Discovery of N-hydroxy-3-alkoxybenzamides as direct acid sphingomyelinase inhibitors using a ligand-based pharmacophore model

Article abstract of DOI:10.1016/j.ejmech.2018.03.065

Acid sphingomyelinase (ASM) has been shown to be involved in many physiological processes, emerging to be a promising drug target. In this study, we constructed a ligand-based pharmacophore model of ASM inhibitors and applied this model to optimize the lead compound α-mangostin, a known inhibitor of ASM. 23 compounds were designed and evaluated in vitro for ASM inhibition, of these, 10 compounds were found to be more potent than α-mangostin. This high hit ratio confirmed that the presented model is very effective and practical. The most potent hit, 1c, was found to selectively and competitively inhibit the enzyme and inhibit the generation of ceramide in a dose-dependent manner. Furthermore, 1c showed favorable anti-apoptosis and anti-inflammatory activity. Interactions with key residues and the Zn²⁺ cofactor of 1c were found by docking simulation. These results provide promising leads and important guidance for further development of efficient ASM inhibitors and drug candidates.

Full text of DOI:10.1016/j.ejmech.2018.03.065

Accepted Manuscript
Discovery of N-hydroxy-3-alkoxybenzamides as direct acid sphingomyelinase
inhibitors using a ligand-based pharmacophore model
Kan Yang, Keyi Nong, Qinlan Gu, Jibin Dong, Jinxin Wang
PII:
S0223-5234(18)30308-8
10.1016/j.ejmech.2018.03.065
EJMECH 10330
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 23 January 2018
Revised Date: 20 March 2018
Accepted Date: 21 March 2018
Please cite this article as: K. Yang, K. Nong, Q. Gu, J. Dong, J. Wang, Discovery of N-hydroxy-3-
alkoxybenzamides as direct acid sphingomyelinase inhibitors using a ligand-based pharmacophore
model, European Journal of Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2018.03.065.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Discovery of N-Hydroxy-3-alkoxybenzamides
as Direct Acid Sphingomyelinase Inhibitors
Using a Ligand-Based Pharmacophore Model
Kan Yang1, Keyi Nong1, Qinlan Gu2, Jibin Dong3*, Jinxin Wang1*
1Jiangsu Key Laboratory of Drug Design and Optimization, Department of Medicinal
Chemistry, School of Pharmacy, China Pharmaceutical University, Nanjing 210009,
China
2 Senior Vocational School, China Pharmaceutical University, Nanjing 210009, China
3Department of Biochemistry, School of Pharmacy, Fudan University, Shanghai
201203, China.
Corresponding author E-mail address: jinxinwangcpu@126.com; jbdong@shmu.edu.cn
Abstract
Acid sphingomyelinase (ASM) has been shown to be involved in many physiological
processes, emerging to be a promising drug target. In this study, we constructed a
ligand-based pharmacophore model of ASM inhibitors and applied this model to
optimize the lead compound α-mangostin, a known inhibitor of ASM. 23 compounds
were designed and evaluated in vitro for ASM inhibition, of these, 10 compounds
were found to be more potent than α-mangostin. This high hit ratio confirmed that the
presented model is very effective and practical. The most potent hit, 1c, was found to
selectively and competitively inhibit the enzyme and inhibit the generation of
ceramide in a dose-dependent manner. Furthermore, 1c showed favorable anti-
apoptosis and anti-inflammatory activity. Interactions with key residues and the Zn2+
cofactor of 1c were found by docking simulation. These results provide promising
leads and important guidance for further development of efficient ASM inhibitors and
drug candidates.
1

ACCEPTED MANUSCRIPT
Keywords: acid sphingomyelinase; inhibitors; pharmacophore model; anti-apoptosis;
anti-inflammatory
Introduction
Acid sphingomyelinase (ASM) is a zinc-dependent phosphodiesterase that catalyzes
the conversion of sphingomyelin (SM) into ceramide (Cers) and phosphocholine at
the cytomembrane or intralysosomal membranes. Cers, which serves as a precursor
for the synthesis of complex sphingolipids and as a versatile signalling lipid in the
regulation of fundamental processes of cell proliferation and death, has been
characterized in recent genomic and lipidomic research as the central signaling and
metabolic relay among sphingolipids [1-3]. Furthermore, the extreme hydrophobicity
of Cers leads to the formation of Cers-enriched membrane microdomains that play an
essential role in transmembrane signaling [4, 5]. The excessively activated ASM-Cers
system can lead to inflammation [6-8], apoptosis [9-11], insulin resistance [3], and
mitochondrial dysfunction [12] under pathologic conditions. Accordingly, in patients
with cardiovascular disease, diabetes or major depression, the activity of ASM was
found to be significantly higher than in normal individuals [13-15]. Higher Cers level
was also found in cystic fibrosis (CF) patients due to the higher ratio of ASM / acid
ceramidase [16]. Functional inhibitors of ASM such as amitriptyline and imipramine
were found to prevent or slow the progression of such pathologies in various animal
studies [17-19]. Meanwhile, a direct inhibitor ARC39 was recently reported to
attenuate murine acute lung injury [20]. On the basis of these significant results, ASM
has received much attention as an attractive therapeutic target for treating Cers-
associated diseases, especially lung diseases [1], metabolic disorders [3, 21], and
central nervous system disease [22, 23]. In several phase-II trials for CF, amitriptyline
showed good therapeutic effects by inhibiting ASM, which reduced Cers
concentrations in lungs [24-26].
In most reported biological studies, the activity of ASM was controlled by genetic
deletion or indirect inhibition by certain antidepressants that lead to its degradation.
At present, direct inhibitors were needed to further examine and verify its
pharmacological potential as a drug target. However, only a limited number of direct
inhibitors have been reported, including analogues of SM [27], bisphosphonates [28],
phosphatidylinositol-3,5-bisphosphates derivatives [29-31], benzopyrone derivatives
[32], xanthone derivatives [33, 34]. The usefulness of these molecules has been
2

ACCEPTED MANUSCRIPT
limited by their lability towards phospholipases and their difficulty in penetrating
biological membranes [35]. Therefore, there is a great need to identify structurally
distinct and more efficient small molecules as direct inhibitors.
There was no available 3D structure of ASM in our early inhibitors studies. We had
constructed an initial ligand-based pharmacophore model to provide insights into
design and development of ASM inhibitors. Directed by the model, we focused our
investigations on synthesizing derivatives of α-mangostin, a known competitive
inhibitor of ASM (IC50 = 14.1 µM) derived from pericarps of the mangosteen fruit
[36]. However, moderate deviation was observed between the predicted and
experimental activities of the synthesized derivatives. This discrepancy may be due to
the four features of the original model were less and too closely ranged: one hydrogen
bond acceptor (HBA), one hydrogen bond donor (HBD), and two hydrophobic groups
(HY), indicating the model was not ideal for further application in the development of
ASM inhibitors [37].
With newly obtained inhibitors in hand [33]. herein we extended the training set
and optimized the pharmacophore model, taking advantage of more accurate features:
one HBA, one HBD, two HYs, one negatively ionizable group (NI), and two excluded
volumes (EV). The model was used to analyze the pharmacophores of α-mangostin.
As a result, a novel class of ASM direct inhibitors were successfully discovered
through pharmacophore-guided and experience-based structural simplification and
optimization of α-mangostin, which involved an interesting transformation from the
natural structure to small-molecule inhibitors. Structure-activity relationship analysis
provided a detailed insight into structural requirements for ASM activity of the new
scaffold. In order to explore the impact on Cers level when ASM was directly
inhibited, the most active compound 1c was selected to test for inhibition of Cers
generation. We also favorably included investigations of 1c as a promising anti-
apoptosis and anti-inflammatory agent by detection of cell viability and the inhibition
of the related inflammatory factor, including IL-6 and IL-8.
It is encouraging that three research groups have reported resolution of the crystal
structure of ASM [38-40]. Thus, the potential binding modes between the compound
and the protein were intensively analyzed by docking simulations, which will provide
more beneficial information for further design of inhibitors.
Results and discussion
3

