Identification of potent orally active factor Xa inhibitors based on conjugation strategy and application of predictable fragment recommender system

Article abstract of DOI:10.1016/j.bmc.2014.11.042

Anticoagulant agents have emerged as a promising class of therapeutic drugs for the treatment and prevention of arterial and venous thrombosis. We investigated a series of novel orally active factor Xa inhibitors designed using our previously reported conjugation strategy to boost oral anticoagulant effect. Structural optimization of anthranilamide derivative 3 as a lead compound with installation of phenolic hydroxyl group and extensive exploration of the P1 binding element led to the identification of 5-chloro-N-(5-chloro-2-pyridyl)-3-hydroxy-2-{[4-(4-methyl-1,4-diazepan-1-yl)benzoyl]amino}benzamide (33, AS1468240) as a potent factor Xa inhibitor with significant oral anticoagulant activity. We also reported a newly developed Free-Wilson-like fragment recommender system based on the integration of R-group decomposition with collaborative filtering for the structural optimization process.

Full text of DOI:10.1016/j.bmc.2014.11.042

Bioorganic & Medicinal Chemistry 23 (2015) 277–289
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Identification of potent orally active factor Xa inhibitors based
on conjugation strategy and application of predictable fragment
recommender system
a
b
a
Tsukasa Ishihara a, , Yuji Koga , Yoshiyuki Iwatsuki , Fukushi Hirayama
⇑
a Drug Discovery Research, Astellas Pharma Inc., 21, Miyukigaoka, Tsukuba-shi, Ibaraki 305-8585, Japan
b Pharmacovigilance, Astellas Pharma Inc., 2-5-1, Nihonbashi-Honcho, Chuo-ku, Tokyo 103-8411, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Anticoagulant agents have emerged as a promising class of therapeutic drugs for the treatment and
prevention of arterial and venous thrombosis. We investigated a series of novel orally active factor Xa
inhibitors designed using our previously reported conjugation strategy to boost oral anticoagulant effect.
Structural optimization of anthranilamide derivative 3 as a lead compound with installation of phenolic
hydroxyl group and extensive exploration of the P1 binding element led to the identification of
5-chloro-N-(5-chloro-2-pyridyl)-3-hydroxy-2-{[4-(4-methyl-1,4-diazepan-1-yl)benzoyl]amino}benzamide
(33, AS1468240) as a potent factor Xa inhibitor with significant oral anticoagulant activity. We also
reported a newly developed Free-Wilson-like fragment recommender system based on the integration of
R-group decomposition with collaborative filtering for the structural optimization process.
Ó 2014 Elsevier Ltd. All rights reserved.
Received 16 October 2014
Revised 28 November 2014
Accepted 28 November 2014
Available online 5 December 2014
Keywords:
Factor Xa
Conjugation
Recommender system
Collaborative filtering
1. Introduction
be more effective in attenuating the coagulation cascade than
inhibitors of thrombin itself. Extensive preclinical data has shown
In the industrialized world, thromboembolic disorders, such as
ischemic stroke, deep venous thrombosis, myocardial infarction,
unstable angina, and pulmonary embolism, are leading causes of
morbidity and mortality. As thromboembolic disorders are trig-
gered by excessive stimulation of the blood coagulation cascade,
anticoagulants are used in the treatment and prevention of these
diseases. The oral anticoagulant warfarin blocks biosynthesis of
vitamin K-dependent coagulation factors and is prescribed for a
growing number of indications. However, this drug takes several
days to reach therapeutic blood concentration and has a narrow
therapeutic index due to its indirect mechanism.1–4 Therefore,
significant medical needs have emerged for orally active direct
anticoagulants with greater efficacy in treating and preventing
blood clotting.
Numerous studies have been conducted to identify inhibitors of
enzymes involved in the blood coagulation cascade. Thrombin was
the initial target, followed by factor Xa (fXa). FXa acts at the junc-
ture between the intrinsic and extrinsic pathways of the coagula-
tion cascade and catalyzes the conversion of prothrombin to
thrombin. Given that this enzyme is upstream from thrombin in
the amplification of the coagulation cascade, inhibitors of fXa could
that inhibition of fXa is effective in both venous and arterial throm-
bosis.5–7 In addition, inhibition of fXa could be less likely to
increase abnormal bleeding than direct inhibition of thrombin,
because specific fXa inhibitors do not affect platelet activation
and aggregation directly.8,9 FXa has therefore emerged as an attrac-
tive target for new therapeutic agents with the potential to treat
and prevent arterial and venous thrombosis.
A number of small molecule fXa inhibitors have been reported
to date,10 and their chemical structures are classified as amidine
derivatives exemplified by the pioneer fXa inhibitor DX-9065a,11
or nonamidine derivatives exemplified by rivaroxaban,12 apix-
aban,13 and edoxaban.14 The benzamidine substituent of amidine
derivatives forms a bidentate salt bridge interaction with the car-
boxylic acid of Asp189 in the S1 site of the enzyme. The high
hydrophilicity of the amidine group increases the overall hydro-
philicity of the molecule, which enables the enzymatic binding
affinity to effectively translate into potent anticoagulant and anti-
thrombotic activity.15,16 However, the highly hydrophilic and basic
profile of the amidine group also reduces membrane permeability
and oral bioavailability of the inhibitor. In contract, non-amidine
inhibitors possess a less basic surrogate for the amidine group,
such as an amine or aminomethyl that interacts with Asp189 or
a small lipophilic substituent such as a methoxy or chloro group
that occupies the small hydrophobic pocket formed by the residues
⇑
Corresponding author. Tel.: +81 29 863 6713; fax: +81 29 854 1519.
E-mail address: tsukasa.ishihara@astellas.com (T. Ishihara).
http://dx.doi.org/10.1016/j.bmc.2014.11.042
0968-0896/Ó 2014 Elsevier Ltd. All rights reserved.

278
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
O
N
N
O
O
N
N
O
N
O
S
Cl
HN
O
O
N
N
H
N
O
N
H
N
O
H
N
O
CONMe2
N
CONH2
Cl
S
O
rivaroxaban
apixaban
edoxaban
N
N
N
N
O
H
O
O
O
R
OH
O
HN
O
O
HN
H
N
OGlu =
3
O
N
OH
COOH
H
O
; darexaban (1)
; YM-222714 (2)
R = OH
R= OGlu
3
Figure 1. Structures of non-amidine type fXa inhibitors.
N
N
R2
NO2
NO2
O
OH
R1
HOOC
R2
b or c, d
a
OH
R1
HN
O
OH
O
N
H
N
H
R1
4 R1 = H
5 R1 = Me
6 R1 = H, R2 = OMe
7 R1 = H, R2 = Me
8 R1 = H, R2 = Cl
9 R1 = Me, R2 = OMe
R1 = H, R2 = OMe
R1 = H, R2 = Me
R1 = H, R2 = Cl
R1 = Me, R2 = OMe
11
12
13
14
Scheme 1. (a) ArNH2, 1-hydroxybenzotriazole, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride, N,N-dimethylformamide (DMF); (b) H2 (1 kgf/cm2), 10% Pd-
C, ethanol; (c) Fe, NH4Cl, methanol, H2O, reflux; (d) Ar0COCl, pyridine.
of Ala190, Val213, and Tyr228 in the S1 site of the enzyme.
Absence of a highly hydrophilic and basic amidine group often
increases oral bioavailability of the inhibitor.10
Here, we describe the identification of highly orally active fXa
inhibitors based on our conjugation strategy and retrospective
application of our newly developed fragment recommender system.
We discovered two clinical candidates—YM46617 and darex-
aban (1, Fig. 1).18 In addition, we proposed a conjugation strategy
to identify orally active inhibitors for enzymes in the coagulation
cascade.19 Our new methodology is based on nonamidine-based
inhibitors that potentially have an improved oral absorption profile
compared to amidine-based inhibitors. The key concept is an
installation of a phenolic hydroxyl group, which works as a trigger
moiety for glucuronide conjugation, into inhibitors. Biotransforma-
tion to glucuronide conjugates lowers lipophilicity of the inhibi-
tors, demonstrating potent in vivo anticoagulant activity after
administration.
2. Inhibitor design
To identify novel potent and orally-active inhibitors of fXa, we
adopted our previously reported conjugation strategy.19 Explora-
tion of orally efficacious inhibitors based on this methodology con-
sists of two steps.
The first step is search for a suitable template inhibitor and the
identification of an appropriate position for the introduction of a
phenolic hydroxyl group into the inhibitor. Figure 1 depicts the
structures of compound 1 and its active glucuronide conjugate 2
(YM-222714).18 We previously reported that the 3-position of
the central phenyl ring of compounds 1 and 2 pointed towards sol-
vent-exposed region, and the sugar residue of compound 2 was
projected out from the enzyme active site into the bulk solvent.
We also identified the related anthranilamide analogue 3.18 The
high structural similarity between compounds 1 and 3 convinced
We also furbished up traditional Free-Wilson analysis20,21 and
framed a novel interpretable and predictable compound design
workflow to recommend promising fragments in the structural
optimization process. This methodology is based on a combination
of R-group decomposition and collaborative filtering in the infor-
matics science field.

