Identification of anthranilamide derivatives as potential factor Xa inhibitors: Drug design, synthesis and biological evaluation

Article abstract of DOI:10.1016/j.ejmech.2015.03.052

The coagulation enzyme factor Xa (fXa) plays a crucial role in the blood coagulation cascade. In this study, three-dimensional fragment based drug design (FBDD) combined with structure-based pharmacophore (SBP) model and structural consensus docking were employed to identify novel fXa inhibitors. After a multi-stage virtual screening (VS) workflow, two hit compounds 3780 and 319 having persistent high performance were identified. Then, these two hit compounds and several analogs were synthesized and screened for in-vitro inhibition of fXa. The experimental data showed that most of the designed compounds displayed significant in vitro potency against fXa. Among them, compound 9b displayed the greatest in vitro potency against fXa with the IC₅₀ value of 23 nM and excellent selectivity versus thrombin (IC₅₀ Combining double low line 40 μM). Moreover, the prolongation of the prothrombin time (PT) was measured for compound 9b to evaluate its in vitro anticoagulant activity. As a result, compound 9b exhibited pronounced anticoagulant activity with the 2 × PT value of 8.7 μM.

Full text of DOI:10.1016/j.ejmech.2015.03.052

European Journal of Medicinal Chemistry 95 (2015) 388e399
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Identification of anthranilamide derivatives as potential factor Xa
inhibitors: Drug design, synthesis and biological evaluation
Junhao Xing a, Lingyun Yang a, Hui Li a, Qing Li a, Leilei Zhao a, Xinning Wang a,
, c, *
Yuan Zhang b, Muxing Zhou a, Jinpei Zhou b, Huibin Zhang a
a Center of Drug Discovery, State Key Laboratory of Natural Medicines, China Pharmaceutical University, 24 Tongjiaxiang, Nanjing 210009, PR China
b Department of Medicinal Chemistry, China Pharmaceutical University, TongjiaXiang 24, 210009 Nanjing, PR China
c Jiangsu Key Laboratory of Drug Discovery for Metabolic Disease, China Pharmaceutical University, Nanjing 210009, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
The coagulation enzyme factor Xa (fXa) plays a crucial role in the blood coagulation cascade. In this
study, three-dimensional fragment based drug design (FBDD) combined with structure-based pharma-
cophore (SBP) model and structural consensus docking were employed to identify novel fXa inhibitors.
After a multi-stage virtual screening (VS) workflow, two hit compounds 3780 and 319 having persistent
high performance were identified. Then, these two hit compounds and several analogs were synthesized
and screened for in-vitro inhibition of fXa. The experimental data showed that most of the designed
compounds displayed significant in vitro potency against fXa. Among them, compound 9b displayed the
greatest in vitro potency against fXa with the IC50 value of 23 nM and excellent selectivity versus
Received 11 November 2014
Received in revised form
21 March 2015
Accepted 23 March 2015
Available online 25 March 2015
Keywords:
Factor Xa
FBDD
Structure-based pharmacophore
Consensus docking
Prothrombin time
Anticoagulant
thrombin (IC50 ¼ 40
compound 9b to evaluate its in vitro anticoagulant activity. As a result, compound 9b exhibited pro-
M.
mM). Moreover, the prolongation of the prothrombin time (PT) was measured for
nounced anticoagulant activity with the 2 ꢀ PT value of 8.7
m
© 2015 Elsevier Masson SAS. All rights reserved.
1. Introduction
prominent roles in various thromboembolic complications. It has
been well demonstrated that the selective inhibition of fXa could
provide an anticoagulant effect by blocking the amplified formation
of thrombin without influencing normal hemostasis [6,] which
means that fXa inhibitors could decrease risk of bleeding. So many
efforts have been devoted to search for novel fXa inhibitors with a
favorable safety profile [7e9]. Currently, three oral, direct, selective
factor Xa inhibitors have been approved: rivaroxaban [10e12]
(Bayer), apixaban [13,14] (Bristol-Myers Squibb/Pfizer) and edox-
aban (Daiichi- Sankyo) [15,16] by the U.S. Food and Drug Admin-
istration (FDA) (Fig. 1). Several other promising fXa inhibitors, such
as betrixaban and eribaxaban, are currently under various stages of
clinical trials [17e20].
FBDD, which adopts an effective strategy that increased activity
is obtained by integrating several independent functional frag-
ments with low-affinity binding together, is a promising paradigm
in lead discovery [21e24]. The preference for FBDD over other
methods, such as high throughput screening (HTS), rests largely on
the enhanced screening of a more impressive conformational space
with a smaller starting number of compounds [25,26]. On the other
hand, the methodology has earned great reputation in rapidly
discovering potential drug leads at relatively lower cost of
Thromboembolic diseases are a leading cause of mortality
worldwide [1,2]. The traditional pharmacotherapies to treat these
diseases such as heparins and vitamin K antagonists have been
proved to be effective in the prevention and treatment of these
thrombotic disorders, but considerable therapeutic limitations
(parenteral administration, numerous food and drug interactions,
inconvenience of frequent monitoring, numerous drug and food
interactions, etc.) restrict their clinical utilization [3], which
prompts investigators to continue the search for novel oral anti-
coagulant not only possessing the same or higher therapeutic ef-
ficacy but also free of the disadvantages typical for the traditional
medicines [4,5]. Factor Xa, which is situated at the juncture of
intrinsic and extrinsic pathways in the coagulation cascade and
catalyzes the conversion of prothrombin to thrombin, plays
* Corresponding author. Center of Drug Discovery, State Key Laboratory of Nat-
ural Medicines, China Pharmaceutical University, 24 Tongjiaxiang, Nanjing 210009,
China.
E-mail addresses: xingjhcpu@163.com (J. Xing), hb_zhang@hotmail.com
(H. Zhang).
http://dx.doi.org/10.1016/j.ejmech.2015.03.052
0223-5234/© 2015 Elsevier Masson SAS. All rights reserved.

