa

Article abstract of DOI:10.1016/j.bmcl.2004.10.045

Syntheses and SAR studies of 5-amidinobenzo[b]thiophene analogs provided compounds with low submicromolar factor IXa potency and equal or slightly better selectivity relative to factor Xa. Syntheses and SAR studies of 5-amidinobenzo[b]thiophene analogs provided compounds with low submicromolar factor IXa activity and equal or slightly better selectivity relative to factor Xa.

Full text of DOI:10.1016/j.bmcl.2004.10.045

Bioorganic & Medicinal Chemistry Letters 15 (2005) 29–35
5-Amidinobenzo[b]thiophenes as dual inhibitors
of factors IXa and Xa
Jennifer X. Qiao,* Xuhong Cheng, Dilip P. Modi, Karen A. Rossi, Joseph M. Luettgen,
Robert M. Knabb, Prabhakar K. Jadhav and Ruth R. Wexler
Bristol-Myers Squibb Pharmaceutical Research Institute, PO Box 5400, Princeton, NJ 08543-5400, USA
Received 8 July 2004; revised 11 October 2004; accepted 14 October 2004
Available online 2 November 2004
Abstract—Syntheses and SAR studies of 5-amidinobenzo[b]thiophene analogs provided compounds with low submicromolar factor
IXa activity and equal or slightly better selectivity relative to factor Xa.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
leaving hemostasis intact via the fVIIa/tissue factor
extrinsic pathway. It has been demonstrated that inhibi-
tion of fIXa either by active site-blocked fIXa or by fIX/
fIXa antibody was effective in prevention of thrombus
formation in animal models without increasing
bleeding.2
Thromboembolic diseases, such as deep vein thrombo-
sis, pulmonary embolism, myocardial infarction, and
thromboembolic stroke, are the leading causes of
morbidity and mortality in the developed countries.
Conventional antithrombotic therapies using either
heparin or warfarin have several limitations, including
excessive bleeding, due to their targeting multiple factors
within the coagulation cascade. To improve the benefit-
to-risk ratio of antithrombotic drugs, small-molecule
direct thrombin inhibitors and selective factor Xa
(fXa) inhibitors are being extensively investigated.
These results suggested that potent and selective small-
molecule fIXa inhibitors may have the potential to be
viable therapeutic agents for the treatment of thrombo-
embolic disorders with a reduced bleeding liability. Our
initial goal was to identify a reversible, small-molecule,
selective fIXa inhibitor as a tool molecule for animal
studies.
The coagulation cascade is a proteolytic sequence of
events, involving several serine proteases, leading to
overall maintenance of hemostasis. The cascade consists
of three systems: the intrinsic and extrinsic pathways
that provide alternative routes for the generation of
fXa, and a final common pathway that leads to throm-
bin formation. Factor IXa (fIXa) in the presence of
cofactor fVIIIa and Ca2+ activates fX to fXa on the sur-
face of platelets or endothelial cells via the intrinsic
pathway.1 FIXa seems to be essential for the amplifica-
tion of coagulation, as indicated by the high bleeding
tendency of hemophilia B patients (fIX deficiency). Inhi-
bition of fIXa may provide ꢀupstreamꢁ inhibition of
intrinsic coagulation and thrombus propagation, while
H2NO2S
HN
R
R'
HN
NH
O
NH
H2N
O
H2N
S
N
1
2
H
fIXa Ki 0.66 µM
fXa Ki 0.00063 µM
High throughput screening of our proprietary fXa inhibi-
tor collection identified amidinoindole 1.3 Compound 1
is a submicromolar lead for fIXa, however it is approxi-
mately 1000-fold more potent for fXa (fIXa Ki 0.66lM,
fXa Ki 0.00063lM).4 In this paper, we describe the
Keywords: Factor IXa; Anticoagulant; Benzo[b]thiophene.