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
279
N
N
MeO
MeO
NH2
X
NH2
Cl
O
HN
O
OH
O
OH
ROOC
OH
c
d
N
H
N
H
Cl
15 X = H
16 X = Cl
17 X = Cl
, R = Et
, R = Et
, R = H
18
19
a
b
Scheme 2. (a) N-Chlorosuccinimide, DMF, 50 °C; (b) HCl aq, reflux; (c) ArNH2, 1-hydroxybenzotriazole, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride,
DMF; (d) Ar0COCl, pyridine.
The next step is the optimization of chemical structures as
enzyme inhibitors. We assumed that the overall interaction of a
glucuronide-conjugated inhibitor with the enzyme is due to the
aglycon unit because of the sugar residue being located in the sol-
vent sphere, and that structural optimization of the aglycon could
therefore result in improved inhibitory activity of the glucuronide-
conjugated inhibitor. On the exploration of the structural modifica-
tion, conformational maintenance of the aglycon should be impor-
tant because conformational change would result in the sugar
residue being located in an undesired orientation. We therefore
focused our efforts on structural modifications of the distal anisi-
dine and the central phenyl moieties of compound 3.
N
N
O
O
NH
O
HN
O
a, b
Ar
O
N
H
20
21
Scheme 3. (a) ArNH2, toluene, reflux; (b) Ar0COCl, pyridine.
3. Chemistry
Preparation of N1-phenylanthranilamide derivatives 11–14 is
illustrated in Scheme 1. 2-Nitrobenzoic acids 4 and 5 were treated
with appropriately substituted anilines in the presence of carbodi-
imide to yield benzanilide derivatives 6–9. Reductions of their
nitro groups were performed using two general procedures. Com-
pound 6 was transformed to aniline by catalytic hydrogenation.
us that the 3-position on the central phenyl ring of compound 3
would point toward the outside of the enzyme in the same way
as that of compound 1 without direct interaction between the
enzyme and a substituent on this position. We therefore selected
the anthranilamide 3 as a template scaffold and planned to intro-
duce a phenolic hydroxyl moiety on its 3-position in central ring.
R1
R1
NO2
NO2
O
NH2
O
a
b
HOOC
OBn
OBn
OH
N
N
H
N
N
H
R2
23 R1 = OMe
24 R1 = Me
25 R1 = OMe R2 = H
26 R1 = Me R2 = H
22
27 R1 = Cl
28 R1 = Cl
R2 = H
R2 = Cl
N
N
c
c or d
Cl
NH2
O
R1
N
N
H
HN
O
O
R3
N
N
H
Cl
R2
30 R1 = OMe, R2 = H, R3 = OH
31 R1 = Me, R2 = H, R3 = OH
32 R1 = Cl, R2 = H, R3 = OH
33 R1 = Cl, R2 = Cl, R3 = OH
34 R1 = Cl, R2 = Cl, R3 = H
29
Scheme 4. (a) (i) Oxaly chloride, cat.DMF, CH2Cl2; (ii) ArNH2, pyridine; (b) H2, 10% Pd-C, ethanol; (c) Ar0COCl, pyridine; (d) (i) Ar0COCl, pyridine; (ii) acetic acid, 60 °C.

280
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
transformed into aminophenols 25 and 26 by one-step catalytic
N
hydrogenation with palladium on carbon, respectively. Aminophe-
nols 25, 26, and 2719 were reacted with acid chloride 10 to give the
desired amides 30–32. Analogous 5-chloro derivative 2819 was
transformed into the target compound 33 via a two-step procedure
involving acylation with acid chloride 10 and subsequent O- to N-
migration of benzoyl moiety under acidic condition. Deshydroxy
analogue 34 was prepared starting from previously reported ani-
line 29.25
N
a
11, 33
O
HN
O
O
OH
R2
OH
OH
N
H
O
Phenol derivatives 11 and 33 were converted into the corre-
sponding glucuronides 35 and 36, respectively (Scheme 5). Treat-
ment of phenols 11 and 33 with acetobromo-a-D-glucuronic acid
R1
COOH
35 R1 = H, R2 = 4-methoxyphenyl
36 R1 = Cl, R2 = 5-chloro-2-pyridyl
methyl ester in the presence of 1,8-diazabicyclo-[5.4.0]-7-unde-
cene followed by basic hydrolysis afforded the desired phenol glu-
curonides 35 and 36, respectively.
Scheme 5. (a) (i) Acetobromo-
[5.4.0]-7-undecene, CHCl3, methanol; (ii) Na2CO3, H2O.
a
-
D
-glucuronic acid methyl ester, 1,8-diazabicyclo-
A
1H nuclear magnetic
resonance (NMR) spectrum of compound 35 showed a signal at
d = 4.95 ppm with an axial–axial coupling of 7.8 Hz in agreement
with its b-configuration of the anomeric position.26–28 The struc-
ture of the b-configuration of compound 36 was also supported
by an anomeric doublet at d = 5.06 ppm with a coupling of 7.3 Hz
characteristics of the expected axial–axial arrangement.
Alternatively, compounds 7–9 were converted to anilines by treat-
ment with reduced iron and ammonium chloride. The intermediate
anilines were acylated with acid chloride, which was prepared
from 4-(4-methyl-1,4-diazepan-1-yl)benzoic acid hydrochloride
(10),22 to afford the desired N1-phenylanthranilamide derivatives
11–14.
4. Results and discussion
Scheme 2 depicts the subsequent steps that were employed to
elaborate compound 19. Treatment of ethyl anthranilate 1523 with
N-chlorosuccinimide provided its 5-chloro analogue 16 which was
then hydrolyzed under acidic condition to give anthranilic acid 17.
Amide bond formation with p-anisidine followed by acylation with
compound 10 provided the target compound 19.
The library of anthranilamide derivatives were prepared by
solution-phase parallel synthesis, which is outlined in Scheme 3.
Coupling of isatoic anhydride with a collection of anilines followed
by treatment with acid chloride gave a set of 57 N1-phenylanthra-
nilamide derivatives 21.
For all synthesized compounds, the inhibition of human fXa in a
purified enzyme system was tested. For selected compounds, sub-
sequent screening against human thrombin for selectivity within
the coagulation cascade and human trypsin for general specificity
against serine proteases was conducted. Prothrombin time (PT)
prolongation effect was evaluated as an indicator of anticoagulant
activity via oral dosing of the test compounds. FXa inhibitors were
dosed to male ICR mice using gastric tube; 0.5 h after oral admin-
istration, blood was collected and platelet poor plasma was pre-
pared to measure PT.
Table 1 shows the results of modification on the central core of
compound 3. As the high structural similarity between compounds
1 and 3 had suggested that the introduction of a substituent into
the 3-position on the central phenyl ring of compound 3 would
maintain fXa inhibitory activity, the phenol derivative 11 and its
corresponding glucuronide 35 demonstrated comparable fXa
Preparation of a collection of N1-(2-pyridyl)anthranilamide
derivatives is shown in Scheme 4. 3-Benzyloxy-2-nitrobenzoic acid
2224 was converted to acid chloride by treatment with oxalyl
chloride and then coupled with appropriately substituted
aminopyridines to afford amides 23 and 24. The methoxy and
methyl-containing benzyloxynitrobenzenes 23 and 24 were
Table 1
Effect of modification of central core
N
N
O
MeO
OH
O
W
3
4
HN
O
O
OGlu =
N
H
OH
HOOC
OH
5
R
Prolongation effectb PT/control PT (fold)
a
Compd
W
R
IC50
(l
M) fXa
3
11
35
14
19
DPC423
H
H
H
H
4-Me
5-Cl
0.14
0.15
0.066
0.061
0.068
1.02
1.97
NTc
0.95
3.91
1.61
OH
OGlu
OH
OH
0.0024
a
Inhibitory activity against human purified enzyme. IC50 value are represented by the average of three separate determinations with an average standard error of the mean
of <20%.
b
Relative prothrombin time (PT) compared with that measured using normal mice plasma at 0.5 h after oral administration (100 mg/kg, n = 3). Each value are represented
by the average with an average standard error of the mean of <20%.
c
Not Tested.

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
281
Table 2
Our efforts next focused to the further optimization step of the
aglycon. An anthranilamide scaffold with a chloro substituent at
the 5-potition in the central phenyl core is a well-known motif
for fXa inhibitors,19,32 which encouraged us to install a 5-chloro
group on the central ring of our inhibitor 11. This installation
resulted in compound 19 demonstrating enhanced fXa inhibition
and potent anticoagulant activity with a PT-prolongation value of
3.9-fold after oral administration.
Effect of modification of P1 binding element
N
N
To identify surrogates for the p-anisidine moiety, we adopted a
rapid search with parallel synthesis methodology to prepare over
50 analogues of the lead compound 3. Table 2 demonstrates the
assay results for a representative set of compounds. The struc-
ture–activity relationships (SARs) in this region were summarized
as follows: (1) 4-substituted phenyl derivatives demonstrated
potent inhibitory activity against fXa compared to the correspond-
ing 2- or 3-substituted phenyl analogues, (2) a small lipophilic
group, such as chloro, methyl, or methoxy, at the 4-position on
the phenyl moiety exhibited potent inhibitory activity, but bulkier
substituents at this position were deleterious, (3) transformation
into 2-aminopyridine was tolerated, but compounds with other
cyclic systems, such as 3-aminopyridines, 5-membered ring sys-
tems, or bicyclic ring systems, were poor fXa inhibitors. Given
these results, we planned to concentrate our efforts on further
investigation of p-anisidine, p-toluidine, p-chloroaniline, and their
matched molecular pairs of 5-substituted 2-aminopyridines as the
P1 binding elements.
The results of replacing the p-anisidine group of compound 11
with the selected motifs described above are shown in Table 3.
Substitution with p-toluidine (12) maintained the fXa inhibitory
activity, and the p-chloroaniline derivative 13 was a highly potent
fXa inhibitor with an IC50 value of 6.9 nM. These two phenols dem-
onstrated potent PT-prolongation effect after oral dosing, and
higher ex vivo anticoagulant effect was observed with an improve-
ment in in vitro anti-fXa activity of aglycon. The 2-aminopyridine
derivatives 30, 31, and 32 demonstrated slightly reduced fXa inhib-
itory activity compared to the corresponding phenyl pairs 11, 12,
and 13, respectively. However, the 2-aminopyridine series demon-
strated a more potent anticoagulant effect via oral dosing than the
corresponding phenyl series. The SAR trend in this anthranilanilide
series was different from the SAR spectrum of our previous phen-
ylenediamine derivatives, in which only compounds with a meth-
oxy group on the distal aromatic ring showed a potent oral
anticoagulant effect.18 We concluded that 5-chloro-2-aminopyr-
idyl unit was the optimal fragment for enhancement of the
ex vivo anticoagulant effect in this series.
O
HN
O
Ar
N
H
Compd Ar
Inh.%a fXa Compd Ar
Inh.%a fXa
24
Me2N
21a
16
21l
Me
Et
21b
21c
88
94
21m
21n
15
23
Me
Me
i-Pr
21d
21e
11
76
21o
21p
38
73
N
Cl
Me
Cl
Cl
N
N
21f
21g
3
100
40
21q
21r
21s
9
Me
N
56
7
MeO
MeO
Me
MeO
85
N
N S
MeO
21h
21i
NEb
39
21t
Me
Me
NEb
NEb
MeO
MeO
H
N
N
N
21u
Cl
F3CO
21j
32
22
21v
NEb
NEb
N
H
PhO
21k
21w
N
a
Our SAR analysis and intuitive perspective prompted us to pre-
pare a novel aglycon with a combination of a chloro-substituent on
the 5-position of the central ring and a 5-chloro-2-aminopyridine
unit as the P1 element, in expectation of boosting oral activity.
Synthesized compound 33 was a potent fXa inhibitor with an
IC50 value of 8.7 nM and also an excellent orally-active anticoagu-
lant with 7.8-fold PT-prolongation effect (Table 3). The correspond-
ing deshydroxy analogue 34 did not demonstrate oral activity
despite its comparable in vitro fXa inhibitory activity to compound
33, reconfirming that the phenolic hydroxyl group was essential to
boost oral anticoagulant effect. The corresponding glucuronide 36
displayed an approximately 20-fold increase in fXa inhibitory
activity over glucuronide 35, a similar improvement in activity to
that observed in the aglycons 33 and 11. These results were consis-
tent with our initial hypothesis that structural optimization of an
aglycon, which would interact with the active site of the enzyme,
would result in an improvement in enzymatic inhibitory activity
of the corresponding glucuronide-conjugated inhibitor.
Inhibitory activity against human purified fXa. Inhibition% means anti–fXa
activity at a final inhibitor concentration of 1.36
a single experimental run.
lM. The values are determined by
b
No effect.
inhibitory activity to the parent inhibitor 3 (IC50; 150 and 66 nM,
respectively). The phenol 11 exhibited improved oral anticoagulant
activity compared to the deshydroxy parent 3, which convinced us
that our conjugation strategy was applicable to the discovery pro-
gram for orally active anticoagulants based on compound 3.
Although 4-methyl-3-hydroxy analogue 14 did exhibit equipotent
in vitro anti-fXa activity to the 3-hydroxy derivative 11, it did not
exhibit PT-prolongation effect after oral administration. Confirma-
tion with
a parallel artificial membrane permeability assay
(PAMPA) demonstrated comparable permeability of compounds
11 and 14 (>30 ꢀ 10ꢁ6 cm/s). Sterically blocked phenols are less
reactive in glucuronide-conjugation compared than less hindered
phenols,29–31 suggesting that substitution in the neighborhood of
the phenolic hydroxyl group interfered with conjugation and
caused a significant loss of oral anticoagulant activity.
The phenol 33 and its glucuronide conjugate 36 were selected
for further evaluation in our advanced profiling assays. Compound
33 exhibited dose-dependent anticoagulant activity following oral