ACCEPTED MANUSCRIPT
Generation and validation of pharmacophore models.
The ligand-based 3D pharmacophore model was generated using 21 training set
molecules (Table S1) in the 3D QSAR Pharmacophore Generation protocol of
Discovery Studio 3.0 (DS 3.0). The protocol uses the Catalyst HypoGen algorithm
and generated ten pharmacophore hypotheses, among which Hypo 1 was identified as
the best pharmacophore model characterized based on the rank values (Table S2). The
expected pharmacophore model showed a total cost (97.039) close to the fixed cost
(84.339) and significant difference (91.655) between null and total cost. Moreover,
Hypo 1 showed the lowest root mean square deviation (RMSD, 1.066) and best
correlation coefficient (0.948). Besides this, Hypo-1 estimated the highest fit value of
8.33 for the most active compounds and 4.42 for the least active compounds (Table
S4). All these data suggest the model is capable of predicting the activity of ASM
inhibitors. This model consists of one HBA, one HBD, two HYs, one NI, and two
EVs (Figure 1A).
Validation of the pharmacophore model was carried out by using the training and
test set compounds (Table S3). Experimental and estimated activities data for the test
set compounds are shown in Figure S1 and Table S5. As expected, Hypo-1 explained
well the activity variation among these compounds with considerably low RMS error.
An error of less than 10 means that the difference in the estimated IC50 value and the
experimental IC50 value is less than one order of magnitude. In addition, there was a
good correlation coefficient (R2 = 0.8557) for the test set. The Fischer’s
randomization test at 95% confidence level was used to evaluate the statistical
significance of the model by scrambling the activities of compounds in the training
set. Results confirm that the cost value of Hypo 1 is lower than the 19 randomly
generated trials and there is a 95% chance that the Hypo1 represents a true correlation
for the training set compounds (Figure S1). Thus, the Hypo 1 model is superior and
was not generated by chance.
Mapping of the most active compound S1 in the training set (experimental activity:
IC50 = 0.02 µM and predicted activity: IC50 = 0.016 µM) revealed that all the features
were mapped to the pharmacophore model with least displacement from the centroid.
As shown in Figure 1B, the only amino group of S1 mapped with the HBD of Hypo-1
and the hydrophobic chain mapped to the two HY groups. One oxygen atom from the
PO3H2 group mapped well with the HBA feature, while the other PO3H2 group
4

ACCEPTED MANUSCRIPT
mapped with the NI group. However, the less active compound S21 (experimental
activity: IC50 = 196.6 µM and predicted activity: IC50 = 132.2 µM) only mapped to the
two HY groups (Figure 1C). It is noteworthy that ASM is fully or partially dependent
on Zn2+ for enzymatic activity [41]. The bisphosphonates in the training set with a
hydroxy or amino group were reported to form stable tridentate complexes with Zn2+
ions at the catalytic center of ASM [28]. Meanwhile, with respect to the mapping of
compound S1 with the pharmacophore model, we hypothesized that the NI together
with HBA and HBD function as a Zn2+ binding site.
Figure 1. Hypo1 and its mapping to the most active and least active compounds from
training set. (A) The topological features of Hypo1: green: HBA, magenta: HBD,
cyan: HY, blue: NI, gray: EV. (B) The most potent compound S1 mapped with Hypo
1. (C) The less active compound S21 mapped with Hypo 1.
Design of novel ASM inhibitors
We mapped α-mangostin to Hypo 1 and found that only part of the molecule was
matched, as shown in the red triangle in Figure 2. The carbonyl and hydroxyl at C-1
were mapped to the HBA and HBD, respectively, with little deviation. The methoxy
at C-7 and the prenyl group at C-8 were mapped with the two HY groups while the NI
group was not included. A previously study had also confirmed that xanthone with a
prenyl group only at the C-8 residue maintained comparable activity to that of α-
mangostin [34]. Thus, we deleted the groups outside the triangle in the structure of α-
mangostin (Figure 2B). Directed by compound S1 in the training set, the prenyl group
was replaced by an alkyl group and removed to the C-7 position to change the
molecule into a chain shape, termed the hydrophobic tail. The carbonyl and hydroxyl
were replaced by a hydroxamic acid, which shares an approximate relative distance
between the two groups. This part was designated as the hydrophilic head and served
5

ACCEPTED MANUSCRIPT
as the H-bond donor and acceptor. Since the designed hydroxamic acid is one of the
most prominent zinc binding groups (ZBGs) used in the design of metallohydrolase
inhibitors, we did not introduce another negatively ionizable group, which has been
discussed as a ZBG above.
This resulted in the design of the first series of inhibitors (1a-1f, Table 1). The
initial compound 1a contained a tail of C4 alkyl group, sharing a similar length to the
prenyl group of α-mangostin. As shown in Figure 2D, the benzene and hydrophobic
tail of compound 1a were mapped to the two HY features of Hypo 1, while its OH
group mapped to HBD. The HBA and NI were missed. As most ASM inhibitors share
a very long hydrophobic chain, we extended the length of the alkyl moiety to make it
more suitable to the pharmacophore (1b-1f). As a result, compound 1b with a C8
hydrophobic chain exhibited a better mapping result than 1a. The benzene and the 3rd
carbon atom at the hydrophobic tail mapped with the two HY groups, while the
hydroxamic acid head mapped well with the HBA and HBD features (Figure 2E).
Figure 2. Pharmacophore based design of novel ASM inhibitors. (A) Mapping of
Hypo 1 with α-mangostin. (B) The structure of α-mangostin. (C) General structure of
the designed inhibitors. (D) Mapping of Hypo 1 with compound 1a. (E) Mapping of
Hypo 1 with compound 1b. The color key is the same as defined in Figure 1.
Various modifications of series 1 were undertaken to develop a deeper
understanding of structure−activity relationships (Table 2). We suspected that a
flexible hydrophilic head carrying a carbonyl and a hydroxyl may be more suitable for
6

ACCEPTED MANUSCRIPT
the HBA and HBD features in Hypo 1. To test this hypothesis, compounds 2a-2d
were prepared with good pharmacophore fit values from 5.34 to 6.87. As the methoxy
group at C-4 of compound 2 seemed redundant in the mapping result of Hypo 1, we
prepared compounds 2h-2j, in which the methoxy was replaced by hydrogen. We next
turned our attention to the hydrophobic tail. Introduction of more rigid groups to
restrict the conformations may improve the activity. Thus, compounds 2k-2o were
designed. Mapping results showed that a benzyl substituent at C-3 of the molecule
made it more suitable for the pharmacophore, such as 2k and 2l (Figure S2). The
carbonyl and hydroxyl from the hydrophilic head mapped with the HBA and HBD,
respectively, while the two benzene rings mapped well with the two HY groups.
Synthesis of title compounds
We began the synthesis from two commercially available materials 3 and 6. As
depicted in Scheme 1, compounds 4 were obtained by substitution with suitable alkyl
bromides in the presence of K2CO3 and KI, while compounds 7 and 9 were selectively
prepared from substrate 6 since the benzoate activated the hydroxyl at C-4.
Compound 9 was further substituted with 1-bromodecane to afford 10. Products 4, 7,
and 10 were allowed to hydrolyze to give the key intermediate 5, 8, and 11,
respectively. In the last step, the carboxylic acid intermediates were activated by
thionyl chloride and were converted to the hydroxamic acid (1a-1d, 12, 1g) by adding
them to a solution of hydroxylamine base. The solution was freshly prepared from
hydroxylamine hydrochloride with NaOH in tetrahydrofuran and water. Compound 1e
were prepared by debenzylation of 13 catalyzed by Pd/C under a hydrogen
atmosphere.
7

ACCEPTED MANUSCRIPT
o
Scheme 1. Reagents and conditions:(a) RBr, K2CO3, KI, acetone, 60 C; (b) NaOH,
o
MeOH, H2O, 100 C; (c) i: SOCl2, DMF, CH2Cl2, ii: hydroxylamine hydrochloride,
NaOH, THF, H2O, rt; (d) BnBr, K2CO3, acetone, 60 oC; (e) Pd/H2, MeOH, rt.
Compounds 2h-2p in the second series were prepared according to the similar
procedure described above, as illustrated in Scheme 2. The key intermediates 5, 14,
and 17 were converted to various amides in the presence of 4-dimethylaminopyridine
and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride in anhydrous
DMF, yielding 2a-2g, and 2q (Scheme 2).
o
Scheme 2. Reagents and conditions: (a) RBr, K2CO3, KI, acetone, 60 C; (b) NaOH,
o
MeOH, H2O, 100 C; (c) EDCI, DMAP, CH2Cl2, rt, or i: SOCl2, DMF, CH2Cl2, ii:
hydroxylamine hydrochloride, NaOH, THF, H2O, rt.
Biological screening and structure−activity relationship studies of the ASM
inhibitors
Enzymatic assays of the target compounds were performed in vitro to confirm the
results predicted by the pharmacophore model (Table 1 and Table 2). Compound 1a
with two features missing did not showed any inhibitory activity. However, when the
lipid tail was lengthened, compounds 1b-1d showed more potent activity than the lead
compound, α-mangostin. Compound 1c was found most active (0.48 µM) among
these molecules, about 50-fold more potent than α-mangostin (24.74 µM), indicating
that the hydrophobic tail with a length of C10 was most feasible. Compound 1e was
less potent than compound 1c, indicating that the free hydroxyl at the C-4 position
had a negative effect on the activity. A certain angle between the tail and head is
needed in the molecule as we found compound 1f with the tail at C-4 position showed
loss of activity. Corresponding carboxylic intermediates of the title compounds were
also tested for their inhibitory activity on ASM. However, none of these intermediates
was active, indicating that the binding mode of Zn2+ was not due to the effects of
anion. On the whole, 4 out of the 6 designed compounds showed more potent activity
that the lead compound, showing a high hit ratio. Taking into account the mapping
8