*
Corresponding author. Tel.: +1 609 818 5298; fax: +1 609 818
6810; e-mail: jennifer.qiao@bms.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.10.045

30
J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
syntheses and structure activity relationships developed
in an effort to optimize 1, which resulted in the 5-amidino-
benzo[b]thiophene series 2 with both improved inhibi-
tory activity for fIXa and improved selectivity relative
to other relevant trypsin-like serine proteases.
provided the common intermediate 10 as a TFA salt.
Coupling of 10 with both aromatic and aliphatic amines
using diisopropylcarbodiimide in pyridine/DMSO or
pyridine/DMF afforded amides 2a in moderate to high
yields. A broad range of functional groups incompatible
with the conditions of the Pinner reaction, such as
ketones and nitriles, were introduced into the final prod-
ucts using this methodology.
2. Chemistry
Scheme 1 depicts the preparation of 5-amidinobenzo[b]-
thiophene analog 3 and its homolog 4. Pd-catalyzed cyan-
ation of methyl 5-cyano-3-benzothiophene acetate with
zinc cyanide gave the methyl ester 5a.5 Alternately, a
three-step sequence provided the ethyl ester 5b, namely,
alkylation of 4-bromothiophenol with ethyl chloroaceto-
acetate followed by cyclization with polyphosphoric acid
using a modified literature method,6 and then displace-
ment of the bromide with copper cyanide. Alkylation
of 5a or 5b with benzyl bromide led to the corresponding
ester 6a or 6b, which was then hydrolyzed to give the acid
7. Coupling of 7 with aniline 8 followed by Pinner chem-
istry/ammonolysis afforded 3. Alternatively, the acid
chloride of 7 reacted with TMS diazomethane gave the
a-TMS diazoketone 9. An Arndt–Eistert reaction of 9
with aniline 8 in 2,4,6-collidine at reflux provided 4.
Enantiomerically pure compounds of this series were
prepared using either Evansꢁ chiral auxiliary synthesis
or chiral HPLC separation of the racemic amidine pre-
cursors. For instance, Scheme 3 illustrates the produc-
tion of isomers of compound 11 using an oxadiazole
as a masked amidine. Condensation of racemic acid 12
with enantiopure methyl 2-amino-1-cyclohexane carb-
oxylate (1R,2S)-13 or (1S,2R)-137 followed by hydro-
genation gave the racemic mixture 11a and 11b,
respectively. Compounds 11c and 11d, the two diastereo-
mers of 11a, were obtained by chiral HPLC separation
of the racemic mixture 14 followed by hydrogenation.
The preparation of 5-amidinobenzo[b]furan analog 15
was carried out using transformations similar to those
described in the literature8 and to those shown in
Scheme 1 for the analogous benzo[b]thiophenes. The
benzo[b]thiophene 1,1-dioxide 16 was prepared by oxi-
dation of 10 with oxoneÒ followed by coupling with
racemic cis-13.
Solution phase parallel synthesis was utilized to prepare
a range of amides (Scheme 2). Conversion of nitrile 6a to
the corresponding amidine followed by acid hydrolysis
OH
Ph
O
Cl
OR
a,b
OR
f
O
S
O
NC
NC
c,d,e
Br
S
S
6a R = Me
5a R = Me
5b R = Et
SH
6b R = Et
Ph
Ph
Ar
OH
O
HN
g
NH
h, i
O
NC
H2N
8
(ArNH2)
S
S
3
7
j, k
Ph
Ph
S
O
l, i
H
N
NH
N2
NC
Ar
8
H2N
SiMe3
O
S
9
4
N
Ar =
N
Scheme 1. Reagents and conditions: (a) MeOH, cat. H2SO4; (b) Zn(CN)2 (1.5equiv), Pd2(dba)3 (5mol%), dppf (10mol%), Zn (30mol%), DMAC,
140°C, 83% for two steps; (c) ClCH2COCH2COOEt, pyridine; (d) PPA, 65°C; (e) CuCN, DMF, 140°C, 35% for three steps; (f) LiN(TMS)2, THF,
then PhCH2Br, ꢀ78°C, 95%; (g) NaOH, EtOH, 100%; (h) BOPCl, NMM, 8, DMF, 59%; (i) HCl, MeOH; then (NH4)2CO3, MeOH, 38%;
(j) (COCl)2, CH2Cl2; (k) TMSCHN2, THF/CH3CN (1:1), 0°C; (l) 8, collidine, 180°C, yield for j–l, 45%.

