X-ray crystal structure guided discovery of new selective, substrate-mimicking sirtuin 2 inhibitors that exhibit activities against non-small cell lung cancer cells

Article abstract of DOI:10.1016/j.ejmech.2018.06.041

Human sirtuin 2 (SIRT2) is a nicotinamide adenine dinucleotide (NAD⁺)-dependent deacylase, and is implicated in human diseases including cancer. Selective small-molecule inhibitors for SIRT2 are sought as chemical tools and potential therapeuti

Full text of DOI:10.1016/j.ejmech.2018.06.041

Accepted Manuscript
X-ray crystal structure guided discovery of new selective, substrate-mimicking sirtuin
2 inhibitors that exhibit activities against non-small cell lung cancer cells
Ling-Ling Yang, Hua-Li Wang, Lei Zhong, Chen Yuan, Si-Yu Liu, Zhu-Jun Yu, Sha
Liu, Yu-Hang Yan, Chengyong Wu, Yuxi Wang, Zhouyu Wang, Yamei Yu, Qiang
Chen, Guo-Bo Li
PII:
S0223-5234(18)30532-4
10.1016/j.ejmech.2018.06.041
EJMECH 10511
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 3 May 2018
Revised Date: 12 June 2018
Accepted Date: 15 June 2018
Please cite this article as: L.-L. Yang, H.-L. Wang, L. Zhong, C. Yuan, S.-Y. Liu, Z.-J. Yu, S. Liu, Y.-H.
Yan, C. Wu, Y. Wang, Z. Wang, Y. Yu, Q. Chen, G.-B. Li, X-ray crystal structure guided discovery of
new selective, substrate-mimicking sirtuin 2 inhibitors that exhibit activities against non-small cell lung
cancer cells, European Journal of Medicinal Chemistry (2018), doi: 10.1016/j.ejmech.2018.06.041.
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
Graphical Abstract

ACCEPTED MANUSCRIPT
X-ray Crystal Structure Guided Discovery of New Selective,
Substrate-Mimicking Sirtuin 2 Inhibitors That Exhibit Activities against
Non-small Cell Lung Cancer Cells
Ling-Ling Yang†,‡,ǁ, Hua-Li Wang†,ǁ, Lei Zhong§,ǁ, Chen Yuan‡, Si-Yu Liu‡, Zhu-Jun
Yu†, Sha Liu†, Yu-Hang Yan†, Chengyong Wu†, Yuxi Wang†, Zhouyu Wang‡, Yamei
Yu†, Qiang Chen*,†, Guo-Bo Li*,†
†
Key Laboratory of Drug Targeting and Drug Delivery System of Ministry of
Education, West China School of Pharmacy, and State Key Laboratory of Biotherapy
and Cancer Center, West China Hospital, West China Medical School, Sichuan
University, Chengdu 610041, China.
‡ College of Food and Bioengineering, Xihua University, Sichuan 610039, China
§ Personalized Drug Therapy Key Laboratory of Sichuan Province, Hospital of the
University of Electronic Science and Technology of China and Sichuan Provincial
People’s Hospital, Sichuan 610072, China
ǁ L.-L. Yang, H.-L. Wang, and L. Zhong contributed equally to this work.
*Corresponding author: liguobo@scu.edu.cn (G.-B. Li); qiang_chen@scu.edu.cn (Q.
Chen).
1

ACCEPTED MANUSCRIPT
Abstract
Human sirtuin 2 (SIRT2) is a nicotinamide adenine dinucleotide (NAD+)-dependent
deacylase, and is implicated in human diseases including cancer. Selective
small-molecule inhibitors for SIRT2 are sought as chemical tools and potential
therapeutics. Here we report the X-ray crystal structure guided structure-activity
relationship studies of new N-(3-(phenoxymethyl)phenyl)acetamide derivatives with
SIRT2, which led to the identification of potent, selective SIRT2 inhibitors.
Crystallographic analyses reveal that the new inhibitors act via inducing the formation
of an enlarged hydrophobic pocket and particularly mimicking the interactions made
by myristoylated-lysine substrates. The most potent inhibitor 24a could
dose-dependently elevate the acetylation level of α-tubulin in the non-small cell lung
cancer H441 cells, which have a high expression level of SIRT2 as determinated by
Western blotting analyses. Further cellular assays reveal that 24a restrains cell growth
mainly through inhibiting cellular proliferation rather than inducing apoptosis.
Moreover, 24a could suppress the migration and invasion of H441 cells. These results
provide an excellent basis for further development of new potent, selective, and cell
active SIRT2 inhibitors as chemical tools and potential therapeutics for SIRT2-driven
non-small cell lung cancers.
2

