The discovery of novel small molecule allosteric activators of aldehyde dehydrogenase 2

Article abstract of DOI:10.1016/j.ejmech.2020.113119

Aldehyde dehydrogenase 2 (ALDH2) plays important role in ethanol metabolism, and also serves as an important shield from the damage occurring under oxidative stress. A special inactive variant was found carried by 35–45% of East Asians. The variant carriers have recently been found at the higher risk for the diseases related to the damage occurring under oxidative stress, such as cardiovascular and cerebrovascular diseases. As a result, ALDH2 activators may potentially serve as a new class of therapeutics. Herein, N-benzylanilines were found as novel allosteric activators of ALDH2 by computational virtual screening using ligand-based and structure-based screening parallel screening strategy. Then a structural optimization was performed and has led to the discovery of the compound C6. It has good activity in vitro and in vivo, which could reduce infarct size by ～70% in ischemic stroke rat models. This study provided good lead compounds for the further development of ALDH2 activators.

Full text of DOI:10.1016/j.ejmech.2020.113119

European Journal of Medicinal Chemistry 212 (2021) 113119
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
The discovery of novel small molecule allosteric activators of aldehyde
dehydrogenase 2
,
b
,
,
,
Wei Tian a 1, Jiapeng Guo a 1, Qingsen Zhang a 1, Shaoyu Fang a, Ruolan Zhou a, Jian Hu a,
Mingping Wang a, Yuefan Zhang c, Jin-Min Guo d, Zhuo Chen e, Ju Zhu a
,
,
**
Canhui Zheng a
,
*
a School of Pharmacy, Second Military Medical University, Shanghai, 200433, China
b General Hospital Of Central Theater Commond, Wuhan, Hubei, 430070, China
c School of Medicine, Shanghai University, Shanghai, 20444, China
d 960 Hospital of the Joint Logistics Support Force, Jinan, Shandong, 250031, China
e School of Pharmacy, East China University of Science and Technology, Shanghai, 200237, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Aldehyde dehydrogenase 2 (ALDH2) plays important role in ethanol metabolism, and also serves as an
important shield from the damage occurring under oxidative stress. A special inactive variant was found
carried by 35e45% of East Asians. The variant carriers have recently been found at the higher risk for the
diseases related to the damage occurring under oxidative stress, such as cardiovascular and cerebro-
vascular diseases. As a result, ALDH2 activators may potentially serve as a new class of therapeutics.
Herein, N-benzylanilines were found as novel allosteric activators of ALDH2 by computational virtual
screening using ligand-based and structure-based screening parallel screening strategy. Then a structural
optimization was performed and has led to the discovery of the compound C6. It has good activity in vitro
and in vivo, which could reduce infarct size by ~70% in ischemic stroke rat models. This study provided
good lead compounds for the further development of ALDH2 activators.
Received 21 September 2020
Received in revised form
1 December 2020
Accepted 17 December 2020
Available online 28 December 2020
Keywords:
Aldehyde dehydrogenase 2
Allosteric activators
Inactive Asian variant
Virtual screening
© 2020 Elsevier Masson SAS. All rights reserved.
Structural optimization
Ischemic stroke
1. Introduction
the highest percentage of the population in Southeast China (up to
65% in Changting County of Fujian province), and 35e45% of East
Aldehyde dehydrogenase (ALDH), one of the most important
superfamilies of phase I oxidizing enzymes besides P-450 super-
family, is responsible for the detoxification of biogenic and xeno-
genic aldehydes [1,2]. ALDH2, one of 19 ALDHs, is well known for its
role in detoxifying acetaldehyde generated from ethanol meta-
bolism. ALDH2 also plays a key role in oxidizing endogenous
aldehydic products that arise from lipid peroxidation under
oxidative stress, such as 4-hydroxy-2-nonenal (4-HNE) and
malondialdehyde (MDA) [1,3]. Therefore, it serves as an important
shield from the damage occurring under oxidative stress.
Asians, while lower than 5% of Caucasians [4,5]. This variant has
recently more and more been recognized as having a major impact
on human health and disease, besides its effect on limiting alcohol
consumption. The variant carrier drinkers are at the higher risk for
several cancers [3,6,7]. In addition, even the variant carrier non-
drinkers are also at the higher risk for the diseases related to the
damage occurring under oxidative stress, such as cardiovascular
and cerebrovascular diseases, diabetes, neurodegenerative disease,
Fanconi anemia, pain, and osteoporosis [1,3,8]. In particular, that an
increased incidence of cardiovascular and cerebrovascular diseases
(myocardial ischemia [9e11] or cerebral ischemia [12e15]) is
associated with the variant was discovered. As a result, ALDH2
activators may potentially serve as a new class of therapeutics, and
they can also serve as important molecular probes to further
explore the function of ALDH2.
A special variant ALDH2*2 with reduced enzymatic activity was
found originated from Chinese thousands years ago. It is carried by
* Corresponding author.
ALDH2 is a tetramer of four identical subunits, and each is
composed of three main domains: the catalytic domain, the
** Corresponding author.
E-mail address: canhuizheng@smmu.edu.cn (C. Zheng).
1
These authors contributed equally.
https://doi.org/10.1016/j.ejmech.2020.113119
0223-5234/© 2020 Elsevier Masson SAS. All rights reserved.

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
coenzyme or NADþ-binding domain (Fig. 1), and the oligomeriza-
tion domain. The ALDH2*2 mutation occurs at amino acid residue
487 locating within the oligomerization domain, where a basic
amino acid lysine substituted for an acidic amino acid glutamic acid
(E487K). This single amino acid substitution disorders a large re-
gion located at the dimer interface, causing an allosteric disruption
of the catalytic and coenzyme-binding sites. This structural insta-
bility of the ALDH2*2 results in the reduced enzymatic activity
compare with wide-type ALDH2 [3,16e18]. The only well studied
family of ALDH2 activators, represented by Alda-1, was discovered
using a high-throughput screening [11]. Cocrystal structures of
activator Alda-1 with wild-type ALDH2 or ALDH2*2 (Fig. 1)
demonstrated that Alda-1 binds at the entrance of the catalytic
domain, and the binding of Alda-1 restores the disordered region
nearly to the native wildtype state even though Alda-1 has no direct
contact with these residues. Alda-1 is therefore an activator which
exerts its allosteric effect to restore the structural defect of a cata-
lytically impaired enzyme [1,16]. It could increase the activity of
ALDH2*2 by ~ 10 folds, and ALDH2 by ~ 2 folds [11,15]. Previously,
structural modification of Alda-1 with N-benzylbenzamide scaffold
was conducted by us. Several new derivatives with improved
water-solubility were obtained, while no ALDH2 activator with
better activity than Alda-1 was found [19].
Nowadays, computational virtual screening is becoming a
powerful tool for the identification of lead compound in drug dis-
covery [20e23]. However, only a few successful examples of the
virtual screening of the allosteric activators have been reported.
The reason maybe is the binding of allosteric activators often results
in a conformational change of the target protein, which reduces the
accuracy of molecular docking and then increases the difficulty of
structure-based virtual screening. In this study, taking advantage of
complementary ligand-based and structure-based virtual
screening techniques and overcoming their limitations, a parallel
screening strategy was used. It identified N-benzylanilines as a
novel family of allosteric activators of ALDH2. Then a structural
optimization was performed and has led to the discovery of the lead
compound C6 with good activity in vitro and in vivo.
2. Results and discussion
2.1. First round of virtual screening: pharmacophore-based and
structure-based parallel screening
3D Pharmacophores are widely used for virtual screening of da-
tabases in order to identify new lead series [24]. The
ReceptorꢀLigand Pharmacophore Generation (RLPG) Protocol gen-
erates pharmacophore models directly from the receptorꢀligand
interactions as revealed in the 3-D structures, which are recom-
mend for profiling targets for which only a few ligands but a 3-D
structure is available [25,26]. RLPG was used to generate the phar-
macophore model of the activators of ALDH2. The model identified
mainly five functional features, including one ring aromatic region,
three hydrophobic regions, hydrogen-bond donor, and several
exclusion regions (Fig. 1). The ring aromatic region feature overlaps
the phenyl group of Alda-1 binding deep into the ALDH2. This
feature is consistent with the Alda-1-ALDH2 interaction mode: one
phenyl group shows p-p stacking interactions with Phe459, Phe170
and Phe296; The three hydrophobic regions overlap with the
methylene in dioxolane group, the other phenyl group and a chlorine
group of Alda-1, respectively, which show hydrophobic interactions
with the active site; The hydrogen-bond donor feature overlap with
the nitrogen in the amide group that have been reported to interact
with the backbone of Asp457. To validate the model,
a
pharmacophore-based virtual screening was undertaken on a test
database, which contains Alda-1 and 200 compounds random
selected from SPECS database. Unfortunately, 15 compounds got
higher score than Alda-1, which indicated the low efficiency of this
screening. After analysis, because of the catalytic site is adjacent to
the binding site of ALDH2 activators most of the false positive
compounds bound deep into it, which should be the binding site of
inhibitors. Therefore, three exclusion region features (Fig. 1) were
added manually to improve the effectiveness of virtual screening.
Using the new pharmacophore model, Alda-1 successfully got the
highest score in the test screening.
Additionally, Glide was selected as the docking method for virtual
screening by comparing four docking programs [27]. This method
was further validated by the test screening described above, which
Fig. 1. Binding modes of Alda-1 with ALDH2 and receptor-ligand based pharmacophore model of ALDH2 (crystal structure PDB ID: 3INL). Alda-1 is showed as purple sticks. The
catalytic domain and the NADþ-binding domain are showed inside the red circles. The model included one ring aromatic region (orange sphere), three hydrophobic regions (cyan
spheres), hydrogen-bond donor (magenta sphere), several exclusion regions (gray spheres) and three exclusion regions added manually (red spheres). (For interpretation of the
references to colour in this figure legend, the reader is referred to the Web version of this article.)
2