ACCEPTED MANUSCRIPT
results, we can conclude that at least four features are essential to produce the most
active compounds.
Table 1. The structures of series 1, pharmacophore mapping values and observed
activities.
No.
R1
R2
Predicted IC50 (µM) Fit value Observed IC50 (µM)
C4H9
C8H17
C10H21
C12H25
C10H21
H
Me
Me
Me
Me
H
18.69
13.89
13.58
13.56
0.91
5.34
5.41
5.41
5.42
6.59
5.27
5.25
-
1a
1b
16.09 ± 1.23
0.48 ± 0.10
18.55 ± 2.34
17.43 ± 0.98
-
1c
1d
1e
C10H21
-
19.1
1f
α-Ma
-
20.06
24.74 ± 6.01
aα-M: α-mangostin. Data are expressed as means ± SD of 3 experiments.
As to series 2, none of the compounds 2a-2d were more effective inhibitors of
ASM. Additionally, compounds 2f and 2g lacking a H-bond donor and 2e with
esterification of the hydroxamic acid were synthesized for comparison. As expected,
these changes resulted in disappearance of activity of such compounds (Table 2).
From the above result, we rationalized that the hydroxamic acid was an essential
group for the inhibitory activity. This outcome may be attributed to its strong
chelation of the Zn2+ in the catalytic center of ASM.
Compound 2i (1.42 µM) was found to be 17-fold more potent than the lead
compound, although it showed less activity than 1c. This result indicated the C-4
methoxy group may be an important factor for the activity. Similarly, compound 2k
(10.49 µM) was found to be 2-fold more potent than α-mangostin, while 2l showed a
great decrease of activity (60.42 µM). Although the two compounds shared a similar
fit value, they showed differences in the pharmacophore mapping mode and the
observed activity. We suspected that restriction by the C-4 methoxy group of 2k
reduced the entropic penalty for achieving the desired transition state and led it to
maintain the preferential conformation to inhibit the enzyme. The same phenomenon
can also be found between compound 2o and 2p. However, this observation needs
9

ACCEPTED MANUSCRIPT
further verification since compounds 2h and 2g showed comparable activity to that of
compounds 1b and 1d, respectively.
In addition, the hydrophobic benzyl tail of 2k shared a similar length (4.971 Å,
predicted by Chem 3D) to that of the inactive compound 1a (4.927 Å), demonstrating
that a rigid tail was much more effective than the flexible alkyl. Compound 2k was
less potent than 1c and 2i. A probable reason is the benzyl tail was much shorter than
the C10 length of 1c. When the benzyl was substituted by either nitrile or methoxy
groups, compounds 2m and 2n showed a decrease of activity. In contrast, a
hydrophobic bromine substituent increased the inhibitory activity, like compound 2o
and 2p, as compared with 2k and 2l, respectively. This result may be caused by the N
or O atom affecting the hydrophobicity of the tail. The absence of activity in
compound 2q came as no surprise, since it shared a much less effective head like 2a
or 2c. From the above results, we can conclude that a hydrophobic tail with
appropriate length and rigidity was very favorable for the inhibitory activity.
Table 2. Structure, pharmacophore mapping and observed activities of series 2
compounds.
Predicted
Fit
Observed
No.
R1
R2
R3
IC50 (µM)
value
IC50 (µM)
C10H21
C10H21
NH(CH2)2OH
OMe
0.48
16.19
14.60
15.17
-
6.87
5.34
5.4
71.00 ± 9.34
2a
2b
2c
2d
2e
2f
NHCH2CH(OH)CH3 OMe
NH(CH2)2OH OMe
NHCH2CH(OH)CH3 OMe
85.20 ± 9.49
C8H17
-
C8H17
5.37
-
-
C10H21
NHOMe
2-aminopyridine
2-aminopyridine
NHOH
OMe
OMe
OMe
H
-
C10H21
-
-
-
C8H17
-
-
-
2g
2h
2i
C8H17
0.69
0.63
14.76
0.57
0.77
35.87
26.15
6.71
6.75
5.38
6.80
6.66
5.00
5.13
17.42 ± 1.45
1.42 ± 0.112
13.06 ± 2.32
10.49 ± 0.87
60.42 ± 4.67
52.87 ± 4.60
42.68 ± 5.26
C10H21
NHOH
H
C12H25
NHOH
H
2j
Benzyl
NHOH
OMe
H
2k
2l
Benzyl
NHOH
4-Cyanobenzyl
4-methxylbenzyl
NHOH
H
2m
2n
NHOH
H
10

ACCEPTED MANUSCRIPT
4-Bromobenzyl
4-Bromobenzyl
Benzyl
NHOH
NHOH
OMe
H
1.38
1.88
6.41
6.28
5.29
7.78 ± 0.81
19.42 ± 1.51
-
2o
2p
2q
NH(CH2)2OH
OMe
17.90
Data are expressed as means ± SD of 3 experiments.
Compound 1c and 2i were also tested their inhibitory potency towards neutral
sphingomyelinase (NSM). As a result, none of the two compounds inhibited NSM at
100 µM (Figure S5), suggesting very high selectivity of 1c and 2i. In addition, we
aimed to discover ASM inhibitors that functioned in direct manners, therefore the
Lineweaver-Burk plots were used for further confirmation about competitive/non-
competitive inhibition of ASM by compound 1c. As shown in Figure 3, compound 1c
displayed competitive inhibition against ASM.
Figure 3. Lineweaver-Burk plots of acid sphingomyelinase activity over a range of
substrate (NBD-SM) concentrations in the absence and presence of 1c at
concentrations 0.5 µM, 1.0 µM, and 5 µM (ASM was almost completely inhibited at 5
µM and the data are beyond this figure). The data are representative of triplicate
experiments.
Ceramide generation inhibition assay
In order to explore the intracellular potency of the inhibitor to block Cers
production, compound 1c was selected to measure inhibition of the generation of
Cers. The NIH3T3 cell line was cultivated and treated with different concentrations of
1c, then the amount of Cers and SM was evaluated by the LC-MS-MS technology. As
shown in Figure 4, upon treatment with 1c (5 µM, 10 µM, and 20 µM), the total
amount of Cers was significantly decreased in a dose dependent manner (26.1 %, 37.0
%, 42.6 %, respectively); while the total amount of SM was increased (31.9 %, 36.4
%, 37.5 %, respectively). Besides, compound 1c showed much better IC50 value (0.48
µM) in the NBD assay. One possible reason is that the NBD assay was performed
directly on proteins, while the ceramide determination was performed on live cells.
11

ACCEPTED MANUSCRIPT
Membrane permeability of 1c and the complex metabolic environment of cells may
affect the activity. Moreover, because of the introduction of NBD group, the NBD-
sphingomyelin might have a little lower affinity compared to natural sphingomyelin.
Since 1c is a competitive inhibitor, it might exhibit better IC50 data in the NBD assay.
This result confirmed that compound 1c is an effective ASM inhibitor in the cell
environment.
Figure 4. Compound 1c inhibited the generation of Cers. Human NIH3T3 cells were
treated with 1c at the doses of 0 µM (Control), 5 µM, 10 µM, and 20 µM. (A) The
amount of total Cers. (B) The amount of total sphingomyelin (SM). Values are mean
± SD from three independent experiments, Student’s t-test **P < 0.01 compared to
control group.
Characterization of compound 1c as potent anti-apoptosis and anti-inflammation
agents
Significantly increased levels of Cers in pathological tissues has the ability to
induce apoptosis. Thus, compound 1c was tested in an anti-apoptosis study to
determine its protective effects on cell. As shown in Figure 5, treatment with
compound 1c effectively increased cell viability even at the low concentration of 10
µM. The apoptosis prevention function of 1c at a lower concentration was more
effective than the above ceramide inhibition activity. It is reasoned that the ceramide
inhibition assay was performed on normal cells, in which ASM is not highly
expressed. While UV induced great increase of the activity of ASM in the apoptosis
assay. Thus, the effect of 1c on apoptotic cells may be more obvious.
However, there was no concentration dependency. Toxicity test of 1c on NIH3T3
cells by MTT assay found an IC50 value of 51.38 ± 3.94 µM (Table S1). This result
may rationalize the phenomenon that higher concentration of 1c didn’t show higher
anti-apoptosis activity.
12