J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
31
Ph
Ph
R
HN
OMe
O
OH
O
R'
NH.TFA
NH.TFA
a, b
c, d
O
NC
H2N
H2N
S
S
S
6a
10
2a R'=CH2Ph
Scheme 2. Solution phase syntheses of 2a. Reagents and conditions: (a) HCl, MeOH; then (NH4)2CO3, MeOH; (b) 6N HCl, reflux; (c) NH2–R, DIC,
pyridine/DMSO (1:4 v/v); (d) RP-HPLC or LC–MS purification, 30–75% overall yields.
a, b
H2N
2
1
c,d
MeO
MeOOC
O
O
Ph
.
NH2 HCl
O
N
HN
Ph
Ph
(1R, 2S)-13
NC
NC
OMe
Ph
S
OEt
O
OH
O
O
N
e, f
g
(1R, 2S)-13
S
12
6b
Ph
Ph
S
HN
HN
O
NH
h
N
CO2Me
CO2Me
O
O
H2N
N
S
11a
14
Ph
HN
g, h
(1S, 2R)-13
NH
CO2Me
12
14
O
H2N
11b
S
i
11c
11d
+
Scheme 3. Reagents and conditions: (a) LDA, THF, 75%; (b) YbOTf, PhH, 72%; (c) NaBH(OAc)3, CH3CN, 78%; (d) H2, Pd–C (10%), HCl, MeOH,
82%; (e) NH2OHÆHCl, Et3N, EtOH; then Ac2O, HOAc, 78%; (f) NaOH, EtOH, 100%; (g) DIC, pyridine/DMSO (1:4 v/v), 78%; (h) H2, Pd–C (5%),
MeOH, 85–90%; (i) chiral HPLC separation, then H2, Pd–C (5%), MeOH.
Using transformations similar to those described in the
literature,9 the 7-cyanonaphthalene derivative 17 was
prepared from a-tetralone (Scheme 4).
following two possibilities. First, the amidine NH forms
an extra hydrogen bond with Ser190 in fIXa, restricting
the orientation of the inhibitor in the S1 pocket such
that the geometrical requirement of hydrogen bond for-
mation between the indole NH and Ser195 may not be
optimal. Second, fIXa has a smaller S1 pocket due to
amino acid residue differences (fIXa Ser190 vs fXa
Ala190, and fIXa Ile213 vs fXa Val213). As a result,
the 5-amidinoindole group may be too large to fit well
into the S1 pocket of fIXa.
3. Results and discussion
The 5-amidinoindole group has been demonstrated,
both computationally and experimentally, to have higher
potency and selectivity for fXa compared with several
other amidino bicyclic P1 groups not containing an
N–H.10 This difference is presumably due to the ability
to form the hydrogen bond between the indole NH
and the oxygen atom of Ser195. In contrast, amidino-
indole 1 showed weaker fIXa potency compared with
fXa potency. This may be due to one or both of the
Initially, to eliminate the possibility of a hydrogen bond
with Ser195, we replaced the indole ring with a
benzo[b]thiophene ring. To our delight, 5-amidino-
benzo[b]thiophene analogs 3 and 21 containing a para
benzimidazole aniline P4 group11 were found to be

32
J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
had much decreased fIXa activity while retaining fXa
COOEt
O
O
potency. Compound 4, a homolog of 3, is 10-fold less
potent for fIXa (fIXa Ki = 1.25lM) compared with 3,
while its fXa potency (fXa Ki = 0.28lM) remains similar
to that of 3. Compound 26 containing a 4-phenylimida-
zole P4 moiety, is less potent for fIXa compared with 3;
however, it is sixfold more selective for fIXa than for
fXa. As previously demonstrated via the overlay of the
Ca trace of the fIXa-bound and the fXa-bound crystal
structures,11 the conformation and geometry of the 99-
loop of fIXa is quite different compared with fXa. The
phenyl group on the imidazole ring in 26 might bump
into the 99-loop of fXa, which resulted in a 20-fold
decrease of fXa potency compared with 3.
