Aminothiazoles as Potent and Selective Sirt2 Inhibitors: A Structure-Activity Relationship Study

Article abstract of DOI:10.1021/acs.jmedchem.5b01517

Sirtuins are NAD⁺-dependent protein deacylases that cleave off acetyl but also other acyl groups from the ε-amino group of lysines in histones and other substrate proteins. Dysregulation of human Sirt2 (hSirt2) activity has been associated with the pathogenesis of cancer, inflammation, and neurodegeneration, which makes the modulation of hSirt2 activity a promising strategy for pharmaceutical intervention. The sirtuin rearranging ligands (SirReals) have recently been discovered by us as highly potent and isotype-selective hSirt2 inhibitors. Here, we present a well-defined structure-activity relationship study, which rationalizes the unique features of the SirReals and probes the limits of modifications on this scaffold regarding inhibitor potency. Moreover, we present a crystal structure of hSirt2 in complex with an optimized SirReal derivative that exhibits an improved in vitro activity. Lastly, we show cellular hyperacetylation of the hSirt2 targeted tubulin caused by our improved lead structure.

Full text of DOI:10.1021/acs.jmedchem.5b01517

Subscriber access provided by La Trobe University Library
Article
Aminothiazoles as potent and selective Sirt2
inhibitors – a structure-activity relationship study
Matthias Schiedel, Tobias Rumpf, Berin Karaman, Attila Lehotzky, Judit Oláh,
Stefan Gerhardt, Judit Ovádi, Wolfgang Sippl, Oliver Einsle, and Manfred Jung
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.5b01517 • Publication Date (Web): 22 Dec 2015
Downloaded from http://pubs.acs.org on January 5, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Aminothiazoles as potent and selective Sirt2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
inhibitors – a structureꢀactivity relationship study
Matthias Schiedel,† Tobias Rumpf,† Berin Karaman,# Attila Lehotzky,§ Judit Oláh,§ Stefan
Gerhardt,∥ Judit Ovádi,§ Wolfgang Sippl,# Oliver Einsle,∥ Manfred Jung†*
† Institute of Pharmaceutical Sciences, AlbertꢀLudwigsꢀUniversity Freiburg, Albertstraße 25,
79104 Freiburg im Breisgau, Germany
# Institute of Pharmacy, MartinꢀLutherꢀUniversity HalleꢀWittenberg, WolfgangꢀLangenbeckꢀ
Straße 4, 06120 Halle (Saale), Germany
§ Institute of Enzymology, Research Centre for Natural Sciences, Hungarian Academy of
Sciences, Magyar Tudósok körútja 2, H 1117 Budapest, Hungary
∥ Institute of Biochemistry and BIOSS Centre for Biological Signalling Studies, Albertꢀ
LudwigsꢀUniversity Freiburg, Albertstraße 21, 79104 Freiburg im Breisgau, Germany
KEYWORDS: epigenetics, histone deacetylases, sirtuins, inhibitors, drug design
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 47
1
2
ABSTRACT: Sirtuins are NAD+ꢀdependent protein deacylases that cleave off acetyl, but also
other acyl groups from the εꢀamino group of lysines in histones and other substrate proteins.
Dysregulation of human Sirt2 (hSirt2) activity has been associated with the pathogenesis of
cancer, inflammation, and neurodegeneration, which makes the modulation of hSirt2 activity a
promising strategy for pharmaceutical intervention. The Sirtuin Rearranging Ligands (SirReals)
have recently been discovered by us as highly potent and isotypeꢀselective hSirt2 inhibitors.