ACCEPTED MANUSCRIPT
Figure 5. Survival of human NIH3T3 cells upon exposure to UV in the presence of
1c. Cells were treated with 1c (from 0 µM to 60 µM) for 20 h, followed by irradiation
with UV light (315 nm) for 5 min. The cells were then maintained for another 24 h.
The apoptosis status of the cells was measured by flow cytometry. (A) Flow
cytometry data (full data are shown in supplementary material). (B) Cell viability as
determined by percentage of Q3 region. Values are expressed as mean ± SD from
three independent experiments, ****p<0.0001 compared to control group, # # #
p<0.001 compared to 0 µM group, # # # # p<0.0001 compared to 0 µM group.
Next, whether ASM inhibitor 1c can effectively prevent the increase of UV-
induced ceramide production in the apoptosis assay was further explored.
Amitriptyline was selected as a positive control since it has been shown to functional
inhibit ASM and decrease cellular Cers content [19]. Initially, one out of the four
repeated experiments was used to determine the viability of cells. Results confirmed
that both amitriptyline and 1c exhibited dose dependent cytoprotection (Figure S7).
Then changes of the ceramide content of the rest samples were investigated by LC-
MS-MS technology. Figure 6 obviously revealed the statistical results that UV
treatment induced a notable increase of Cers (about 200 %), a phenomenon that was
consistent with the previous reports [42, 43]. It is noteworthy that amitriptyline and 1c
could dose dependently inhibit this change. Furthermore, 1c with a concentration of
50 µM restored Cers to a normal level (105.777 pmol/mg), comparing to the control
group (107.686 pmol/mg). This result further validated that the mechanism by which
compound 1c protected cells from apoptosis was to reduce the content of ceramide by
inhibiting ASM.
13

ACCEPTED MANUSCRIPT
Figure 6. UV induced an increase of Cers, while compound 1c could block this
phenomenon. Ami, amitriptyline. Values are expressed as mean ± SD from three
independent experiments, **p<0.01 compared to control group, # # p<0.01 compared
to UV group, # p<0.05 compared to UV group.
Additionally, ASM-Cers-related cell inflammation has been found to be involved in
several inflammatory diseases, such as atherosclerosis [44], cystic fibrosis [45], and
other lung diseases [1]. Here, compound 1c was selected to investigate the inhibitory
effects of IL-6, IL-8 production in LPS-induced NIH3T3 cells. As shown in Figure 7,
LPS induced a robust increase in transcription of IL-6 and IL-8. Treatment with
compound 1c significantly reduced the upregulated cytokine in a dose-dependent
manner, although it did not show a significant effect at the concentration of 2.5 µM in
the IL-8 assay. These data suggest that direct inhibition of ASM can reduce the
pathological inflammatory response.
Figure 7. Acid sphingomyelinase inhibitor 1c reduces inflammation in cells. Normal
NIH3T3 cells were induced by LPS and then treated with different concentrations of
1c. Cytokines were evaluated by ELISA. **p<0.01 compared to control group,
****p<0.0001 compared to control group, # # #p<0.001 compared to LPS group, # #
# #p<0.0001 compared to LPS group.
Molecular docking
The crystal structure (PDB code: 5fi9) of ASM was resolved recently and
investigations showed that its catalytic domain adopts a calcineurin-like fold with two
zinc ions. Furthermore, two key residues His 280 and His 317 are essential for Cers to
14

ACCEPTED MANUSCRIPT
leave [38]. Comparation of the pharmacophore and the X-ray structure revealed some
consistent features. The catalytic center of ASM, which is consist of ions and polar
amino acids such as His and Asp, may be the potential corresponding site of the HBD,
NI, and HBA features of the pharmacophore. A hydrophobic shallow groove of the
protein also satisfies the HY groups of the pharmacophore (Figure S8). To validate
this assumption and illustrate the potential binding modes between the inhibitors and
the enzyme, docking simulations of compounds 1a, 1c, and 2e were performed in the
binding pocket of ASM using the CCDC Gold Suite 5.3.0 software.
Compound 1c (green) exhibited a similar binding mode with 1-aminodecylidene
bis-phosphonic acid (AbPA) (blue), a ligand co-crystallized with ASM (Figure 8A).
The hydroxamic acid group was inserted into the innermost binding sites of the active
pocket while the lipid tail stretched into a shallow groove. The carbonyl group at the
head formed one hydrogen bond with residue Asp 276. This group was found to
coordinate with one zinc. The hydroxyl at the head formed three hydrogen bonds with
the residues His 206 and His 280 and was found to coordinate with the second zinc.
Meanwhile, Pi conjugation was found between the benzene ring of 1c and His 455
(Figure 8B).
It is worth noting that shortening the tail (1a) or esterification of the head (2e) led to
a remarkable loss of the potency. This finding was also supported by the docking
studies. The head group of 1a matched the catalytic center, but its tail was too short to
fill the groove. In contrast, 2e maintained a suitable tail, while its head deviated far
from the catalytic center (Figure S9).
In conclusion, the docking results clearly indicated that the hydroxamic acid group
was critical for the inhibitors to bind to the protein and to effect inhibitory potency,
which was consistent with the SAR analyses. These results provide great assistance
for the rational design of new ASM inhibitors that will better exploit the shape of the
active site.
15

ACCEPTED MANUSCRIPT
Figure 8. The binding modes of 1c with ASM. (A) Superposition of 1c and the
original bisphosphonate ligand in the crystal. (B) Binding mode of the active
inhibitors. Green: 1c; blue: AbPA. Hydrogen bonds are shown as green lines, metal
contact in yellow, and Pi contact in blue.
Conclusions
In the present study, we generated an excellent ligand-based pharmacophore model
using a series of known ASM inhibitors. The model showed good statistical
parameters in its validation process and was used to perform the simplification and
optimization of the lead compound, α-mangostin. Enzymatic assays found that four
out of the six compounds in series 1 showed greater potency than α-mangostin,
indicating the pharmacophore model was highly effective. Further mapping-guided
optimization together with empirically-based structural modifications provided a
detailed structure-activity relationship investigation, which identified compound 1c as
the most active and selective ASM inhibitor. Treatment by 1c on the NIH3T3 cell line
efficiently decreased the total amount of Cers generation in a dose-dependent manner.
Significantly, 1c exhibited very excellent in vitro results against apoptosis and
inflammation, two events that are typical in the pathological process of Cers-
associated diseases. MTT assay indicated a much higher cytotoxic IC50 (51.38 µM)
than its inhibitory activity (0.48 µM). In the apoptosis assay, Cers was found
significant increased under irradiation of UV, while amitriptyline or 1c treatment
16

ACCEPTED MANUSCRIPT
could attenuate this change and protected cells from UV induced apoptosis. Docking
simulations directly provided detailed information about binding between 1c and
ASM. Taken together, the pharmacophore model and the SAR analyses make new
and significant contributions toward understanding ASM inhibition. The information
gained here offers practical guidance in the search for novel lead compounds that will
be more potent ASM inhibitors. Furthermore, the promising compounds described
here may serve as valuable pharmacological tools, which can be applied to verify the
role of ASM as a potential drug target in various ceramide-associated diseases.
Experimental Section
Pharmacophore modelling
Biological activity data represented as IC50 (µM) were obtained from literature and
the PubChem BioAssay data in NCBI. Chemical structures of selected ASM
inhibitors were separated into training and test sets on the basis of activity and
structural diversity (Supplementary Table S1 and S4, respectively). The training set
compounds were used to generate pharmacophore models using the 3D QSAR
Pharmacophore Generation protocol of DS 3.0. Four features were selected as the
most found features viz. hydrogen bond acceptor (HBA), hydrogen bond donor
(HBD), hydrophobic (HY) and negatively ionizable (NI). The conformation
generation was chosen as “best”, which ensures the best coverage of conformational
space. The uncertainty value was defined as 2, representing the ratio of the
uncertainty range of the measured activity against the actual activity for each
compound. Other settings were kept the default. In the protocol run, ten
pharmacophore hypotheses were generated on the basis of a high correlation
coefficient (r), low total cost, and low RMSD. Further quality assessment and
validation of the pharmacophore model was done using the test set compounds in DS
3.0.
Molecular docking
The docking study was performed with CCDC GOLD 5.3.0 software (Cambridge,
UK). The co-crystal structures of ASM with an inhibitor (AbPA) (PDB code: 5fi9)
was chosen as the receptor.34 Small molecules were drawn in ChemDraw and the
initial lowest energy conformations were calculated in Chem3D. Protein was
preliminary prepared in DS 3.0 with adding hydrogens, deleting glycosyls and water
and conducting the “Clean Protein” protocol. The prepared protein was then imported
17