a
b
Br
Br
18a 7-Br
18b 5-Br
18a: 18b = 1:1
19a endo
19b exo
19a: 19b = 3:1
c, d
Ph
COOEt
COOEt
NC
e, f
Br
17
Scheme 4. Reagents and conditions: (a) AlCl3, Br2, 80°C; column
chromatography followed by crystallization from hexane, 46% for 18a;
(b) (C2H5O)2POCH2COOC2H5, NaH, THF, 0°C reflux, 59%; (c) Br2,
CCl4; (d) DBU, benzene, 90% for two steps; (e) CuCN, DMF, 140°C,
65%; (f) LiN(TMS)2, THF, then PhCH2Br, ꢀ78°C, 69%.
We anticipated that modification in the P4 region might
provide further opportunity to improve selectivity.
Combining the tools of computational library design
and automated synthesis, 300 amines were selected via
informative library design14 and manually picked to
allow for maximum spatial diversity of the P4 region.
Several compounds showed submicromolar fIXa affinity
and 6- to 10-fold selectivity for fIXa compared to fXa.
However, none of the compounds were significantly
more potent or more selective than 26. Of the 300 com-
pounds prepared, compounds 27 and 28 were the two
most potent and selective fIXa inhibitors among com-
pounds prepared with Ar–X–Ar type of P4 groups
(Table 2). Compound 29, containing a benzothiazole
group, was the most potent and selective compound
containing a 5,6-, or 6,6-fused heterocyclic P4 group.
Because of the much decreased fXa potency observed
with 29, it is possible that this compound does not bind
the same way as the earlier mentioned benzothiophenes
such as 3. Modeling of 29 suggests an alternate binding
mode that places the benzothiazole ring system above
the S1 pocket near the disulfide of Cys191–Cys220.
essentially equipotent for fIXa and fXa (Table 1). Rela-
tive to the corresponding indoles 20 and 22, the fIXa
potency was retained while the fXa potency decreased
10- to 20-fold. The decreased fXa Ki in the benzo[b]thio-
phene series confirms the important role of a hydrogen
bond with Ser195 for fXa potency. The retained fIXa
potency suggests that the structure around the S1 site
of fIXa may be more ꢀplasticꢁ than that around the S1
site of fXa. The hydrophobic interaction of the benzo-
thiophene ring with the wall of the S1 pocket and the
polarizable and sterically larger sulfur atom appear to
be favorable in improving selectivity relative to fXa.
Compound 23 bearing a 1-naphthylmethyl group, com-
pared to compound 24 with an ethyl group, though
equipotent in fIXa, provided slightly better fIXa/fXa
selectivity. On the other hand, cyclopropyl analog 25
Table 1.
N
Y
Ph
N
H
N
NH.TFA
NH.TFA
NH.TFA
N
R'
H2N
HN
O
N
Ph
HN
H2N
H2N
Ph
N
N
O
O
26
X
S
4
S
Compound
X
Y
R0 (+/ꢀ)
fIXa Ki, lM
fXa Ki, lM
fIXa/fXa
1048
0.8
25
1.6
19
0.3
3.1
223
Trypsin Ki, lM
Thrombin Ki, lM
1
0.66
0.00063
0.12
0.15
1.02
0.34
>4.2
>4.2
3.90
0.27
0.06
>4.2
>4.2
0.60
>6.3
13
3
S
H
F
CH2Ph
CH2Ph
0.098
0.063
0.074
0.075
0.23
20
21
22
23
24
25
4
NH
S
0.0025
0.047
0.004
0.74
H
H
H
H
H
CH2Ph-p-NH2
CH2Ph-p-NH2
CH2-1-naphthyl
CH2CH3
4.2
4.2
NH
S
>6.3
0.27
0.39
4.2
S
0.22
4.90
0.071
0.022
0.28
S
–CH2CH2–
1.25
0.39
4.5
0.2
26
2.02
>6.3
All compounds were purified by either reverse phase HPLC or preparative LC/MS (water/acetonitrile gradient + 0.5% TFA). The basic compounds
were isolated as TFA salts following lyophilization. All compounds gave satisfactory spectral and analytical data.