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
gave Alda-1 second highest score, which indicated that it was also a
suitable method for structure-based virtual screening. Then in the
first round, a pharmacophore-based and structure-based parallel
screening was carried out to search the SPECS database [22], the
schematic of which is presented in Fig. 2A. After extra precision
docking, visual inspection and with the consideration of the chem-
ical diversity, 55 candidates were selected and purchased finally
from top ranked compounds (Table S1).
Considered that the ALDH2 activators was reported to have the
activation activity both on ALDH2 and ALDH2*2,11 and only ALDH2
protein is commercially available, the activities of the title com-
pounds were evaluated in a fluorescent ALDH2 enzyme assay [11].
The maximum fold activation was taken as the primary indicator,
and the activities were corrected by the control Alda-1. The
maximum fold activation of Alda-1 was 1.6e1.8 in our different
tests, which was assigned as 100%. All purchased compounds were
evaluated, and 2 hits showed detectable activities (Table 1).
Because compound A1 has an ester group inside its scaffold which
is unstable in vivo, compound A2 with N-benzylaniline scaffold,
which showed 8.8% maximum fold activation relative to Alda-1,
was selected to further studies.
benzodioxole group resulted in compound B1, which had slightly
better activation activity than A2. Replacement of morpholine or
carboxylic acid group with chlorine group got compounds B2 or B3
respectively, and replacement or removing of these two groups
simultaneously got compounds B4 or B5. These four compounds
could not show detectable activities, indicating that morpholine
and carboxylic acid groups were important for their activation ac-
tivity. In addition, introduction of ethyl ester group on the car-
boxylic acid group of compound B1 obtained compound B6, which
showed much better activation activity than B1 and A2, indicating
that hydrophobic substitutes on the carboxylic acid group were
desirable for achieving a better activity. Compound B6 displayed
63.7% maximum fold activation corrected by that of Alda-1, and it
could be a good starting point for further optimization studies.
2.3. Structural optimization
The docking mode of B6 in complex with ALDH2 (Fig. 3B) was
obtained using crystal structure 3INJ as templet. One phenyl group
showed
p-p stacking interactions and the nitrogen in the benzy-
lamino group formed hydrogen-bond with ALDH2, similar with the
same groups of Alda-1 (Fig. 3A). However, the other groups of B6
displayed quite different binding mode to ALDH2 with Alda-1: the
ester group showed hydrogen-bond with the backbone of Phe459
as a hydrogen-bond acceptor. The ethyl group in the ester group
formed hydrophobic interactions with a narrow hydrophobic
pocket on the surface of ALDH2. The morpholine group bound to a
polar pocket on the surface of ALDH2. These two sites haven’t been
utilized by Alda-1 for binding. This docking mode was consistent
with the preliminary SAR of this type of compound from the second
round screening, and could guide the further structural optimiza-
tion of B6 (Table 3).
Considered that the ester group is unstable in vivo, the
replacement of it of B6 with amide group was performed. In
addition, the 1,3-benzodioxole group of Alda-1 could be replaced by
3-fluoro-4-methoxyphenyl group with more stability [28].
Considered that the same group of B6 was predicted to bind to
ALDH2 in the same site, the replacement was also carried out to
generate compound C1. It had comparable activation activity with
B6, and was used as the new template compound for subsequent
modifications.
2.2. Second round of virtual screening: hit-based substructure
search
Considered that compounds in many commercial chemical da-
tabases are inherently enriched with analogues, a structure-activity
relationship (SAR) of the hits identified through virtual screening
could easily be established and better hits could be found through
subsequent extraction of analogues in the same database [20]. Then
the second round hit-based substructure search, another ligand-
based screening, was undertake to search the Specs database
(showed in Fig. 2B). Because 3,4-dichlorophenyl group of A2 was
predicted to bind to ALDH2 in the same site with the 1,3-
benzodioxole group of Alda-1, which has better binding capacity,
N-(benzo[d][1,3]dioxol-5-ylmethyl)aniline (showed in Fig. 2 as A2
scaffold) was used as the scaffold of the following substructure
search. 6 Analogues of A2 were selected and purchased for bio-
logical evaluation finally. The activation activities of all compounds
were evaluated and displayed in Table 2.
Replacement of the 3,4-dichlorophenyl group of A2 with 1,3-
Fig. 2. The strategy of two rounds of virtual screening: pharmacophore-based and structure-based parallel screening (A), and hit-based substructure search (B).
3

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
Table 1
Structures and activation activities of the hits obtained from the first round of virtual screening.
No.
Structures
Maximum fold activation corrected by that of Alda-1 (%)
A1
6
1
A2
9
2
Table 2
Structures and activation activities of the hits obtained from the second round of virtual screening..
No.
Structures
R [1]
Maximum fold activation corrected by that of Alda-1 (%)
R [2]
B1
23
1
B2
B3
N. A.
N. A.
B4
B5
B6
N. A.
N. A.
64
/
2
N. A.: No activity.
Firstly, we performed modifications of the R1 group on com-
pound C1 (Table 3). A series of compounds C2 e C11 containing
different aliphatic R1 groups were synthesized. Compounds C3, C4,
C5 and C6 with substitutes containing 3e4 carbons had good
activation activities. Compound C6 showed 104.1% maximum fold
activation corrected by that of Alda-1. These substitutes may be the
groups which fitted with the R1 binding pocket of ALDH2 very well
(Fig. 3C). Then, the fluorine or hydroxyl group was introduced into
R1 of C5 and C6 to get compounds C12, C13 and C14, which also had
good activation activities. The compound C14 had the best activity,
with 119.9% maximum fold activation corrected by that of Alda-1.
The hydroxyl group of C14 was predicted to stretch to the open-
ing of the active site, and might form hydrogen bonds potentially
with the side chain of Lys127 considered its flexibility (Fig. 3D).
We next employed compound C6 as the template compound for
further modifications of R2 group (Table 3). Removing of the
morpholine group or the nitrogen atom in it got compounds C15
and C20, which displayed reduced activation activity than C6. In
addition, replacement of the morpholine group by phenyl, or the
oxygen atom by methylene with or without fluorine substituted
resulted in compounds C21, C16 e C18. These compounds showed
very low or no detectable activity. However, compound C19
generated from moving the oxygen atom out of the morpholine
ring, displayed slightly weaker activity than C6. Above results
indicated that the morpholine group, especially the distal polar
atom, was important for the activity of this type of compound.
Finally, we introduced methyl or acetyl as R3 group on the ni-
trogen in the benzylamino group of compounds C6, C13 and C14 to
get compounds C22, C23 and C24 respectively, which displayed no
detectable activity. It was consistent with the predicted binding
model, in which the nitrogen atom of this type of compound
formed important hydrogen-bonds with ALDH2. In addition, we
4