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
389
Fig. 1. The structures of rivaroxaban, Apixaban, Edoxaban, Betrixaban and Eribaxaban.
economic investment than biophysical approaches such as nuclear
magnetic resonance (NMR), X-ray diffraction, surface plasmon
resonance (SPR) and mass spectrometry (MS) [27]. Therefore, in our
study, the strategy for three-dimensional fragment-based lead
discovery [26] was employed to discover novel compounds with
fXa inhibitory potency. Specifically, some known structurally
diverse fXa inhibitors with co-crystallized conformations were
firstly overlaid and then deconstructed into small groups according
to the predefined rules. The resulted fragments that occupy
different regions of the active site were then merged or linked to
generate the novel, potential fXa inhibitors. Several screening
technologies, such as structure-based key pharmacophore con-
straints, consensus docking screening and ADMET, were carried out
to screen the new molecular library generated through chemical
space analysis to identify the potential active molecules. The root-
mean square deviation (RMSD) values between the docking and
combining conformations of the new compounds were calculated
and chosen as the main standards for lead identification. The
flowchart of each procedure was shown in Fig. 2. Finally, two hit
compounds combined with several analogs were synthesized and
subsequently evaluated for their inhibitory activity against fXa, and
the compound showing the most potent activity of fXa inhibition
was further assessed the degree of selectivity versus thrombin and
extension of the prothrombin time (PT). The results obtained from
FBDD and biological experiment would provide noteworthy infor-
mation for further development of fXa inhibitors.
2. Results and discussion
2.1. Fragment library
There were approximately 114 fXa-inhibitor cocrystal structures
[28e34] available in the protein data bank (PDB) when this article
was prepared. Thirty-nine representative fXa inhibitors were
selected based on the diverse structures of the co-crystallized li-
gands (Fig. 3) and then overlaid with the reference structure.
Therefore, the active conformations of the ligands were also well
overlaid within the same coordinate system. The generated
Fig. 2. The flowchart of the novel hits discovery strategy.

390
J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
fragment library was comprised with 305 fragments based on the
39 known ligands under the established setting rules. The infor-
mation about the spatial distribution in the target was included in
the resultant fragment library, as well as about their chemical
structures. The binding conformations of inhibitor rivaroxaban
(PDB code 2w26) and its fragments were shown in Fig. 4, and it was
found that the inhibitor was deconstructed into three component
fragments which maintained the initial orientations and spatial
positions of the ligand.
docked poses in each program. During screening, the recom-
mended criterion for similar docking results was slightly modified
from RMSD < 2.0 Å to RMSD < 3.0 Å because of limited number of
hit compounds when the proposed criteria applied. The 380 top-
ranking hit compounds from last step were independently
docked to the fXa structure (PDB code 2w26) by LigandFit and
Glide. And the top 100 compounds resulting from each docking
program were combined, and their poses were compared visually
followed by the calculation of corresponding RMSD value (RMSD_1
in Fig. 7). As a result, 43 compounds containing similar docked
poses generated from two different docking programs were
considered to be valid docking results.
2.2. Evaluation of fragment-based compound library and hits
identification
2.2.1. Construction of new molecular library
2.2.5. RMSD-based scoring function
According to the setting rules, All the new molecules should be
constructed by the fragments from different initial sets (Fig. 5). As a
result, 11858 compounds were developed from 305 fragments by
chemical space analysis.
To further evaluate the quality of the 43 compounds, the RMSD
values between the combining conformations and the predicted
binding conformations which were generated from molecular
docking were then calculated to re-rank hit compounds. The RMSD
values of the 43 compounds were showed in Fig. 7 (RMSD_2), and it
was found that most of them were distributed around 2.00 Å, with
five compounds <1.00 Å and three >3.00 Å. Table 1 exhibits all the
detailed RMSD values of the 43 potential fXa inhibitors in
ascending order, and Table 2 lists the structures of top-ten new
compounds generated by this stage. Among them, three known
2.2.2. Lipinski's rule and ADMET profile
To effectively abandon the redundant compounds and generate
a library with high quality, some virtual screening, including Lip-
inski's rule and ADMET profile, was firstly performed to filter the
new molecules. Finally, 814 compounds were selected from the
library.
compounds (1876, RMSD
¼
0.1487, Ki
¼
0.14 nM; 2398,
RMSD ¼ 0.3021, IC50 ¼ 27 nM; 643, RMSD ¼ 1.1324, Ki ¼ 90 nM)
with good fXa inhibitory activity were also constructed with the
fragments from the different initial inhibitors. From the compari-
son of RMSD values and activity data of the three compounds, we
can find that the smaller deviations between their binding and
combining conformations that they obtain, the higher inhibitory
activity they will display. This correlation is consistent with our
assumption. The RMSD values of top-ten new compounds were all
small and in the range from 0.1487 to 1.3938, indicating that all
fragments maintained their original conformation and binding
mode in the newly constructed molecules which would show good
activity against fXa with a high probability. As for the selection of
hits compounds, synthetic tractability and cost were firstly
considered. As a result, compounds 3780 and 319 were chosen as
hit compounds.
2.2.3. Structural based pharmacophore generation
Twelve fXa-inhibitor cocrystal structures which were included in
the above mentioned thirty-nine representative fXa inhibitors were
superimposed with the same coordinate. The Ludi interaction map
was used to generate pharmacophore features which consisted of
hydrogen bond acceptor (HBA), hydrogen bond donor (HBD), aro-
matic ring and hydrophobic features (Fig. 6A and B). Then some
common pharmacophoric constraints in the active site of fXa were
retained and some unrepresentative features were deleted. As a
result, six sets were considered as the key pharmacophoric features
(Fig. 6C). Subsequently, the central pharmacophore feature of each
set was defined as the common feature and constructed the Hypo1
(Fig. 6D). Hypo1 was then aligned to rivaroxaban. As were shown in
Fig. 6E and F, one HBA and HBD, one aromatic ring in the S4 pocket,
two hydrophobic cores in S1 pocket, one HBA and one HBD were all
included in the SBP model. Eventually, Hypo 1 was employed as the
query to quickly identify compounds in the library generated from
last stage, which have an excellent mapping. As a result, 380 com-
pounds were well matched this model.
The interaction mode of two hit compounds with fXa predicted
by glide in Schrodinger 2009 were illustrated in Fig. 8. For the
interaction mechanism of 3780 with fXa (Fig. 8a), 5-
chlorothiophene occupies hydrophobic S1pocket formed by
Ala190, Val213 and Tyr228 and forms a Cl-
p interaction between
the chloro atom and the phenyl ring of Tyr228, which played a part
in the interaction between inhibitors and fXa. The NH of scaffold
carboxamide forms a hydrogen bond with carbonyl oxygen of
Gly219. The P4 phenyllactam of 3780 is flanked by the phenyl
groups of Phe174 and Tyr99 in the S4 aryl binding pocket and forms
an edge to face interaction with Trp215. The carbonyl oxygen
(scaffold carboxamide) also forms hydrogen bond with Gly216. For
Compound 319 (Fig. 8B), chloropyridine occupies the S1 pocket
more deeply than 3780 because the carboxamide group extends P1
2.2.4. Consensus docking
The compounds resulting pharmacophore screening were then
analyzed by docking study. Presently, there are many potential
pitfalls and inherent limitations remaining in docking program.
There is no single docking program that is competent in all cases,
which poses a pressing question regarding the optimal protocol to
select for docking-based virtual screening. Recent studies have re-
ported a novel consensus docking strategy, which instead of
combining various docking scores, comprehensively integrated the
results of docking poses from separate docking programs [35e37].
The use of consensus docking poses could remarkably improve the
reliability of docking and remedy the drawback existing in single
docking program, which always provide similar docking scores
with no way of distinguishing which one is correct [36]. Previously,
we have successfully adopted this method in identification of
dipeptidyl peptidase IV inhibitors [37]. Now, consensus docking
strategy is also employed by using the following two different
docking programs: LigandFit in DS2.5 and glide in Schrodinger
2009. DockScore and GScore were selected to separately rank
by 2-bonds length, but Cl-p interaction with Tyr228 was not
formed because the chlorine atom deviated from the phenyl ring of
Tyr228. The other interaction mode is similar to 3780.
To confirm the reliability of the strategy, compounds 3780 and
319 were firstly synthesized and evaluated for in vitro inhibition of
fXa. To our delight, the two compounds exhibited inhibitory activity
against fXa with the IC50 value of 67 nM and 298 nM, respectively,
which encouraged us to further explore the efficacy of analogs of
these two hit compounds. As a result, a series of analogs of com-
pounds 3780 and 319 were designed, synthesized and evaluated for
their fXa inhibitory activity.