Human purified enzymes were used. Values are averages from multiple determinations (n > 2). Ki values were calculated based on the inhibition
observed over a range of inhibitor concentrations (0.001–50lM) as described in Ref. 12 and Ref. 13.

J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
33
Table 2.
Table 3.
Ph
S
NH.TFA
Ph
HN R
O
NH.TFA
N
HN
O
H2N
H2N
S
S
Compound R (+/ꢀ)
fIXa Ki, fXa Ki, fIXa/
lM
Compound Stereochemistry fIXa Ki, fXa Ki, fIXa/fXa
lM
lM
fXa
lM
Cl
29
(+/ꢀ)
R-(ꢀ)
S-(+)
0.78
0.48
>8
<0.1
0.2
O2
27
S
0.93
5.36
0.2
N
29-R
29-S
2.7
5.7
N
H
N
14.22
2.5
O
Ph
28
29
0.25
0.78
0.45
0.6
The simple substituted arylamides, 30 and 31, although
less potent for fIXa relative to benzimidazole phenyl-
amides such as 3, showed slightly enhanced fIXa selecti-
vity relative to fXa. Compound 32 bearing a 4-
methylcyclohexyl group demonstrated a sixfold fIXa/
fXa selectivity. Compounds 33 and 11 were the most
potent fIXa inhibitors synthesized in this library; how-
ever, they showed less selectivity than 29. Compound
11, containing a mixture of four diastereomers with a
cis-cyclohexyl ring conformation, also had fIXa potency
in the hundred nanomolar range although it had simi-
lar inhibitory activity against fXa. Molecular modeling
suggests that the cyclohexyl group may interact with
the Tyr99 region of fIXa, which is at the entrance of
the S3/S4 pocket. Compound 11c, the most potent single
diastereomer of 11, had similar fIXa potency and selec-
tivity as 11.
S
>8
<0.1
N
30
31
1.73
0.96
>8
>8
<0.2
<0.1
CN
F
H3C
32
0.58
0.13
0.14
0.31
3.24
0.2
0.3
1.2
1.0
Me
33
0.41
0.12
0.3
N
N
CF3
Efforts to modify the methyl ester group in 11 did not
improve fIXa potency or selectivity (Table 4). Replace-
ment of the methyl ester in 11 with a larger group such
as benzylamide or hydrolysis of the ester to the acid
resulted in compounds with decreased fIXa potency
(37, 38). On the other hand, the smaller ethyl ester, ethyl
amide, and glycine amide were tolerated (34, 35, 36).
11
(cis)
MeOOC
HN
11a
MeOOC
(+/ꢀ at benzyl position)
Variation of the heteroatoms in bicyclic P1 group
(X = O, NH, SO2) resulted in compounds with
decreased fIXa potency (Table 5). Interestingly,
compound 40 containing a 7-amidinonaphthalene P1
group showed potency and selectivity for fIXa similar
HN
11b
11c
11d
0.50
0.28
0.09
0.27
5.6
1.0
—
MeOOC
(+/ꢀ at benzyl position)
HN
MeOOC
(R at benzyl position)
Table 4.