ACCEPTED MANUSCRIPT
into GOLD and edited with extracting ligand. The binding site was defined at the
position of the original included ligand AbPA. Goldscore_P450_csd was chosen as
the configuration template and GoldScore was chosen as scoring function. Other
parameters kept default. Docking poses were loaded into Gold software for a
visualization of ligand-receptor interactions and analysis.
Chemistry
General: Starting materials and reagents were obtained from commercially available
sources with high–grade purity. Thin-layer chromatography (TLC) was performed on
Merck silica gel plates (Merck silica gel plates GF254, Germany) and visualized
under UV light (254 nm). Column chromatography was performed with Merck
Kieselgel 60 (200-300 mesh, Germany). Melting points were determined on a MEL-
TEMP Ⅱ apparatus (Laboratory Devices, MA, USA). 1H NMR and 13C NMR spectra
were recorded on a Bruker AV-300 MHz instrument (Bruker Biospin AG,
Switzerland) in CDCl3 or DMSO-d6 with TMS as the internal standard. MS were
obtained on Waters Q-Tof microTM (Waters, MA, USA) with the ESI (+, -)
protonation interface. High–resolution mass spectra (HRMS) were recorded on Q-Tof
Premier hybrid mass spectrometer with electron spray ionization (ESI). HPLC
analysis was performed on a Shimadzu SPD-20A/20AV HPLC system (Shimadzu,
Tokyo, Japan). The column used was a C18 column (5 µ micron, 250 × 4.60 mm) at a
temperature of 30 °C and a flow rate of 1.0 mL/min. The purity of title compounds
was higher than 95%, which was assessed at 254 nM under a condition of methanol
/water (0.1% AcOH) (a ratio of 65:35 for 1a, 1e, 2k-2q; a ratio of 90:10 for 1b-1d, 1f,
2a-2j). NBD-Cers were determined at 525 nM under a condition of methanol (0.1%
TFA) /water (89:11). NBD–SM was purchased from Avanti (AL, USA). All title
compounds were examined for PAINS and passed. Anhydrous reactions were carried
out in dried glassware under a nitrogen atmosphere. The boiling range for petroleum
ether was 60–90 °C. Key intermediates were synthesized followed the common used
methods in literatures.
General procedure for the Preparation of compounds 1a-1d, 13, 1f and 2h-2p
A mixture of the substrate (100 mg), dimethyl formamide (1 drop) and thionyl
chloride (3 equiv.) in anhydrous dichloromethane (10 mL) was heated to reflux for 3
h and concentrated to remove the thionyl chloride and solvent. The concentrate was
then redissolved by anhydrous tetrahydrofuran (10 mL) and slowly added into a
18

ACCEPTED MANUSCRIPT
solution of hydroxylamine, which was freshly prepared by dissolving
hydroxylammonium chloride (5 equiv.) and NaOH (5 equiv.) into tetrahydrofuran (4
mL) and water (0.5 mL). After the drop add, the mixture was stirred at room
temperature for 5 min, concentrated in vacuum and purified by chromatography on
silica gel to afford the title product.
3-Butoxy-N-hydroxy-4-methoxybenzamide (1a)
White solid (72 mg, 68.4 %), M.P. 149-150 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.06 (s, 2H), , 8.90 (s, 2H), 7.38-7.35 (m, 2H), 7.01-6.98 (m, 1H), 4.00 (t, J = 6.30
Hz, 2H), 3.80 (s, 3H), 1.75-1.66 (m, 2H), 1.50-1.43 (m, 2H), 0.96 (t, J = 7.20 Hz,
3H); 13C NMR (75 MHz, DMSO-d6): δ 164.42, 151.84, 148.12, 125.36, 120.50,
111.78, 111.69, 68.39, 56.07, 31.24, 19.20, 14.14; HRMS (ESI-TOF): calcd for
C12H18NO4 [M + H]+ 240.1230, found 240.1233.
N-Hydroxy-4-methoxy-3-(octyloxy)benzamide (1b)
1
White solid (63 mg, 60 %), M.P. 168-171 °C, H-NMR (300 MHz, DMSO-d6): δ
11.06 (s, 1H), 8.89 (s, 1H), 7.38-7.34 (m, 2H), 7.01-6.98 (m, 1H), 3.99 (t, J = 6.60 Hz,
2H), 3.79 (s, 3H), 1.74-1.67 (m, 2H), 1.41-1.39 (m, 2H), 1.27 (m, 8H), 0.87-0.84 (m,
3H); 13C NMR (75 MHz, DMSO-d6): δ 164.41, 151.85, 148.12, 125.37, 120.49,
111.81, 111.71, 68.72, 56.08, 31.70, 29.17, 29.15, 29.13, 25.97, 22.54, 14.41; HRMS
(ESI-TOF): calcd for C16H26NO4 [M + H]+ 296.1856, found 296.1860.
3-(Decyloxy)-N-hydroxy-4-methoxybenzamide (1c)
White solid (71 mg, 67.7 %), M.P. 118-120 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.04 (s, 1H)，8.87 (s, 1H), 7.38-7.36 (m, 2H), 7.01-6.98 (m, 1H), 3.99 (t, J = 6.30
Hz, 2H), 3.80 (s, 3H), 1.74-1.70 (m, 2H), 1.41-1.39 (m, 2H), 1.26 (m, 12H), 0.86-0.84
13
(m, 3H); C NMR (75 MHz, DMSO-d6): δ 164.44, 151.87, 148.14, 125.38, 120.50,
111.86, 111.72, 68.94, 56.08, 31.76, 29.48, 29.42, 29.22, 29.15, 25.97, 22.54, 14.39;
HRMS (ESI-TOF): calcd for C18H30NO4 [M + H]+ 324.2169, found 324.2174.
3-(Dodecyloxy)-N-hydroxy-4-methoxybenzamide (1d)
White solid (78 mg, 74.7 %), M.P. 128-129 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.05 (s, 1H), 8.88 (s, 1H), 7.36 (m, 2H), 7.00 (m, 1H), 3.97 (m, 2H), 3.80 (s, 3H),
1.72 (m, 2H), 1.25 (m, 18H), 0.86 (m, 3H); 13C NMR (75 MHz, DMSO-d6): δ 164.04,
151.88, 148.15, 125.39, 111.88, 111.74, 68.74, 56.08, 31.76, 29.48, 29.17, 25.97,
22.54, 14.37; HRMS (ESI-TOF): calcd for C20H34NO4 [M + H]+ 352.2482, found
352.2482
19

ACCEPTED MANUSCRIPT
4-(Benzyloxy)-3-(decyloxy)-N-hydroxybenzamide (13)
1
White solid (57 mg, 54.9 %), H-NMR (300 MHz, CDCl3): δ 7.44-7.31 (m, 6H),
7.20 (m, 1H), 6.89 (m, 1H), 5.18 (s, 2H), 4.05 (m, 2H), 1.84 (m, 2H), 1.47 (m, 2H),
1.27 (m, 12H), 0.90 (t, J = 6.00 Hz, 3H); ESI-MS m/z: 422 [M + Na]+.
4-(Decyloxy)-N,3-dihydroxybenzamide (1f)
White solid (56 mg, 53.3 %), M.P. 128-130 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.45 (s, 1H), 9.21 (s, 1H), 7.40-7.35 (m, 2H), 6.97-6.95 (m, 1H), 4.02 (t, J = 6.30 Hz,
2H), 1.74-1.70 (m, 2H), 1.41-1.39 (m, 2H), 1.25 (m, 12H), 0.88 (t, J = 6.30 Hz, 3H);
13C NMR (75 MHz, DMSO-d6): δ 167.63, 151.42, 146.70, 123.39, 122.02, 116.57,
112.72, 68.64, 31.75, 29.48, 29.42, 29.25, 29.16, 29.09, 25.88, 22.55, 14.40; ESI-MS
m/z: 332 [M + Na]+.
N-Hydroxy-3-(octyloxy)benzamide (2h)
1
White solid (73 mg, 68.9 %), M.P. 85-87 °C, H-NMR (300 MHz, DMSO-d6): δ
11.18 (s, 1H), 9.01 (s, 1H), 7.34-7.29 (m, 3H), 7.06-7.04 (m, 1H), 4.01 (t, J = 6.3 Hz,
2H), 1.74-1.69 (m, 2H), 1.41 (m, 2H), 1.26 (m, 8H), 0.86-0.38 (m, 3H); 13C NMR (75
MHz, DMSO-d6): δ 164.37, 159.03, 134.58, 129.94, 119.43, 117.86, 113.02, 68.05,
31.69, 29.19, 29.12, 29.09, 25.96, 22.54, 14.48; HRMS (ESI-TOF): calcd for
C15H24NO3 [M + H]+ 266.1751, found 266.1747.
3-(Decyloxy)-N-hydroxybenzamide (2i)
White solid (82 mg, 77.8 %), M.P. 107-108 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.16 (s, 1H), 9.01 (s, 1H), 7.38-7.29 (m, 3H), 7.07-7.03 (m, 1H), 4.01 (t, J = 6.30 Hz,
2H), 1.74-1.69 (m, 2H), 1.41-1.37 (m, 2H), 1.25 (m, 12H), 0.88 (t, J = 6.30 Hz, 3H);
13C NMR (75 MHz, DMSO-d6): δ 164.34, 159.03, 134.57, 129.93, 119.42, 117.85,
113.01, 68.04, 31.75, 29.47, 29.41, 29.22, 29.15, 29.09, 25.95, 22.54, 14.38; HRMS
(ESI-TOF): calcd for C17H28NO3 [M + H]+ 294.2064, found 294.2067.
3-(Dodecyloxy)-N-hydroxybenzamide (2j)
White solid (76 mg, 72.4 %), M.P. 115-117 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.18 (s, 1H), 9.02 (s, 1H), 7.36-7.29 (m, 2H), 7.06-7.04 (m, 1H), 4.00 (t, J = 6.60 Hz,
13
2H), 1.73-1.69 (m, 2H), 1.41 (m, 2H), 1.24 (m, 16H), 0.85-0.83 (m, 3H); C NMR
(75 MHz, DMSO-d6): δ 164.34, 159.03, 134.57, 129.93, 119.42, 117.85, 112.99,
68.03, 31.76, 29.48, 29.24, 29.18, 29.10, 25.96, 22.56, 14.39; HRMS (ESI-TOF):
calcd for C19H32NO3 [M + H]+ 322.2377, found 322.2384.
3-(Benzyloxy)-N-hydroxy-4-methoxybenzamide (2k)
20