Here, we present a wellꢀdefined structureꢀactivity relationship study, which rationalizes the
unique features of the SirReals and probes the limits of modifications on this scaffold regarding
inhibitor potency. Moreover, we present a crystal structure of hSirt2 in complex with an
optimized SirReal derivative that exhibits an improved in vitro activity. Lastly, we show cellular
hyperacetylation of the hSirt2 targeted tubulin caused by our improved lead structure.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
2

Page 3 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION: Until today, 18 different histone deacetylases (HDACs) have been
identified and grouped into four classes, according to their homology to yeast HDACs, which
were the first to be discovered.1 Class I, II, and IV HDACs are Zn2+ꢀdependent, while sirtuins,
initially described as class III HDACs or Sir2 proteins, constitute a unique class of this enzyme
super family. Sirtuins are dependent on NAD+ as a cofactor to remove acetyl,2 but also other acyl
groups, such as myristoyl,3 palmitoyl,4 and succinyl,5 from the εꢀamino group of lysines. Beyond
histones, a multitude of nonꢀhistone substrates has been identified in recent years, e.g. αꢀtubulin,6
NFκB,7 p53,8 and BubR1.9 Apart from deacylation, sirtuins were shown to catalyze ADPꢀ
ribosylation as well.10 By regulating the acylation or ADPꢀribosylation state of their substrate
proteins, sirtuins have been implicated to influence a wide range of cellular processes like
ageing,11 metabolic sensing, apoptosis,12 inflammation,13 and transcription.14 Sirtuins have been
conserved from bacteria to eukaryotes and share a catalytic domain of approximately 260 amino
acids with a high degree of sequence similarity. While bacteria and archaebacteria possess only
one or two sirtuins, in eukaryotes this number is higher. The human genome encodes seven
sirtuin isotypes, which differ in their catalytic activity and their subcellular localization.15 The
human isotype Sirtuin 2 (hSirt2) is predominantly localized in the cytoplasm, however, it has
also been found in the nucleus. hSirt2 was shown to have a major impact on cell cycle
regulation,6 peripheral myelination,16 autophagy,17 and immune and inflammatory response.18 A
dysregulation of hSirt2 activity was reported to play a critical role in the pathogenesis of
cancer,19 neurodegenerative diseases,20 type II diabetes,21 and bacterial infections.18b, 18c To
further investigate the effects of hSirt2ꢀdependent deacylation, and its impact on downstream
signaling, modulators of hSirt2 activity are urgently needed. A number of hSirt2 modulators
have been discovered thus far, and selected examples are depicted in Figure 1: the physiological
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 47
1
2
sirtuin inhibitor nicotinamide (1)22 and its derivatives 5ꢀ((5ꢀbenzamidonaphthalenꢀ1ꢀ
yl)oxy)nicotinamide (2)23 and the 3′ꢀphenethyloxyꢀ2ꢀanilinobenzamide analogue 3,24 the highly
potent but unselective ELT inhibitor 31 (4),25 the highly selective 5,6,7,8ꢀ
tetrahydrobenzo[4,5]thieno[2,3ꢀd]pyrimidinꢀ4(3H)ꢀone (5),26 AGK2 (6),27 salermide (7),28 and
the macrocyclic peptide S2iL5 (8).29 However, isotypeꢀselective and drugꢀlike inhibitors of
hSirt2 with proven cellular activity are still scarce. Recently, we discovered a novel class of
potent and highly isotypeꢀselective hSirt2 inhibitors.30 Due to a major rearrangement of the
active site of hSirt2 upon ligand binding, these inhibitors were termed Sirtuin Rearranging
Ligands (SirReals). The core of the SirReals is an acylated 2ꢀaminothiazole scaffold, which
connects an arylmethyl moiety with a 4,6ꢀdimethylpyrimidine (Figure 1).
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
4

Page 5 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Chemical structures and inhibition data of selected hSirt2 inhibitors, including
SirReal1 (9) and SirReal2 (10). The intramolecular hydrogen bonds are shown as dashed lines.
By inducing a rearrangement of the active site of hSirt2, the 4,6ꢀdimethylpyrimidine moiety is
bound to a yetꢀunexploited binding pocket, which lays the foundation for the excellent isotypeꢀ
selectivity of the SirReals. Therefore, we named the newly formed binding site ‘selectivity
pocket’. The arylmethyl moiety protrudes towards the substrate channel pushing the acylꢀlysine
out of its physiological position, and thereby enlarging the distance between substrate and
NAD+. This efficiently blocks the acyl transfer. A preliminary structure activity relationship
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 47
1
2
(SAR) study has already been reported,30 revealing an additive contribution of the arylmethyl
and the 4,6ꢀdimethylpyrimidine moiety, and the importance of the intramolecular hydrogen bond
for activity. Here, we systematically probe further SARs to analyze the limits of modification
within this scaffold.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS: Guided by the structural insights obtained from hSirt2ꢀSirReal complexes,30 we
aimed to systematically probe the limits of variation within the scaffold of the SirReals and to
established a wellꢀdefined SARꢀmodel. To study the effect of an alteration of the aminothiazole
core, we initially generated a few synthetically easily accessible aminothiadiazole derivatives.
The synthesis of these inhibitors is outlined in Scheme 1. A condensation of a carboxylic acid
and thiosemicarbazide yielded the aminothiadiazoles 11a-e,31 which were subsequently
chloroacetylated to obtain 12a-g.32 A reaction of the alkyl chlorides with aromatic thiols
generated the desired aminothiadiazoles 13a-h (Table 1).
Scheme 1. Synthesis of aminothiadiazolesa
aReagents and conditions: (a) thiosemicarbazide, H2SO4, 80 ꢀ 90 °C, 7 h; (b) acetyl chloride,
DIPEA, acetonitrile, 0 °C to rt, 2 h; (c) aromatic thiol, Na2CO3, KI, DMSO, 2 h.