ACCEPTED MANUSCRIPT
Introduction
The lysine deacetylases (KDACs) are important epigenetic modification enzymes,
which are involved in multiple physiological processes as well as the development
and progression of human diseases (e.g. cancer).[1-3] KDACs can be classified into
those employing one zinc ion to promote catalysis (termed histone deacetylases,
HDACs) and those employing the nicotinamide adenine dinucleotide (NAD+) as a
cofactor in catalysis (termed sirtuins, SIRTs) (Figure 1a).[3] Several clinically useful
inhibitors for the HDACs, including Vorinostat, Romidepsin, Chidamide,
Panobinostat, and Belinostat, have been established for treating cutaneous T cell
lymphoma, peripheral T-cell lymphoma or multiple myeloma. In contrast, there are no
such inhibitors for the sirtuins.[1]
The human genome encodes seven isotypes of sirtuins (SIRT1-7), which differ in
their catalytic activity and subcellular localization (Table S1). Sirtuins are able to
erase various acyl groups, such as acetyl, propionyl, crotonyl, succinyl, glutaryl,
decanoyl, myristoyl and palmitoyl, from histones and many other protein
substrates;[4,5] nevertheless, sirtuin isoforms exhibit different preference for
acyl-lysine substrates (Table S1). The SIRT2, mainly located in cytoplasm, can
catalyze deacetylation and defatty-acylation for various protein substrates, including
transcription factors (e.g., p53, Foxo1, p300, NFκB, and HIF1α), differentiation
proteins (e.g., tubulin, keratin 8, PAR3, and PRLR), and metabolic enzymes (e.g.,
G6PD, LDH, PEPCK1, S6K1, and PGAM).[2,5,6] The dysregulation or overexpression
of SIRT2 is often described as a key factor contributing to human diseases, including
3

ACCEPTED MANUSCRIPT
cancers. For example, Chen et al found that SIRT2 was overexpressed in
hepatocellular carcinoma cell lines, and it promoted epithelial to mesenchymal
transition via the deacetylation of protein kinase B and the activation of the glycogen
synthase kinase-3β/β-catenin signaling pathway.[7] SIRT2 was also observed to be
up-regulated in pancreatic adenocarcinoma and breast cancer, and SIRT2 deletion or
inhibition led to the suppression of cancer cell growth, invasion, and metastasis.[8-10]
Thus, the development of potent and selective SIRT2 inhibitors would be an impetus
to investigate associated molecular mechanisms and to develop potential agents
combating cancer or other relevant diseases.
To date, a number of structurally diverse small-molecule SIRT2 inhibitors have
been reported (Figure S1).[11-20] Notably, aminothiazoles and thienopyrimidinones
were discovered as new types of selective SIRT2 inhibitors, which work via inducing
the formation of an allosteric hydrophobic binding pocket, hence providing new
insights into inhibitor design.[21-24] With the aid of LEADOPT,[25] an automatic
structure-based molecule design tool, we successfully identified the
N-(3-(phenoxymethyl)phenyl)acetamide derivatives (Figure 1b and Figure S2) as
selective SIRT2 inhibitors, which exhibited inhibitory activity against human breast
cancer cell growth.[26] Here we describe the X-ray crystal structure guided
structure-activity
relationship
(SAR)
studies
of
new
N-(3-(phenoxymethyl)phenyl)acetamide derivatives and their activities against
non-small cell lung cancer.
4

ACCEPTED MANUSCRIPT
Figure 1. Crystallographic analysis reveals the binding mode of
1 with SIRT2.
(a) Outline mechanisms for sirtuin catalysed deacylations. (b) View from the SIRT2:1
complex structure (PDB ID 5YQL) showing that 1 binds to make hydrophobic
interactions with residues Phe119, Phe131, Leu134, Leu138, Leu206, Tyr139,
Pro140, Phe190, Phe96, and Phe143. (c) Schematic showing aromatic rings (AR)
or R-substituents on the N-(3-(phenoxymethyl)phenyl)acetamide scaffold that were
the focus of structural modifications.
We first carried out the crystallographic analyses of SIRT2 with
2-((4,6-dimethylpyrimidine-2-yl)thio)-N-(3-(phenoxymethyl)phenyl)acetamide
(1),
which reveal that similar to aminothiazoles and thienopyrimidinones, 1 induces the
formation of an enlarged hydrophobic pocket at the interface of the Rossmann fold
domain and the Zn-binding domain (Figure 1b and Figure S2), but which does not
occupy the substrate-lysine binding pocket (Figure 1b), suggesting that there is still a
considerable possibility for structural optimization. Therefore, we designed and
synthesized a series of new derivatives of 1, and performed SAR studies with them,
5

ACCEPTED MANUSCRIPT
which led to the discovery of new potent SIRT2 inhibitors. The subsequent
crystallographic analyses revealed that the new inhibitors bound in the induced
hydrophobic pocket, and mimicked the interactions made by the myristoylated-lysine
substrates. Since there are few reports of SIRT2 inhibitors against non-small cell lung
cancer (NSCLC), we examined the anti-NSCLC effects for the most potent inhibitor.
Chemistry.
A series of new N-(3-(phenoxymethyl)phenyl)acetamide derivatives were synthesized
according to the synthetic routes outlined in Schemes 1-3. Briefly, commercially
available N-(4-hydroxyphenyl)acetamide or N-(3-hydroxyphenyl)acetamide (3)
underwent a five-step process involving K2CO3-promoted nucleophilic substitution,
SOCl2-mediated deacetylation reaction, condensation reaction of acyl halide with
amine, Fe-mediated reduction reaction of aromatic nitro compound, and condensation
reaction of carboxylic acid with amine to yield the key intermediate 8 in high yield
(Scheme 1), which were described in our previous work.[26] Reactions of the key
intermediate 8 with commercially available nucleophiles (e.g. aromatic thiol and
phenolic hydroxyl) or synthesized nucleophile ([1,2,4]triazolo[4,3-a]pyridine-3-thiol,
see Scheme S1) in the presence of potassium tert-butoxide as acid binding agent to
afford the desired target compounds 9a and 9d-9i. Compound 9a was further reacted
with 2-bromopropane to produce the target compounds 9b (45% yield) and 9c (36%
yield).
6