282
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
Table 3
Effect of modification of central core and P1 binding element
N
N
O
X
OH
OH
O
W
HN
R
O
O
OGlu =
N
H
Q
HOOC
OH
Prolongation effecta PT/control PT (fold)
a
Compd
W
R
X
Q
IC50 (lM) fXa
11
12
13
30
31
32
33
34
OH
OH
OH
OH
OH
OH
OH
H
H
H
H
H
H
H
Cl
Cl
Cl
OMe
Me
Cl
OMe
Me
Cl
Cl
Cl
Cl
CH
CH
CH
N
N
N
N
N
N
0.15
1.97
2.11
2.67
2.03
2.13
3.61
7.80
1.21
NTa
0.068
0.0069
0.22
0.29
0.016
0.0087
0.0036
0.0015
0.0024
36
DPC423
OGlu
1.61
a
See the corresponding footnotes in Table 1.
The first step is the preparation of ‘Fragment Matrices’ via com-
mon scaffold detection and following R-group decomposition for
target sets of compounds. Figure 2 shows a brief example of our
concept for restructuring SAR information. SAR analysis is a gen-
eral process in drug discovery research, in which almost all medic-
inal chemists utilize SAR tables. The first step for preparation of
fragment matrices involves transforming SAR tables to matrices
in which rows corresponded to scaffolds and columns to substitu-
ents. Both rows and columns are then ordered by hierarchical
Table 4
In vitro inhibitory activities and anticoagulant activities
M) PTc
Mouse
a
b
Compd
IC50
(l
M)
CT2 (l
fXa
Thrombin
Trypsin
Human
36
DPC423
0.0015
0.0024
>100
>100
>100
>100
0.27
0.67
0.28
0.70
a
See the corresponding footnotes of Table 1.
CT2 values are defined as the concentration required to double clotting time and
b
represent the average of three separate determinations with the average standard
errors of the mean being <10%.
c
Prothrombin time.
administration, and demonstrated over 10 times more potency in
PT-prolongation effect than DPC423, the clinical candidate as an
antithrombotic agent at the time of our study (Fig. 4). Pharmacoki-
netic evaluation revealed that conjugation to the corresponding
glucuronide was observed when compound 33 was dosed orally,
and that plasma concentration of the parent phenol 33 was below
the limit of quantification (data not shown). Subsequent screening
against human thrombin for selectivity within the coagulation cas-
cade and human trypsin for general specificity against serine pro-
teases, as well as in vitro anticoagulant activities in human and
mice plasma was carried out for the potent fXa inhibitor 36
(Table 4). Compound 36 exhibited highly selectivity for both
thrombin and trypsin (>10,000-fold), and also showed potent
in vitro PT-prolongation activity. These results indicated that glu-
curonide 36 demonstrated ex vivo PT-prolongation effect due to
its specific fXa inhibitory activity after oral dosing of the phenol 33.
5. Fragment recommender system
For structural optimization process, we developed a novel inter-
pretable and predictable compound design workflow that recom-
mends promising fragments on the basis of common scaffold
detection, R-group decomposition, and recommender system in
the informatics science field.33 In this section, we describe the con-
cept and retrospective application to our exploratory study.
Figure 2. Conceptual example of restructuring SARs information. Figures in
parentheses of the top box represent hypothetical activity.

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
283
O
Y
N
H
N
N
OMe
MeO
X
COOH
Y
Y
N
H
O
X
N
H
Y
Y
X
O
N
H
OH
Y
O
HO
X
N
H
X
O
N
H
O
X
N
Y
H
O
OH
X
Cl
Y
N
OH
O
H
O
X
OH
Y
Y
Y
N
X
H
O
N
H
O
OMe
X
N
H
Y
Y
Y
COOH
X
O
N
O
Y
H
O
X
N
H
OH
O
X
N
H
F
O
X
COOH
N
H
Y
X
O
N
H
Y
O
OH
X
N
H
O
X
X
X
X
X
More Potent
Less Potent
N
S
S
MeO
Br
Me
MeO
Cl
X
X
X
X
X
N
N
S
Cl
MeO
Cl
Me
Br
Figure 3. Fragment matrix for oral anticoagulant effect.
clustering based on structural similarity, and dendrograms are
arranged in the outer area of the matrix to resemble a heatmap.
Intersections of rows and columns, namely a combination of a scaf-
fold and a substituent, equals one compound, and its similar ana-
logues are located in the neighborhood in this fragment matrix.
Bajorath et al. pioneered the visualization of SARs and recently
they reported a relative workflow.34,35
matrices with collaborative filtering functions into our fragment
recommender system that thereby enables novel interpretation
and prediction of promising substituents.
Here, we first explored SARs around the central core and P1
binding element and then investigated their combination effect.
The second step is quantitative estimation based on a similar
concept to Free-Wilson analysis.20,21 This analysis and traditional
SAR study by medicinal chemists estimates the changes in biolog-
ical activities or physicochemical properties as a linear combina-
tion of the effects of modification of scaffolds and substituents.
Analogical estimation of the activity or property of a newly
designed compound could therefore be well-implemented by lin-
ear regression of substituent transformation with the relative
weighting of scaffold similarity. In the case of the previous
instance, 6-chloro and 6-methoxyquinoline would be estimated
to show activity values of 3 and 5, respectively. This is the same
concept as collaborative filtering of recommender system in the
informatics science field. Therefore, the integration of fragment
Figure 4. Anticoagulant activity of compound 33 0.5 h after oral dosing in mice.
Relative prothrombin time (PT) compared to that measured using normal mice
plasma (mean + standard deviation, n = 3).