Our main focus was placed on the modifications of the arylmethyl and pyrimidine moieties,
which were shown to be crucial for the interaction with the cofactor or ligand binding,
respectively.30 First, we explored modifications of the arylmethyl moiety. Molecular docking
studies, based on the hSirt2ꢀSirReal complexes,30 indicated that substitutions at the naphthyl
group of the parent compound 10 are beneficial for the potency of the ligands. Compound 14a
ACS Paragon Plus Environment
6

Page 7 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(Figure 2a), a chloroꢀsubstituted derivative of the parent compound 10, was predicted to interact
with the two backbone carbonyl groups of His187 and Val233 (Figure 2b). By means of a
halogenation of the naphthyl residue, we aimed to gain ligand affinity via a σꢀhole interaction
with the peptide backbone. To follow up on this hypothesis we synthesized the halogenated
naphthyl derivatives 14a, 14d, 14f, 14g, 14i (Table 2). Additionally, novel derivatives of the
parent compound 9, with substituted benzyl moieties, were generated to consider the impact of
further structural changes of the arylmethyl moiety, as well. Encouraged by molecular docking
studies, that showed a high docking score for the (S)ꢀenantiomer of 14b (Figure 2a,c), we aimed
to synthesize SirReals with an additional alkylation in αꢀposition to the carbonyl of the amide
and eventually separated the enantiomers by chiral HPLC. The docking revealed that the methyl
group ((S)ꢀconfiguration) is orientated towards the exit of the selectivity pocket, which can
accommodate longer alkyl groups. On the other hand, the (R)ꢀconfiguration was predicted to be
less favorable due to the close proximity of the conserved water molecule and Ile93. This is also
reflected in the less favorable binding energies calculated for the (R)ꢀenantiomer (Table S1).
Furthermore, we wanted to investigate the effects of structural changes of different substitution
patterns of the pyrimidine moiety. Initial structural data and SAR studies have shown that the
4,6ꢀdimethylpyrimidine moiety perfectly fits into the newly formed ‘selectivity pocket’, and that
a loss of the methyl groups leads to a decay in ligand potency.30 Additionally, molecular
docking, based on hSirt2ꢀSirReal complexes,30 demonstrated by the example of 14c (Figure 2a),
that larger alkyl substituents, e.g. ethyl, in position 4 and 6 of the pyrimidine moiety cause a
steric clash with residues Ala135, Tyr139, and Leu206 of the ‘selectivity pocket’, as well as with
Thr171 of the Cꢀpocket. These studies predicted a loss of the Hꢀbond to the conserved water
molecule, and thereby lowering the ligand affinity to hSirt2 (Figure 2d). To show the accuracy of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 47
1
2
3
4
our docking studies we still wanted to actually test, whether hSirt2 would be able to
5
6
7
8
accommodate its ‘selectivity pocket’ to other substitution patterns of the pyrimidine moiety, as
well.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Docking poses derived for the envisaged SirReals 14a-c. (a) Chemical structures used
for docking studies. (b) Docking pose derived for 14a (colored magenta). Distances between the
halogen atom and the two carbonyl groups of His187 and Val233 are shown as blue lines with
distances given in Angstrom. (c) Docking pose derived for (S)ꢀ14b (colored cyan). The methyl
group ((S)-configuration) is orientated towards the exit of the ‘selectivity pocket’ which can
accommodate longer alkyl groups. The molecular surface of the binding pocket is displayed and
colored according to the hydrophobicity (green=hydrophobic, magenta= hydrophilic). (d)
Docking pose derived for 14c (colored green) in comparison with the Xꢀray structure of 10
ACS Paragon Plus Environment
8

Page 9 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(colored orange). The larger ethyl groups of the pyrimidine ring cause steric clashes with Ala135
and Leu206 and, as a consequence, the Hꢀbond to the conserved water molecules is lost.
Hydrogen bonds are shown as dashed cyan colored lines, water molecules as red spheres.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To establish a wellꢀdefined SAR model, we set up a synthesis platform to generate a
2ꢀaminothiazole library that was directed to yield compounds with a broad structural variety,
particularly in the arylmethyl and pyrimidine parts of the ligand (Scheme 2).
Scheme 2. Synthesis platform utilized to generate the 2ꢀaminothiazole librarya
aReagents and conditions: (a) NaNO2, HCl, water, ꢀ5 – 0 °C, 10 min; (b1) acrolein, CuCl2 x 2
H2O, acetone, 3 h; (b2) NaHCO3, MgO; then acrolein, CuCl2 x 2 H2O, acetone, 3 h; (b3) FeCl3 x
6 H2O, HCl, water, ꢀ5 – 0 °C; then CuCl2 x 2 H2O, HCl, acetone/ethanol, ꢀ5 – 0 °C; then
acrolein, acetone/water, 4h; (c) thiourea or Nꢀmethylthiourea, ethanol, reflux, 2 h; (d) acyl
chloride, DIPEA, acetonitrile, 0 °C to rt, 2 h; (e) aromatic thiol, Na2CO3, KI, DMSO, 2 h.