ACCEPTED MANUSCRIPT
Scheme 1. Syntheses of compounds 9a-9i. a
a Reagents and conditions: (a) 1-(bromomethyl)-3-nitrobenzene, K2CO3, DMF, RT, 4h,
81%; (b) SOCl2, MeOH, 65℃, 3h, 100%; (c) thiophene-2-carbonyl chloride, Et3N,
DCM, 0℃, 2h, 70%; (d) Fe, NH4Cl, EtOH/H2O =2:1, 90℃, 1.0h, 92%; (e)
2-bromoacetic acid, HOBT, EDCI, DIPEA, DCM, RT, 12h, 67%; (f) t-BuOK, DMF,
RT, 3-6h, 30-81%; (g) 2-bromopropane, K2CO3, DMF, 60℃, 4h, 45% for 9b, 36% for
9c.
Synthetic access to structurally diverse N-(3-(phenoxymethyl)phenyl)acetamide
derivatives 22-30 was achieved using the similar method as that for compound 2.[26]
The desired target compounds were prepared in a four-step sequence starting from the
synthetic intermediates 5a or 5b (see Scheme 2). Firstly, the 3- or 4- free amino group
of 5a or 5b was transformed to amide group derivatives 10-15 by the standard
condensation reaction of acyl halide/carboxylic acid with amine (Scheme 2). Then,
Fe-mediated reduction reaction was performed to reduce the nitro group of
7

ACCEPTED MANUSCRIPT
intermediates 10-15 to the corresponding amines, followed by coupling with
2-bromoacetic acid to give the key intermediates 16-21. Finally, nucleophilic
substitution of 16-21 with 4,6-dimethylpyrimidine-2-thiol provided the target
compounds 22-27 in good overall yields (Scheme 2). Similarly, coupling of
N-(4-((3-(2-bromoacetamido)benzyl)oxy)phenyl)-1-methyl-1H-pyrazole-4-carboxami
de (18a) with [1,2,4]triazolo[4,3-a]pyridine-3-thiol or naphthalen-1-ol in the presence
of potassium tert-butoxide at room temperature afforded the desired compounds 28
and 29. The target compounds 30a and 30b were further prepared by removal of the
N-Boc protecting group of 27a and 27b in the presence of trifluoroacetic acid at room
temperature, respectively.
The target compounds 33, 35a-35f and 36-41 were prepared as described in
Scheme 3. The commercially available tert-butyl(4-hydroxyphenyl)carbamate (31)
was subjected to the same synthetic transformations as shown in Scheme 1 to afford
the target compound 33 in four steps in 52% over yield. Subsequently, 33 was
converted to the corresponding intermediate 34 by removing the N-Boc protecting
group, which was then reacted with different acyclic carboxylic acids to afford the
target compounds 35a-35f in good yields (Scheme 3). The N-Boc protecting group of
35e and 35f was further removed to give the target compounds 36 (92% yield) and 37
(90% yield), respectively (Scheme 3). The 1-methylthiourea analogues 38 and 39
were synthesized through the condensation reactions of 36 and 37 with
isothiocyanatomethane, respectively, in the presence of triethylamine at room
temperature (Scheme 3). Finally, the (1R,2R)-2-aminocyclohexan-1-ol group was
8

ACCEPTED MANUSCRIPT
introduced to the free amino group of 36 and 37 under basic conditions using
7-oxabicyclo[4.1.0]heptane to give compounds 40 (48% yield) and 41 (41% yield),
respectively (Scheme 3).
Scheme 2. Syntheses of compounds 22-30.a
a
Reagents and conditions: (a) Different acyl chlorides, Et3N, DCM, 0℃, 2h-4h,
70-75%; (b) Different carboxylic acids, HOBT, EDCI, DIPEA, DCM, RT, 12h,
65-78%; (c) Fe, NH4Cl, EtOH/H2O =2:1, 90℃, 0.5-2.0 h, 90-96%; (d) HOBT, EDCI,
DIPEA, DCM, 2-Bromoacetic acid, RT, 12h, 67%-70%; (e) t-BuOK, DMF, RT, 3-6h,
78-81%; (f) TFA, DCM, RT, 4h, 85% for 30a, 88% for 30b.
9