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
Fig. 3. Binding modes of Alda-1 (A) and docking mode of B6 (B), C6 (C) and C14 (D) in complex with ALDH2 (crystal structure PDB ID: 3INJ). Ligands are showed as purple sticks.
ALDH2 is shown in solvent-accessible surface or gray ribbons representation, and important interacting residues are shown in gray sticks. Hydrogen bonds are showed as cyan
dotted lines. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
replaced the methylene in the benzylamino group of compound C6
with carbonyl group to yield compound C25. It also showed no
detectable activity, which indicated that the N-benzylaniline scaf-
fold of this type of compound was essential for their activity.
were constructed by the reductive amination reaction with 3-fluoro-
4-methoxybenzaldehyde. Then they were hydrolyzed using NaOH in
MeOH/THF/H2O to obtain the key intermediates IV. Finally, title
compounds C1eC19 were got by condensation with different amines
under the condition of EDCI/DMAP/DCM. Another title compound
C22wasobtainedbyreductiveaminationreactionwithformaldehyde
aqueous solution from compound C6. Title compounds C23 and C24
were obtained by acylation reaction with acetic anhydride from
compounds C13 and C14 separately.
2-Bromine-5-nitrobenzoic acid (3) was condensation with cyclo-
propane methyl amine to produce compound 4. Then the bromine
group of 4 was displaced by tetrahydropyran group by an onepot
Suzuki-Miyaura-hydrogenation reaction to get compound 5 [29]. It
could also be displaced by phenyl group by Suzuki-Miyaura coupling
reaction and a following reduction reaction to get compound 7.
Finally, title compounds C20eC21 were constructed by the reductive
amination reaction with 3-fluoro-4-methoxybenzaldehyde sepa-
rately. In addition, compound 1a was reacted with morpholine to
produce compound 8. Then, compound 9 were got by condensation
with cyclopropane methyl amine under the condition of HATU/
DIPEA/DMF. It was reduced to form compound 10. Finally, title com-
poundC25wasgotbycondensationwith3-fluoro-4-methoxybenzoic
acid.
2.4. Further biological evaluation of C6
Representative ALDH2 activator C6 showed good dose-response
relationship in the ALDH2 activation assay for different concen-
trations (Fig. 4A). The EC50 value could be determined from the
curve as the second indicator of the activation activities. Compound
C6 had the EC50 value of 17.9 1.0
mM, which slight weaker than
that of Alda-1 (4.3 1.0 M). Furthermore, to preliminary evaluate
m
its effect in vivo, middle cerebral artery occlusion (MCAO) Sprague-
Dawley (SD) rat model was used, which could simulate ischemic
stroke-like events [15]. A potent ALDH2 activator C6 was intra-
cerebroventricular (i.c.v.) injected and the infarct size of the brain
was measured by the TTC staining assay. The injection of C6 (64
in 5 l) could reduce infarct size significantly by ~ 70% (Fig. 4B and
C), which achieved comparable good effect with Alda-1 in the same
molar dosage (50 g in 5 l). In conclusion, this type of compound
mg
m
m
m
showed good activity in vitro and in vivo, therefore they could be
the valuable lead compound of ALDH2 activator for further research
and development.
3. Conclusion
2.5. Chemistry
In this study, N-benzylanilines were found as novel allosteric
activators of ALDH2 by two rounds of virtual screening,
pharmacophore-based and structure-based parallel screening, and
hit-based substructure search. Then a structural optimization was
performed and has led to the discovery of compound C6 with good
activity in vitro and in vivo. It could reduce infarct size by ~70% in
ischemic cerebral injury rat models. This study provided good lead
compounds for the further development of the ALDH2 activator,
The compounds in Table 3 were synthesized using the method
outlined in Scheme 1. Firstly, 2-fluoro-5-nitrobenzoic acid (1a) or 3-
nitrobenzoic acid (1b) was esterified by SOCl2 in EtOH to get com-
pound 2a or 2b. The fluorine group of 2a was displaced by different
amines by heating with K2CO3 in DMSO to produce compounds I.
Then the nitro group of compounds 2b or I were reduced by heating
with Fe/NH4Cl in EtOH/H2O to form compounds II. Intermediates III
5

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
Table 3
Structures and activation activities of the compounds generated from the structural optimization..
No.
Structures
R [1]
Maximum fold activation corrected by that of Alda-1 (%)
R [2]
R [3]
-H
R [4]
-H
C1
63
52
94
97
1
1
3
6
C2
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
-H
C3
C4
C5
92 0.0
C6
104
59
2
C7
2
C8
41 10
C9
51
80
2
5
C10
C11
C12
C13
C14
66 12
101
97
6
6
5
120
8
C15
C16
-H
-H
-H
-H
-H
56
N. A.
C17
C18
-H
-H
-H
-H
N. A.
12
1
6

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
Table 3 (continued)
No.
Structures
R [1]
Maximum fold activation corrected by that of Alda-1 (%)
R [2]
R [3]
-H
R [4]
-H
C19
C20
C21
C22
80
53
3
4
3
-H
-H
-H
-H
-H
7
-CH3
N. A.
N. A.
N. A.
N. A.
C23
C24
C25
-H
-H
-H
N. A.: No activity.
Fig. 4. (A) Dose-response curve of assay for ALDH2 activation for C6 and Alda-1. The activities were showed as the maximum fold activation corrected by the activities of Alda-1; (B)
The protective effect of C6 and Alda-1 with same molar dosage against ischemic stroke in MCAO male rats. The infarct size in MCAO rats are shown as mean SD, **P < 0.01, n ¼ 6;
(C) Representative 1% TTC staining of 6 corresponding coronal brain sections of rats with vehicle, C6 and Alda-1 treatment. The ischemic infarct region is white.
further extensive SAR study was ongoing. The potent ALDH2 acti-
vators would serve as a new class of therapeutics for several dis-
eases related to the damage occurring under oxidative stress, and
they could also serve as important molecular probes to further
explore the function of ALDH2.
manual mode. The models were validated by a pharmacophore-
based virtual screening on a test database, which contains Alda-1
and 200 compounds random selected from SPECS database.
4.1.2. Molecular docking
The ligand Alda-1 extracted from the crystal structure complex
3INL was docked back into ALDH2 by Libdock [30], Gold [31], MVD
[32] and Glide [27]. The highest scoring conformations in the
docking results were compared with the original conformation
within the crystal structure complex to calculate the RMSD values,
which showed that Glide and MVD offered the best results (0.3933
and 0.4191 respectively). Then, these two methods were further
validated by the test screening described above, which gave Alda-1
second and 28th highest score respectively. The result indicated
that Glide was the better method for structure-based virtual
screening in this study.
4. Experimental section
4.1. Molecular modeling
4.1.1. Pharmacophore model construction
The X-ray crystal structure of ALDH2-Alda-1 complex was ob-
tained from the Brookhaven Protein Databank entry 3INL. All
crystallographic water molecules were removed from the coordi-
nate set. The pharmacophore model of ALDH2 activators was
constructed on Discovery Studio software package [30]. Most of the
functional features in pharmacophore model were generated by the
ReceptorꢀLigand Pharmacophore Generation Protocol. The exclu-
sion region features were generated by the Exclusion Constraint
Protocol and clustered by the Cluster Current Features Protocol or
Docking runs of Libdock were performed using the Receptor-
ligand interactions Protocol on Discovery Studio. The Fast confor-
mation method and high quality docking preferences, and best
conformation method were used, and other parameters were
7

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
Scheme 1. Synthesis of Title Compounds. Reagents and conditions: (a) SOCl2/EtOH, r. t., 98%; (b) K2CO3, DMSO, 80ꢁC, 95%; (c) Fe/NH4Cl, EtOH/H2O, 80ꢁC, ~ 80%; (d) NaBH(OAc)3,
DCE, 80ꢁC to r. t., 80-85%; (e) NaOH, MeOH/THF/H2O, r. t., ~ 90%; (f) EDCI/DMAP, DCM, r. t., 70-80%; (g) DMAP, DCM, C5H5N, rt, 87%; (h) PdXPhosG2 (1 mol%), Pd/C(12mol%),
K3PO4(3eq), Dioxane:H2O(4:1), 80 ℃, 4h; then MeOH, NH4COOH , rt to 60℃, 16h; 75%; (i) Pd(dppf)Cl2, K2CO3, Dioxane:H2O(3:1), 100 ℃, 12h, 70%.
default; For docking runs of Gold, GoldScore was used as the
scoring function, most accurate genetic algorithm search options
was set, and other parameters were default; For docking runs of
MVD, MolDock Score and MolDock Score [Grid] were used as
scoring functions, the default settings were used; Docking runs of
Glide were performed by the Ligand Docking Protocol on
4.1.3. Virtual screening
In the first round, pharmacophore-based and structure-based
parallel screening was carried out, which completed by Screen Li-
brary Protocol installed on Discovery Studio and Glide respectively.
The respective top 500 compounds obtained by two methods (978
compounds totally) was subjected to extra precision docking by
Glide. Then the results were visual inspected to select the candidate
compounds based on their binding mode and chemical diversity.
Finally, 55 candidates were selected and purchased. The second
round (hit-based substructure search was completed by Align to
Selected Substructure Protocol on Discovery Studio. 23 Analogues
€
Schrodinger. 3 Å around Cys301, Cys302, and Cys303 were set as
excluded volumes for preparation of the active cavity grid files.
Standard precision and flexible ligand sampling was employed, and
other parameters were default. In addition, extra precision was
used to obtain more accurate docking results.
8