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
391
Fig. 3. Chemical structures of initial 39 known inhibitors that were cocrystallized with fXa.

392
J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
2.3. Chemistry
of 8.7 mM. Although compound 9b was less active than betrixaban
in the fXa binding assay, it showed similar selectivity against
thrombin and a comparable magnitude in anticoagulant activity
in vitro with betrixaban.
The synthetic routes of intermediates and target compounds
9aee and 12aeg are depicted in Scheme 1. The commercially
starting material available 1 was treated with chloroacetyl chloride
to generate compound 2 [38], which was further converted to in-
termediate 4 (X ¼ O) through treating with nitric acid in concen-
trated sulfuric acid at 0 ꢁC [39]. Additionally, treatment of the
commercially available 3 with 5-chlorovaleryl chloride using excess
sodium hydroxide afforded 4 (X ¼ CH2). Hydrogenation of 4 using
Pd/C (10%) and hydrazine hydrate (80%) in anhydrous EtOH pro-
vided aniline 5, which was acidylated by 6 to afforded 7. Nitro-
reduction of 7, followed by acylation with aromatic acid afforded
corresponding target compounds 9aee. Alternatively, treatment of
compound 8 (X ¼ O) with ethyl oxalylmonochloride provided
oxoacetate analog 10. Hydrolysis (KOH (1 N) in THF/water) of 10
gave oxoacetic acid intermediate 11, which was subsequently con-
3. Conclusion
In summary, we have reported a drug design strategy which
integrated FBDD, structure-based key pharmacophore and struc-
tural consensus docking for identification of novel fXa inhibitors.
Then two hit compounds together with their analogs were syn-
thesized and evaluated for potential inhibitory activity against fXa.
In vitro evaluation demonstrated that most of test compounds
displayed evident fXa inhibition. Of these derivatives, compound
9b was found to possess the best fXa inhibitory activity with an IC50
value of 23 nM and excellent selectivity over thrombin (48
In vitro anticoagulant activity, compound 9b also exhibited pro-
M.
mM).
verted to compounds 12aeg by acylation with aromatic amines.
nounced anticoagulant activity with the 2 ꢀ PT value of 8.7
m
1H
The physical characteristics, IR,
NMR, 13C NMR, MS and
Therefore it could be considered as lead compound for obtaining
even more potent fXa inhibitors and may be eventually developed
into new anticoagulant drugs candidates against fXa.
elemental analysis data for all intermediates and target com-
pounds, were reported in the supporting information.
2.4. Biological activities and discussion
4. Experimental section
2.4.1. In vitro inhibition activities studies of fXa
4.1. fXa inhibitors set collection
All the targeted compounds were evaluated in vitro for their fXa
enzyme inhibitory activity, and betrixaban has been selected as the
positive control. Tables 3 and 4 summarized the fXa inhibitory ac-
tivity of designed compounds. The assay results showed that most
of the novel compounds exhibited significant inhibitory activity
against fXa with the IC50 value from 1120 to 23 nM (Table 4),
particularly for compound 9b, which was the most potent fXa in-
hibitor in these series with the IC50 value of 23 nM.
All the compounds were selected according to the following
rules: (i) Compounds should bind to the same active site of fXa; (ii)
The known inhibitors selected should cover diverse chemical
structures. Most fXa inhibitors collected could be found their
crystal complex structures in PDB. Without the experimentally
resolved structures of the inhibitors within the complex of the
target protein, accurate molecular docking simulations were
employed to generate their binding conformations. All the com-
pounds were well-aligned the reference crystal structure keeping
the original technical parameter of the protein structural alignment
software default, and then their binding conformations were
extracted from the complex structures to generate three-
dimensional fragments [26].
The experimental data showed that the compound 12a, which
was non-substituted at the benzene ring, showed weak fXa inhi-
bition activity with the IC50 value of 1120 nM. The mono-chloro
substituted compounds demonstrated regiochemical preferences
of para > meta > ortho (12b ¼ 798 nM, 12c ¼ 621 nM, and
12d ¼ 452 nM). Of the synthesized compounds 12aeg, a chlorine
group (12d) at the paraposition exhibited most potent compound
of this series. An activity difference between chloro- and fluoro-
substitution was also found in this series. 4-fluoro substituent
analog 12e, which was more active than non-substituted 12a, was
1.5-fold less potent in the binding assay compared with the 4-
chloro substituted compound 12d. Additionally, methoxyphenyl
analogs 12f, 12g were also more active than 12a, and an activity
order of para > ortho was also observed. Replacement of oxalamide
group in 12aeg with amide group increased activity dramatically
(12aeg vs 9aee). In a similar trend to that observed in 12aeg, the
unsubstituted phenyl analog 9a exhibited moderate fXa inhibitory
activity with the IC50 value of 512 nM. Chlorophenyl analogs 9c and
9d were more potent than 9a, and 9b (4-Cl) exhibited the most
potent fXa inhibition in vitro with the IC50 of 23 nM. Introduction of
methoxy group at para or meta-position could slightly improve fXa
inhibitory activity and an activity order of para > meta was
observed (9d ¼ 286 nM, 9e ¼ 378 nM).
4.2. Three-dimensional fragment generation
The corresponding fragment located in different regions of the
target were generated by deconstructing the superimposed com-
pounds with their binding conformations according to some simple
rules, similar to those described in the original report on the ret-
rosynthetic combinatorial analysis procedure (RECAP) [40]. Ac-
cording to this principle, ring connecting bonds and hydrogen
bonds should not be cut, and molecules with more than the defined
“maximum atoms”(i.e., maximum number of atoms allowed in the
compound to be cleaved) were not cut. The minimum number of
non-hydrogen atoms in a fragment was set to 2. It was an option to
set whether the bonds between heteroatoms and carbons are
allowed to cleave. The resultant fragments were denominated in
the same way described in previously reported literature [26]. All
generated fragments in the library were not only well matched
with separate regions in the target, but also maintained their
original conformations and binding modes of the selected in-
hibitors with fXa.
2.4.2. Selectivity versus thrombin and prothrombin time (PT) assay
In view of its excellent bioactivity in the primary fXa assay, 9b
was chosen to assess the degree of selectivity versus thrombin and
the extension of the prothrombin time (PT) to determine its in vitro
anticoagulant activity. The assay results were showed in Table 5,
which demonstrated that 9b not only exhibited weak activity
4.3. Construction of new molecules
The construction of new compounds was implemented auto-
matically through linking or merging the neighboring three-
against thrombin with an IC50 value of 48
mM but also showed
excellent in vitro anticoagulant activity, judging by its 2 ꢀ PT value
dimensional fragments through
a series of bonding rules