ACCEPTED MANUSCRIPT
Scheme 3. Syntheses of compounds 33, 35a-35f and 36-41.a
OH
O
b
O
c
a
O
NO2
d
O
N
H
O
N
H
32
31
O
O
O
S
N
O
S
N
O
N
H
N
H
e
N
N
O
N
H
H2N
33
34
O
O
O
35a: R=
35b: R=
35c: R=
O
S
N
35d: R=
Ph
O
N
O
N
H
f
H
O
N
35e: R= t-Bu
O
O
N
R
N
H
H
H
O
35a-35f
N
t-Bu
35f: R=
N
O
H
O
O
S
N
N
O
N
H
g
36: n=2, 37: n=3
H2N
35e,35f
N
H
n
36, 37
O
h
O
O
S
S
N
N
O
N
N
H
N
H
H
38: n=2, 39: n=3
N
N
N
n
38, 39
40, 41
H
S
O
OH
O
N
H
H
N
40: n=2, 41: n=3
i
N
H
n
a Reagents and conditions: (a) 1-(bromomethyl)-3-nitrobenzene, K2CO3, DMF, RT, 4h,
87%; (b) Fe, NH4Cl, EtOH/H2O =2:1, 90℃, 1.0 h, 93%; (c) 2-bromoacetic acid,
HOBT, EDCI, DIPEA, DCM, RT, 12h, 82%; (d) 4,6-dimethylpyrimidine-2-thiol,
t-BuOK, DMF, RT, 5h, 78%; (e) TFA, DCM, RT, 4h, 95%; (f) Different carboxylic
acids, HOBT, EDCI, DIPEA, DCM, RT, 12h, 70-86%; (g) TFA, DCM, RT, 3.5h, 92%
for 36, 90% for 37; (h) isothiocyanatomethane, Et3N, DCM, RT, 12h, 80% for 38, 81%
for 39; (i) 7-oxabicyclo[4.1.0]heptane, Et3N, H2O, RT, 12h, 48% for 40, 41% for 41.
Results and Discussion
X-ray Crystal Structure Guided SAR Analyses. We began by performing
crystallographic
analyses
of
SIRT2
with
the
hit
compound
10

ACCEPTED MANUSCRIPT
2-((4,6-dimethylpyrimidine-2-yl)thio)-N-(3-(phenoxymethyl)phenyl)acetamide
(1,
Figure 1b), which was identified by using LEADOPT and had an IC50 value of 25.9
µM against SIRT2; under this assay conditions, the known inhibitor SirReal2 showed
IC50 of 3.38 µM (Figure S2). Co-crystallization experiments yielded a high-quality
X-ray crystal structure of SIRT2:1 complex (1.64 Å, PDB ID 5YQL); details of
crystallization conditions and structure determinations are given in Experimental
section and Table S2. SIRT2:1 crystallized in the P1211 space group with one
molecule per asymmetric unit (ASU) (Table S3). We observed that 1 could be
confidently modelled into the clear Fo-Fc density at the interface of the Rossmann fold
domain and the zinc binding domain (Figure S3).
The crystal structure reveals that 1 likely acts as a pseudo-substrate and
specifically induces the structural rearrangement of the active site and the formation
of the hydrophobic acetyl-lysine-binding tunnel, which is similar as the aminothiazole
inhibitors such as SirReal2 (PDB ID 4RMG) (r.m.s.d. of 0.30 Å over 289 Cα atoms,
Figure S4)[22,23]; 1 is positioned to make hydrophobic interactions with Phe96 (loop
N), Phe119 (α4), Phe131 (α5), Leu134 (α5), Leu138 (α5), Tyr139 (α5), Phe190 (loop
L3), Pro140 (loop L1), and Phe143 (loop L1), and π-π stacking interactions with
Phe190 (loop L3) (Figure 1b and Figure S5). Comparison of the structures of SIRT2:1
and SIRT2:H3K9myr (PDB ID 4Y6L)[27] reveals the binding of 1 may mimic
interactions made by the myristoyl moiety of H3K9myr as it binds into the
hydrophobic pocket of SIRT2 (Figure S6); nevertheless, 1 does not occupy the
substrate binding pocket, e.g. where the lysine motif of H3K9myr binds (Figure 1b).
11

ACCEPTED MANUSCRIPT
Molecular
docking
analyses
indicate
that
the
compound
N-(4-((3-(2-((4,6-dimethylpyrimidine-2-yl)thio)acetamido)benzyl)oxy)phenyl)thiophe
ne-2-carboxamide (2, Figure S2), which is identified by our preliminary SAR
studies,[26] appears to occupy the substrate binding pocket, thereby leading to more
potent inhibition to SIRT2 (IC50 = 3.41 µM, Figure S2). We attempted
co-crystallization but failed to obtain the structure of SIRT2 in complex with 2
(probably due to its poor water solubility). According to the crystallographic and
preliminary SAR results, we designed and synthesized a series of new
N-(3-(phenoxymethyl)phenyl)acetamide derivatives (Scheme 1-3) and performed
relatively systematic SAR studies for them with SIRT2 (Table 1-3).
Variation of Aromatic Rings to Replace the Dimethylpyrimidine Moiety. As revealed
by crystallographic analyses, the dimethylpyrimidine moeity of 1 is positioned to
make π-π stacking interactions with Phe190 and hydrophobic interactions with
Leu138, Leu206, Phe190, Tyr139, Pro140 and Phe143 (Figure 1b); an intramolecular
hydrogen bond between dimethylpyrimidine and amide was observed (Figure 1c),
which may contribute to the formation of the preferred binding conformer with the
hydrophobic pocket. We therefore used various aromatic rings to replace the
dimethylpyrimidine moiety, and synthesized compound 9a-9i (Table 1) via the
synthetic routes as shown in Scheme 1. First, 1H-1,2,4-triazole (9a),
1-isopropyl-1H-1,2,4-triazol-3-yl (9b), 1-isopropyl-1H-1,2,4-triazol-5-yl (9c),
1-methyl-1H-tetrazole-5-yl (9d), and [1,2,4]triazolo[4,3-a]pyridine (9e) were
introduced (Table 1), all of which have potential to form an intramolecular hydrogen
12