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
were found, and subjected to extra precision docking by Glide.
Finally, 6 compounds were selected and purchased for biological
evaluation after visual inspection. No PAINS liability was found in
compounds reported by analysis in the online filter (http://zinc15.
docking.org/patterns/home).
reaction was quenched with the saturated sodium bicarbonate
solution, extracted with ethyl acetate, concentrated by rotary
evaporation and separated by silica gel column (PE/EA system) to
obtain compound III with 80e85% yield.
4.2.5. General method for step e
4.2. Chemistry
Compound III (1.0 mmol) was dissolved with 10 ml of mixed
solution with the ratio of MeOH: THF at 1:1. Sodium hydroxide
(400 mg, 10.0 mmol) was dissolved in 8 ml of mixed solvents with
the ratio of MeOH: H2O at 5:3, poured into the reaction mixture,
and reacted at room temperature for 4 h. Completion of the reac-
tion was confirmed by TLC. The reaction solution was concentrated
by rotary evaporation and separated via a silica gel column (PE/EA
system) to afford the pure compound IV with ~90% yield.
1H NMR and 13C NMR spectra were recorded on Bruker AC-300P
or AVANCE II600 spectrometer (Bruker Biospin, Swiss), using TMS
as an internal standard and CDCl3 or DMSO‑d6 as solvents. Chemical
shifts are given in ppm (d), and the spectral data are consistent with
the assigned structures. The mass spectra were recorded on a
Micromass Qtof-Micro LCꢀMS instrument. Silica gel thin-layer
chromatography was performed on precoated plates GF-254. Sil-
ica gel column chromatography was performed with silica gel 60 G.
All compounds were routinely checked by TLC by using silica gel
plates GF-254. All solvents and reagents were analytically pure and,
when necessary, were purified and dried by standard protocols. All
starting materials were commercially available unless otherwise
indicated. Yields of purified products were not optimized. The pu-
rity of key compounds (>95%) was determined on an Agilent 1100
series liquid chromatography (LC) system (column, Diamonsil Plus
4.2.6. General method for step f
Compound IV (1 mmol), appropriate amines (1.1 mmol, 1.1 eq.),
EDCI (0.96 g, 5.0 mmol) and DMAP (244.3 mg, 2.0 mmol) were
dissolved in anhydrous dichloromethane and allowed to react at
room temperature overnight. Completion of the reaction was
confirmed by TLC. The reaction solution was washed with 1 M
diluted hydrochloric acid and saturated sodium bicarbonate solu-
tion successively, concentrated by rotary evaporation and purified
via a silica gel column (PE/EA system) to afford the products with
70e80% yield.
5
m
m C18-A 4.6250 ꢂ 4.6 mm; mobile phase, CH3CN/H2O; UV
wavelength, absorbance at 254 nm). The candidates of virtual
screening (Table S1) were purchased from the commercial source
(1H NMR and LC-MS spectra are available on the website: http://
www.specs.net).
4.2.7. General method for step g
Compound C13 or C14 (1.0 mmol), acetic anhydride (2.0 mmol, 2
eq.), and DMAP (0.50 mmol, 0.5 eq.) was added into 5 ml pyridine.
The reaction mixture was stirred for 3 h at room temperature.
Completion of the reaction was confirmed by TLC. The mixture was
extracted with ethyl acetate, and concentrated by rotary evapo-
rator, then purified by silica gel column (DCM/MeOH system) to
provide the products with ~80% yield.
4.2.1. General method for step a
Compound 1a or 1b (10.8 mmol) was added to ethanol solution
(15 ml), into which thionyl chloride (1.54 g, 12.9 mmol) was slowly
dropped under ice-cooling and stirred for 5 min. Subsequently, the
solution was refluxed at 80 ꢁC for 3 h. The reaction was monitored
with TLC until it completed. Then the reaction solution was
concentrated by rotary evaporation and separated by silica gel
column (PE/EA system) to afford compound 2a or 2b with 98%
yield.
4.2.8. Method for step h
Compound 4 (50 mg, 1.0 mmol), 2-(3,6-dihydro-2H-pyran-4-
yl)- 4,4,5,5-tetramethyl-1,3,2-dioxaborolane (52 mg, 1.0 mmol),
PdXPhosG2 (1 mg, 0.01 mmol), Pd/C(22 mg, 0.12 mmol) and K3PO4
(107 mg, 3.0 mmol) was added into 4 ml of the mixed solvent of
dioxane/H2O (3:1) under argon atmosphere. The reaction mixture
was heated at 80 ꢁC for 4 h, then cooled to room temperature.
Subsequently, the reaction solution was added with ammonium
formate (106 mg, 10 mmol) dissolved in methol (2 ml) and reacted
at room temperature for 16 h. Completion of the reaction was
confirmed by TLC. The reaction was filtered with celite, concen-
trated by rotary evaporation and purified by silica gel column (PE/
EA system) to afford white solid with 75% yield.
4.2.2. General method for step b
Compound 2a (2 g, 9.4 mmol), appropriate amines (9.4 mmol, 1
eq.) and potassium carbonate (2.6 g, 18.8 mmol) were dissolved in
15 ml of DMSO and heated under reflux at 80 ꢁC for 2 h. The re-
action was monitored with TLC until it completed. The reaction
solution was poured into ice water (300 ml), extracted with EA in
triplicate, concentrated by rotary evaporation and separated by
silica gel column (PE/EA system) to get compound I with 95% yield.
4.2.3. General method for step c
Compound 2b or I (9.4 mmol) was dissolved in ethanol (15 ml),
which was added with ammonium chloride (1.76 g, 32.9 mmol)
dissolved in 5 ml of water. The solution was heated up to 80 ꢁC and
added with iron powder (1.76 g, 31.5 mmol) in batches. After 2 h of
reaction, the reaction was confirmed by TLC to have completed.
After the filtration with Celite, the filtrate was concentrated by
rotary evaporation and separated via the silica gel column (PE/EA
system) to afford compound II with ~80% yield.
4.2.9. Method for step i
Compound 4 (50 mg, 1.0 mmol), 4,4,5,5-tetramethyl-2-phenyl-
1,3,2-dioxaborolane (34 mg, 1.0 mmol), Pd(dppf)Cl2 (12 mg,
0.10 mmol), and K2CO3 (69 mg, 3.0 mmol) was added into 4 ml the
mixed solvent of dioxane:H2O (3:1) under Argon atmosphere.
Then, the reaction mixture was heated to 100 ꢁC for 12 h.
Completion of the reaction was confirmed by TLC. The reaction was
filtered with celite, concentrated by rotary evaporation and purified
by silica gel column (PE/EA system) to afford white solid with 70%
yield.
4.2.4. General method for step d
Compound II (1.0 mmol) and aldehydes (1.0 mmol, 1.1 eq.) were
dissolved in dichloroethane (10 ml), stirred at 80 ꢁC for 1 h and
allowed to cool down to room temperature. The solution was added
with sodium triacetoxyborohydride (338 mg, 1.6 mmol) and
allowed to react overnight at room temperature under the argon
atmosphere. Completion of the reaction was confirmed by TLC. The
4.2.10. N-ethyl-5-((3-fluoro-4-methoxybenzyl)amino)-2-
morpholinobenzamide (C1)
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
9.96 (t, J ¼ 5.3 Hz,
1H), 7.17e7.06 (m, 5H), 6.62 (dd, J ¼ 8.7, 3.0 Hz, 1H), 6.32 (t,
J ¼ 6.1 Hz, 1H), 4.19 (d, J ¼ 6.2 Hz, 2H), 3.79 (s, 3H), 3.76e3.66 (m,
9