Ph
HN
NH.TFA
H2N
HN
R
O
41.3
>9.0
S
MeOOC
(S at benzyl position)
Compound Ra
fIXa Ki, fXa Ki, fIXa/fXa
lM lM
11
34
35
36
37
38
COOMe
COOEt
CONHEt
0.14 0.12
1.2
0.7
3.3
4.3
Unfortunately, a crystal structure of 29 complexed with
fIXa has not been obtained. The two enantiomers of 29
had very different fIXa potency (Table 3). The R isomer
(ee >98%, [a]D = ꢀ80° (MeOH)) was the more potent
fIXa inhibitor and showed fivefold selectivity for fIXa,
similar to that observed with a related indole inhibitor.4
0.12
0.43
0.28
0.18
0.13
0.065
CONHCH2CONH2
CONHCH2Ph
COOH
>32
3.50
>9
0.20
—
18
a Stereochemistry of two substituents on the cyclohexyl is cis-racemic.

34
J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
Table 5.
Ph
NH
HN
H2N
O
COOR1
X
Compound
X
R1
fIXa Ki, lM
fXa Ki, lM
Trypsin Ki, lM
fIXa/fXa
39
34
15
16
40
NH
S
Me
Et
Et
Et
Et
0.34
0.12
1.19
8.53
0.24
0.019
0.18
0.27
1.5
2.5
18
0.7
4.4
—
O
SO2
>4.2
>4.2
>4.2
>9.0
0.53
–HC@CH–
0.5
Bugelski, P.; Nichols, A. J.; Church, W. R.; Valocik, R.;
Koster, P.; Baker, A.; Blackburn, M. N. Arterioscl.
Throm. Vas. 1999, 19, 2554; (d) Feuerstein, G. Z.;
Toomey, J. R.; Valocik, R.; Koster, P.; Patel, A. Thromb.
Haemostasis 1999, 82, 1443; (e) Spanier, T.; Oz, M. C.;
Minanov, O. P.; Simantov, R.; Kisiel, W.; Stern, D. M.;
Rose, E. A.; Schmidt, A. M. J. Thorac. Cardiovasc. Surg.
1998, 115, 1179; (f) Refino, C. J.; Jeet, S.; DeGuzman, L.;
Bunting, S.; Kirchhofer, D. Thromb. Haemostatsis 1999,
82, 1188.
to amidinobenzothiophene 34. SAR studies with a vari-
ety of replacements for 5-amidinobenzo[b]thiophene
demonstrated, once again, that the S1 pocket is a finger-
print region of serine proteases.15
In the above discussion, we did not test the second pos-
sibility: fIXa has a smaller S1 pocket, and the larger
bicyclics may not fit well into it. Changing the bicyclic
P1 groups to certain preferable monocyclic P1 groups
may provide an opportunity for further improvement
of fIXa potency and selectivity over fXa.
3. Dominguez, C.; Han, Q.; Duffy, D. E.; Park, J. M.; Quan,
M. L.; Rossi, K.; Wexler, R. U.S. 5,886,191, 1999.
4. The modification of 5-amidinoindole analogs toward
potent and selective fIXa inhibitors is described in the
following paper: Batt, D. G.; Qiao, J. X.; Modi, D. P.;
Houghton, G. C.; Pierson, D. A.; Rossi, K. A.; Luettgen,
J. M.; Knabb, R. M.; Jadhav, P. K.; Wexler, R. R. Bioorg.
Med. Chem. Lett. 2004, 14, 5269.
5. Jin, F.; Confalone, P. N. Tetrahedron Lett. 2000, 41, 3271.
6. Ple, P. A.; Marnett, L. J. J. Heterocycl. Chem. 1988, 25,
1271.
7. Aoyama, T.; Shioiri, T. Chem. Pharm. Bull. 1981, 29,
3249.
4. Conclusion
We identified a series of 5-amidinobenzo[b]thiophenes as
low submicromolar fIXa inhibitors, some of which
showed improved selectivity over fXa. The most selec-
tive compounds prepared herein were 6- to 10-fold more
selective for fIXa relative to fXa (26 and 29-R); however,
they were less potent (fIXa Ki 0.4–0.8lM) than the com-
pounds containing the benzimidazole phenylamide P4
group. Further varying the P4 substituents did not result
in compounds with improved fIXa potency and selecti-
vity. In addition, we have identified several compounds
that are dual fIXa and fXa inhibitors, the therapeutic
benefits of which are yet to be validated.