ACCEPTED MANUSCRIPT
bond with the amide. Comparing with 2, although these compounds have improved
clogP and clogS properties, they showed much lower inhibitory activities to SIRT2
(Table 1 and Figure S7). Similarly, the introduced moieties including
3-methoxyphenyl (9f), 3-methylphenyl (9g), 3,5-dimethylphenyl (9h), and
1-naphthalene (9i) led to reduced inhibitory activity (Table 1 and Figure S7). These
results indicate that the dimethylpyrimidine moiety is an optimal substituent for
further structural optimization.
Table 1. Inhibitory activities and calculated clogP/clogS values of compounds with
various aromatic rings to replace the dimethylpyrimidine moiety of the
2-((4,6-dimethylpyrimidine-2-yl)thio)-N-phenylacetamide scaffold against human
SIRT2.
Sirt2 (IC50 (µM) /
Cpd. ID
R
pIC50 / s.e. pIC50) cLogP b cLogS b
a
3.41 / 5.47 /
5.05
-6.07
2
0.081
>300 / <3.52 / ~
3.57
4.41
-5.25
-4.68
9a
9b
N
141.6 / 3.85 /
0.096
S
N
N
269.7 / 3.57 /
0.437
4.16
-4.65
9c
13

ACCEPTED MANUSCRIPT
44.45 / 4.35 /
0.079
3.27
4.39
5.32
5.60
-4.52
-5.95
-6.33
-6.46
9d
9e
9f
>300 / <3.52 / ~
>300 / <3.52 / ~
>300 / <3.52 / ~
9g
>300 / <3.52 / ~
>300 / <3.52 / ~
5.68
6.11
4.63
-6.61
-6.75
-5.68
9h
9i
3.38 / 5.47 /
0.111
SirReal2
a
The method for measuring IC50 / pIC50 (n=3) values is described in Experimental
Section. IC50 curves are given in Figure S7.
b
The clogP and clogS values were calculated using the ALOGPS 2.1 program
(http://www.vcclab.org/lab/alogps/; available August 5, 2017).[28]
Variation of Cyclic Substituents on 3- or 4-Position of the Phenyl B Moiety. The
influence of various cyclic substituents at 3- or 4-position of the phenyl B moiety
(Table 2) on the inhibitory activity were then investigated. The compounds 2, 22b,
23a-23b, 24a-24b, 25a-25b, 26a-26b, 28-29, and 30a-30b were synthesized using the
routes shown in Scheme 2. From the calculated clogP and clogS values (Table 2),
compounds 24a, 24b, 25a, 25b, 26a, 26b, 30a, and 30b appear to have improved
physiochemical properties, compared with 2 (cLogP = 5.05 and cLogS = -6.07).
14

ACCEPTED MANUSCRIPT
Comparison of the inhibitory activities between the pairs (2 vs 22b), (23a vs 23b),
(24a vs 24b), (25a vs 25b), (26a vs 26b) and (30a vs 30b) reveals that the compounds
with substituents at the 4-position of the phenyl B moiety have much better potency
aganist SIRT2 than that at the 3-position (Table 2). Compounds 23a, 24a, and 26a,
which have furan, 1-methyl-1H-pyrazole, and tetrahydro-2H-pyran at the 4-position
of the phenyl B moiety (Table 2), respectively, showed IC50 values of 3.10, 0.815, and
1.32 µM (Table 2 and Figure S8), respectively, better than 2 (IC50 = 3.41 µM, Figure
S2) and SirReal2 (IC50 = 3.38 µM, Figure S2). Interestingly, inconsistent with above
compound pairs, 27a and 27b exhibited IC50 of 76.1 and 15.2 µM, respectively.
Comparison of the inhibitory activity of 28 and 29 with 24a revealed that replacement
of
the
moieties
[1,2,4]triazolo[4,3-a]pyridine
and
naphthalene
with
dimethylpyrimidine led to significant reduction of potency against SIRT2, further
reflecting the good compatibility of dimethylpyrimidine with the induced
hydrophobic pocket.
Table 2. Inhibitory activities and calculated clogP/clogS values of compounds with
various cyclic substituents on 3- or 4-position of the phenyl B moiety against human
SIRT2.
SIRT2 (IC50 (µM)
ID
Ring
Pos.
R
cLogP
cLogS
/ pIC50 / s.e.pIC50)a
15

ACCEPTED MANUSCRIPT
4
3
4
3
4
3
4
3
4
3
4
3.41 / 5.47 / 0.081
5.05
4.99
4.66
4.59
3.52
3.48
4.23
4.19
4.08
4.03
4.31
-6.07
-6.05
-4.42
-4.42
-5.09
-5.10
-5.72
-5.70
-5.59
-5.57
-5.42
2
127.7 / 3.89 /
0.500
22b
23a
23b
24a
24b
25a
25b
26a
26b
27a
3.10 / 5.51 / 0.075
122.6 /3.91 / 0.224
0.815 / 6.09 /
0.021
93.7 / 4.03 / 0.045
29.4 / 4.53 / 0.057
>300 / <3.52 / ~
1.32 / 5.88 / 0.029
168.8 / 3.77 /
0.225
76.1 / 4.12 / 0.036
16