284
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
Table 5
Measured and simulated oral anticoagulant effect by slope one estimationa
Y
(1.00)
2.03
(1.02)
2.13
(1.05)
3.61
1.02
1.97
(1.02)
2.11
(1.03)
2.67
X
X
O
O
N
H
OH
Y
N
H
O
Y
N
H
OH
Y
N
N
(1.00)
(4.09)
(1.00)
(4.39)
(1.00)
(8.83)
0.95
(1.00)
(1.00)
X
O
N
H
OH
Cl
Y
3.91
(4.33)
(6.01)
X
O
N
H
X
N
X
N
X
N
X
X
X
MeO
Me
Cl
MeO
Me
Cl
a
Figures in parentheses represent estimated values.
For retrospective application of our fragment recommender system
design workflow, we selected the combination step as a case study.
We used a collection of diazepane derivatives 1, 3, 11–14, 19, 30–
32 and additional 16 analogues described in our previous work as a
training set.18 Figure 3 shows this set of compounds tabled in a
fragment matrix in which columns corresponded P1 binding ele-
ments and rows to scaffolds. To grasp SARs instantly, this matrix
visualizes compound potency with colored cells that range from
red for low potency to green for high potency. This diagram
promptly highlights important SARs. The scaffolds are classified
into three groups, as follows: 4- and 5-substituted phenylenedi-
amine scaffold (rows 1–5), 3-substituted anthranylamide (rows
6–9, enclosed in the dotted box), and 3-substituted phenylenedi-
amine (rows 10–16). The compounds in the second group fre-
quently demonstrate higher activity than the others, and the
SARs around P1 elements of second group differ from those of
the third one. These observations are consistent with our findings
by traditional SAR analysis.
Our focus on the second group with the aim of boosting oral
activity led us to estimate values of blank cells, namely to predict
the activity of practicable compounds. Slope One algorithm36 used
for collaborative filtering for the second cluster afforded an all-
filled matrix (Table 5) and recommended a compound with a com-
bination of chloro-substituent on the 5-position of the central ring
and 5-chloro-2-aminopyridine for the P1 element as the most
potent compound, with an estimated PT-prolongation value of
8.8-fold. The proposed compound was closely to compound 33,
as were the estimated and measured activities.
phenol 11 demonstrated enhanced anticoagulant activity com-
pared to the parent deshydroxy analogue 3 after oral administra-
tion. Based on our hypothesis that an interaction of
a
glucuronide-conjugated inhibitor with the active site of the enzyme
would be formed by its aglycon and that structural optimization of
the aglycon could improve the inhibitory activity of the conjugated
inhibitor, we conducted structural modifications of compound 11.
Our efforts to focus on the central phenyl ring and the P1 binding
element culminated in the identification of compound 33
(AS1468240), which exhibited potent fXa inhibitory activity and
excellent PT-prolongation effect when administered orally.
We also developed a new Free-Wilson-like fragment recom-
mender system for the interpretation and prediction in the com-
pound design process based on an integration of common scaffold
detection, R-group decomposition, and collaborative filtering. We
retrospectively applied this workflow to our study, which afforded
computational analytical results that were consistent with our find-
ings by traditional SAR analysis and recommended a combination of
fragments for a promising candidate. The details of our newly devel-
oped fragment recommender system will be published in due
course.
7. Experimental
7.1. Chemistry
1H NMR spectra were measured with a JEOL JNM-LA300 or JEOL
EX400 spectrometer. Chemical shifts are expressed in d units using
tetramethylsilane as the standard (in the NMR description, s = sin-
glet, d = doublet, t = triplet, m = multiplet, and br = broad peak).
Mass spectra were recorded on a JEOL JMS-LX2000 spectrometer.
For salts, assignments of ion peaks are based on the basic compo-
nent. The elemental analyses were performed with a Yanaco MT-5
microanalyzer (C, H, and N) and a Yokogawa IC-7000S ion chro-
matographic analyzer (Cl). Melting points were measured using a
Yanaco MP-500D melting point apparatus without correction. All
reagents purchased were used without further purification. C-18
reversed-phase silica gel (ODS) column chromatography was per-
formed on YMC gel (ODS-A 120–230/70).
6. Conclusion
We designed and synthesized a series of novel orally active fXa
inhibitors based on our conjugation strategy. Our previously
reported fXa inhibitor 3 was selected as a lead compound. The high
structural similarity between compounds 1 and 3 led us to select
substitution point for introduction of a sugar residue, which affor-
ded a glucuronide-conjugated fXa inhibitor 35 with comparable
in vitro inhibitory activity to compound 3. The corresponding

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
285
7.1.1. 3-Hydroxy-40-methoxy-2-nitrobenzanilide (6)
7.1.4. 40-Chloro-3-hydroxy-2-{[4-(4-methyl-1,4-diazepan-1-yl)
To a stirred solution of 3-hydroxy-2-nitrobenzoic acid (4)
(1.83 g, 10.0 mmol), p-anisidine (1.23 g, 10.0 mmol), and 1-hydroxy
benzotriazole hydrate (1.35 g, 10.0 mmol) in 20 mL of N,N-dimeth-
ylformamide (DMF) was added 1-[3-(dimethylamino)propyl]-3-
ethylcarbodiimide hydrochloride (2.50 g, 13.0 mmol) at ambient
temperature. After 3 days, the reaction mixture was concentrated
in vacuo. The residue was diluted with ethyl acetate and washed
with H2O. The organic layer was dried over magnesium sulfate
and concentrated in vacuo. The residue was triturated with CHCl3
and the resulting solid was filtered off to yield the title compound
as a colorless solid (2.04 g, 71%): mp 180–181 °C; 1H NMR
(300 MHz, DMSO-d6) d 3.74 (3H, s), 6.92 (2H, d, J = 8.8 Hz), 7.21–
7.26 (2H, m), 7.50 (1H, t, J = 8.6 Hz), 7.58 (2H, d, J = 8.8 Hz), 10.45
(1H, s), 11.20 (1H, br s); FAB-MS m/z 289 [M+H]+.
benzoyl]amino}benzanilide Hydrochloride (13)
A mixture of compound 8 (1.43 g, 4.89 mmol), reduced iron
(2.80 g, 49.0 mmol), ammonium chloride (2.80 g, 49.0 mmol),
methanol (50 mL), and H2O (5 mL) was stirred at 60 °C for 3 h.
The reaction mixture was filtered through a pad of CeliteÒ and con-
centrated in vacuo. The residue was subjected to chromatography
over silica gel eluting with n-hexane/ethyl acetate (1:1 by volume)
to yield 2-amino-40-chloro-3-hydroxybenzanilide (270 mg, 21%).
To a stirred solution of 2-amino-40-chloro-3-hydroxybenzanilide
(270 mg, 1.03 mmol) in pyridine (11 mL) was added compound
37 (447 mg, 1.55 mmol). After 2 days, the reaction mixture was
concentrated in vacuo. The residue was diluted with CHCl3 and
washed with 5% aqueous sodium bicarbonate. The organic layer
was dried over magnesium sulfate and concentrated in vacuo.
The residue was subjected to chromatography over silica gel elut-
ing with CHCl3/MeOH/c.NH3 (100:5:0.5 by volume). The resulting
material was dissolved in EtOH (7.3 mL). To this solution was
added 1 N hydrochloric acid (1.1 mL), and the whole was stirred
for 0.5 h. The resulting precipitate was filtered off and dried to
yield the title compound as a colorless solid (299 mg, 56%): mp
197–199 °C; 1H NMR (400 MHz, DMSO-d6) d 2.12–2.21 (1H, m),
2.26–2.39 (1H, m), 2.78 (3H, s), 3.03–3.20 (2H, m), 3.30–3.54 (4H,
m), 3.72–3.78 (1H, m), 3.89–3.96 (1H, m), 6.84 (2H, d, J = 8.8 Hz),
7.10–7.13 (1H, m), 7.15–7.18 (1H, m), 7.22–7.26 (1H, m), 7.36
(2H, d, J = 8.8 Hz), 7.71 (2H, d, J = 8.8 Hz), 7.85 (2H, d, J = 8.8 Hz),
9.96 (1H, s), 9.99 (1H, s), 10.40 (1H, s), 10.76 (1H, br s); FAB-MS
m/z 479 [M+H]+; Anal. calcd for C26H27N4O3Clꢂ0.9HClꢂ0.5H2O: C,
59.96; H, 5.59; N, 10.76; Cl, 12.93. Found: C, 59.66; H, 5.95; N,
9.94; Cl, 12.74.
7.1.2. 3-Hydroxy-40-methoxy-2-{[4-(4-methyl-1,4-diazepan-1-yl)
benzoyl]amino}benzanilide Hydrochloride (11)
To a mixture of compound 6 (1.15 g, 4.00 mmol) and methanol
(50 mL) was added 10% Pd-C (300 mg) and the whole was stirred
under hydrogen at atmospheric pressure for 1 h. The reaction mix-
ture was filtered through a pad of CeliteÒ and concentrated in
vacuo to yield 966 mg of 2-amino-3-hydroxy-40-methoxybenzani-
lide (966 mg, 94%). The resulting material was used in subsequent
reactions without purification. A mixture of compound 1022
(812 mg, 3.00 mmol) and thionyl chloride (8.0 mL) was stirred at
60 °C for 30 min. The reaction mixture was concentrated in vacuo
to yield 4-(4-methyl-1,4-diazepan-1-yl)benzoyl chloride hydro-
chloride (37). This material and 2-amino-3-hydroxy-40-methoxy-
benzanilide (774 mg) were combined in pyridine (15 mL) and the
whole was stirred at ambient temperature for 2 h. The reaction
mixture was concentrated in vacuo. The residue was diluted with
ethyl acetate and washed with 5% aqueous sodium bicarbonate.
The resulting precipitate was filtered off. The organic layer was
dried over magnesium sulfate and concentrated in vacuo. The res-
idue was combined with the filtrate described above, and the
whole was subjected to chromatography over silica gel eluting
with CHCl3/MeOH (95:5 by volume). The resulting material was
suspended in EtOH (10 mL). To this suspension was added 4 N
hydrogen chloride in ethyl acetate (0.7 mL) and the whole was stir-
red for 0.5 h. The resulting precipitate was filtered off and dried in
vacuo to yield the title compound as a pale yellow solid (896 mg,
58%): mp 169–171 °C; 1H NMR (400 MHz, DMSO-d6) d 2.10–2.41
(2H, m), 2.78 (3H, s), 3.02–3.22 (2H, m), 3.35–3.57 (4H, m), 3.67–
3.81 (4H, m), 3.87–3.99 (1H, m), 6.80–6.95 (4H, m), 7.11 (1H, d,
J = 7.3 Hz), 7.17–7.28 (2H, m), 7.57 (2H, d, J = 8.8 Hz), 7.85 (2H, d,
J = 8.8 Hz), 10.02 (1H, s), 10.19 (1H, s), 10.41 (1H, s), 10.64 (1H,
br s); FAB-MS m/z 475 [M+H]+; Anal. calcd for C27H30N4O4ꢂHClꢂH2-
O: C, 61.30; H, 6.29; N, 10.59; Cl, 6.70. Found: C, 61.37; H, 6.47; N,
10.33; Cl, 6.59.
7.1.5. 3-Hydroxy-40-methyl-2-nitrobenzanilide (7)
The title compound was obtained from p-toluidine using the
methods described for the synthesis of compound 8 as a beige
amorphous powder (quant): 1H NMR (300 MHz, CDCl3) d 2.35
(3H, s), 7.09 (1H, dd, J = 1.3 and 7.3 Hz), 7.19 (2H, d, J = 8.1 Hz),
7.26 (1H, dd, J = 1.3 and 8.4 Hz), 7.47 (2H, d, J = 8.1 Hz), 7.57–7.62
(1H, m), 8.00 (1H, s), 10.49 (1H, s); FAB-MS m/z 273 [M+H]+.
7.1.6. 3-Hydroxy-40-methyl-2-{[4-(4-methyl-1,4-diazepan-1-yl)
benzoyl]amino}benzanilide Hydrochloride (12)
The title compound was obtained from compound 7 using the
methods described for the synthesis of compound 13 as a colorless
solid (12%): mp 216–217 °C; 1H NMR (400 MHz, DMSO-d6) d 2.08–
2.36 (5H, m), 2.78 (3H, s), 3.03–3.20 (2H, m), 3.32–3.54 (4H, m),
3.70–3.78 (1H, m), 3.89–3.96 (1H, m), 6.86 (2H, d, J = 8.8 Hz),
7.09–7.12 (3H, m), 7.18–7.26 (2H, m), 7.54 (2H, d, J = 8.8 Hz),
7.85 (2H, d, J = 9.3 Hz), 10.00 (1H, s), 10.20 (1H, s), 10.27 (1H, s),
10.64 (1H, br s); FAB-MS m/z 459 [M+H]+; Anal. calcd for C27H30N4-
O3ꢂ0.9HClꢂH2O: C, 63.66; H, 6.51; N, 11.00; Cl, 6.26. Found: C,
63.99; H, 6.67; N, 10.35; Cl, 6.57.
7.1.3. 40-Chloro-3-hydroxy-2-nitrobenzanilide (8)
To a stirred solution of compound 4 (2.00 g, 10.9 mmol), 4-chlo-
roaniline (1.53 g, 12.0 mmol), and 1-hydroxybenzotriazole hydrate
(2.21 g, 16.4 mmol) in DMF (100 mL) was added 1-[3-(dimethyl-
amino)propyl]-3-ethylcarbodiimide hydrochloride (3.15 g, 16.4
mmol) at ambient temperature. After 2 days, the reaction mixture
was concentrated in vacuo. The residue was diluted with CHCl3 and
washed with brine. The organic layer was dried over magnesium
sulfate and concentrated in vacuo. The residue was subjected to
chromatography over silica gel eluting with CHCl3/MeOH (100:1
by volume) to yield the title compound as a beige amorphous pow-
der (2.97 g, 93%): 1H NMR (300 MHz, CDCl3) d 7.08 (1H, dd, J = 1.1
and 7.1 Hz), 7.26 (1H, dd, J = 1.1 and 8.6 Hz), 7.34 (2H, d, J = 8.6 Hz),
7.53–7.62 (3H, m), 7.97 (1H, s), 10.48 (1H, br s); FAB-MS m/z 293
[M+H]+.
7.1.7. 3-Hydroxy-40-methoxy-4-methyl-2-nitrobenzanilide (9)
The title compound was obtained from 3-hydroxy-4-methyl-2-
nitrobenzoic acid (5) using the methods described for the synthesis
of compound 8 as a beige amorphous powder (79%): 1H NMR
(300 MHz, CDCl3) d 2.36 (3H, s), 3.82 (3H, s), 6.92 (2H, d,
J = 9.0 Hz), 7.00 (1H, d, J = 7.5 Hz), 7.35 (1H, br s), 7.45 (1H, d,
J = 7.5 Hz), 7.50 (2H, d, J = 9.0 Hz); FAB-MS m/z 303 [M+H]+.
7.1.8. 3-Hydroxy-40-methoxy-4-methyl-2-{[4-(4-methyl-1,4-
diazepan-1-yl)benzoyl]amino}benzanilide Hydrochloride (14)
The title compound was obtained from compound 9 using the
methods described for the synthesis of compound 13 as a colorless
solid (24%): mp 259–261 °C; 1H NMR (400 MHz, DMSO-d6) d 2.11–
2.21 (1H, m), 2.27–2.38 (4H, m), 2.78 (3H, s), 3.04–3.22 (2H, m),