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
4H), 3.31e3.27 (m, 2H), 2.80e2.75 (m, 4H), 1.15 (t, J ¼ 7.2 Hz, 3H);
HRMS (ESI) m/z found 388.2043 (MþHþ), while C21H26FN3O3
(MþHþ) requires 388.2031.
4.2.17. N-(cyclobutylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-morpholinobenzamide (C8)
White solid; 1H NMR (300 MHz, Chloroform-d)
d 10.32 (s, 1H),
7.58 (d, J ¼ 2.9 Hz, 1H), 7.11e7.01 (m, 3H), 6.88 (t, J ¼ 8.4 Hz, 1H),
6.62 (dd, J ¼ 8.6, 2.9 Hz, 1H), 4.26 (s, 2H), 3.91e3.74 (m, 7H), 3.47
(dd, J ¼ 7.4, 5.5 Hz, 2H), 3.01e2.82 (m, 4H), 2.69e2.46 (m, 1H),
2.18e2.00 (m, 2H), 1.98e1.85 (m, J ¼ 7.3, 6.6 Hz, 2H), 1.83e1.62 (m,
2H); HRMS (ESI) m/z found 428.2345 (MþHþ), while C24H30FN3O3
(MþHþ) requires 428.2344.
4.2.11. 5-((3-fluoro-4-methoxybenzyl)amino)-N-methyl-2-
morpholinobenzamide (C2)
Light yellow solid; 1H NMR (600 MHz, DMSO‑d6)
d 9.74 (d,
J ¼ 4.7 Hz, 1H), 7.17e7.09 (m, 4H), 7.04 (d, J ¼ 8.6 Hz, 1H), 6.62 (dd,
J ¼ 8.6, 2.9 Hz, 1H), 6.30 (t, J ¼ 6.1 Hz, 1H), 4.18 (d, J ¼ 6.1 Hz, 2H),
3.79 (s, 3H), 3.73e3.68 (m, 4H), 2.81 (d, J ¼ 4.8 Hz, 3H), 2.78e2.72
(m, 4H); HRMS (ESI) m/z found 374.1876 (MþHþ), while
4.2.18. N-(cyclopentylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-morpholinobenzamide (C9)
C
20H24FN3O3 (MþHþ) requires 374.1874.
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.10 (t, J ¼ 5.7 Hz,
1H), 7.19 (d, J ¼ 2.9 Hz, 1H), 7.17e7.13 (m, 1H), 7.12e7.07 (m, 3H),
6.63 (dd, J ¼ 8.7, 3.0 Hz, 1H), 6.34 (t, J ¼ 6.2 Hz, 1H), 4.19 (d,
J ¼ 6.1 Hz, 2H), 3.79 (s, 3H), 3.72e3.68 (m, 4H), 3.23 (dd, J ¼ 7.1,
5.9 Hz, 2H), 2.80e2.76 (m, 4H), 2.14e2.08 (m, 1H), 1.75e1.67 (m,
2H), 1.62e1.55 (m, 2H), 1.54e1.47 (m, 2H), 1.24e1.21 (m, 2H); HRMS
(ESI) m/z found 442.36 (MþHþ), while C25H32FN3O3 (MþHþ) re-
quires 442.2500.
4.2.12. N-cyclopropyl-5-((3-fluoro-4-methoxybenzyl)amino)-2-
morpholinobenzamide (C3)
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.07 (d, J ¼ 4.6 Hz,
1H), 7.18e7.03 (m, 5H), 6.62 (dd, J ¼ 8.7, 2.9 Hz,1H), 6.35 (s,1H), 4.19
(s, 2H), 3.79 (s, 3H), 3.66 (dd, J ¼ 5.9, 3.3 Hz, 4H), 2.88e2.80 (m, 1H),
2.75 (t, J ¼ 4.5 Hz, 4H), 0.76e0.70 (m, 2H), 0.55e0.49 (m, 2H); HRMS
(ESI) m/z found 400.2028(MþHþ), while C22H26FN3O3 (MþHþ) re-
quires 400.2031.
4.2.19. N-(cyclohexylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-morpholinobenzamide (C10)
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.16 (d, J ¼ 7.7 Hz,
4.2.13. 5-((3-fluoro-4-methoxybenzyl)amino)-2-morpholino-N-
propylbenzamide (C4)
1H), 7.21 (d, J ¼ 3.0 Hz, 1H), 7.17e7.08 (m, 4H), 6.62 (dd, J ¼ 8.7,
3.0 Hz, 1H), 6.36 (t, J ¼ 6.2 Hz, 1H), 4.19 (d, J ¼ 6.1 Hz, 2H), 3.79 (s,
3H), 3.75e3.65 (m, 5H), 2.85e2.72 (m, 4H), 1.96e1.86 (m, 2H),
1.77e1.68 (m, 2H),1.63e1.55 (m,1H),1.38e1.16 (m, 5H); HRMS (ESI)
m/z found 442.2502 (MþHþ), while C25H32FN3O3 (MþHþ) requires
442.2500.
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.02 (t, J ¼ 5.6 Hz,
1H), 7.17 (d, J ¼ 3.0 Hz, 1H), 7.16e7.06 (m, 4H), 6.62 (dd, J ¼ 8.7,
3.0 Hz, 1H), 6.33 (t, J ¼ 6.2 Hz, 1H), 4.19 (d, J ¼ 6.1 Hz, 2H), 3.79 (s,
3H), 3.73e3.66 (m, 4H), 3.24 (dd, J ¼ 13.0, 6.9 Hz, 2H), 2.81e2.74 (m,
4H), 1.58e1.51 (m, 2H), 0.92 (t, J ¼ 7.4 Hz, 3H); HRMS (ESI) m/z
found 402.2203 (MþHþ), while C22H28FN3O3 (MþHþ) requires
402.2187.
4.2.20. (5-((3-fluoro-4-methoxybenzyl)amino)-2-
morpholinophenyl)(piperidin-1-yl)methanone (C11)
Yellow solid; 1H NMR (600 MHz, DMSO)
d
7.15 (d, J ¼ 12.5 Hz,