ACCEPTED MANUSCRIPT
3
4
3
15.2 / 4.82 / 0.063
3.92
3.18
3.13
-5.30
-5.36
-5.35
27b
30a
30b
41.3 / 4.38 / 0.029
141.6 / 3.85 /
0.096
S
N
N
4
4
>300 / <3.52 / ~
>300 / <3.52 / ~
3.63
4.85
-5.14
-5.48
28
29
N
SirR
eal2
3.38 / 5.47 / 0.111
4.63
-5.68
a
The method for measuring IC50 / pIC50 (n=3) values is described in Experimental
Section. IC50 curves are given in Figure S8.
Variation of Acyclic Substituents on 4-Position of the Phenyl B Moiety. The inhibitory
activities of compounds with various acyclic substituents on 4-position of the phenyl
B moiety were also investigated (Table 3). Compounds 33, 35a, 35b, 36 and 37,
bearing the 2-methylpropan-2-ol, methoxyethane, prop-1-ene moiety, ethanamine and
propan-1-amine, respectively, showed IC50 of ~100 µM (Table 3 and Figure S9),
whereas compounds 35d-35f, which have larger substituents (e.g., benzyl
methylcarbamate, tert-butyl ethylcarbamate and tert-butyl propylcarbamate, Table 3),
displayed lower inhibitory activity to SIRT2 than 33, 35a, 35b, 36 and 37 (Table 3
and Figure S9). Compounds 37 and 38, bearing the 1-methyl-3-ethylthiourea and
17

ACCEPTED MANUSCRIPT
1-methyl-3-propylthiourea moiety at 4-position of the phenyl B moiety, manifested
IC50 of 11.7 µM and 9.56 µM, respectively (Table 3 and Figure S9), which is less
potent than 24a (IC50 = 0.815 µM, Table 3). Similar to compounds 35d-35f,
compounds 40 and 41 displayed low inhibitory potency to SIRT2 (Table 3), probably
due to the blocking effects of the large substituents at 4-position of the phenyl B
moiety. Overall, these SAR results revealed very potent SIRT2 inhibitors (e.g. 24a),
and reflected the high requirements for specific compounds to perfectly bind with the
induced hydrophobic pocket.
Table 3. Inhibitory activities and calculated clogP/clogS values of compounds with
acyclic substituents on 4-position of the phenyl B moiety against human SIRT2.
SIRT2 (IC50 (µM) /
ID
24a
33
Acyclic
cLogP
3.52
cLogS
-5.09
-5.74
pIC50 / s.e. pIC50)a
0.815 / 6.09 / 0.021
100.2 / 4.00 / 0.027
4.62
O
109.5 / 3.96 / 0.108
89.3 / 4.05 / 0.477
3.86
4.68
-5.5
35a
35b
-5.76
>300 / <3.52 / ~
>300 / <3.52 / ~
3.10
4.27
-5.34
-5.92
35c
35d
18

ACCEPTED MANUSCRIPT
~300 / ~3.52 / ~
4.15
4.41
-5.73
-5.82
35e
35f
>300 / <3.52 / ~
~300 / ~3.52 / ~
2.55
2.83
-5.26
-5.42
36
37
70.6 / 4.15 / 0.043
11.7 / 4.93 / 0.047
9.56 / 5.02 / 0.052
3.61
3.89
-5.32
-5.44
38
39
>300 / <3.52 / ~
3.35
3.62
-5.40
-5.55
40
41
175.0 / 3.76 / 0.047
a
The method for measuring IC50 / pIC50 (n=3) values is described in Experimental
Section. IC50 curves are given in Figure S9.
Crystallographic Studies. In order to investigate how the most potent inhibitor 24a
(IC50 = 0.815 µM) binds with SIRT2, we sought to obtain the crystal structure of the
complex of SIRT2 with 24a. The SIRT2:24a structure was obtained by
co-crystallization (crystal conditions see Table S2) and determined to 1.52 Å
resolution (Table S3). This structure crystallized in the P1211 space group with one
monomer per asymmetric unit, which is same as the SIRT2:1 structure (Table S3). We
observed that there was clear Fo–Fc density at the interface of the Rossmann fold
domain and the zinc binding domain (Figure S10), into which 24a could be
19

ACCEPTED MANUSCRIPT
confidently modelled. The crystal structure reveals that 24a is positioned to make
hydrogen-bonding interactions with Val233 (loop S) and Arg97 (loop N), π-π stacking
interactions with Phe190 (loop L3) and Phe235 (loop S), and hydrophobic interactions
with Phe119 (α4), Phe131 (α5), Leu134 (α5), Leu138 (α5), Pro140 (loop L1), Phe 143
(loop L1), Ile169 (loop L2), Phe96 (loop N), and Phe235 (loop S) (Figure 2a and
Figure S11).
Figure 2. Crystallographic analyses of SIRT2 with 24a and 24b. (a) View from a
crystal structure of SIRT2:24a complex (PDB ID 5YQO) reveals that the inhibitor
forms hydrophobic and electrostatic interactions with residues on the α4/α5 helixes
and the S/N loops. (b) Comparison of complex structures of SIRT2:1 and SIRT2:24a
shows that both induce/bind to the hydrophobic pocket. (c) View from a crystal
structure of SIRT2:24b complex (PDB ID 5YQN) reveals that 24b binds via a similar
mode with that of 24a. (d) 24b appears not able to induce the formulation of stable
20