The biggest challenge within the synthesis of the SirReals was the preparation of
αꢀchloropropanals via Meerwein reaction,33 due to the instability of some arenediazonium salts,
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 47
1
2
especially the naphthalenediazonium salts.34 In order to overcome these issues, we had to use
three different methods to obtain sufficient amounts of the appropriate αꢀchloropropanals.35
Tolueneꢀ and methoxybenzenediazonium salts could only be converted into the appropriate αꢀ
chloropropanals under standard Meerwein conditions in a neutral medium,35a while halogenated
benzenediazonium salts, as well as the diazonium salt from 4ꢀphenylaniline, successfully reacted
in an acidic medium. For the synthesis of the 5ꢀ(naphthylmethyl)thiazolꢀ2ꢀamine derivatives
15a-f, we followed a modified version of the Meerwein reaction published by Obushak et al.35b
Initially, naphthalenediazonium chlorides were treated with iron(III) chloride in an aqueous
solution to obtain the naphthalenediazonium tetrachloroferrates(III). An exchange reaction with
CuCl2 in acetone resulted in the precipitation of the naphthalenediazonium
tetrachlorocuprates(II), which were isolated and used as a fine crystalline powder for the
subsequent reaction with acrolein.35b With this protocol, we were able to generate the desired
naphthyl derivatives 15b-c, which could not be synthesized under standard Meerwein conditions,
neither in an acidic nor in a neutral medium. For those naphthylamines that could be transformed
into the corresponding αꢀchloropropanals using standard Meerwein conditions to a small
extent,35a the modified version of the Meerwein reaction enabled us to increase the yield of this
reaction largely. However, due to the timeꢀconsuming protocol of the modified Meerwein
reaction, it was exclusively used to synthesize the naphthyl derivatives 15a-f and compound 16g,
which could not be generated in satisfying yields applying standard Meerwein conditions. The
αꢀchloropropanals were immediately converted into the aminothiazole scaffold (15a-f, 16a-g) by
condensation with thiourea.35a The aminothiazoles were subsequently chloroacetylated32 to yield
17a-j and 18a-i, followed by a nucleophilic substitution with an aromatic thiol to generate
compounds 14a-r, 19a-q and 20a-d. The synthesis of the halogenated 1ꢀnaphthylamines (21a-c),
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
10

Page 11 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
which were needed as starting materials for the synthesis of the halogenated inhibitors (14a, 14d,
14f, 14g, 14i), is illustrated in Scheme 3 by the example of the chlorinated 1ꢀnaphthylamines
21a-b. Starting with the nitration of 2ꢀaminonaphthaleneꢀ1ꢀsulfonic acid followed by a
desulfonation36 the isomers of the respective nitronaphthalenamines 22a-b were separated by
flash chromatography. Halonitronaphthalenes 23a-c were synthesized by applying Sandmeyer
conditions.37 Finally, the halogenated 1ꢀnaphthylamines 21a-c were obtained by a reduction of
the nitro group with SnCl2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3. Synthesis of chlorinated 1ꢀnaphthylamines as starting material for further
synthesesa
aReagents and conditions: (a) H2SO4, KNO3, ꢀ15 °C, 40 min; (b) H2SO4/water (1:1), reflux, 40
min; (c) glacial acid, H2SO4, NaNO2, 15ꢀ20 °C, 15 min; then CuCl and HCl, rt, 1 ꢀ 4 h; (d) SnCl2
x 2 H2O, ethanol, rt, 4 – 24 h.
The synthesized SirReal derivatives were evaluated for their inhibitory activity against hSirt1
and hSirt2 in a biochemical in vitro assay that was previously described.38 The in vitro inhibition
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 47
1
2
3
4
5
6
7
8
9
data of the aminothiadiazoles (13a-h) and the aminothiazole derivatives 20a-d, 24a-b, 25, 26a-b
and 27-32, which strongly deviate from the original scaffold is summarized in Table 1.