ACCEPTED MANUSCRIPT
White solid (86 mg, 81.3 %), M.P. 178-179 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.06 (s, 1H), 8.90 (s, 1H), 7.47-7.34 (m, 7H), 7.34-7.02 (m, 1H), 5.12 (s, 2H), 3.82(s,
3H); 13C NMR (75 MHz, DMSO-d6): δ 164.37, 152.03, 147.79, 137.39, 128.87,
128.34, 128.26, 125.39, 120.88, 112.59, 111.85, 70.48, 56.16; HRMS (ESI-TOF):
calcd for C17H16NO4 [M + H]+ 274.1074, found 274.1079.
3-(Benzyloxy)-N-hydroxybenzamide (2l)
White solid (70 mg, 65.7 %), M.P. 174-176 °C, 1H-NMR (300 MHz, DMSO-d6): δ
13
11.22 (s, 1H), 9.06 (s, 1H), 7.48-7.31 (m, 8H), 7.18-7.14 (m, 1H), 5.15 (s, 2H); C
NMR (75 MHz, DMSO-d6): δ 164.33, 158.71, 137.32, 134.66, 130.03, 128.92,
128.35, 128.15, 119.77, 118.19, 113.55, 69.81; HRMS (ESI-TOF): calcd for
C14H14NO3 [M + H]+ 244.0968, found 244.0969.
3-((4-Cyanobenzyl)oxy)-N-hydroxybenzamide (2m)
White solid (75 mg, 70.8 %), M.P. 160-162 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.21 (s, 1H), 9.05 (s, 1H), 7.89-7.86 (m, 2H), 7.67-7.64 (m, 2H), 7.41-7.37 (m, 3H),
13
7.19-7.15 (m, 1H), 5.27 (s, 2H); C NMR (75 MHz, DMSO-d6): δ 164.21, 158.36,
143.16, 134.73, 132.90, 130.12, 128.51, 120.04, 119.20, 118.21, 113.58, 111.00,
68.85; HRMS (ESI-TOF): calcd for C15H13N2O3 [M + H]+ 269.0921, found 269.0923.
N-Hydroxy-3-((4-methoxybenzyl)oxy)benzamide (2n)
White solid (73 mg, 70.0 %), M.P. 170-172 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.20 (s, 1H), 9.04 (s, 1H), 7.40-7.34 (m, 5H), 7.16-7.13 (m, 1H), 6.97-6.94 (m, 2H),
13
5.06 (s, 2H), 3.76 (s, 3H); C NMR (75 MHz, DMSO-d6): δ 164.34, 159.50, 158.76,
134.61, 129.97, 129.16, 119.19, 118.19, 114.30, 113.55, 69.60, 55.55; HRMS (ESI-
TOF): calcd for C15H16NO4 [M + H]+ 274.1074, found 274.1076.
N-Hydroxy-3-((4-methoxybenzyl)oxy)benzamide (2o)
White solid (90 mg, 85.8 %), M.P. 149-151 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.06 (s, 1H), 8.99 (s, 1H), 7.62-7.42 (m, 5H), 7.05-7.03 (m, 1H), 5.11 (m, 2H), 3.82
(s, 3H); 13C NMR (75 MHz, DMSO-d6): δ 164.26, 151.99, 147.55, 136.88, 131.82,
130.31, 125.36, 121.46, 120.99, 112.63, 111.85, 69.65, 56.17; HRMS (ESI-TOF):
calcd for C15H16NO4 [M + H]+ 352.0185, found 352.0179.
N-Hydroxy-3-((4-methoxybenzyl)oxy)benzamide (2p)
White solid (86 mg, 72.3 %), M.P. 160-161 °C, 1H-NMR (300 MHz, DMSO-d6): δ
11.18 (s, 1H), 9.00 (s, 1H), 7.62-7.59 (m, 2H), 7.44-7.35 (m, 4H), 7.16-7.14 (m, 1H),
5.14 (s, 2H); 13C NMR (75 MHz, DMSO-d6): δ 164.26, 158.51, 136.80, 134.66,
21

ACCEPTED MANUSCRIPT
131.85, 130.24, 121.46, 119.88, 118.19, 113.58, 68.99; HRMS (ESI-TOF): calcd for
C15H16NO4 [M + Na]+ 343.9897, found 343.9893.
General procedure for the preparation of compounds 2a - 2g, 2q
A mixture of the carboxylic acid substrate (100 mg), 1-Ethyl-3-(3-
dimethylaminopropyl)
carbodiimide
hydrochloride
(3
equiv.),
4-
dimethylaminopyridine (0.5 equiv.) and amine (3 equiv.) in anhydrous DMF (5 mL)
was stirred at room temperature for 6 h under nitrogen atmosphere. The mixture was
diluted with ethyl acetate (20 mL), washed with water (15 mL × 3) and dried over
Na2SO4, concentrated and purified by chromatography on silica gel to afford the
target products.
3-(Decyloxy)-N-(2-hydroxyethyl)-4-methoxybenzamide (2a)
1
White solid (62 mg, 54.4 %), M.P. 98-99 °C, H-NMR (300 MHz, CDCl3): δ 7.43
(s, 1H), 7.28 (s, 1H), 6.90-6.87 (m, 1H), 6.60-6.58 (m, 1H), 4.10 (t, J = 7.05 Hz, 2H),
3.93 ( s, 3H), 3.86 (m, 2H), 3.65 (m, 2H), 1.90-1.85 (m, 2H), 1.76-1.71 (m, 4H), 1.47
(m, 2H), 1.28 (m, 8H), 0.92 (t, J = 6.12 Hz, 3H); 13C NMR (75 MHz, CDCl3): δ
168.37, 152.27, 148.57, 126.60, 119.51, 112.01, 110.62, 69.18, 62.46, 56.02, 42.96,
31.88, 29.54, 29.38, 29.30, 29.11, 25.93, 22.66, 14.10; HRMS (ESI-TOF): calcd for
C20H34NO4 [M + H]+ 352.2482, found 352.2486.
3-(Decyloxy)-N-(2-hydroxypropyl)-4-methoxybenzamide (2b)
1
White solid (69 mg, 58.2 %), M.P. 89-91 °C, H-NMR (300 MHz, CDCl3): δ 7.43
(s, 1H), 6.89-6.87 (m, 1H), 6.52 (s, 1H), 4.10 (t, J = 6.69 Hz, 2H), 3.93 (s, 3H), 3.68-
3.65 (m, 1H), 3.34-3.31 (m, 1H), 1.90-1.83 (m, 4H), 1.49-1.45 (m, 2H), 1.29-1.27 (m,
13H), 0.92 (t, J = 6.12 Hz, 3H); 13C NMR (75 MHz, CDCl3): δ 168.20, 152.29,
148.62, 126.73, 119.44, 112.15, 110.69, 69.21, 67.65, 56.04, 47.57, 31.87, 29.53,
29.37, 29.29, 29.12, 25.92, 22.65, 21.04, 14.07; HRMS (ESI-TOF): calcd for
C21H36NO4 [M + H]+ 366.2639, found 366.2643.
N-(2-Hydroxyethyl)-4-methoxy-3-(octyloxy)benzamide (2c)
White solid (79 mg, 68.5 %), M.P. 85-87 °C, 1H-NMR (300 MHz, CDCl3): δ 7.39-
7.29 (m, 2H), 6.98 (s, 1H), 6.80-6.77 (m, 1H), 4.01 (t, J = 6.6 Hz, 2H), 3.96 (s, 3H),
3.76 (m, 2H), 3.56 (m, 2H), 1.81-1.76 (m, 2H), 1.40 (m, 2H), 1.24 (m, 8H), 0.85 (m,
13
3H); C NMR (75 MHz, CDCl3): δ 168.29, 152.24, 148.52, 126.63, 119.70, 112.12,
110.68, 56.00, 42.98, 31.78, 29.32, 29.12, 25.92, 22.61, 14.05; HRMS (ESI-TOF):
calcd for C18H30NO4 [M + H]+ 324.2169, found 324.2173.
22