ACCEPTED MANUSCRIPT
interactions with the loop N, e.g., its binding leads to 3.3 Å movement of the C-α
atom of Phe96 (loop N) from that observed in the SIRT2:24a structure.
Comparison of structures of SIRT2:24a and SIRT2:1 reveals that 24a and 1
induce the formation of a similar hydrophobic pocket (r.m.s.d. of 0.08 Å over 286 Cα
atoms)
(Figure
2b
and
Table
S4);
the
2-((4,6-dimethylpyrimidine-2-yl)thio)-N-phenylacetamide moiety of 24a and 1 is
positioned to make similar hydrophobic interactions with the residues around the
induced pocket. Notably, compared with 1, 24a is likely able to make more tight
interactions with SIRT2, e.g. forming hydrophobic interactions with Arg97 and
Val233 and π-π stacking interactions and Phe235 (Figure 2a-b), which may explain
why 24a (IC50 = 0.815 µM) has more potent inhibition to SIRT2 than 1 (IC50 = 25.9
µM).
We then performed crystallographic analyses for the complex of SIRT2 with 24b,
which has the same moiety as 24a (i.e. 1-methyl-1H-pyrazole-4-carboxamide) at
3-position of the phenyl B moiety (Table 2), but has low inhibitory activity to SIRT2
(IC50 = 93.7 µM). Pleasingly, we obtained a high-resolution SIRT2:24b structure
(1.63 Å) by co-crystallization experiments (Table S2). The structure reveals that 24b
is positioned to make hydrogen-bonding and hydrophobic interactions with the
residues on loops N/S/L1/L2/L3 and helixes α4/α5 around the induced hydrophobic
pocket, similar as that observed in the SIRT2:1 and SIRT2:24a structures (Figure 1b,
Figure 2a-b and Figure S12-13). Nevertheless, 24b appears not able to perfectly
interact with loop N, of which the binding leads to 3.3 Å movement of the Cα atom of
21

ACCEPTED MANUSCRIPT
Phe96 (loop N) from that observed in the SIRT2:24a and SIRT2:1 structure (Figure
2d and Figure S14), reflecting that the flexibility of the loop N makes a high
requirement for inhibitors to accommodate with the hydrophobic pocket. The results
may explain why 24b inhibits SIRT2 less potently than 24a or 1.
Comparison of crystal structures of SIRT2:24a, SIRT2:24b, SIRT2:1,
SIRT2:SirReal2 (PDB ID 4RMG)[22], SIRT2:SirReal probe (PDB ID 5DY5)[21]
SIRT2:thienopyrimidinone (PDB ID 5MAT)[24], SIRT2:H3K9myr (PDB ID 4Y6L)[27]
,
,
and
SIRT2:ADP-ribose
(PDB
ID
3ZGV)[29]
reveals
that
these
N-(3-(phenoxymethyl)phenyl)acetamide inhibitors induce structural rearrangement of
the active site (r.m.s.d. of >1.0 Å for Cα atoms comparing with the structures of SIRT2
with H3K9myr and ADP-ribose, Table S4), and reveals evidence for the flexibility in
the conformation of the loop N, suggesting that Phe96 and Arg97 (loop N, Figure 2)
may be important in capturing these specific inhibitors (Figure 2).
On superimposing the SIRT2:24a structure with the crystal structures of SIRT2
complexed with different substrates, we observed that 24a appears mimic interactions
made by the myristoylated-lysine substrates (e.g. H3K9myr[27]) as they bind to SIRT2
(Figure
3
and
Figure
S15).
For
example,
the
2-((4,6-dimethylpyrimidine-2-yl)thio)-N-(3-(phenoxymethyl)phenyl)acetamide
moiety of 24a is positioned to make hydrophobic interactions with SIRT2, similar
with the myristoyl moiety of H3K9myr observed in the SIRT2:H3K9myr structure
(PDB
ID
4Y6L)[27]
.
More
interestingly,
the
amide
of
the
1-methyl-1H-pyrazole-4-carboxamide moiety of 24a forms hydrogen-bonding
22

ACCEPTED MANUSCRIPT
interactions with Val233, which is likely equivalent to the amide on the lysine side
chain of H3K9myr. These results provide insights into inhibitor design, e.g., via
mimicking the interaction features of the myristoylated-lysine substrates.
Figure 3. Superimposition of SIRT2:24a (PDB ID 5YQO) with SIRT2:H3K9myr
(PDB ID 4Y6L)[27] revealing that 24a may mimic interactions made by the
myristoylated-lysine substrate.
Sirtuin Selectivity Analyses. We then tested the most potent compound 24a against
other sirtuin isotypes, including SIRT1, SIRT3, SIRT5, SIRT6, and SIRT7, to
investigate its selectivity profiles (see Experimental Section). Notably, no or low
23