8. Chan, H. T. J.; Elix, J. A.; Ferguson, B. A. Synth.
Commun. 1972, 2, 409.
9. Meyer, J. H.; Bartlett, P. A. J. Am. Chem. Soc. 1998, 120,
4600.
10. Zhou, Y.; Johnson, M. E. J. Mol. Recognit. 1999, 12, 235.
11. SAR and crystal structure of a benzimidazole analog of
factor Xa inhibitor SN-429 in factor IXa is described in
the following paper: Smallheer, J. M.; Alexander, R. A.;
Wang, J.; Wang, S.; Nakajima, S.; Rossi, K. A.; Small-
wood, A.; Barbera, F.; Burdick, D.; Luettgen, J. M.;
Knabb, R. M.; Wexler, R. R.; Jadhav, P. K. Bioorg. Med.
Chem. Lett. 2004, 14, 5263.
12. All Ki assays are based on chromogenic substrates
specific for the enzyme being evaluated by using analo-
gous protocols as described by Kettner, C.;
Mersinger, L.; Knabb, R. J. Biol. Chem. 1990, 265,
18289.
Acknowledgements
We would like to thank Dr. Leslie Robinson and Dr.
Jinglan Zhou for their assistance in computational
library design and automated synthesis. We would also
like to thank Dr. Joanne Smallheer and Dr. Douglas
Batt for their suggestions on the manuscript.
13. Factor IXa inhibition was assayed using human fIXa
with a chromogenic substrate in the absence (substrate
S229) or presence (substrate S299) of ethylene glycol, as
described by Sturzebecher, J.; Kopetzki, E.; Bode, W.;
Hopfner, K.-P. FEBS Lett. 1997, 412, 295, FIXa
substrate: S229 with a sequence of H-(D)Leu-Ph0Gly-
Arg-pNA, Km = 700lM and a low Kcat can only be used
for fIXa inhibitors Ki > 50nM. The other fIXa substrate
S299, has a sequence of MeSO2-chGly-Gly-Arg-pNA,
Km = 1800lM.
References and notes
1. (a) van Dieijen, G.; Tans, G.; Rosing, J.; Hemker, H. C. J.
Biol. Chem. 1981, 256, 3433; (b) Duffy, E. J.; Parker, E. T.;
Mutucumarana, V. P.; Johnson, A. E.; Lollar, P. J. Biol.
Chem. 1992, 267, 17006; (c) Duffy, E. J.; Lollar, P. J. Biol.
Chem. 1992, 267, 7821.
2. (a) Benedict, C. R.; Ryan, J.; Wolitzky, B. J. Clin. Invest.
1991, 88, 1760; (b) Himber, J.; Refino, C. J.; Burcklen, L.;
Roux, S.; Kirchhofer, D. Thromb. Haemostasis 2001, 85,
475; (c) Feuerstein, G. Z.; Patel, A.; Toomey, J. R.;
14. We employed an in-house protocol16 that generated
millions of 2, 3, and 4 point pharmacophores from the
entire fIXa dataset. Once the pharmacophores were

J. X. Qiao et al. / Bioorg. Med. Chem. Lett. 15 (2005) 29–35
35
generated, the most ꢀinformativeꢁ pharmacophores were
selected based on activity versus inactivity as well as
diversity. A library of amines were selected through
ꢀinfopickingꢁ of the most informative pharmacophores,
resulting in a diverse and informative set.
15. Katz, B. A.; Mackman, R.; Luong, C.; Radika, K.;
Martelli, A.; Sprengeler, P. A.; Wang, J.; Chan, H.; Wong,
L. Chem. Bio. 2000, 7, 299–312.
16. Miller, J. L.; Bradley, E. K.; Teig, S. L. J. Chem. Inf.
Comp. Sci. 2003, 43, 47–54.