4.2.14. 5-((3-fluoro-4-methoxybenzyl)amino)-N-isobutyl-2-
morpholinobenzamide (C5)
1H), 7.13e7.04 (m, 2H), 6.87 (d, J ¼ 8.7 Hz, 1H), 6.53 (dd, J ¼ 8.7,
2.4 Hz, 1H), 6.32 (s, 1H), 6.13 (s, 1H), 4.16 (s, 2H), 3.79 (s, 3H),
3.68e3.46 (m, 6H), 3.14e2.91 (m, 4H), 2.54 (s, 2H), 1.59e1.25 (m,
6H); HRMS (ESI) m/z found 428.2352 (MþHþ), while C24H30FN3O3
(MþHþ) requires 428.2344.
Yellow solid; 1H NMR (600 MHz, DMSO‑d6)
d 10.12 (t, 1H), 7.19
(d, J ¼ 2.9 Hz, 1H), 7.16e7.08 (m, 4H), 6.63 (dd, J ¼ 8.7, 2.9 Hz, 1H),
6.34 (t, J ¼ 6.1 Hz,1H), 4.19 (d, J ¼ 6.1 Hz, 2H), 3.79 (s, 3H), 3.74e3.66
(m, 4H), 3.13 (t, J ¼ 6.4 Hz, 2H), 2.81e2.75 (m, 4H), 1.85e1.77 (m,
1H), 0.91 (d, J ¼ 6.7 Hz, 6H); HRMS (ESI) m/z found 416.2353
(MþHþ), while C23H30FN3O3 (MþHþ) requires 416.2344.
4.2.21. N-(2-fluoro-2-methylpropyl)-5-((3-fluoro-4-
methoxybenzyl)amino)-2-morpholinobenzamide (C12)
Yellow solid; 1H NMR (300 MHz, Chloroform-d)
d 11.05 (s, 1H),
7.63 (d, J ¼ 3.0 Hz, 1H), 7.19e7.02 (m, 3H), 6.90 (t, J ¼ 8.4 Hz, 1H),
6.71 (dd, J ¼ 8.7, 2.8 Hz, 1H), 4.29 (s, 2H), 3.89e3.81 (m, 7H), 3.64
(dd, J ¼ 22.9, 5.7 Hz, 2H), 3.01e2.85 (m, 4H),1.44 (s, 3H),1.36 (s, 3H).
4.2.15. N-(cyclopropylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-morpholinobenzamide (C6)
White solid; 1H NMR (600 MHz, Chloroform-d)
d 9.88 (s, 1H),
HRMS
(ESI)
m/z
found
434.2260(MþH)þ,
While
7.78 (d, J ¼ 2.8 Hz, 1H), 7.18 (d, J ¼ 8.8 Hz, 1H), 7.10e7.01 (m, 2H),
6.91 (t, J ¼ 8.4 Hz, 1H), 6.69 (dd, J ¼ 8.9, 2.6 Hz,1H), 4.31 (s, 2H), 4.02
(t, J ¼ 4.5 Hz, 4H), 3.86 (s, 3H), 3.36e3.24 (m, 6H), 1.19e1.06 (m, 1H),
0.63e0.51 (m, 2H), 0.38e0.25 (m, 2H); 13C NMR (150 MHz, Chlo-
C
23H29F2N3O3(MþH)þ requires 434.2177.
4.2.22. 5-((3-fluoro-4-methoxybenzyl)amino)-N-((1-
hydroxycyclopropyl)methyl)-2-morpholinobenzamide (C13)
roform-d)
d
165.94,152.45 (d, J ¼ 246.1 Hz), 146.71 (d, J ¼ 10.7 Hz),
White solid; 1H NMR (600 MHz, Chloroform-d)
d 9.70 (s, 1H),
145.44, 141.19, 132.21 (d, J ¼ 5.5 Hz), 128.63, 123.01 (d, J ¼ 3.5 Hz),
122.49, 115.69, 115.58, 115.14 (d, J ¼ 18.5 Hz), 113.48 (d, J ¼ 2.2 Hz),
67.52, 56.30, 53.77, 47.42, 44.61, 10.79, 3.92; HRMS (ESI) m/z found
414.2200 (MþHþ), while C23H28FN3O3 (MþHþ) requires 414.2187.
7.88 (s,1H), 7.15 (d, J ¼ 8.4 Hz,1H), 6.97 (d, J ¼ 8.3 Hz,1H), 6.94e6.89
(m, 2H), 6.84 (t, J ¼ 8.4 Hz, 1H), 4.79 (s, 2H), 3.94e3.90 (m, 4H), 3.85
(s, 3H), 3.82e3.80 (m, 2H), 3.06e3.00 (m, 4H), 2.02 (s, 3H),
1.02e0.97 (m, 2H), 0.97e0.91 (m, 2H). HRMS (ESI) m/z found
470.2081(M-H)-, While C25H30FN3O5(MþH)þ requires 472.2169.
4.2.16. 5-((3-fluoro-4-methoxybenzyl)amino)-2-morpholino-N-
neopentylbenzamide (C7)
4.2.23. 5-((3-fluoro-4-methoxybenzyl)amino)-N-(2-hydroxy-2-
methylpropyl)-2-morpholinobenzamide (C14)
Light yellow solid; 1H NMR (600 MHz, DMSO‑d6)
d 10.13 (t,
J ¼ 6.4 Hz, 1H), 7.22 (d, J ¼ 3.0 Hz, 1H), 7.18e7.07 (m, 4H), 6.63 (dd,
J ¼ 8.7, 3.0 Hz, 1H), 6.35 (t, J ¼ 6.2 Hz, 1H), 4.19 (d, J ¼ 6.1 Hz, 2H),
3.79 (s, 3H), 3.72e3.67 (m, 4H), 3.15 (d, J ¼ 6.4 Hz, 2H), 2.84e2.77
(m, 4H), 0.91 (s, 9H); HRMS (ESI) m/z found 430.2508 (MþHþ),
while C24H32FN3O3 (MþHþ) requires 430.2500.
White solid; 1H NMR (600 MHz, Chloroform-d)
d 9.95 (s, 1H),
7.96 (s,1H), 7.17 (d, J ¼ 8.4 Hz,1H), 6.97 (d, J ¼ 8.5 Hz,1H), 6.94e6.89
(m, 2H), 6.84 (t, J ¼ 8.5 Hz, 1H), 4.80 (s, 2H), 3.93e3.89 (m, 4H), 3.85
(s, 3H), 3.49 (d, J ¼ 5.6 Hz, 2H), 3.03e2.98 (m, 4H), 1.88 (s, 3H), 1.30
(s, 6H). HRMS (ESI) m/z found 474.2417(MþH)þ, While
10