286
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
3.38–3.56 (4H, m), 3.74–3.80 (4H, m), 3.91–3.99 (1H, m), 6.90–6.94
(4H, m), 7.18 (1H, d, J = 7.8 Hz), 7.29 (1H, d, J = 7.8 Hz), 7.59 (2H, d,
J = 8.8 Hz), 7.86 (2H, d, J = 9.3 Hz), 10.11 (1H, s), 10.35 (1H, s), 10.72
(1H, br s), 11.42 (1H, s); FAB-MS m/z 489 [M+H]+; Anal. calcd for
7.1.13. Solution-phase synthesis of library compounds 21
Isatoic anhydride (20) (0.1 mmol) was partitioned into reaction
vials and treated with aniline (0.10 mmol) in toluene (1.0 mL) at
100 °C for 12 h. The mixture in each vessel was allowed to cool
to ambient temperature, and the solvent was removed under
reduced pressure. The residue in each vessel was dissolved in pyr-
idine (1.0 mL) and treated with compound 37 (0.10 mol) at ambi-
ent temperature for 12 h. The mixture was then treated with
polymer-bound tris(2-aminoethyl)amine for 12 h and filtered off.
The filtrate was collected in vials and the solutions were evapo-
rated in vacuo to afford anthranilamide derivatives 21, which were
analyzed by HPLC and MS. 21a; FAB-MS m/z 443 [M+H]+, 21b; FAB-
MS m/z 443 [M+H]+, 21c; FAB-MS m/z 443 [M+H]+, 21d; FAB-MS
m/z 463 [M+H]+, 21e; FAB-MS m/z 463 [M+H]+, 21g; FAB-MS m/z
459 [M+H]+, 21h; FAB-MS m/z 489 [M+H]+, 21i; FAB-MS m/z 493
[M+H]+, 21j; FAB-MS m/z 513 [M+H]+, 21k; FAB-MS m/z 521
[M+H]+, 21l; FAB-MS m/z 472 [M+H]+, 21m; FAB-MS m/z
457 [M+H]+, 21n; FAB-MS m/z 471 [M+H]+, 21o; FAB-MS m/z 430
[M+H]+, 21p; FAB-MS m/z 444 [M+H]+, 21q; FAB-MS m/z 444
[M+H]+, 21r; FAB-MS m/z 444 [M+H]+, 21s; FAB-MS m/z
460 [M+H]+, 21t; FAB-MS m/z 449 [M]+, 21u; FAB-MS m/z 433
[M+H]+, 21v; FAB-MS m/z 469 [M+H]+, 21w; FAB-MS m/z 480
[M+H]+. Compound 21f was obtained from 2-amino-5-chloro-3-
hydroxy-40-methoxybenzanilide using the methods described for
the synthesis of compound 19 as a colorless solid (423 mg, 56%):
1H NMR (400 MHz, DMSO-d6) d 2.15–2.25 (1H, m), 2.27–2.42
(1H, m), 2.78 (3H, d, J = 4.9 Hz), 3.05–3.21 (2H, m), 3.39–3.56
(3H, m), 3.72–3.80 (1H, m), 3.90–4.06 (2H, m), 6.91 (2H, d,
J = 9.3 Hz), 7.22–7.26 (1H, m), 7.45 (2H, d, J = 8.7 Hz), 7.57–7.62
(1H, m), 7.75–7.80 (4H, m), 7.93 (1H, d, J = 7.8 Hz) , 8.54 (1H, d,
J = 8.3 Hz), 10.68 (1H, s), 10.81 (1H, br s), 11.54 (1H, s); FAB-MS
m/z 463 [M+H]+.
C
28H32N4O4ꢂHClꢂ0.2H2O: C, 63.62; H, 6.37; N, 10.60; Cl, 6.71.
Found: C, 63.57; H, 6.42; N, 10.54; Cl, 6.67.
7.1.9. Ethyl 2-Amino-5-chloro-3-hydroxybenzoate (16)
A mixture of ethyl 2-amino-3-hydroxybenzoate (15)23 (7.02 g,
38.8 mmol), N-chlorosuccinimide (5.18 g, 38.8 mmol), and DMF
(50 mL) was stirred at 50 °C for 1.5 h. The reaction mixture was
cooled to ambient temperature. The mixture was diluted with
ethyl acetate and washed with 5% aqueous sodium bicarbonate,
H2O, and then brine. The organic layer was dried over magnesium
sulfate and concentrated in vacuo. The residue was dissolved in
EtOH, decolorized with activated charcoal, filtered through a pad
of CeliteÒ, and concentrated in vacuo. The resulting solid was crys-
tallized from EtOH/H2O to yield the title compound as a brown
solid (7.92 g, 95%): mp 121–122 °C; 1H NMR (300 MHz, CDCl3) d
1.38 (3H, t, J = 7.3 Hz), 4.33 (2H, q, J = 7.3 Hz), 5.26 (1H, br s),
5.87 (2H, br s), 6.81(1H, d, J = 2.4 Hz), 7.48(1H, d, J = 2.4 Hz); FAB-
MS m/z 216 [M+H]+.
7.1.10. 2-Amino-5-chloro-3-hydroxybenzoic Acid (17)
A mixture of compound 16 (3.23 g, 15.0 mmol) and 3 N hydro-
chloric acid (160 mL) was stirred at 80 °C for 5 days. The reaction
mixture was cooled to ambient temperature and filtered. To the fil-
trate was added 1 N aqueous NaOH (320 mL) and the whole was
stirred for 1 h. The resulting precipitate was filtered off, washed
with H2O, and dried in vacuo to yield the title compound as a beige
solid (1.55 g, 55%): mp >300 °C (dec); 1H NMR (400 MHz, DMSO-
d6) d 3.37 (2H, br s), 6.78 (1H, d, J = 2.4 Hz), 7.17 (1H, d,
J = 2.4 Hz), 8.34 (1H, br s), 10.19 (1H, s); FAB-MS m/z 186 [MꢁH]+.
7.1.14. 3-Benzyloxy-N-(5-methoxy-2-pyridyl)-2-
nitrobenzamide (23)
7.1.11. 2-Amino-5-chloro-3-hydroxy-40-methoxybenzanilide
(18)
A mixture of 3-benzyloxy-2-nitrobenzoic acid (22)24 (2.07 g,
7.57 mmol), three drops of DMF, and oxalyl chloride (1.92 g,
15.1 mmol) in CH2Cl2 (33 mL) was stirred at ambient temperature
for 3 h. The reaction mixture was concentrated in vacuo. This
material and 2-amino-5-methoxypyridine (940 mg, 7.57 mmol)37
were combined in pyridine (37 mL) and stirred at ambient temper-
ature for 20 h. The reaction mixture was concentrated in vacuo.
The residue was diluted with CHCl3 and washed with 5% aqueous
sodium bicarbonate. The organic layer was dried over magnesium
sulfate and concentrated in vacuo. The residue was subjected to
chromatography over silica gel eluting with CHCl3/MeOH (100:1
by volume) to give the title compound as a colorless amorphous
powder (2.13 g, 74%): 1H NMR (300 MHz, CDCl3) d 3.84 (3H, s),
5.23 (2H, s), 7.19–7.49 (9H, m), 7.92 (1H, d, J = 2.9 Hz), 8.22 (1H,
d, J = 9.0 Hz), 8.56 (1H, br s); FAB-MS m/z 380 [M+H]+.
The title compound was obtained from compound 17 using the
methods described for the synthesis of compound 6 as a brown
solid (55%): mp 215–217 °C; 1H NMR (400 MHz, DMSO-d6) d 3.74
(3H, s), 5.93 (2H, br s), 6.78 (1H, d, J = 2.5 Hz), 6.91 (2H, d,
J = 9.3 Hz), 7.23 (1H, d, J = 2.5 Hz), 7.59 (2H, d, J = 9.3 Hz), 9.90
(1H, s), 10.09 (1H, br s); FAB-MS m/z 293 [M+H]+.
7.1.12. 5-Chloro-3-hydroxy-40-methoxy-2-{[4-(4-methyl-1,4-
diazepan-1-yl)benzoyl]amino}benzanilide hydrochloride (19)
To a stirred solution of compound 18 (430 mg, 1.47 mmol) in
pyridine (10 mL) was added compound 37 (577 mg, 2.00 mmol).
After 24 h, the reaction mixture was concentrated in vacuo. The res-
idue was diluted with CHCl3 and washed with 5% aqueous sodium
bicarbonate. The organic layer was dried over magnesium sulfate
and concentrated in vacuo. The residue was subjected to chroma-
tography over silica gel eluting with CHCl3/MeOH (100:5 by vol-
ume). The resulting material was dissolved in EtOH (10 mL) and
added 1 N hydrochloric acid (0.8 mL), and the whole was stirred
for 0.5 h. The resulting precipitate was filtered off and dried to yield
the title compound as a colorless solid (250 mg, 31%): mp 248–
251 °C; 1H NMR (400 MHz, DMSO-d6) d 2.10–2.34 (2H, m), 2.81
(3H, s), 3.01–3.25 (2H, m), 3.35–3.60 (4H, m), 3.62–3.79 (4H, m),
3.82–4.00 (1H, m), 6.84 (2H, d, J = 9.3 Hz), 6.88 (2H, d, J = 8.8 Hz),
7.12 (1H, d, J = 2.4 Hz), 7.18 (1H, d, J = 2.4 Hz), 7.54 (2H, d,
J = 9.3 Hz), 7.84 (2H, d, J = 8.8 Hz), 9.86 (1H, br s), 9.96 (1H, s),
10.16(1H, s), 10.43 (1H, s); ESI-MS m/z 509 [M+H]+; Anal. calcd
for C27H29N4O4Clꢂ0.9HClꢂH2O: C, 57.93; H, 5.74; N, 10.01; Cl,
12.03. Found: C, 57.76; H, 5.38; N, 10.05; Cl, 12.05.
7.1.15. 2-Amino-3-hydroxy-N-(5-methoxy-2-pyridyl)benzamide
(25)
To a mixture of compound 23 (720 mg, 1.90 mmol) and metha-
nol (19 mL) was added 10% Pd/C (70 mg) and the whole was stirred
under hydrogen at 3 kgf/cm2 for 1 h. The reaction mixture was fil-
tered through a pad of CeliteÒ and concentrated in vacuo to give
the title compound as a brown amorphous powder (480 mg,
97%): 1H NMR (300 MHz, DMSO-d6) d 3.83 (3H, s), 5.89 (2H, br s),
6.43 (1H, t, J = 7.9 Hz), 6.80 (1H, dd, J = 1.2 and 7.9 Hz), 7.27 (1H,
dd, J = 1.2 and 7.9 Hz), 7.46 (1H, dd, J = 3.1 and 9.2 Hz), 7.99 (1H,
d, J = 9.2 Hz), 8.08 (1H, d, J = 3.1 Hz), 9.54 (1H, br s), 10.14 (1H,
s); FAB-MS m/z 260 [M+H]+.