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
393
Fig. 4. (A): Binding conformation of rivaroxaban (PDB code 2w26). (B) Conformations of the three constituted fragments deconstructed from rivaroxaban: fragment 1 (yellow),
fragment 2 (green), and fragment 3 (cycan). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 5. Compound 1170, a potential inhibitor of fXa, constructed from fragments of three different known inhibitors.
Fig. 6. (A) All 12 pharmacophore queries aligned to define the common features Interactions of rivaroxaban with the active site of fXa (PDB code 2w26). (B) The pharmacophoric
constraints in the active site of fXa. (C) The key features were retained according to the clustering rule; (D) The SBP model consisted of common features; (E) Best pharmacophore
model Hypo1 aligned to rivaroxaban; (F) Interactions of rivaroxaban with the active site of fXa (PDB code 2w26).
described in the literature [26]. The bonds which could be formed
between the fragments have been firstly identified by the following
criteria: (a) the angle between the bond directions should be less
than 15ꢁ, and (b) the distance between the atom that remains in
one fragment and the atom that leaves the other fragment should
be less than a threshold 1 Å. The first rule guarantees the adequate
rotational alignment of the generated fragments, and the second
criterion ensures the proper translational alignment of the frag-
ments. Both internal and peripheral bonds could be considered for
disconnection to form a new bond with another fragment, and the

394
J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
Fig. 7. RMSD value distribution of the 43 potential fXa inhibitors.
Table 1
Ranking of the 43potentially fXa inhibitors using the RMSD values.
Index
Compound name
RMSDa
RMSDb
Ranking
Index
Compound name
RMSDa
RMSDb
Ranking
6
1876
8769
432
2398
752
3780
643
1771
7425
319
0.4525
0.6459
1.7625
0.8172
1.6580
0.8762
1.3897
1.0346
0.9071
1.9872
1.2398
1.4891
2.5091
1.8616
2.2546
2.3712
1.7832
2.8109
2.1092
2.3876
2.0136
2.4580
0.1487
0.1921
0.2539
0.3021
0.5124
1.0841
1.1324
1.2234
1.3324
1.3938
1.5823
1.6984
1.7425
1.7942
1.8482
1.8887
1.9912
2.1432
2.1789
2.2341
2.3014
2.3237
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
21
38
4
36
16
43
27
42
8
28
39
13
20
5
22
14
26
31
9
254
4913
470
532
6841
791
4592
35
609
7812
10061
211
2.1234
1.6723
1.7825
1.3526
0.7652
2.0941
1.7624
0.9851
1.8724
2.0116
1.3240
1.5217
0.3561
0.7653
1.2901
1.3924
1.2093
2.3218
2.9174
0.4238
1.6218
2.3436
2.3449
2.4137
2.4286
2.4762
2.4987
2.5132
2.5856
2.6142
2.6354
2.6437
2.6483
2.6637
2.7381
2.7420
2.7414
2.8216
2.8937
3.0824
3.1417
3.7276
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
30
23
40
10
29
35
24
41
18
1
11
33
7
37
2
853
4612
6783
289
1082
3147
10342
2651
726
890
9217
8203
335
17
12
25
3
15
34
168
3245
5872
901
3128
122
913
19
32
861
a
The RMSD values were calculated between the docking poses generated from LigandFit and Glide.
The RMSD values were calculated between the combining conformations and the predicted binding conformations.
b
fragments could overlap. However, if the distance between the
fragment centroids was below the refined threshold of 3.0 Å, then
the fragments could not join because they were considered to
occupy the same location in the receptor. The self-assembly of
fragments was performed after four steps. The first step connects
pairs of fragments. In the second step, the results of the first pro-
cedure were used as input, so that the generated compounds can be
the combinations of up to four fragments, and so on. Four rounds
were carried out so that the new compounds could be generated by
the constituent molecules from the previous step. The new mole-
cules should not be constructed with the fragments from the same
inhibitor to guarantee the novelty of the generated compounds,
judging from the fragment denomination. For fragments that
contained no atoms which were close enough to generated new
bonds, methylene linkers (at most two) were added to combine the
adjacent fragments and formed a single drug-like compound.
Additionally, fragment merging was performed if the carbon-
ehydrogen or carbon-halogen bonds in the different fragments
bound in proximal regions overlapped. Generally, a total molecule
obtained by combining adjacent fragments was handled by energy
minimization. However, all the new compounds were not subjected
to the energy minimization procedure in this strategy, so that the
constituent fragments of the constructed compounds would keep
their original conformations in the protein-binding pocket.
4.4. Evaluation of constructed molecular library
4.4.1. In silico evaluation of the Lipinski's rule and ADMET profile
The Lipinski's rule of five parameters-molecular weight (MW),
AlogP, number of hydrogen bond donors (HBD), number of
hydrogen bond acceptors (HBA), and molecular fractional polar
surface area were firstly employed to qualitatively estimated their
potential “drug-like” properties. ADMET descriptors can be added if
other criteria are required. The generated molecular library is
subjected to this filtration through employing the Calculate Mo-
lecular Properties protocol available in discovery studio 3.0, and all
the parameters in the Parameters Explorer were in default.
4.4.2. Structure-based key pharmacophore generation
Structure-based key pharmacophore models were generated
based on 12 complex structures (4BTI, 3Q3K, 3M36, 3LIW, 3KQC,
3IIT, 2Y81, 2Y5G, 2P16, 2W26, 1WU1 and 1G32) with a Ludi inter-
action map using receptor-ligand pharmacophore generation pro-
tocol with default parameters in DS3.0. This protocol created a
pharmacophore query from a Ludi interaction map which was
created in a receptor active site sphere and consists of hydrogen
bond acceptor, hydrogen bond donor, and hydrophobic features.
Finally 12 pharmacophore models were generated, which were
then aligned and reserved the most common features to generate a