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
C25H32FN3O5(MþH)þ requires 474.2326.
C
24H29FN2O3 (MþHþ) requires 413.2235.
4.2.24. N-(cyclopropylmethyl)-3-(3-fluoro-4-
methoxybenzylamino)benzamide (C15)
White solid; 1H NMR (600 MHz, DMSO‑d6)
1H), 7.20e7.16 (m, 1H), 7.15e7.06 (m, 4H), 7.01 (d, J ¼ 7.6 Hz, 1H),
6.69 (dd, J ¼ 8.1, 2.4 Hz, 1H), 6.39 (t, J ¼ 6.1 Hz, 1H), 4.25 (d,
J ¼ 6.0 Hz, 2H), 3.80 (s, 3H), 3.11 (t, J ¼ 6.2 Hz, 2H), 1.08e0.95 (m,
1H), 0.46e0.33 (m, 2H), 0.21 (t, J ¼ 4.7 Hz, 2H); HRMS (ESI) m/z
found 329.1673 (MþHþ), while C19H21FN2O2 (MþHþ) requires
329.1600.
4.2.30. N-(cyclopropylmethyl)-4-((3-fluoro-4-methoxybenzyl)
amino)-[1,10-biphenyl]-2-carboxamide (C21)
d
8.33 (t, J ¼ 5.8 Hz,
White solid; 1H NMR (600 MHz, Chloroform-d)
d 7.36 (d,
J ¼ 3.6 Hz, 4H), 7.29 (m, 1H), 7.17 (d, J ¼ 8.3 Hz, 1H), 7.13e7.06 (m,
2H), 7.03 (d, J ¼ 2.5 Hz, 1H), 6.92 (t, J ¼ 8.4 Hz, 1H), 6.72 (dd, J ¼ 8.4,
2.6 Hz, 1H), 5.27 (t, J ¼ 5.5 Hz,1H), 4.32 (s, 2H), 3.87 (s, 2H), 2.95 (dd,
J ¼ 7.3, 5.3 Hz, 2H), 0.62e0.48 (m, 1H), 0.30e0.20 (m, 2H), ꢀ0.07
to ꢀ0.18 (m, 2H). HRMS (ESI) m/z found 405.1975 (MþHþ); while
C
25H25FN2O2 (MþHþ) requires 405.1973.
4.2.25. N-(cyclopropylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-(piperidin-1-yl)benzamide (C16)
4.2.31. N-(cyclopropylmethyl)-5-((3-fluoro-4-
methoxybenzyl)(methyl)amino)-2- morpholinobenzamide (C22)
White solid; 1H NMR (600 MHz, Chloroform-d)
d
11.04 (t,
Light yellow solid; 1H NMR (600 MHz, DMSO‑d6)
d 10.24 (t,
J ¼ 5.3 Hz, 1H), 7.59 (d, J ¼ 3.0 Hz, 1H), 7.15e7.03 (m, 3H), 6.92 (q,
J ¼ 8.1 Hz,1H), 6.63 (dd, J ¼ 8.6, 3.0 Hz, 1H), 4.27 (s, 2H), 3.88 (s, 3H),
3.31 (dd, J ¼ 7.2, 5.1 Hz, 2H), 2.85 (t, J ¼ 5.2 Hz, 4H), 1.81e1.70 (m,
4H), 1.14e1.00 (m, 1H), 0.64e0.50 (m, 2H), 0.35e0.24 (m, 2H).
HRMS (ESI) m/z found 412.2396 (MþHþ); while C24H30FN3O2
(MþHþ) requires 412.2395.
J ¼ 5.3 Hz, 1H), 7.31 (d, J ¼ 3.2 Hz, 1H), 7.21 (d, J ¼ 8.9 Hz,1H), 7.09 (t,
J ¼ 8.7 Hz,1H), 7.02 (dd, J ¼ 12.4, 2.1 Hz,1H), 6.98e6.93 (m,1H), 6.83
(dd, J ¼ 8.9, 3.3 Hz, 1H), 4.48 (s, 2H), 3.79 (s, 3H), 3.77e3.74 (m, 4H),
3.16 (dd, J ¼ 7.1, 5.3 Hz, 2H), 2.85e2.80 (m, 4H), 1.09e0.97 (m, 1H),
0.53e0.46 (m, 2H), 0.29e0.20 (m, 2H); HRMS (ESI) m/z found
428.2355 (MþHþ); while C24H30FN3O3 (MþHþ) requires 428.2344.
4.2.26. N-(cyclopropylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-(4-fluoropiperidin-1-yl)benzamide (C17)
4.2.32. 5-(N-(3-fluoro-4-methoxybenzyl)acetamido)-N-((1-
hydroxycyclopropyl)methyl)-2-morpholinobenzamide (C23)
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.40 (t, J ¼ 5.3 Hz,
White solid; 1H NMR (600 MHz, Chloroform-d)
d 9.70 (s, 1H),
1H), 7.23 (d, J ¼ 3.0 Hz, 1H), 7.16e7.12 (m, 1H), 7.11e7.06 (m, 3H),
6.62 (dd, J ¼ 8.6, 3.0 Hz, 1H), 6.33 (t, J ¼ 6.2 Hz, 1H), 4.89e4.75 (m,
1H), 4.18 (d, J ¼ 6.1 Hz, 2H), 3.78 (s, 3H), 3.14 (dd, J ¼ 7.1, 5.3 Hz, 2H),
2.91 (t, J ¼ 9.6 Hz, 2H), 2.76e2.71 (m, 2H), 2.06e1.96 (m, 2H),
1.94e1.84 (m, 2H), 1.04e0.97 (m, 1H), 0.49e0.45 (m, 2H), 0.24e0.21
(m, 2H). HRMS (ESI) m/z found 430.2306 (MþHþ); while
7.88 (s,1H), 7.15 (d, J ¼ 8.4 Hz,1H), 6.97 (d, J ¼ 8.3 Hz,1H), 6.94e6.89
(m, 2H), 6.84 (t, J ¼ 8.4 Hz, 1H), 4.79 (s, 2H), 3.94e3.90 (m, 4H), 3.85
(s, 3H), 3.82e3.80 (m, 2H), 3.06e3.00 (m, 4H), 2.02 (s, 3H),
1.02e0.97 (m, 2H), 0.97e0.91 (m, 2H). HRMS (ESI) m/z found
470.2081(M-H)-, While C25H30FN3O5(MþH)þ requires 472.2169.
C
24H29F2N3O2 (MþHþ) requires 430.2301.
4.2.33. 5-(N-(3-fluoro-4-methoxybenzyl)acetamido)-N-(2-
hydroxy-2-methylpropyl)-2-morpholinobenzamide (C24)
4.2.27. N-(cyclopropylmethyl)-2-(4,4-difluoropiperidin-1-yl)-5-
((3-fluoro-4-methoxybenzyl)amino)benzamide (C18)
White solid; 1H NMR (600 MHz, Chloroform-d)
d 9.95 (s, 1H),
7.96 (s,1H), 7.17 (d, J ¼ 8.4 Hz,1H), 6.97 (d, J ¼ 8.5 Hz,1H), 6.94e6.89
(m, 2H), 6.84 (t, J ¼ 8.5 Hz, 1H), 4.80 (s, 2H), 3.93e3.89 (m, 4H), 3.85
(s, 3H), 3.49 (d, J ¼ 5.6 Hz, 2H), 3.03e2.98 (m, 4H), 1.88 (s, 3H), 1.30
(s, 6H). HRMS (ESI) m/z found 474.2417(MþH)þ, While
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
9.90 (t, J ¼ 5.4 Hz,
1H), 7.18e7.12 (m, 2H), 7.12e7.05 (m, 3H), 6.61 (dd, J ¼ 8.7, 3.0 Hz,
1H), 6.35 (t, J ¼ 6.2 Hz, 1H), 4.19 (d, J ¼ 6.1 Hz, 2H), 3.79 (s, 3H), 3.15
(dd, J ¼ 7.1, 5.4 Hz, 2H), 2.91 (t, J ¼ 5.7 Hz, 4H), 2.21e2.07 (m, 4H),
1.08e0.97 (m, 1H), 0.50e0.41 (m, 2H), 0.30e0.19 (m, 2H). HRMS
(ESI) m/z found 448.2214 (MþHþ); while C24H28F3N3O2 (MþHþ)
requires 448.2206.
C
25H32FN3O5(MþH)þ requires 474.2326.
4.2.34. N-(cyclopropylmethyl)-5-(3-fluoro-4-methoxybenzamido)-
2-morpholinobenzamide (C25)
White solid; 1H NMR (600 MHz, DMSO‑d6)
d
10.34 (t, J ¼ 5.3 Hz,
4.2.28. N-(cyclopropylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-(4-hydroxypiperidin-1-yl)benzamide (C19)
1H), 7.22 (d, J ¼ 2.9 Hz, 1H), 7.08 (d, J ¼ 8.5 Hz, 1H), 6.65 (dd, J ¼ 8.5,
2.9 Hz, 1H), 5.12 (s, 2H), 3.75 (dd, J ¼ 5.6, 3.6 Hz, 4H), 3.16 (dd,
J ¼ 7.0, 5.3 Hz, 2H), 2.80 (t, J ¼ 4.6 Hz, 4H), 1.07e1.00 (m, 1H),
0.52e0.47 (m, 2H), 0.28e0.23 (m, 2H). HRMS (ESI) m/z found
428.1986(MþH)þ, While C23H26FN3O4(MþH)þ requires 428.1907.
White solid; 1H NMR (300 MHz, Chloroform-d)
d 10.48 (s, 1H),
7.58 (s, 1H), 7.14e6.99 (m, 3H), 6.91 (t, J ¼ 8.4 Hz, 1H), 6.73e6.56 (m,
1H), 4.28 (s, 2H), 3.87 (s, 3H), 3.59 (s, 4H), 3.30 (dd, J ¼ 7.1, 4.7 Hz,
2H), 2.88 (s, 4H), 1.49 (s, 9H), 1.17e0.94 (m, 1H), 0.64e0.45 (m, 2H),
0.37e0.16 (m, 2H). HRMS (ESI) m/z found 513.2877 (MþHþ); while
4.3. Biology
C
28H37FN4O4 (MþHþ) requires 513.2872.
4.3.1. Assay for ALDH2 activation
4.2.29. N-(cyclopropylmethyl)-5-((3-fluoro-4-methoxybenzyl)
amino)-2-(tetrahydro-2H-pyran-4-yl)benzamide (C20)
The ALDH2 test kit (ab115348) and active human ALDH2 full
length protein (ab87415) were purchased from Abcam (Cambridge,
United Kingdom), and the assay was performed by BioTek Synergy
2 Multi-Mode Microplate Readers (Winooski, Vermont, United
States of America). ALDH2 can oxidize acetaldehydes to acetic acids
with the aid of NADþ while reducing NADþ to NADH. Then, NADH
production was coupled with the dye to form a yellow product and
the absorbance value of the yellow product could be monitored at
an excitation wavelength of 370 nm and an emission wavelength of
White solid; 1H NMR (600 MHz, Chloroform-d)
d 7.11 (d,
J ¼ 8.5 Hz, 1H), 7.08e7.01 (m, 2H), 6.90 (t, J ¼ 8.4 Hz, 1H), 6.63 (dd,
J ¼ 8.6, 2.3 Hz, 1H), 6.56 (d, J ¼ 2.5 Hz, 1H), 5.89 (t, J ¼ 5.5 Hz, 1H),
4.22 (s, 2H), 3.99 (dd, J ¼ 11.3, 4.1 Hz, 2H), 3.86 (s, 3H), 3.47 (td,
J ¼ 11.7, 2.4 Hz, 2H), 3.26 (dd, J ¼ 7.1, 5.5 Hz, 2H), 3.11e3.00 (m, 1H),
1.79e1.65 (m, 4H), 1.06e0.94 (m, 1H), 0.57e0.49 (m, 2H), 0.28e0.20
(m, 2H). HRMS (ESI) m/z found 413.2230 (MþHþ); while
11