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
287
7.1.16. 3-Hydroxy-N-(5-methoxy-2-pyridyl)-2-{[4-(4-methyl-
1,4-diazepan-1-yl)benzoyl]amino}benzamide Hydrochloride
(30)
J = 8.7 Hz), 8.35 (1H, d, J = 2.5 Hz), 9.71 (1H, s), 9.95 (1H, s), 10.58
(1H, s), 10.75 (1H, br s); FAB-MS m/z 480 [M+H]+; Anal. calcd for
C
25H26N5O3ClꢂHClꢂ0.7H2O: C, 56.76; H, 5.41; N, 13.24; Cl, 13.40.
The title compound was obtained from compound 25 using the
methods described for the synthesis of compound 19 as a colorless
solid (55%): mp 218–220 °C; 1H NMR (400 MHz, DMSO-d6) d 2.12–
2.21 (1H, m), 2.29–2.41 (1H, m), 2.77 (3H, d, J = 4.9 Hz), 3.03–3.19
(2H, m), 3.36–3.82 (8H, m), 3.89–3.97 (1H, m), 6.84 (2H, d,
J = 8.8 Hz), 7.11 (1H, dd, J = 2.4 and 6.8 Hz), 7.20–7.25 (2H, m),
7.45 (1H, dd, J = 2.9 and 9.3 Hz), 7.85 (2H, d, J = 8.8 Hz), 7.99–8.06
(2H, m), 9.98 (2H, s), 10.36 (1H, s), 10.92 (1H, br s); FAB-MS m/z
476 [M+H]+; Anal. calcd for C26H29N5O4ꢂHClꢂ0.5H2O: C, 59.94; H,
6.00; N, 13.44; Cl, 6.80. Found: C, 59.77; H, 5.79; N, 13.41; Cl, 7.13.
Found: C, 57.05; H, 5.18; N, 13.27; Cl, 13.02.
7.1.21. 5-Chloro-N-(5-chloro-2-pyridyl)-3-hydroxy-2-{[4-(4-
methyl-1,4-diazepan-1-yl)benzoyl]amino}benzamide Hydro-
chloride (33)
Compound 37, which was prepared from 3.00 mmol of com-
pound 10, and compound 2819 (891 mg, 2.99 mmol) were com-
bined in pyridine (10 mL) and stirred at ambient temperature for
13 h. The reaction mixture was concentrated in vacuo. The residue
was dissolved in acetic acid (20 mL) and the whole was stirred at
ambient temperature for 17 h. The reaction mixture was concen-
trated in vacuo. The residue was diluted with CHCl3 and washed
with 5% aqueous sodium bicarbonate. The organic layer was dried
over magnesium sulfate and concentrated in vacuo. The residue
was subjected to chromatography over silica gel eluting with
CHCl3/MeOH/c.NH3 (100:5:0.5 by volume). The resulting material
was dissolved in 1 N hydrochloric acid and concentrated in vacuo.
The residue was subjected to chromatography over ODS gel eluting
with CH3CN/0.002 N hydrochloric acid (3:7 by volume) to give the
title compound as a colorless amorphous powder (492 mg, 31%):
1H NMR (400 MHz, DMSO-d6) d 2.10–2.21 (1H, m), 2.23–2.37
(1H, m), 2.79 (3H, d, J = 4.9 Hz), 3.02–3.21 (2H, m), 3.37–3.56
(4H, m), 3.66–3.95 (2H, m), 6.81 (2H, d, J = 8.8 Hz), 7.15 (2H, s),
7.82 (2H, d, J = 8.8 Hz), 7.89 (1H, dd, J = 2.4 and 8.8 Hz), 8.08 (1H,
d, J = 8.8 Hz), 8.36 (1H, d, J = 2.4 Hz), 9.51 (1H, s), 10.49 (1H, br s),
10.68 (2H, br s); FAB-MS m/z 514 [M+H]+; Anal. calcd for C25H25N5-
O3Cl2ꢂHClꢂ1.1H2O: C, 52.62; H, 4.98; N, 12.27; Cl, 18.64. Found: C,
52.48; H, 4.69; N, 12.23; Cl, 18.49.
7.1.17. 3-Benzyloxy-N-(5-methyl-2-pyridyl)-2-nitrobenzamide
(24)
The title compound was obtained from 2-amino-5-picoline
using the methods described for the synthesis of compound 23
as a colorless amorphous powder (74%): 1H NMR (400 MHz, CDCl3)
d 2.28 (3H, s), 5.23 (2H, s), 7.20–7.23 (1H,m), 7.26–7.28 (1H,m),
7.31–7.39 (5H,m), 7.45 (1H, t, J = 8.3 Hz), 7.56 (1H, dd, J = 2.5 and
8.8 Hz), 8.00 (1H, d, J = 2.5 Hz), 8.17 (1H, d, J = 8.8 Hz), 8.75 (1H,
br s); FAB-MS m/z 364 [M+H]+.
7.1.18. 2-Amino-3-hydroxy-N-(5-methyl-2-pyridyl)benzamide
(26)
The title compound was obtained from compound 24 using the
methods described for the synthesis of compound 25 as a brown
amorphous powder (87%): 1H NMR (400 MHz, DMSO-d6) d 2.27
(3H, s), 5.90 (2H, br s), 6.43 (1H, t, J = 8.0 Hz), 6.82 (1H, dd, J = 1.2
and 8.0 Hz), 7.27 (1H, dd, J = 1.2 and 8.0 Hz), 7.93 (1H, dd, J = 2.0
and 8.4 Hz), 7.99 (1H, d, J = 8.4 Hz), 8.19 (1H, d, J = 2.0 Hz), 9.56
(1H, br s), 10.16 (1H, s); FAB-MS m/z 244 [M+H]+.
7.1.22. 5-Chloro-N-(5-chloro-2-pyridyl)-2-{[4-(4-methyl-1,4-
diazepan-1-yl)benzoyl]amino}benzamide Hydrochloride (34)
The title compound was obtained from 2-amino-5-chloro-N-(5-
chloro-2-pyridyl)benzamide 2925 using the methods described for
the synthesis of compound 19 as a colorless solid (87%): mp
254–257 °C; 1H NMR (400 MHz, DMSO-d6) d 2.12–2.20 (1H, m),
2.32–2.43 (1H, m), 2.78 (3H, d, J = 4.8 Hz), 3.05–3.20 (2H, m),
3.39–3.56 (4H, m), 3.73–3.82 (1H, m), 3.91–3.97 (1H, m), 6.90
(2H, d, J = 8.7 Hz), 7.65 (1H, dd, J = 2.4 and 8.8 Hz), 7.79 (2H, d,
J = 8.7 Hz), 7.99–8.02 (2H, m), 8.11 (1H, d, J = 8.8 Hz), 8.43 (1H, d,
J = 8.8 Hz), 8.48 (1H, d, J = 2.4 Hz), 10.94 (1H, br s), 11.23 (1H, s),
11.29 (1H, s); FAB-MS m/z 498 [M+H]+; Anal. calcd for C25H25N5O2
Cl2ꢂHClꢂ0.3H2O: C, 55.58; H, 4.96; N, 12.96; Cl, 19.69. Found: C,
55.68; H, 5.02; N, 12.73; Cl, 19.37.
7.1.19. 3-Hydroxy-2-{[4-(4-methyl-1,4-diazepan-1-
yl)benzoyl]amino}-N-(5-methyl-2-pyridyl)benzamide
Hydrochloride (31)
Compound 37, which was prepared from 0.949 mmol of com-
pound 10, and compound 26 (210 mg, 0.864 mmol) were com-
bined in pyridine (5 mL) and stirred at ambient temperature for
24 h. The reaction mixture was concentrated in vacuo. The residue
was diluted with CHCl3 and washed with 5% aqueous sodium
bicarbonate. The organic layer was dried over magnesium sulfate
and concentrated in vacuo. The residue was subjected to chroma-
tography over silica gel eluting with CHCl3/MeOH (94:6 by vol-
ume). The resulting material was dissolved in 1 N hydrochloric
acid and concentrated in vacuo. The residue was subjected to chro-
matography over ODS gel eluting with CH3CN/0.001 N hydrochlo-
ric acid (10:90 by volume) to give the title compound as a pale
yellow amorphous powder (103 mg, 24%): 1H NMR (400 MHz,
DMSO-d6) d 2.10–2.21 (1H, m), 2.28–2.40 (4H, m), 2.78 (3H, d,
J = 4.9 Hz), 3.03–3.19 (2H, m), 3.38–3.54 (4H, m), 3.72–3.79 (1H,
m), 3.83–3.94 (1H, m), 6.82 (2H, d, J = 9.3 Hz), 7.13–7.26 (3H, m),
7.84–7.91 (4H, m), 8.18 (1H, s), 9.78 (1H, s), 10.04 (1H, br s),
10.85 (1H, br s), 11.04 (1H, br s); FAB-MS m/z 459 [M+H]+; Anal.
calcd for C26H29N5O3ꢂ1.7HClꢂ2H2O: C, 56.01; H, 6.27; N, 12.56; Cl,
10.81. Found: C, 55.86; H, 6.39; N, 12.73; Cl, 10.70.
7.1.23. 3-[(4-Methoxyphenyl)carbamoyl]-2-{[4-(4-methyl-1,4-
diazepan-1-yl)benzoyl]amino}phenyl b-D-glucopyranosiduronic
acid (35)
Compound 11 (366 mg, 0.716 mmol) was suspended in CHCl3
(50 mL) and washed with 5% aqueous sodium bicarbonate. The
organic layer was dried over magnesium sulfate and concentrated
in vacuo to yield a colorless powder. To a stirred mixture of the this
material, methanol (3.5 mL), and CHCl3 (3.5 mL) was added 1,8-
diazabicyclo[5.4.0]undec-7-ene (436 mg, 2.86 mmol) at ambient
temperature. After 30 min, to the reaction mixture was added acet-
obromo-a-D-glucuronic acid methyl ester (853 mg, 2.15 mmol) at
7.1.20. N-(5-Chloro-2-pyridyl)-3-hydroxy-2-{[4-(4-methyl-1,4-
diazepan-1-yl)benzoyl]amino}benzamide Hydrochloride (32)
The title compound was obtained from compound 2719 using
the methods described for the synthesis of compound 19 as a col-
orless solid (37%): mp 229–231 °C; 1H NMR (400 MHz, DMSO-d6) d
2.17–2.26 (2H, m), 2.74 (3H, s), 3.00–3.60 (6H, m), 3.77–3.88 (2H,
m), 6.82 (2H, d, J = 8.8 Hz), 7.10–7.25 (3H, m), 7.83 (2H, d,
J = 8.8 Hz), 7.90 (1H, dd, J = 2.5 and 8.7 Hz), 8.13 (1H, d,
ambient temperature, and the whole was stirred for 21 h. To the
reaction mixture were added sodium carbonate (750 mg,
7.08 mmol) and H2O (4 mL) at ambient temperature, and the
whole was stirred for 24 h. The reaction mixture was concentrated
in vacuo. The residue was diluted with H2O, washed with CHCl3,
and then extracted with n-butanol. The organic layer was concen-
trated in vacuo. The residue was diluted with H2O (5 mL), neutral-
ized with acetic acid and concentrated in vacuo. The residue was