ACCEPTED MANUSCRIPT
inhibitory activities to all of the tested sirtuin isotypes were observed for 24a even at
the concentration of 100 µM (Table 4). From the structural perspective of sirtuins,
SIRT2 has a similar overall domain organization and fold as SIRT1, SIRT3, SIRT4
and SIRT5 (Figure S16), but there are some differences in residues of the catalytic site
and α4/α5 helixes between SIRT2 and other sirtuins;[30-32] in contrast, SIRT2 has
overall domain organization and fold with SIRT6 especially in zinc binding domain
and catalytic site (Figure S17).[30,33] More specially, SIRT2 has a distinct hydrophobic
pocket at the interface of the Rossmann fold domain and the zinc-binding domain
(near the α4 and α5 helixes and loop N), which can accommodate the
myristoylated-lysine substrates[27] and specific small-molecule inhibitors, such as
aminothiazoles,[21]
thienopyrimidinones,[24]
and
(3-(phenoxymethyl)phenyl)acetamides (e.g. 24a). These differences may illustrate
why 24a manifests substantial selectivity for SIRT2.
Table 4. The selectivity of 24a against sirtuin isotypes.
Enzyme
SIRT2
SIRT1
SIRT3
SIRT5
SIRT6
SIRT7
IC50 (µM)
0.815
>100
>100
>100
>100
>100
Antitumor Activity Studies. Since 24a has excellent inhibitory potency and
selectivity to SIRT2, we used 24a as a chemical probe to investigate its activities
against non-small cell lung cancer (NSCLC) cells; NSCLC cells were chosen here
24

ACCEPTED MANUSCRIPT
because there are few reports of SIRT2 inhibitors against this type of cancer. We first
detected the expression of SIRT2 in a panel of NSCLC cell lines, including HCC827,
H1993, H441, H1975, Calu6, and A549, by Western blot assays. The results reveal
that SIRT2 has varied expression levels in the tested NSCLC cell lines (Figure 4a),
e.g., having 3.6 times expression level in H441 cells than that in HCC827 cells
(Figure 4b), indicating that SIRT2 is likely to have different roles in NSCLC
cells.[34-36]
Figure 4. (a) Western blot analyses of SIRT2 expression in a panel of NSCLC cells.
(b) SIRT2 abundance was assessed as ratios relative to HCC827 cells normalized to
α-tubulin protein loading control.
We then examined whether 24a is an active inhibitor of SIRT2 in H441 cells,
which have a high expression level of SIRT2 (Figure 4), by detecting the acetylation
level of α-tubulin, a known protein substrate of SIRT2.[22,23] As shown in Figure 5, the
level of Ac-α-tubulin was dose-dependently elevated in the presence of 24a in H441
25

ACCEPTED MANUSCRIPT
cells, while the structurally similar compound 24b, which has low potency against
SIRT2 (Table 2), had little effects on α-tubulin acetylation; the expression level of
Ac-α-tubulin showed a 2.2-fold increase in the presence of 24a at 30 µM compared
with untreated control, which was set to 1.0. These results indicated that 24a could
suppress the catalytic activity of SIRT2 in H441 cells, and it could be used as a
cell-active chemical tool to probe the biological activity of SIRT2 inhibition in H441
cells.
Figure 5. (a) Western blot detection of Ac-α-tubulin levels in H441 cells after 72 h
incubation with 24a or 24b. (b) Ac-α-tubulin expression was calculated as ratios
relative to untreated control normalized to α-tubulin protein loading control.
Anti-viability Activity of 24a. The anti-viability activity of 24a against H441 cells was
tested using MTT assay. The results reveal that 24a potently inhibited the viability of
H441 cells with IC50 value of 3.93 µM (Figure 6a), which is more potent than that of
26

ACCEPTED MANUSCRIPT
24b (IC50 > 100 µM), consistent with their SIRT2 inhibitory activities (Table 2). To
complement the results from short-term treatments, long-term colony formation assay
was also performed to visually assess the cytoreductive activity of 24a. As indicated
in Figure 6b, 24a efficiently reduced the formation of colonies at concentrations
higher than 3 µM, while 24b showed no significant influence on H441 cell colonies.
Figure 6. (a) The profile of cell viability after treatment with compound 24a or 24b.
(b) Long-term colony formation assay on H441 cells.
Anti-proliferation Activity of 24a. We further evaluated the anti-proliferation activity
of 24a using EdU incorporation assays. Compound 24a could markedly decrease the
number of proliferating H441 cells (red nuclei), compared with both DMSO vehicle
and 24b (Figure 7a). Meanwhile, flow cytometry assays were carried out to detect the
effects of 24a on cellular apoptosis. No significant difference in the apoptotic rate of
different treatment groupswas observed (Figure 7b). These data suggest that 24a
restrains the H441 cell growth mainly through inhibiting cellular proliferation.
27

ACCEPTED MANUSCRIPT
Figure 7. (a) The fluorescence microscopic appearance of EdU and DAPI on H441
cells after treatment with 30 µM compound 24a or 24b. (b) Annexin V/PI co-staining
on compound 24a or 24b treated H441 cells.
Migration and Invasion Inhibition of 24a. Since it has been reported that SIRT2 is
closely related to tumor metastasis,[7,37,38] we further assessed the ability of 24a to
suppress cell migration and invasion (pivotal steps of tumor metastasis) in H441 cells.
In wound healing assay, cell migration was effectively suppressed by 24a at
concentrations higher than 10 µM (Figure 8). The inhibitory effect was also observed
in transwell invasion assays, in which the number of invaded cells was visibly
decreased by 24a in a dose-dependent manner (Figure 8). Of note, compound 24b
showed little activity to inhibit both cell migration and invasion, partly reflecting 24a
works via specifically inhibiting SIRT2.
28