Aminothiazoles (9-10, 14a-r, 19a-q) and their in vitro inhibition of hSirt1 and hSirt2 are shown
in Table 2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. In vitro inhibition of hSirt1 and hSirt2 by aminothiadiazoles 13a-h and aminothiazoles
20a-d, 24a-b, 25, 26a-b and 27-32
hSirt1 inhibition
%@conc. [µM]
or IC50 ± SE [µM]
n.i.@ 200 µMc
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
12% @ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
hSirt2 inhibition
%@conc. [µM]
or IC50 ± SE [µM]
1.89 ± 1.50
Shared scaffold or complete
chemical structure
compd
R1
R2
13a
13b
13c
13d
13e
13f
1ꢀnaphthylmethyl
phenyl
4ꢀchlorobenzyl
4ꢀbromobenzyl
trifluoromethyl
phenyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
5,6ꢀdiphenylꢀ1,2,4ꢀtriazinꢀ3ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
4,6ꢀdimethylpyrimidinꢀ2ꢀyl
19.9 ± 31.7
30.9 ± 20.5
167.7 ± 30.29
502.8 ± 43.95
19.9% @ 50 µM
17.8% @ 50 µM
n.i. @ 50 µM
13g
13h
benzyl
methyl
20a
3ꢀhydroxyphenyl
n.i.@ 200 µM
143 ± 27.4
3,5ꢀ
dimethylphenyl
20ba
16% @ 200 µM
207 ± 27.4
20c
20d
2ꢀaminophenyl
4ꢀchlorophenyl
n.i.@ 200 µM
n.i.@ 200 µM
36% @ 50 µM
n.i. @ 50 µM
84% @ 500 µM
n.i. @ 50 µM
77% @ 500 µM
48% @50 µM
n.i. @ 10 µM
24a
24b
CH3
n.i.@ 200 µM
n.i.@ 200 µM
C5H11
25
39% @ 200 µM
289 ± 127
4,6ꢀ
dimethylpyrimidin
ꢀ2ꢀyl
4ꢀhydroxyꢀ6ꢀ
propylꢀpyrimidinꢀ
2ꢀyl
26a
26bb
27
16% @ 200 µM
n.i.@ 200 µM
77.8 ± 7.62
n.i. @ 50 µM
18% @ 50 µM
35% @ 50 µM
28% @ 200 µM
23% @ 200 µM
28
29
n.i.@ 200 µM
33.0 ± 15.4
ACS Paragon Plus Environment
12

Page 13 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
30
31
n.i.@ 200 µM
n.i.@ 200 µM
24% @ 250 µM
n.i. @ 50 µM
7
8
9
32
n.i.@ 200 µM
n.i. @ 50 µM
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aCompound previously published by our group30
bSirtuin inhibitor presented by Kazantsev et al.39
cn.i.: inhibition @ conc. [µM] < 10%
Among the aminothiadiazoles, compound 13a, bearing a 1ꢀnaphthylmethyl and a 4,6,ꢀ
dimethylpyrimidine moiety, displays the most potent hSirt2 inhibition. This is consistent with the
previously published SirRealꢀmediated inhibition of hSirt2.30 However, a switch from the
aminothiazole to the aminothiadiazole scaffold was shown to lead to a loss in potency, and was
therefore not further investigated. Furthermore, we observed a substantial decrease in potency for
those compounds that strongly deviate from the original scaffold (20a-d, 24a-b, 25, 26a-b and
27-32, Table 1). Thus, we hypothesize that the unique mechanism of SirRealꢀmediated hSirt2
inhibition is restricted to compounds that display a high extent of similarity to the parent
compounds 9 or 10. By disassembling our lead structure 10 into fragments (24a, 30-32, Table 1),
we clearly demonstrate that the presence of all three functional elements, namely: arylmethyl,
aminothiazole, and 4,6ꢀdimethylpyrimidine is crucial for efficient inhibition of hSirt2 activity.
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 47
1
2
3
4
Table 2. In vitro inhibition of hSirt1 and hSirt2 by aminothiazoles 9-10, 14a-r, 19a-q
5
6
7
8
9
Shared scaffold:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
hSirt1 inhibition
%@conc. [µM]
or IC50 ± SE [µM]
15% @ 200 µM
n.i.@ 200 µM
29% @ 200 µM
17% @ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µMb
n.i.@ 200 µM
n.i.@ 200 µM
20% @ 200 µM
35% @ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
21% @ 200 µM
17% @ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
22% @ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
n.i.@ 200 µM
18% @ 200 µM
n.i.@ 200 µM
16% @ 200 µM
27% @ 200 µM
12% @ 200 µM
14% @ 200 µM
29% @ 200 µM
10% @ 200 µM
14% @ 200 µM
n.i.@ 200 µM
17% @ 200 µM
11% @ 200 µM
18% @ 200 µM
18% @ 200 µM
11% @ 200 µM
n.i.@ 200 µM
hSirt2 inhibition
compd
Aryl
R1
R2
R3
R4
R5
%@conc. [µM]
or IC50 ± SE [µM]
3.75 ± 0.83
0.44 ± 0.08
0.18 ± 0.02
0.42 ± 0.04
0.26 ± 0.03
9.77± 4.78
45.6 ± 24.6
0.21 ± 0.02
0.31 ± 0.01
0.32 ± 0.05
0.48 ± 0.05
0.54 ± 0.06
0.54 ± 0.08
1.45± 0.18
1.92 ± 0.42
2.34 ± 0.42
15.0 ± 2.11
65.0 ± 31.7
127.2± 13.4
>100
35% @ 50 µM
25% @ 50 µM
1.33 ± 0.15
1.64 ± 0.40
3.40 ± 0.49
4.72 ± 5.55
14.6 ± 9.24
16.8 ± 4.96
33.0 ± 13.8
164.5 ± 23.4
31% @ 50 µM
52% @ 50 µM
48% @ 50 µM
48% @ 50 µM
46% @ 50 µM
40% @ 50 µM
38% @ 50 µM
n.i. @ 50 µM
n.i. @ 50 µM
9a
phenyl
H
H
H
H
CH3
CH3
CH3
CH3
CH3
CH3
C2H5
CH3
CH3
CH3
CH3
CH3
CH3
CH3
H
H
CH3
CH3
CH3