288
T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
subjected to chromatography over ODS gel eluting with CH3CN/
H2O (30:70 by volume) to give the title compound as a colorless
amorphous powder (208 mg, 45%): 1H NMR (400 MHz, DMSO-d6)
d 1.94–2.02 (2H, m), 2.44 (3H, s), 2.70–2.75 (2H, m), 2.84–2.90
(2H, m), 3.22–3.48 (6H, m), 3.62–3.65 (2H, m), 3.70 (3H, s), 3.78
(1H, d, J = 9.3 Hz), 4.95 (1H, d, J = 7.8 Hz), 5.17 (1H, br s), 5.39
(1H, br s), 6.75 (2H, d, J = 8.8 Hz), 6.84 (2H, d, J = 8.8 Hz), 7.29–
7.34 (3H, m), 7.54 (2H, d, J = 8.8 Hz), 7.80 (2H, d, J = 8.8 Hz), 9.40
(1H, s), 10.04 (1H, s); FAB-MS m/z 651 [M+H]+; Anal. calcd for C33-
H38N4O10ꢂ3H2O: C, 56.24; H, 6.29; N, 7.95. Found: C, 56.36; H, 6.18;
N, 7.96.
the clotting time (CT2) was estimated from each individual con-
centration–response curve. Each measurement was performed
three times and represented as the mean value.
7.2.4. Ex vivo anticoagulant assays in mice
The test drug was dissolved or suspended in 0.5% methyl cellu-
lose and orally administered to male ICR mice (mass range:
30ꢁ37 g) at a dose of 100 mg/kg using a gastric tube. Citrated
blood was collected from the inferior vena cava 0.5 and 2.0 h after
oral administration, and platelet-poor plasma was prepared by
centrifugation for measurement of PT. All data were expressed as
relative-fold values, compared with the baseline value of vehicle-
treated mice.
7.1.24. 5-Chloro-3-[(5-chloro-2-pyridyl)carbamoyl]-2-{[4-(4-
methyl-1,4-diazepan-1-yl)benzoyl]amino}phenyl b-D-gluco-
pyranosiduronic acid (36)
7.3. Computational simulations
The title compound was obtained from compound 33 using the
methods described for the synthesis of compound 35 as a colorless
amorphous powder (15%): 1H NMR (400 MHz, DMSO-d6) d 1.93–
2.01 (2H, m), 2.45 (3H, s), 2.72–2.77 (2H, m), 2.87–2.91 (2H, m),
3.22–3.66 (8H, m), 3.84 (1H, d, J = 9.3 Hz), 5.06 (1H, d, J = 7.3 Hz),
5.21 (1H, br s), 5.49 (1H, br s), 6.73 (2H, d, J = 8.8 Hz), 7.37 (1H,
d, J = 1.5 Hz), 7.43 (1H, d, J = 1.5 Hz), 7.77 (2H, d, J = 8.8 Hz), 7.87
(1H, dd, J = 2.4 and 8.8 Hz), 8.07 (1H, d, J = 8.8 Hz), 8.34 (1H, d,
J = 2.4 Hz), 9.37 (1H, s), 10.81 (1H, br s); FAB-MS m/z 690
[M+H]+; Anal. calcd for C31H33N5O9Cl2ꢂ3H2O: C, 50.01; H, 5.28; N,
9.41; Cl, 9.52. Found: C, 50.27; H, 5.02; N, 9.55; Cl, 9.69.
7.3.1. Preparation of fragment matrix
Common substructures were extracted in the set of 26 com-
pounds, and R-group decomposition was executed for the common
substructure as a query. A collection of 26 combinations of scaf-
folds and fragments were tabled as all of them were allocated in
one matrix, in which rows corresponded to scaffolds and columns
to substituents. Rows and columns were ordered by hierarchical
clustering (average linkage) by Tanimoto distance with RDKit38
morgan fingerprint (1024 bits, radius = 2) in KNIME workflow39
,
respectively. The dendrogram of clustering for the scaffolds was
depicted on the right outer side, and that for fragments was on
the upper side. Scaffold and fragment was expressed with the
smiles notation on the left outer side and the lower side, respec-
tively. Each cell was colored with red as low potency via yellow
to green as high potency.
7.2. Pharmacology
7.2.1. In vitro assay for inhibition of factor Xa
The hydrolysis rates of synthetic substrates were assayed by
continuously measuring absorbance at 405 nm at 37 °C with a
microplate reader (model 3550, Bio-Rad, U.S.). Reaction mixtures
7.3.2. Estimation with slope one algorithm
Measured values of oral PT–prolongation effect in fragment
matrix were transformed into PT indexes with following equation:
PT index = log10(PT/controlPT-1). Imputation for missing data in
the matrix was executed with Slope One algorithm36 and following
inverse transformation to afford estimated values of PT-prolonga-
tion effect for nonsynthesized compounds.
(125
substrates (S-2222) and an inhibitor in either 0.05 M Tris-HCl, pH
8.4, or 0.15 M NaCl. Reactions were initiated with 25 L of enzyme
lL) were prepared in 96-well plates containing chromogenic
l
solution. The concentration of inhibitor required to inhibit enzyme
activity by 50% (IC50) was calculated from dose–response curves in
which the logit transformation of residual activity was plotted
against the logarithm of inhibitor concentration.
Acknowledgments
7.2.2. Enzyme selectivity
The authors deeply acknowledge Drs. Shuichi Sakamoto, Tom-
ihisa Kawasaki, and Takeshi Shigenaga for useful discussion. The
authors also thank to Dr. Takeshi Kadokura for pharmacokinetic
analysis. The authors are grateful to Ms. Yumiko Moritani for per-
forming pharmacological experiments. The authors would like to
express our gratitude to Mr. Norio Seki for solution-phase synthe-
sis of library compounds and the staff of the Division of Analytical
& Pharmacokinetics Research Laboratories for the elemental anal-
ysis and spectral measurements.
Reaction mixtures were prepared in 96-well plates containing
the chromogenic substrate and test compound. The reaction was
initiated by the addition of enzyme, and the color was continu-
ously monitored at 405 nm using a microplate reader SpectraMax
340PC (Molecular Devices, CA, U.S.) at 37 °C. Each enzyme was
used at final concentration as follows: 0.20 U mLꢁ1 thrombin and
1.0 U mLꢁ1 trypsin. The enzymatic activities were assessed by the
amidolysis of the following chromogenic substrates for the corre-
sponding protease: S-2222 for trypsin and S-2238 for thrombin.
The rate of substrate hydrolysis (mOD minꢁ1) was measured at
37 °C. The mode of inhibition was estimated from a Lineweaver–
Burk plot. The Ki was determined from a Dixon plot by plotting
the reciprocal of the initial reaction velocities at different substrate
concentrations against different inhibitor concentrations.
References and notes
1. Hirsh, J. N. Engl. J. Med. 1865, 1991, 324.
2. Stein, P. D.; Grandison, D.; Hua, T.; Slettehaugh, P. M.; Henry, J. W.; Turlapaty,
P.; Kerwin, R. Postgrad. Med. J. 1994, 70, S72.
3. Hirsh, J.; Poller, L. Arch. Int. Med. 1994, 154, 282.
4. Bona, R. D.; Hickey, A. D.; Wallace, D. M. Thromb. Haemost. 1997, 78, 137.
5. Kaiser, B. Cell. Mol. Life Sci. 2002, 59, 189.
7.2.3. Prothrombin time assays in vitro
6. Samama, M. M. Thromb. Res. 2002, 106, V267.
7. Walenga, J. M.; Jeske, W. P.; Hoppensteadt, D.; Fareed, J. Curr. Opin. Invest. Drugs
2003, 4, 272.
8. Sitko, G. R.; Ramjit, D. R.; Stabilito, I. I.; Lehman, D.; Lynch, J. J.; Vlasuk, G. P.
Circulation 1992, 85, 805.
9. Lynch, J. J.; Sitko, G. R.; Lehman, E. D.; Vlasuk, G. P. Thromb. Haemost. 1995, 74,
640.
10. Pinto, D. J. P.; Smallheer, J. M.; Cheney, D. L.; Knabb, R. M.; Wexler, R. R. J. Med.
Chem. 2010, 53, 6243. and references cited therein.
After collection of citrated blood samples, platelet-poor plasma
was prepared by centrifugation at 3000 rpm for 10 min and stored
at ꢁ40 °C until use. Plasma clotting times were measured using a
KC10A coagulometer (Amelung Co., Lehbrinsweg, Germany) at
37 °C. Prothrombin time (PT) was measured using Orthobrain
thromboplastin (OrthoDiagnostic Systems Co., Tokyo, Japan), and
values for each test sample were compared with coagulation times
of a distilled water control. The concentration required to double
11. Nagahara, T.; Yokoyama, Y.; Inamura, K.; Katakura, S.; Komoriya, S.;
Yamaguchi, H.; Hara, T.; Iwamoto, M. J. Med. Chem. 1994, 37, 1200.