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
395
Table 2
The structures of top-ten new compounds generated by this stage.
Compound name
1876
Structures
RMSDa
0.1487
Fragment 1 (blue)
Fragment 2 (green)
Fragment 3 (red)
IC50/Ki(nM)
3Q3K_
fragment2
2P95_
fragment3
2XBV_
fragment4
Ki ¼ 0.14
8769
0.1921
2p16_
2XBV_
2XBV_
__
fragment1
fragment1
fragment3
432
0.2539
3M36_
2XC5_
3KQC_
__
fragment1
fragment5
fragment3
2398
0.3021
3IIT_
fragment5
2W26_
fragment3
IC50 ¼ 27
752
0.5124
3SW2_
3Q3K_
2XC5_
__
fragment2
fragment3
fragment5
3780
1.0841
Berixaban_
fragment1
2p16_
fragment3
2w26_
fragment3
__
643
1.1324
2J94_
fragment1
1WU1_
fragment4
Ki ¼ 90
(continued on next page)

396
J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
Table 2 (continued )
Compound name
1771
Structures
RMSDa
1.2234
Fragment 1 (blue)
Fragment 2 (green)
Fragment 3 (red)
IC50/Ki(nM)
__
2BOK_
2WYG_
1G32_
fragment2
fragment1
fragment3
7425
1.3324
1.3938
4BTU_
fragment2
1IOE_
fragment1
2P3T_
fragment6
__
__
319
Berixaban_
fragment1
Edoxaban_
fragment 2
2p16_
fragment2
a
The RMSD values were calculated between the combining conformations and the predicted binding conformations.
Fig. 8. The docking model of compounds 3780 and 319 with the binding site of fXa. (A): The interactions of compound 3780 with the active site of fXa. (B): The interactions of
compound 319 with the active site of fXa. (C): Overlay of binding conformations of compounds 3780 and 319 with the active site of fXa. Hydrogen atoms have been hided for clarity.
The figures were prepared using PyMol (www.pymol.org).
SBP model, Hypothesis 1 (Hypo 1). Subsequently, Hypo 1 was
employed to filter the new molecular library obtained by last
strategy.
4.4.4. Hit selection
The generated combining conformations of the novel com-
pounds maintained the original binding modes of their constituent
fragments because the energy minimization procedure was not
applied in this strategy. Glide in Schrodinger 2009 was employed to
simulate the binding conformations of the new compounds with
fXa. To confirm whether the constituent fragments in the generated
molecules maintained their original binding modes, the RMSD
value between the combining and docking conformations is used as
a criterion to assess the conformational distortions of the compo-
nent fragments. The lower RMSD values molecules obtain, the
higher potential activity they will display. When the RMSD value
between the combining and docking conformations below 3 Å, it
suggests that the fragments are combined without remarkable
distortions of their individual binding modes in the final molecules.
The combining and docking conformations are imported into
molecule window in DS3.0. Calculate Docked Rmsd calculates the
RMSD of atoms in the combining conformation with respect to
docking conformation. The RMSD values are calculated between
4.4.3. Docking simulation
All the molecules, which passed ADMETand the pharmacophore
screening, were prepared using the LigPrep module in Schrodinger
2009 and two parallel docking studies were independently con-
ducted employing LigandFit available in DS2.5 and Glide in Schro-
dinger 2009 to determine a consensus docked pose for each
molecule, as described in our previous study [37], and fXa structure
used as receptor in docking was retrieved from the protein data
bank (PDB entry: 2w26) [32]. For each molecule, the RMSD value
was calculated between the docked poses from Glide and Ligandfit.
This value gives us a direct measurement to assess the accuracy of
docking results. When the RMSD value between two docking re-
sults is larger than 3 Å, there is a high probability to be false positive
hit.

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
397
Scheme 1. Synthesis routes and structures of intermediates and target compounds. Reagents and conditions: (a) Chloroacetyl chloride, EtOH/H2O, NaOH, 38e45 ꢁC, 1 h; (b) sulfuric
acid; nitric acid ꢂ10 ꢁC, 1 h; (c) 5-Chlorovaleryl chloride, NaOH, THF/H2O, 0-rt, overnight; (d) hydrazine hydrate (80%), Pd/C (10%), EtOH, reflux, 1 h; (e) Oxalyl chloride, DMF, CH2Cl2,
0-rt, 1 h, then CH2Cl2, triethylamine, 0-rt, 16 h; (f) hydrazine hydrate (80%), Pd/C (10%), EtOH, reflux, 1.5 h; (g) aromatic carboxylic acid, Oxalyl chloride, DMF, CH2Cl2, 0-rt, 1 h, then
CH2Cl2, triethylamine, 0-rt,14 h; (h) ethyl oxalylmonochloride, CH2Cl2, N,N-Diisopropylethylamine, 0-rt, 1 h; (i) KOH, H2O/THF, Hydrochloric acid (1 N), 1 h, 0-rt; (j) EDCI, DMAP,
aromatic amines, DMF, rt, 120 h.
Table 3
Table 4
fXa inhibitory effect of designed compounds 319 and 12aeg.
fXa inhibitory effect of designed 3780 and 9aee.
Compound
R
IC50 (nM)a
67 18
Compound
X
R
IC50 (nM)a
3870
319
12a
12b
12c
12d
12e
12f
N
5-chloro
H
298
CH
CH
CH
CH
CH
CH
CH
1120 145
798 92
621 67
452 46
678 72
828 101
612 59
9a
9b
521 59
2-chloro
3-chloro
4-chloro
4-fluorine
2-methoxy
4-methoxy
23
8
12g
a
The data represent the mean of at least three independent determinations.
9c
9d
9e
183 22
286 34
378 49
two conformations, including hydrogen atoms.
4.5. Chemistry
All starting materials, solvents and reagents were commercial
sources and used without further purification. All the reactions
mentioned in this article were monitored by thin layer chroma-
tography (TLC) at 254 nm under a UV lamp with the following
eluent system: n-hexane/ethyl acetate or dichloromethane/meth-
anol. Column chromatography separations were obtained on silica
gel (300e400 mesh) eluting withCH2Cl2/MeOH (60/1). %Purity of
the target compounds (>97%) were determined by HPLC analysis
a
The data represent the mean of at least three independent determinations.
(UV detector, wavelength: 230 nm). 1H NMR and 13C NMR spectra
on a Bruker AV 300 MHz spectrometer were recorded in DMSO-d6
or CDCl3. Chemical shifts are recorded in
d (ppm) units relative to