CH3
CH3
CH3
C2H5
CH3
CH3
CH3
CH3
CH3
CH3
H
10a
naphthalenꢀ1ꢀyl
7ꢀchloronaphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
7ꢀbromonaphthalenꢀ1ꢀyl
2ꢀmethylnaphthalenꢀ1ꢀyl
7ꢀchloronaphthalenꢀ1ꢀyl
6ꢀchloronaphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
7ꢀchloronaphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ2ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
naphthalenꢀ1ꢀyl
3ꢀethoxyphenyl
3ꢀmethylphenyl
4ꢀchlorophenyl
4ꢀmethylphenyl
4ꢀmethoxyphenyl
phenyl
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
CH3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
CH3
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
14a
rac-14b
(S)-14b
(R)-14b
14c
14d
14e
rac-14f
14g
rac-14h
rac-14i
14j
CH3
(S)ꢀCH3
(R)ꢀCH3
H
H
H
C2H5
H
C2H5
CH3
H
rac-14k
CH3
H
H
H
14la
H
14m
14n
H
H
H
H
H
H
H
H
CH3
CH3
OH
CH3
OH
CH3
CH3
C3H7
CH3
CH3
NH2
CH3
CH3
CH3
CH3
CH3
H
CH3
CH3
C2H5
CH3
CH3
C2H5
C6H5
C2H5
C2H5
H
14o
14pa
14q
14r
19a
19b
19c
NH2
CH3
CH3
CH3
CH3
CH3
H
CH3
CH3
C2H5
CH3
CH3
C2H5
C6H5
C2H5
C2H5
H
H
H
19d
rac-19e
CH3
H
CH3
H
19fa
rac-19g
19h
19i
19j
19k
4ꢀbiphenyl
4ꢀbiphenyl
4ꢀbiphenyl
H
H
H
H
2ꢀmethylphenyl
4ꢀmethoxyphenyl
4ꢀmethoxyphenyl
4ꢀchlorophenyl
4ꢀmethoxyphenyl
4ꢀchlorophenyl
2ꢀmethylphenyl
3ꢀmethylphenyl
19l
19m
rac-19n
19o
19p
19q
H
CH3
H
H
H
C2H5
C2H5
aCompounds previously published by our group30
bn.i.: inhibition @ conc. [µM] < 10%
Analyzing the data of Table 2, we can confirm the findings of our preliminary SAR model,30
highlighting the importance of the 4,6ꢀdimethylpyrimidine moiety for the binding of the ligand to
ACS Paragon Plus Environment
14

Page 15 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the newly formed ‘selectivity pocket’. On the one hand, we show that ligands lose their affinity
without the methyl groups at the pyrimidine ring in position 4 and/or 6 (14j-l), on the other hand
we reveal that a higher degree of methylation (14m) or bulkier substituents (14c, 19l, 19nꢀo)
cause a decrease in potency, as predicted by molecular docking studies (Figure 2d). Furthermore,
we could show that polar substituents at the pyrimidine ring, e.g. ꢀOH or ꢀNH2 (14q-r), are not
beneficial in terms of potency. The inhibition data of the novel ligands with 2ꢀnaphthyl (14n),
biphenyl (19g-i) or substituted phenyl moieties (19a-e, 19j-q), which were generated to study the
impact of structural changes of the arylmethyl moiety, clearly indicates the superiority of the 1ꢀ
naphthyl substituted SirReals. We were able to rationalize the observed loss of in vitro potency
caused by replacing the 1ꢀnaphthyl moiety with other aryl substituents e.g. biphenyl, by means of
docking studies. These indicate a steric clash with the residues Val233, Phe234 and Phe235 of
the binding pocket (data not shown). Comparing the substituted phenyl ligands with their parent
compound 9, we conclude that a substitution in position 3 is beneficial in terms of hSirt2
inhibition (19a-b), whereas other substitution patterns lead to a decrease in potency (e.g. 19c-d,
19j). This observation is reflected by the docking results (Table S1) where we propose that the 4ꢀ
position of the phenyl ring is unfavorable for substitution due to the close proximity of Phe234,
whereas in the 3ꢀposition the substituent protrudes into the acylꢀlysine channel (not shown). As
predicted by molecular docking, an alkylation in the αꢀposition of the amide leads to increased
potency of the (S)ꢀenantiomer in the case of the methyl group ((S)-14b). Compared to racꢀ14b,
an ethyl group already led to a decrease in potency (rac-14h). As the ethyl compounds (rac-14h,
racꢀ14f) additionally turned out to be poorly soluble, we therefore did not separate the
enantiomers of those compounds. Most importantly, however, we improved the in vitro activity
of our lead structure by a chlorination or bromination of the naphthyl residue in position 7 (14a,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 47
1
2
3
4
5
6
7
8
9
14d). We were able to rationalize these results by binding free energy studies in combination
with molecular docking (See Methods section for further details). Using docking poses
calculated with the program GLIDE (Schrödinger LLC, New York, USA) and subsequent
refinement using the program AMBER and a GBSA solvation model implemented in MOE
2012.10 (Chemical Computing Group, Montreal, Canada) we observed a correlation between the