T. Ishihara et al. / Bioorg. Med. Chem. 23 (2015) 277–289
289
12. Roehrig, S.; Straub, A.; Pohlmann, J.; Lampe, T.; Pernerstorfer, J.; Schlemmer, K.
H.; Reinemer, P.; Perzborn, E. J. Med. Chem. 2005, 48, 5900.
13. Pinto, D. J. P.; Orwat, M. J.; Koch, S.; Rossi, K. A.; Alexander, R. S.; Smallwood, A.;
Wong, P. C.; Rendina, A. R.; Luettgen, J. M.; Knabb, R. M.; He, K.; Xin, B.; Wexler,
R. R.; Lam, P. Y. S. J. Med. Chem. 2005, 50, 5339.
23. Reisch, J.; Gunaherath, G. M. K. B. Monatsh. Chem. 1988, 119, 1169.
24. Yoshino, H.; Tsuchiya, Y.; Saito, I.; Tsujii, M. Chem. Pharm. Bull. 1987, 35, 3438.
25. Beight, D. W.; Craft, T. J.; Denny, C. P.; Franciskovich, J. B.; Goodson, T. Jr.; Hall,
S. E.; Herron, D. K.; Joseph, S. P.; Klimlowski, V. J.; Masters, J. J.; Mendel, D.;
Milot, G.; Pineiro-Nunez, M. M.; Sawyer, J. S.; Shuman, R. T.; Smith, G. F. Tebbe,
A. L.; Tinsley, J. M.; Weir, L. C.; Wikel, J. H.; Wiley, M. R.; Yee, Y. K. (Eli Lilly and
Co, USA). PCT Int. Appl. WO2000039118, 2000; Chem. Abstr. 2000, 133, 89437h.
26. Berrang, B.; Brine, G. A.; Carrol, F. I. Synthesis 1997, 1165.
27. Schmidt, R. R.; Grundler, G. Synthesis 1981, 886.
28. Badman, G. T.; Green, D. V. S.; Voyle, M. J. Organomet. Chem. 1990, 388, 117.
29. Holloway, C. J.; Zierhut, M. IRCS Med. Sci.: Library Compendium 1981, 9, 96.
30. Ebner, T.; Burchell, B. Drug Metab. Dispos. 1993, 21, 50.
31. Shimizu, M.; Matsumoto, Y.; Tatsuno, M.; Fukuoka, M. Biol. Pharm. Bull. 2003,
26, 216.
14. Chung, N.; Jeon, H. K.; Lien, L. M.; Lai, W. T.; Tse, H. F.; Chung, W. S.; Lee, T. H.;
Chen, S. A. Thromb. Haemost. 2011, 105, 535.
15. Masters, J. J.; Franciskovich, J. B.; Tinsley, J. M.; Campbell, C.; Campbell, J. B.;
Craft, T. J.; Froelich, L. L.; Gifford-Moore, D. S.; Hay, L. A.; Herron, D. K.;
Klimkowski, V. J.; Kurz, K. D.; Metz, J. T.; Ratz, A. M.; Shuman, R. T.; Smith, G. F.;
Smith, T.; Towner, R. D.; Wiley, M. R.; Wilson, A.; Yee, Y. K. J. Med. Chem. 2000,
43, 2087.
16. 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.
32. Adler, M.; Kochanny, M. J.; Ye, B.; Rumennik, G.; Light, D. R.; Biancalana, S.;
Whitlow, M. Biochemistry 2002, 41, 15514.
17. Hirayama, F.; Koshio, H.; Katayama, N.; Kurihara, H.; Taniuchi, Y.; Sato, K.;
Hisamichi, N.; Sakai-Moritani, Y.; Kawasaki, T.; Matsumoto, Y.; Yanagisawa, I.
Bioorg. Med. Chem. 2002, 10, 1509.
18. Hirayama, F.; Koshio, H.; Ishihara, T.; Hachiya, S.; Sugasawa, K.; Koga, Y.; Seki,
N.; Shiraki, R.; Shigenaga, T.; Iwatsuki, Y.; Moritani, Y.; Mori, K.; Kadokura, T.;
Kawasaki, T.; Matsumoto, Y.; Sakamoto, S.; Tsukamoto, S. J. Med. Chem. 2011,
54, 8051.
19. Ishihara, T.; Koga, Y.; Sugasawa, K.; Mori, K.; Iwatsuki, Y.; Hirayama, F.
Accepted in Bioorg. Med. Chem.
20. Free, S. M.; Wilson, J. W. J. Med. Chem. 1964, 7, 395.
33. Francesco Ricci, F.; Rokach L.; Shapira, B., Introduction to Recommender Systems
Handbook, Springer, 2011, pp. 1–35 and references cited therein.
34. Wassermann, A. M.; Haebel, P.; Weskamp, N.; Bajorath, J. J. Chem. Inf. Model.
2012, 52, 1769.
35. Gupta-Ostermann, D.; Shanmugasundaram, V.; Bajorath, J. J. Chem. Inf. Model.
2014, 54, 801.
36. Lemire, D., Maclachlan, A. SIAM Data Mining (SDM’05), Newport Beach,
California, April 21–23, 2005.
37. Coyne, W. E.; Cusic, J. W. J. Med. Chem. 1968, 11, 1208.
38. RDKit: Open-Source Cheminformatics. http://www.rdkit.org.
39. KNIME: The Konstanz Information Miner. http://www.knime.org.
21. Kubinyi, H. Quant. Struct.-Act. Relat. 1988, 7, 121.
22. Koshio, H.; Hirayama, F.; Ishihara, T.; Shiraki, R.; Shigenaga, T.; Taniuchi, Y.;
Sato, K.; Moritani, Y.; Iwatsuki, Y.; Kaku, S.; Katayama, N.; Kawasaki, T.;
Matsumoto, Y.; Sakamoto, S.; Tsukamoto, S. Bioorg. Med. Chem. 2005, 13, 1305.