398
J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
Interdisciplinary Council on Quality of, R. Outcomes, Guidelines for the pre-
Table 5
vention of stroke in patients with stroke or transient ischemic attack: a
guideline for healthcare professionals from the American heart association/
American stroke association, Stroke 42 (2011) pp.227e276.
Selectivity versus thrombin and anticoagulant activity of 9b.
Compound
Thrombin IC50
(
m
M)a
2 ꢀ PT (
m
M)b
[3] J. Fareed, I. Thethi, D. Hoppensteadt, Old versus new oral anticoagulants: focus
on pharmacology, Annu. Rev. Pharmacol. Toxicol. 52 (2012) pp.79e99.
[4] A. Maan, R. Padmanabhan, A.Y. Shaikh, M. Mansour, J.N. Ruskin, E.K. Heist,
Newer anticoagulants in cardiovascular disease: a systematic review of the
literature, Cardiol. Rev. 20 (2012) pp.209e221.
9b
48
18
8.7
2.4
Betrixaban
a
The data represent the mean of at least three independent determinations.
2 ꢀ PT value is defined as the inhibitor concentration required to double the
b
[5] U. Harbrecht, Old and new anticoagulants, Hamostaseologie 31 (2011)
pp.21e27.
time to fibrin formation.
[6] R.J. Leadley Jr., Coagulation factor Xa inhibition: biological background and
rationale, Curr. Top. Med. Chem. 1 (2001) pp.151e159.
TMS. Mass spectrometry (MS): HewlettePackard 1100 LC/MSD
spectrometer, in m/z; Elemental analyses: CHNeO-Rapid
instrument.
[7] T. Xue, S. Ding, B. Guo, Y. Zhou, P. Sun, H. Wang, W. Chu, G. Gong, Y. Wang,
X. Chen, Y. Yang, Design, synthesis, and structure-activity and structure-
pharmacokinetic relationship studies of novel [6,6,5] tricyclic fused oxazoli-
dinones leading to the discovery of a potent, selective, and orally bioavailable
FXa inhibitor, J. Med. Chem. 57 (2014) pp.7770e7791.
[8] U. Trstenjak, J. Ilas, D. Kikelj, Low molecular weight dual inhibitors of factor Xa
and fibrinogen binding to GPIIb/IIIa with highly overlapped pharmacophores,
Eur. J. Med. Chem. 64 (2013) pp.302e313.
[9] M. Ilic, P. Dunkel, J. Ilas, E. Chabielska, A. Zakrzeska, P. Matyus, D. Kikelj, To-
wards dual antithrombotic compounds - balancing thrombin inhibitory and
fibrinogen GPIIb/IIIa binding inhibitory activities of 2,3-dihydro-1,4-
benzodioxine derivatives through regio- and stereoisomerism, Eur. J. Med.
Chem. 62 (2013) pp.329e340.
4.6. Inhibition of fXa and thrombin in vitro
The synthesized compounds were measured in vitro for their
factor Xa/thrombin inhibition using achromogenic substrates. The
targeted compounds were prepared at 10 mM concentration and
then serially diluted to give twelve different aliquots spanning a
[10] E. Perzborn, S. Roehrig, A. Straub, D. Kubitza, F. Misselwitz, The discovery and
development of rivaroxaban, an oral, direct factor Xa inhibitor, Nat. Rev. Drug
Discov. 10 (2011) pp.61e75.
[11] F. Misselwitz, S.D. Berkowitz, E. Perzborn, The discovery and development of
rivaroxaban, Ann. N. Y. Acad. Sci. 1222 (2011) pp.64e75.
[12] J.S. Paikin, J.J. Manolakos, J.W. Eikelboom, Rivaroxaban for stroke prevention
in atrial fibrillation: a critical review of the ROCKET AF trial, Expert Rev.
cardiovasc. Ther. 10 (2012) pp.965e972.
[13] M.R. Lassen, B.L. Davidson, A. Gallus, G. Pineo, J. Ansell, D. Deitchman, The
efficacy and safety of apixaban, an oral, direct factor Xa inhibitor, as throm-
boprophylaxis in patients following total knee replacement, J. Thromb. Hae-
most. 5 (2007) pp.2368e2375.
range of 0.010e500 mMin buffer (50 mM Tris-Cl, 100 mM NaCl, 0.1%
BSA, pH 7.4). The final enzyme concentrations were 0.003 U/mL
(human fXa from Enzyme Research Laboratories) and 0.125 U/
mL(human thrombin from Sigma Chemical Co.). The reagents
containing compound dilutions, buffer and enzyme were mixed,
centrifuged, and incubated for 30 min at 37 ꢁC in 96-well microtiter
plates. The enzyme reaction was initiated by adding appropriate
substrate (for factor Xa, S-2222; for thrombin, S-2238). The time
course of the reaction was monitored continuously for 20 min at
405 nm in amicrotiter plate reader. The IC50 was calculated from the
mean of duplicates from a dilution series of the compound with
graphpad prism 5.
[14] A. De Smedt, M. Cambron, K. Nieboer, M. Moens, R.J. Van Hooff, L. Yperzeele,
K. Jochmans, J. De Keyser, R. Brouns, Intravenous thrombolysis with recom-
binant tissue plasminogen activator in a stroke patient treated with apixaban,
Int. J. Stroke 9 (2014) pp.E31.
[15] T. Furugohri, K. Isobe, Y. Honda, C. Kamisato-Matsumoto, N. Sugiyama,