experimental pIC50 values and the calculated binding energies EGBSA (r2=0.67, RMSE 0.60,
q2LOO=0.62, Figure S1). An improvement of the model was derived by including two topological
descriptors describing the ligand structures (“diameter” and surface descriptor “PEOE_VSA4”,
see Methods section). The resulting model showed the following equation:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
pIC50 = ꢀ1.474 – 0.192 “diameter” – 0.032 “PEOE_VSA4” – 0.195 “EGBSA
”
“Diameter” is the largest value in the distance matrix dimension of the molecules and is an
indicator of the dimension of the molecules. Increasing the “diameter” of the inhibitors is
unfavorable as well as increasing the surface descriptor “PEOE_VSA4”. The final quantitative
structure activity relationship (QSAR) model is able to rationalize the observed in vitro activities
of the developed inhibitors (r2=0.81, RMSE 0.45, q2LOO=0.76, Figure S2). The QSAR model was
further tested by 10ꢀfold cross validation using randomly selected 20% of the compounds as test
set. The calculated q2L20%O value of 0.74 supports the robustness of the model.
Moreover, we were able to rationalize these results with a crystal structure of hSirt2 in complex
with 14d and NAD+ (Figure 3).
ACS Paragon Plus Environment
16

Page 17 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. 14d of the hSirt2ꢀ14dꢀNAD+ complex binds to hSirt2 in a similar fashion as observed
for 10 of the hSirt2ꢀ10ꢀNAD+ complex. (a) Superposition of the hSirt2ꢀ10ꢀNAD+ complex
(cartoon: slate blue, PDBꢀID 4RMG) with the crystal structure of hSirt2ꢀ14dꢀNAD+ (cartoon:
pale green; 14d: raspberry sticks; NAD+: pale yellow sticks). Both complexes adopt the ‘locked
open’ꢀconformation (RMSD of all Cαꢀatoms: 0.33 Å). (b) 14d and NAD+ of the hSirt2ꢀ14dꢀ
NAD+ complex assume a very similar position within the active site of hSirt2 as 10 and NAD+ of
the hSirt2ꢀ10ꢀNAD+ structure. (c) 14d as well as NAD+ are wellꢀdefined by the electron density.
σꢀweighted iterative OMIT maps are shown as green mesh and contoured at 3.0σ. (d)
Interactions of 14d with hSirt2. Interacting residues are represented as sticks (green). Phe190,
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 47
1
2
3
4
5
6
Ile232 are not labeled and Leu206 is not shown for the sake of clarity. The water molecule is
represented as a black sphere and hydrogen bonds as dashes yellow lines.
7
8
9
The crystal structure reveals that 14d, as well as NAD+, assume a very similar position within the
active site of hSirt2 compared to 10 and NAD+ of the hSirt2ꢀ10ꢀNAD+ structure (Figure 3aꢀb).
We were not able to observe a σꢀhole interaction in our coꢀcrystal structure as suggested by the
docking study. Binding of 14d is mainly driven by hydrophobic interactions with the side chains
of the residues forming the extended Cꢀsite (ECS), the acylꢀlysine binding site as well as the
‘selectivity pocket’ at the hinge region. The 7ꢀbromonaphthyl substituent fills the lipophilic
naphthyl binding site more efficiently than the unsubstituted naphthyl moiety, and thereby allows
further hydrophobic interactions. The carbonylꢀO of the amide also forms a waterꢀmediated
hydrogen bond to the backbone carbonylꢀO of Pro94. As it was already observed in the crystal
structure of hSirt2ꢀ10ꢀNAD+, 14d also forms an intramolecular hydrogen bond between the
amide NꢀH and one of the nitrogen atoms of the dimethylpyrimidine ring. (Figure 3d). To see,
whether the combination of beneficial substitution patterns would lead to a further increase in
potency, we combined a 7ꢀchloro substituent with an αꢀmethylation in the amide part. Yet, we
did not detect an additive effect by combining a 7ꢀhalonaphthyl moiety with a methylation in the
αꢀposition of the amide (rac-14i).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
18

Page 19 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To assess the cellular activity of the 7ꢀhalonaphthyl derivatives, we utilized immunofluorescence
microscopy and western blot experiments, showing an enhanced tubulin hyperacetylation for
14a, when compared to DMSO control (Figure 4, for raw images see Figure S3, for western blots
Figure S4).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Optimized aminothiazole 14a induces tubulin hyperacetylation in cultured HeLa cells.