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
R.O. Juvonen, D. Petersen, R.A. Deitrich, T.D. Hurley, V. Vasiliou, Aldehyde
dehydrogenase inhibitors: a comprehensive review of the pharmacology,
mechanism of action, substrate specificity, and clinical application, Pharm.
Rev. 64 (2012) 520e539.
450 nm. The assay is conducted using 5 mg/ml, 25 mM and 1 mM as
the starting concentration of ALDH2, acetaldehyde and NADþ under
room temperature. The activity of the ALDH2 can be expressed as
the change in absorbance per minute (SlopeOD450). The maximum
fold activation was calculated as the activity of the ALDH2 with
[3] E.R. Gross, V.O. Zambelli, B.A. Small, J.C. Ferreira, C.H. Chen, D. Mochly-Rosen,
A personalized medicine approach for Asian Americans with the aldehyde
dehydrogenase 2*2 variant, Annu. Rev. Pharmacol. Toxicol. 55 (2015)
107e127.
[4] H.R. Luo, G.S. Wu, A.J. Pakstis, L. Tong, H. Oota, K.K. Kidd, Y.P. Zhang, Origin and
dispersal of atypical aldehyde dehydrogenase ALDH2487Lys, Gene 435 (2009)
96e103.
[5] H. Li, S. Borinskaya, K. Yoshimura, N. Kal’ina, A. Marusin, V.A. Stepanov, Z. Qin,
S. Khaliq, M.Y. Lee, Y. Yang, A. Mohyuddin, D. Gurwitz, S.Q. Mehdi, E. Rogaev,
L. Jin, N.K. Yankovsky, J.R. Kidd, K.K. Kidd, Refined geographic distribution of
the oriental ALDH2*504Lys (nee 487Lys) variant, Ann. Hum. Genet. 73 (2009)
335e345.
[6] J.I. Garaycoechea, G.P. Crossan, F. Langevin, L. Mulderrig, S. Louzada, F. Yang,
G. Guilbaud, N. Park, S. Roerink, S. Nik-Zainal, M.R. Stratton, K.J. Patel, Alcohol
and endogenous aldehydes damage chromosomes and mutate stem cells,
Nature 553 (2018) 171e177.
target compound at 100
treatment (SlopeOD450compound/SlopeOD450min). After corrected
by the activities of Alda-1 (SlopeOD450max as 100% [(Slo-
peOD450compound SlopeOD450min)/(SlopeOD450max Slo-
mM treatment divided by that without
)
-
-
peOD450min) * 100], it was taken as the primary indicator. The EC50
and standard error of selected compounds were determined from
the activity of the ALDH2-dose curves by Graphpad Prism
(GraphPad Software, Inc., San Diego, California, United States of
America), which was taken as the second indicator.
[7] M. Matejcic, M.J. Gunter, P. Ferrari, Alcohol metabolism and oesophageal
4.3.2. Assay for protective effect against ischemic stroke in MCAO
rats
cancer:
a systematic review of the evidence, Carcinogenesis 38 (2017)
859e872.
Animals were maintained in accordance with institutional
guidelines as defined by Institutional Animal Care and Use Com-
mittee for U.S. institutions. Male SD rats were provided by Shanghai
Experimental Animal Center of Chinese Academy of Sciences
(Shanghai, China) were housed in controlled conditions and
received a standard rat chow and tap water ad libitum. Male SD rats
(200e250 g) were randomized into the 3 groups: dimethyl sulf-
[8] C.H. Chen, L. Sun, D. Mochly-Rosen, Mitochondrial aldehyde dehydrogenase
and cardiac diseases, Cardiovasc. Res. 88 (2010) 51e57.
[9] Y. Mizuno, S. Hokimoto, E. Harada, K. Kinoshita, K. Nakagawa, M. Yoshimura,
H. Ogawa, H. Yasue, Variant aldehyde dehydrogenase 2 (ALDH2*2) is a risk
factor for coronary spasm and ST-segment elevation myocardial infarction,
J. Am. Heart Assoc. 5 (2016), e003247.
[10] X.J. Luo, B. Liu, Q.L. Ma, J. Peng, Mitochondrial aldehyde dehydrogenase, a
potential drug target for protection of heart and brain from ischemia/reper-
fusion injury, Curr. Drug Targets 15 (2014) 948e955.
[11] C.H. Chen, G.R. Budas, E.N. Churchill, M.H. Disatnik, T.D. Hurley, D. Mochly-
Rosen, Activation of aldehyde dehydrogenase-2 reduces ischemic damage to
the heart, Science 321 (2008) 1493e1495.
[12] Y.F. Sung, C.C. Lu, J.T. Lee, Y.J. Hung, C.J. Hu, J.S. Jeng, H.Y. Chiou, G.S. Peng,
Homozygous ALDH2*2 is an independent risk factor for ischemic stroke in
Taiwanese men, Stroke 47 (2016) 2174e2179.
[13] Loci associated with ischaemic stroke and its subtypes (SiGN): a genome-wide
association study, Lancet Neurol. 15 (2016) 174e184.
[14] W. Wang, L.L. Lin, J.M. Guo, Y.Q. Cheng, J. Qian, J.L. Mehta, D.F. Su, P. Luan,
A.J. Liu, Heavy ethanol consumption aggravates the ischemic cerebral injury
by inhibiting ALDH2, Int. J. Stroke 10 (2015) 1261e1269.
[15] J.M. Guo, A.J. Liu, P. Zang, W.Z. Dong, L. Ying, W. Wang, P. Xu, X.R. Song, J. Cai,
S.Q. Zhang, J.L. Duan, J.L. Mehta, D.F. Su, ALDH2 protects against stroke by
clearing 4-HNE, Cell Res. 23 (2013) 915e930.
[16] S. Perez-Miller, H. Younus, R. Vanam, C.H. Chen, D. Mochly-Rosen, T.D. Hurley,
Alda-1 is an agonist and chemical chaperone for the common human alde-
hyde dehydrogenase 2 variant, Nat. Struct. Mol. Biol. 17 (2010) 159e164.
[17] H.N. Larson, J. Zhou, Z. Chen, J.S. Stamler, H. Weiner, T.D. Hurley, Structural
and functional consequences of coenzyme binding to the inactive asian
variant of mitochondrial aldehyde dehydrogenase: roles of residues 475 and
487, J. Biol. Chem. 282 (2007) 12940e12950.
[18] H.N. Larson, H. Weiner, T.D. Hurley, Disruption of the coenzyme binding site
and dimer interface revealed in the crystal structure of mitochondrial alde-
hyde dehydrogenase "Asian" variant, J. Biol. Chem. 280 (2005) 30550e30556.
[19] J. Hu, W. Tian, R. Zhou, Y. Zhang, J. Lv, J. Zhu, X. Chen, X. Pan, C. Zheng, Design,
synthesis, and biological evaluation of new ALDH2 activators, J. Saudi Chem.
Soc. 23 (2019) 255e262.
oxide (DMSO) (n ¼ 6); C6 (n ¼ 6, 64
mg in 5 ml). These rats were
injected (i.c.v.) with C6 or saline 5 min before transient middle
cerebral artery occlusion (MCAO) operation.
Anesthesia was induced by 1% isoflurane in 30% oxygen and 70%
nitrous oxide with a face mask. Ischemia was induced by placing an
11-mm silicone-coated 8e0 filament from the left common carotid
artery into the internal carotid artery for 2 h. Rectal temperature
was maintained at 37 ꢁC 0.5 ꢁC with a control unit (Frederick Haer
and Co) during anesthesia, thereafter animals were maintained in
an incubator at 31 ꢁC. Following reperfusion 24 h, rats were rean-
esthetized and brains were immediately removed. Infarct size was
measured by the TTC staining assay. Briefly, the brains were cut into
1e1.5 mm thick slices perpendicular to the brain axis. The slices
were put in 0.8% triphenyltetrazolium chloride (TTC) for 15 min at
37 ꢁC to distinguish the viable brain from the necrotic. After 1 h of
incubation in 4% formaldehyde, slices were scanned and the areas
were determined by planimetry of computer images (Image-Pro
Plus 6.0; Bethesda, MD, USA).
Declaration of competing interest
[20] C. Zhuang, S. Narayanapillai, W. Zhang, Y.Y. Sham, C. Xing, Rapid identification
of Keap1-Nrf2 small-molecule inhibitors through structure-based virtual
screening and hit-based substructure search, J. Med. Chem. 57 (2014)
1121e1126.
[21] T. Zhu, S. Cao, P.C. Su, R. Patel, D. Shah, H.B. Chokshi, R. Szukala, M.E. Johnson,
K.E. Hevener, Hit identification and optimization in virtual screening: practical
recommendations based on a critical literature analysis, J. Med. Chem. 56
(2013) 6560e6572.
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
Acknowledgments
[22] X. Liu, G. Dong, J. Zhang, J. Qi, C. Zheng, Y. Zhou, J. Zhu, C. Sheng, J. Lu, Dis-
covery of novel human acrosin inhibitors by virtual screening, J. Comput.
Aided Mol. Des. 25 (2011) 977e985.
[23] G. Dong, C. Sheng, S. Wang, Z. Miao, J. Yao, W. Zhang, Selection of evodiamine
as a novel topoisomerase I inhibitor by structure-based virtual screening and
hit optimization of evodiamine derivatives as antitumor agents, J. Med. Chem.
53 (2010) 7521e7531.
[24] A.R. Leach, V.J. Gillet, R.A. Lewis, R. Taylor, Three-dimensional pharmacophore
methods in drug discovery, J. Med. Chem. 53 (2010) 539e558.
[25] D. Pathak, N. Chadha, O. Silakari, Identification of non-resistant ROS-1 in-
hibitors using structure based pharmacophore analysis, J. Mol. Graph. Model.
70 (2016) 85e93.
This work was supported by Shanghai Pujiang Program
(2019PJD058) and Shanghai Biological and Pharmaceutical Science
and Technology Support Program (20S11903000).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.ejmech.2020.113119.
[26] J. Meslamani, J. Li, J. Sutter, A. Stevens, H.O. Bertrand, D. Rognan, Protein-
ligand-based pharmacophores: generation and utility assessment in compu-
tational ligand profiling, J. Chem. Inf. Model. 52 (2012) 943e955.
[27] Glide, Schrodinger, LLC: New York, NY, United States of America.
[28] W. Yang, C.H. Chen, D. Mochly-Rosen, Preparation of benzamides and pyr-
idinecarboxamides as mitochondrial aldehyde dehydrogenase-2 modulators,
References
[1] C.H. Chen, J.C. Ferreira, E.R. Gross, D. Mochly-Rosen, Targeting aldehyde de-
hydrogenase 2: new therapeutic opportunities, Physiol. Rev. 94 (2014) 1e34.
[2] V. Koppaka, D.C. Thompson, Y. Chen, M. Ellermann, K.C. Nicolaou,
12

W. Tian, J. Guo, Q. Zhang et al.
European Journal of Medicinal Chemistry 212 (2021) 113119
PCT Int. Appl. (2014). WO 2014160185 A2.
America.
[29] P.S. Campbell, C. Jamieson, I. Simpsonb, A.J.B. Watson, Practical synthesis of
pharmaceutically relevant molecules enriched in sp3 character, Chem. Com-
mun. 54 (2018) 46e49.
[31] GOLD. The Cambridge Crystallographic Data Centre: Cambridge, United
Kingdom.
[32] R. Thomsen, M.H. Christensen, MolDock: a new technique for high-accuracy
molecular docking, J. Med. Chem. 49 (2006) 3315e3321.
[30] Discovery Studio. Accelrys, Inc.: San Diego, California, United States of
13