ACCEPTED MANUSCRIPT
N-(2-Hydroxypropyl)-4-methoxy-3-(octyloxy)benzamide (2d)
1
White solid (83 mg, 69.0 %), M.P. 109-111 °C, H-NMR (300 MHz, CDCl3): δ
7.42 (s, 1H), 7.34-7.29 (m, 1H), 6.85-6.82 (m, 2H), 4.06-4.01 (m, 3H), 3.65-3.60 (m,
1H), 3.30 (m, 2H), 1.86-1.81 (m, 2H), 1.45 (m, 2H), 1.29-1.22 (m, 11H), 0.89-0.87
(m, 3H); 13C NMR (75 MHz, CDCl3): δ 168.19, 152.24, 148.46, 126.71, 119.57,
112.16, 110.67, 56.02, 47.61, 31.78, 29.32, 29.17, 25.92, 22.62, 21.01, 14.05; HRMS
(ESI-TOF): calcd for C19H32NO4 [M + H]+ 338.2326. found 338.2333.
3-(Decyloxy)-N,4-dimethoxybenzamide (2e)
1
White solid (80 mg, 73.1 %), M.P. 115-117 °C, H-NMR (300 MHz, CDCl3): δ
7.36 (s, 1H), 7.24 (m, 1H), 6.88-6.85 (m, 1H), 4.09 (t, J = 6.90 Hz, 2H), 3.92-3.90 (s,
13
6H), 1.90-1.85 (m, 2H), 1.47-1.43 (m, 2H), 1.28 (m, 12H), 0.90-0.87 (m, 3H); C
NMR (75 MHz, CDCl3): δ 166.49, 152.64, 148.74, 124.22, 119.66, 111.85, 110.70,
69.20, 64.56, 56.03, 31.88, 29.54, 29.37, 29.30, 29.08, 25.92, 22.66, 14.09; HRMS
(ESI-TOF): calcd for C19H32NO4 [M + H]+ 338.2326, found 338.2330.
3-(Decyloxy)-4-methoxy-N-(pyridin-2-yl)benzamide (2f)
White solid (104 mg, 83.4 %), M.P. 83-83 °C, 1H-NMR (300 MHz, CDCl3): δ 9.23
(s, 1H), 8.50 (d, J = 8.43 Hz, 1H), 8.29-8.28 (m, 1H), 7.87-7.82 (m, 1H), 7.59-7.57
(m, 2H), 7.15-7.10 (m, 1H), 6.97 (d, J = 8.88 Hz, 1H), 4.14 (t, J = 6.84 Hz, 2H), 3.95
(s, 3H), 1.94-1.85 (m, 2H), 1.49-1.44 (m, 2H), 1.32 (m, 12H), 0.91 (t, J = 5.82 Hz,
3H); 13C NMR (75 MHz, CDCl3): δ 165.40, 152.84, 151.78, 148.75, 147.41, 138.71,
126.52, 120.13, 119.67, 114.32, 111.96, 110.72, 69.20, 56.08, 31.89, 29.54, 29.37,
29.30, 29.09, 25.92, 22.66, 14.10; HRMS (ESI-TOF): calcd for C23H33N2O3 [M + H]+
385.2486, found 385.2494.
3-(Decyloxy)-4-methoxy-N-(pyridin-3-yl)benzamide (2g)
White solid (108 mg, 84.9 %), M.P. 70-73 °C, 1H-NMR (300 MHz, CDCl3): δ 9.27
(s, 1H), 8.48-8.45 (m, 1H), 8.26-8.25 (m, 1H), 7.84-7.79 (m, 1H), 7.58-7.56 (m, 2H),
7.12-7.08 (m, 1H), 6.95-6.92 (m, 1H), 4.12 (t, J = 6.90 Hz, 2H), 3.93 (s, 3H), 1.92-
1.83 (m, 2H), 1.47-1.42 (m, 2H), 1.31 (m, 8H), 0.88-0.86 (m, 3H); 13C NMR (75
MHz, CDCl3): δ 165.42, 152.88, 151.74, 148.74, 147.05, 138.95, 126.41, 120.25,
119.64, 114.45, 111.97, 110.74, 69.21, 56.09, 31.80, 29.33, 29.20, 29.09, 25.93,
22.64, 21.24, 14.08; HRMS (ESI-TOF): calcd for C21H29N2O3 [M + H]+ 357.2173,
found 357.2179.
3-(Benzyloxy)-N-(2-hydroxyethyl)-4-methoxybenzamide (2q)
23

ACCEPTED MANUSCRIPT
1
White solid (65 mg, 55.7 %), M.P. 132-133 °C, H-NMR (300 MHz, CDCl3): δ
7.46-7.37 (m, 3H), 7.34-7.28 (m, 4H), 6.85-6.78 (m, 2H), 5.13 (s, 2H), 3.89 (s, 3H),
3.78 (m, 2H), 3.57 (m, 2H), 3.26 (s, 1H); 13C NMR (75 MHz, CDCl3): δ 168.19,
152.57, 148.07, 136.57, 128.58, 128.04, 127.56, 126.56, 120.32, 113.05, 110.86,
71.07, 62.34, 56.04, 42.92; HRMS (ESI-TOF): calcd for C17H20NO4 [M + H]+
302.1387, found 302.1392.
Preparation of 3-(decyloxy)-N,4-dihydroxybenzamide (1e)
Compound 13 (50 mg) was dissolved in methanol and 10 % wet Pd/C (5 mg) were
added. The mixture was stirred at room temperature for 10 h under H2 atmosphere. As
the reaction finished, the mixture was filtered, concentrated in vacuum and purified by
chromatography on silica gel to give 1e. White solid (69 mg, 89.1 %), M.P. 100-102
°C, 1H-NMR (300 MHz, DMSO-d6): δ 9.39 (s, 1H), 7.74 (s, 1H), 7.43-7.34 (m, 2H),
7.08 (s, 1H), 6.81-6.78 (m, 1H), 3.99 (t, J = 6.60 Hz, 2H), 1.75-1.68 (m, 2H), 1.42 (m,
2H), 1.26 (m, 12H), 0.88 (t, J = 5.40 Hz, 3H); 13C NMR (75 MHz, DMSO-d6): δ
168.07, 150.20, 146.79, 125.73, 121.50, 115.21, 113.32, 68.81, 31.76, 29.52, 29.44,
29.29, 29.22, 29.18, 25.93, 22.55, 14.41; HRMS (ESI-TOF): calcd for C17H28NO4 [M
+ H]+ 310.2013, found 310.2013.
Biology
In vitro ASM inhibition assay
ASM was obtained as follows. The human hepatocellular carcinoma cells (Huh 7)
were grown in a medium (DMEM; 10% Gibco) at 37 °C in a humidified 5.0 % CO2
incubator. As the cells grown to a confluence of 95 %, they were washed with
phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM
KH2PO4) and lysed in lysisbuffer (25 mM Tris-HCl, PH 7.4, 5 mM EDTA). The cells
were gathered to an eppendorf tube and broken by Ultrasonic Processor with a
strength of 10%. Cell debris was removed by centrifugation at 300 g for 5 min at 4
°C. The supernatant was transferred to a new eppendorf tube and the concentra-tion of
protein was detected by the BCA method.
Quantitative analysis of ASM activity was performed on 400 µg of total cellular
protein extract. NBD-sphingomyelin (1 mg/mL, 4 µL) was suspended in 250 µL of
enzyme buffer (100 mM sodium acetate pH 4.7, 1% Triton X-100, 40 mM CaCl2, 100
µM ZnCl2) and added to the protein extract. Inhibitors were added in various
24

ACCEPTED MANUSCRIPT
concentrations in DMSO. The total volume of the reaction was adjusted to 500 µL by
addition of water and incubated at 37 °C for 30 min. The reaction was stopped by
addition of 800 µL of chloroform / methanol (v/v = 2:1), and phases were separated
by centrifugation. 400 µL of organic phase was extracted and dried by N2. The
products were dissolved by 30 µL chloroform and analyzed by thin layer
chromatography. Inhibitory rates were determined by measuring the fluorescence
intensity of produced NBD-Cers, which was quantified by the ImageJ2x software.
The IC50 was calculated by the inhibitory rate of different compounds concentration in
the GraphPad Prism 6 software. The results refer to the mean values resulting from
three independent experiments.
Kinetic studies of ASM inhibition
The kinetics studies were performed according to the protocol reported above. The
substrate BND-SM was dissolved with DMSO and diluted before use. The product
BND-Cers was detected by HPLC. Initially different protein concentration and
substrate concentration with a different reaction time were attempted to determine the
reaction condition that the reaction product can be obviously detected. Next, a
standard curve for NBD-Cers was established, which covering the estimated
concentration spot of the product. For each concentration of the test compound, NBD-
SM was used at 1.35 µM, 1.89 µM, 2.70 µM, 4.54 µM, and 13.5 µM in the reactions.
The reaction was stopped after 10 min. Lineweaver-Burk plots were calculated using
linear regression. Each experiment was performed in triplicate.
Ceramide generation inhibition assay
Human NIH3T3 cells were maintained in DMEM medium supplemented with 10 %
fetal bovine serum under 37 °C and 5 % CO2. As the cells grown to a confluence of
90 %, cells were cultured with fresh medium and inhibitor was added into the cell
culture. The cells were incubated for another 20 h. Quantification was carried out with
the LC-MS-MS technology.
Cell apoptosis assay
Human NIH3T3 cells were maintained in DMEM medium supplemented with 10 %
fetal bovine serum under 37 °C and 5 % CO2. Cells were seeded at 510 cells per well
in 6-well plates and incubated overnight. Following, inhibitor was added into the cell
culture, and incubated for another 20 h until UV irradiation. Cells without treatment
of UV and inhibitors were chosen as control group. Then cells were cultured with
25