T. Nagahara, Y. Morishima, T. Shibano, DU-176b, a potent and orally active
factor Xa inhibitor: in vitro and in vivo pharmacological profiles, J. Thromb.
Haemost. 6 (2008) pp.1542e1549.
4.7. Prothrombin time (PT) assay
Commercially available kits (ThrombopLastin-DS from Fisher
Diagnostics, a division of Fisher Scientific, Co.L.L.C.) were employed
to measure PT. Clotting times were measured using an automatic
coagulometer (BE Compact-X), in accordance with the manufac-
turer's instructions. Increasing concentrations of inhibitor or sol-
vent were added to plasma and incubated for 3 min at 37 ꢁC. The
plasma clotting times were calculated and compared with those
from the appropriate control plasma.
[16] D.A.
Parasrampuria,
T.
Marbury,
N.
Matsushima,
S.
Chen,
P.K. Wickremasingha, L. He, V. Dishy, K.S. Brown, Pharmacokinetics, safety,
and tolerability of edoxaban in end-stage renal disease subjects undergoing
haemodialysis, Thromb. Haemost. 113 (2015).
[17] S.A. Doggrell, Is there evidence to support the use of direct factor Xa inhibitors
in coronary artery disease? Rev. Recent Clin. Trials 6 (2011) pp.147e157.
[18] D.J. Pinto, M.J. Orwat, M.L. Quan, Q. Han, R.A. Galemmo Jr., E. Amparo, B. Wells,
C. Ellis, M.Y. He, R.S. Alexander, K.A. Rossi, A. Smallwood, P.C. Wong,
J.M. Luettgen, A.R. Rendina, R.M. Knabb, L. Mersinger, C. Kettner, S. Bai, K. He,
R.R. Wexler, P.Y. Lam, 1-[3-Aminobenzisoxazol-50-yl]-3-trifluoromethyl-6-
[20-(3-(R)-hydroxy-N-pyrrolidin
yl)methyl-[1,10]-biphen-4-yl]-1,4,5,6-
tetrahydropyrazolo-[3,4-c]-pyridin-7-one(BMS-740808) a highly potent, se-
lective, efficacious, and orally bioavailable inhibitor of blood coagulation
factor Xa, Bioorg. Med. Chem. Lett. 16 (2006) pp.4141e4147.
Acknowledgment
[19] N.C. Chan, J. Hirsh, J.S. Ginsberg, J.W. Eikelboom, Betrixaban (PRT054021):
pharma- cology, dose selection and clinical studies, Future Cardiol. 10 (2014)
pp.43e52.
This study was supported by the Natural Science Foundation of
Jiangsu Province (No. BK 20141349) and the China National Key Hi-
Tech Innovation Project for the R&D of Novel Drugs (No.
2013ZX09301303-002).
[20] P.C. Wong, E.J. Crain, C.A. Watson, A.M. Zaspel, M.R. Wright, P.Y. Lam,
D.J. Pinto, R.R. Wexler, R.M. Knabb, Nonpeptide factor Xa inhibitors III: effects
of DPC423, an orally-active pyrazole antithrombotic agent, on arterial
thrombosis in rabbits, J. Pharmacol. Exp. Ther. 303 (2002) pp.993e1000.
[21] A.C. Gibbs, M.C. Abad, X. Zhang, B.A. Tounge, F.A. Lewandowski, G.T. Struble,
W. Sun, Z. Sui, L.C. Kuo, Electron density guided fragment-based lead dis-
covery of ketohexokinase inhibitors, J. Med. Chem. 53 (2010) pp.7979e7991.
[22] A.C. Good, J. Liu, B. Hirth, G. Asmussen, Y. Xiang, H.P. Biemann, K.A. Bishop,
T. Fremgen, M. Fitzgerald, T. Gladysheva, A. Jain, K. Jancsics, M. Metz,
A. Papoulis, R. Skerlj, J.D. Stepp, R.R. Wei, Implications of promiscuous Pim-1
kinase fragment inhibitor hydrophobic interactions for fragment-based drug
design, J. Med. Chem. 55 (2012) pp.2641e2648.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.ejmech.2015.03.052.
References
[23] L. Wang, M. Stanley, J.W. Boggs, T.D. Crawford, B.J. Bravo, A.M. Giannetti,
S.F. Harris, S.R. Magnuson, J. Nonomiya, S. Schmidt, P. Wu, W. Ye, S.E. Gould,
L.J. Murray, C.O. Ndubaku, H. Chen, Fragment-based identification and opti-
mization of a class of potent pyrrolo[2,1-f][1,2,4]triazine MAP4K4 inhibitors,
Bioorg. Med. Chem. Lett. 24 (2014) pp.4546e4552.
[1] R. Carrillo-Esper, E. Alcantar-Luna, M.A. Herrera-Cornejo, D. Jaimovich,
M.A. Ramos-Corrales, A. Villagomez-Ortiz, Practice guidelines for the imple-
mentation of a quality program in thromboprophylaxis and treatment man-
agement in patients with venous thromboembolic disease, Cir. Cir. 80 (2012)
pp.92e105.
[2] K.L. Furie, S.E. Kasner, R.J. Adams, G.W. Albers, R.L. Bush, S.C. Fagan,
J.L. Halperin, S.C. Johnston, I. Katzan, W.N. Kernan, P.H. Mitchell, B. Ovbiagele,
Y.Y. Palesch, R.L. Sacco, L.H. Schwamm, S. Wassertheil-Smoller, T.N. Turan,
D. Wentworth, C.o.C.N.C. o.C.C., American Heart Association Stroke Council, C.
[24] E.H. Mashalidis, P. Sledz, S. Lang, C. Abell, A three-stage biophysical screening
cascade for fragment-based drug discovery, Nat. Protoc.
8
(2013)
pp.2309e2324.
[25] L.B. Salum, A.D. Andricopulo, Fragment-based QSAR: perspectives in drug
design, Mol. Divers. 13 (2009) pp.277e285.