ACCEPTED MANUSCRIPT
fresh medium during UV irradiation. The irradiation was performed at a peak
emission wavelength of 315 nm, at a dose as 0.29 mW/cm2 for 5 min. After
irradiation, corresponding cells were continuously maintained at different
concentration of inhibitors or solvent DMSO for another 24 h. Cells’ apoptosis status
was measured by flow cytometry (Becton Dickinson FACSAria III, NJ, USA).
Cytotoxic assay
Human NIH3T3 cells were maintained in DMEM medium supplemented with 10 %
fetal bovine serum under 37 °C and 5 % CO2. As the cells grown overnight, inhibitor
was added into the cell culture. The cells were incubated for another 24 h. Viability of
cells was tested using the MTT method.
Anti-inflammation assay
Human NIH3T3 cells were maintained in DMEM medium supplemented with 10 %
fetal bovine serum under 37 °C and 5 % CO2. Cells were seeded at 2*105 cells per
well in 96-well plates and incubated overnight. Following, inhibitor was added into
the cell culture, and incubated for another 20 h. Each well was added LPS to a
concentration of 50 ng/mL and cells were continuously maintained for another 6 h.
Then IL-6 or IL-8 was measured by ELISA method.
Notes: The authors declare no competing financial interest. These authors contributed
equally to this work.
Funding: This work was supported financially by the key research & development
program in Jiangsu [NO. BE2015683], by the introduction program of leading
scientific and technological entrepreneurship in Nanjing [NO. 2013B14007], by the
innovative research project of graduate student in Jiangsu [NO. KYLX16_1170].
References
[1] I. Petrache, E.V. Berdyshev, Ceramide Signaling and Metabolism in
Pathophysiological States of the Lung, in: D. Julius (Ed.) Annu Rev Physiol, 78
(2016) 463-480.
[2] Y.A. Hannun, L.M. Obeid, The Ceramide-centric universe of lipid-mediated cell
regulation: stress encounters of the lipid kind, J Biol Chem, 277 (2002) 25847-25850.
[3] J.A. Chavez, S.A. Summers, A ceramide-centric view of insulin resistance, Cell
Metab, 15 (2012) 585-594.
26

ACCEPTED MANUSCRIPT
[4] B. Stancevic, R. Kolesnick, Ceramide-rich platforms in transmembrane signaling,
FEBS Lett, 584 (2010) 1728-1740.
[5] M. Miyaji, Z.X. Jin, S. Yamaoka, R. Amakawa, S. Fukuhara, S.B. Sato, T.
Kobayashi, N. Domae, T. Mimori, E.T. Bloom, T. Okazaki, H. Umehara, Role of
membrane sphingomyelin and ceramide in platform formation for Fas-mediated
apoptosis, J Exp Med, 202 (2005) 249-259.
[6] A. Gomez-Munoz, N. Presa, A. Gomez-Larrauri, I.G. Rivera, M. Trueba, M.
Ordonez, Control of inflammatory responses by ceramide, sphingosine 1-phosphate
and ceramide 1-phosphate, Prog Lipid Res, 61 (2016) 51-62.
[7] P.P. Sordillo, D.C. Sordillo, L. Helson, Review: The Prolonged QT Interval: Role
of Pro-inflammatory Cytokines, Reactive Oxygen Species and the Ceramide and
Sphingosine-1 Phosphate Pathways, In Vivo, 29 (2015) 619-636.
[8] G.F. Nixon, Sphingolipids in inflammation: pathological implications and
potential therapeutic targets, Br J Pharmacol, 158 (2009) 982-993.
[9] C. Garcia-Ruiz, A. Morales, J.C. Fernandez-Checa, Glycosphingolipids and cell
death: one aim, many ways, Apoptosis, 20 (2015) 607-620.
[10] S.A. Morad, M.C. Cabot, Ceramide-orchestrated signalling in cancer cells, Nat
Rev Cancer, 13 (2013) 51-65.
[11] I. Petrache, V. Natarajan, L. Zhen, T.R. Medler, A.T. Richter, C. Cho, W.C.
Hubbard, E.V. Berdyshev, R.M. Tuder, Ceramide upregulation causes pulmonary cell
apoptosis and emphysema-like disease in mice, Nat Med, 11 (2005) 491-498.
[12] L. Ciarlo, V. Manganelli, T. Garofalo, P. Matarrese, A. Tinari, R. Misasi, W.
Malorni, M. Sorice, Association of fission proteins with mitochondrial raft-like
domains, Cell Death Differ, 17 (2010) 1047-1058.
[13] E.L. Smith, E.H. Schuchman, The unexpected role of acid sphingomyelinase in
cell death and the pathophysiology of common diseases, Faseb j, 22 (2008) 3419-
3431.
[14] J. Kornhuber, C. Rhein, C.P. Muller, C. Muhle, Secretory sphingomyelinase in
health and disease, Biol Chem, 396 (2015) 707-736.
[15] J. Kornhuber, A. Medlin, S. Bleich, V. Jendrossek, A.W. Henkel, J. Wiltfang, E.
Gulbins, High activity of acid sphingomyelinase in major depression, J Neural
Transm (Vienna), 112 (2005) 1583-1590.
[16] V. Teichgraber, M. Ulrich, N. Endlich, J. Riethmuller, B. Wilker, C.C. De
27

ACCEPTED MANUSCRIPT
Oliveira-Munding, A.M. van Heeckeren, M.L. Barr, G. von Kurthy, K.W. Schmid, M.
Weller, B. Tummler, F. Lang, H. Grassme, G. Doring, E. Gulbins, Ceramide
accumulation mediates inflammation, cell death and infection susceptibility in cystic
fibrosis, Nat Med, 14 (2008) 382-391.
[17] S. Koka, M. Xia, Y. Chen, O.M. Bhat, X. Yuan, K.M. Boini, P.L. Li, Endothelial
NLRP3 inflammasome activation and arterial neointima formation associated with
acid sphingomyelinase during hypercholesterolemia, Redox Biol, 13 (2017) 336-344.
[18] N. Beckmann, D. Sharma, E. Gulbins, K.A. Becker, B. Edelmann, Inhibition of
acid sphingomyelinase by tricyclic antidepressants and analogons, Front. Physio., 5
(2014) 1-14.
[19] E. Gulbins, M. Palmada, M. Reichel, A. Luth, C. Bohmer, D. Amato, C.P. Muller,
C.H. Tischbirek, T.W. Groemer, G. Tabatabai, K.A. Becker, P. Tripal, S. Staedtler, T.F.
Ackermann, J. van Brederode, C. Alzheimer, M. Weller, U.E. Lang, B. Kleuser, H.
Grassme, J. Kornhuber, Acid sphingomyelinase-ceramide system mediates effects of
antidepressant drugs, Nat Med, 19 (2013) 934-938.
[20] M.J. McVey, M. Kim, A. Tabuchi, V. Srbely, L. Japtok, C. Arenz, O. Rotstein, B.
Kleuser, J.W. Semple, W.M. Kuebler, Acid sphingomyelinase mediates murine acute
lung injury following transfusion of aged platelets, Am J Physiol Lung Cell Mol
Physiol, 312 (2017) L625-l637.
[21] T. Hla, A.J. Dannenberg, Sphingolipid signaling in metabolic disorders, Cell
Metab, 16 (2012) 420-434.
[22] E. Gulbins, S. Walter, K.A. Becker, R. Halmer, Y. Liu, M. Reichel, M.J.
Edwards, C.P. Muller, K. Fassbender, J. Kornhuber, A central role for the acid
sphingomyelinase/ceramide system in neurogenesis and major depression, J
Neurochem, 134 (2015) 183-192.
[23] W.Y. Ong, D.R. Herr, T. Farooqui, E.A. Ling, A.A. Farooqui, Role of
sphingomyelinases in neurological disorders, Expert Opin Ther Targets, 19 (2015)
1725-1742.
[24] J. Riethmuller, J. Anthonysamy, E. Serra, M. Schwab, G. Doring, E. Gulbins,
Therapeutic efficacy and safety of amitriptyline in patients with cystic fibrosis, Cell
Physiol Biochem, 24 (2009) 65-72.
[25] L. Nahrlich, J.G. Mainz, C. Adams, C. Engel, G. Herrmann, V. Icheva, J. Lauer,
C. Deppisch, A. Wirth, K. Unger, U. Graepler-Mainka, A. Hector, S. Heyder, M.
28