J. Xing et al. / European Journal of Medicinal Chemistry 95 (2015) 388e399
399
[26] H. Yuan, T. Lu, T. Ran, H. Liu, S. Lu, W. Tai, Y. Leng, W. Zhang, J. Wang, Y. Chen,
Novel strategy for three-dimensional fragment-based lead discovery, J. Chem.
Inf. Model 51 (2011) pp.959e974.
[27] M. Sandor, R. Kiss, G.M. Keseru, Virtual fragment docking by glide: a validation
study on 190 protein-fragment complexes, J. Chem. Inf. Model 50 (2010)
1165e1172.
M.M. Morrissey, K.L. Sacchi, S.T. Sakata, K.J. Shaw, S.C. Wu, Z. Zhao, M. Adler,
S. Cheeseman, W.P. Dole, J. Ewing, R. Fitch, D. Lentz, A. Liang, D. Light,
J. Morser, J. Post, G. Rumennik, B. Subramanyam, M.E. Sullivan, R. Vergona,
J. Walters, Y.X. Wang, K.A. White, M. Whitlow, M.J. Kochanny, Thiophene-
anthranilamides as highly potent and orally available factor Xa inhibitors,
J. Med. Chem. 50 (2007) pp.2967e2980.
[28] H. Liu, T.J. Smith, W.M. Lee, A.G. Mosser, R.R. Rueckert, N.H. Olson, R.H. Cheng,
T.S. Baker, Structure determination of an Fab fragment that neutralizes human
rhinovirus 14 and analysis of the Fab-virus complex, J. Mol. Biol. 240 (1994)
pp.127e137.
[29] K. Yoshikawa, S. Kobayashi, Y. Nakamoto, N. Haginoya, S. Komoriya,
T. Yoshino, T. Nagata, A. Mochizuki, K. Watanabe, M. Suzuki, H. Kanno, T. Ohta,
Design, synthesis, and SAR of cis-1,2-diaminocyclohexane derivatives as
potent factor Xa inhibitors. Part II: exploration of 6-6 fused rings as alternative
S1 moieties, Bioorg. Med. Chem. 17 (2009) pp.8221e8233.
[30] J.R. Corte, T. Fang, D.J. Pinto, W. Han, Z. Hu, X.J. Jiang, Y.L. Li, J.F. Gauuan,
M. Hadden, D. Orton, A.R. Rendina, J.M. Luettgen, P.C. Wong, K. He, P.E. Morin,
C.H. Chang, D.L. Cheney, R.M. Knabb, R.R. Wexler, P.Y. Lam, Structure-activity
relationships of anthranilamide-based factor Xa inhibitors containing piper-
idinone and pyridinone P4 moieties, Bioorg. Med. Chem. Lett. 18 (2008)
pp.2845e2849.
[31] L. Anselm, D.W. Banner, J. Benz, K.G. Zbinden, J. Himber, H. Hilpert, W. Huber,
B. Kuhn, J.L. Mary, M.B. Otteneder, N. Panday, F. Ricklin, M. Stahl, S. Thomi,
W. Haap, Discovery of a factor Xa inhibitor (3R,4R)-1-(2,2-difluoro-ethyl)-
pyrrolidine-3,4- dicarboxylic acid 3-[(5-chloro-pyridin-2-yl)-amide] 4-[[2-
[34] D.J. Pinto, M.J. Orwat, S. Koch, K.A. Rossi, R.S. Alexander, A. Smallwood,
P.C. Wong, A.R. Rendina, J.M. Luettgen, R.M. Knabb, K. He, B. Xin, R.R. Wexler,
P.Y. Lam, Discovery of 1-(4-methoxyphenyl)-7-oxo-6-(4-(2-oxopiperidin-1-
yl)phenyl)-4,5,6,7- tetra- hydro-1H -pyrazolo[3,4-c]pyridine-3-carboxamide
(apixaban, BMS-562247), a highly potent, selective, efficacious, and orally
bioavailable inhibitor of blood coagulation factor Xa, J. Med. Chem. 50 (2007)
pp.5339e5356.
[35] D. Plewczynski, M. Lazniewski, M. von Grotthuss, L. Rychlewski, K. Ginalski,
VoteDock: consensus docking method for prediction of protein-ligand in-
teractions, J. Comput. Chem. 32 (2011) pp.568e581.
[36] D.R. Houston, M.D. Walkinshaw, Consensus docking: improving the reliability
of docking in a virtual screening context, J. Chem. Inf. Model 53 (2013)
pp.384e390.
[37] J. Xing, Q. Li, S. Zhang, H. Liu, L. Zhao, H. Cheng, Y. Zhang, J. Zhou, H. Zhang,
Identification of dipeptidyl peptidase IV inhibitors: virtual screening, syn-
thesis and biological evaluation, Chem. Biol. Drug Des. 84 (2014) pp.364e377.
ꢀ
[38] U. Trstenjak, J. Ilas, D. Kikelj, Studies towards the synthesis of alkyl N-(4-
nitrophenyl)- 3/2-oxomorpholine-2/3-carboxylates, Helvetica Chim. Acta. 96
(2013) pp.2160e2172.
fluoro-4-(2-oxo-2H-pyridin-1- yl)-phenyl]-amide] as
Bioorg. Med. Chem. Lett. 20 (2010) pp.5313e5319.
a
clinical candidate,
[39] P.L.W. Werner, W.K.R. Mederski, Markus Woissyk, Practical and efficient
processes for the preparation of 4-(4-aminophenyl)morpholin-3-ones on a
larger scale: precursor of factor Xa inhibitors, Heterocycles 74 (2007)
pp.437e445.
[40] X.Q. Lewell, D.B. Judd, S.P. Watson, M.M. Hann, RECAPeretrosynthetic
combinatorial analysis procedure: a powerful new technique for identifying
privileged molecular fragments with useful applications in combinatorial
chemistry, J. Chem. Inf. Comput. Sci. 38 (1998) pp.511e522.
[32] S. Roehrig, A. Straub, J. Pohlmann, T. Lampe, J. Pernerstorfer, K.H. Schlemmer,
P. Reinemer, E. Perzborn, Discovery of the novel antithrombotic agent 5-
chloro-N-({(5S)-2-oxo-3-[4-(3-oxomorpholin-4-yl)phenyl]-1,3-oxazolidin-5-
yl}methyl) thiophene-2-carboxamide (BAY 59-7939): an oral, direct factor Xa
inhibitor, J. Med. Chem. 48 (2005) pp.5900e5908.
[33] B. Ye, D.O. Arnaiz, Y.L. Chou, B.D. Griedel, R. Karanjawala, W. Lee,