Acetylation level of the microtubule network (red) in the presence or absence of sirtuin inhibitors
(10 µM). Treatment of HeLa cells with 14a leads to higher acetylation levels of the microtubule
network as compared to the DMSO treated cells (negative control). 10 was used as a positive
control. Nuclei were DAPIꢀstained (blue). The scale bar represents 10 µm.
Treatment with the optimized SirReal derivative 14a leads to hyperacetylation of the microtubule
network in a similar manner as observed for 10, which was used as a positive control. While
compounds 14a and 10 both lead to a significant increase in tubulin hyperacetylation at a
concentration of 20 µM, 14a also induces a significant gain in tubulin acetylation at only 10 µM
(Figure S4). Thus, we could show that the improved in vitro potency of 14a is also relevant
under physiological conditions. Of note, the halogenated derivatives were badly soluble in cell
culture media and may therefore be limited in their efficacy at higher concentrations.
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 47
1
2
3
4
5
6
7
8
9
DISCUSSION AND CONCLUSION: Starting out from compounds 9 and 10 as potent and
selective hSirt2 inhibitors, we established a synthesis platform to systematically probe the limits
of modification within this scaffold. We were able to elaborate a wellꢀdefined SAR model for
both, the arylmethyl and the pyrimidine moiety. Guided by the structural knowledge, revealed by
hSirt2ꢀ9/10 coꢀcrystals as well as molecular docking studies, we were able to improve the in
vitro hSirt2 inhibition of both lead structures, 9 and 10. Moreover, the cellular activity of the
improved aminothiazole inhibitor 14a was validated by tubulin hyperacetylation in HeLa cells.
In combination with the herein reported coꢀcrystal structure, our SAR model will be the
foundation for further developments of the Sirtuin Rearranging Ligands as valuable biological
tool compounds to gain deeper insight into sirtuin biology and to probe the druggability of
sirtuins.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
20

Page 21 of 47
Journal of Medicinal Chemistry
1
2
3
4
EXPERIMENTAL SECTION:
5
6
1. Protein expression and purification: hSirt1133ꢀ747 was expressed as a GSTꢀtagged enzyme
7
8
9
41
and purified as described previously.40 hSirt225ꢀ389 was expressed Nꢀterminally tagged with His6
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
with minor modifications.40 hSirt256ꢀ356 was expressed and purified according to Rumpf et al.30
2. In Vitro Testing: Potency of hSirt1 and hSirt2 inhibition was determined with a fluorescenceꢀ
based homogeneous assay using the substrate ZMAL (CbzꢀLys(acetyl)ꢀAMC).38 hSirt1133ꢀ747 or
hSirt225ꢀ389 were mixed with assay buffer (50 mM Tris, 137 mM NaCl, 2.7 mM KCl, pH 8.0),
NAD+ (final assay concentration 500 ꢁM), the substrate ZMAL (final assay concentration 10,5
ꢁM), the inhibitor dissolved in DMSO at different concentrations, or DMSO only as a control
(final DMSO concentration 5% (v/v)). To ensure initial state conditions, total substrate
conversion of controls was adjusted to approximately 15% ꢀ 30%. The assay was performed in
96ꢀwell plates with a reaction volume of 60 ꢁL per well. All determinations were performed at
least in duplicates. After an incubation of 4 h at 37 °C and 140 rpm, deacetylation reaction was
stopped by the addition of 60 ꢁL of a stop solution containing trypsin and nicotinamide (50 mM
Tris, 100 mM NaCl, 6.7% (v/v) DMSO, trypsin 5.5 U ꢁLꢀ1, 8 mM nicotinamide, pH 8.0). The
reaction mixture was further incubated for 20 min at 37 °C and 140 rpm. Fluorescence intensity
was measured in a microplate reader (BMG Polarstar, λex = 390 nm, λem = 460 nm). Rates of
inhibition were determined by using the controls, containing no inhibitor, as a reference.
Graphpad Prism software (La Jolla, CA) was employed to determine IC50 values.
3. Data collection, structure solution and refinement: Data were collected at 100 K at X06SA
beamline of the Swiss Light Source (Villigen, Switzerland) equipped with a Pilatus 6M detector
at a wavelength of 1.0 Å with oscillations of 0.5°. Data were processed with XDS42 and scaled
based on the CC1/2 criterion43 using Aimless.44 Data collection statistics are shown in Table 5.
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 47
1
2
The structure was solved by molecular replacement with MOLREP45 using the hSirt2ꢀ10ꢀNAD+ꢀ
complex (PDBꢀID 4RMG)30 as a search model. The structural model was built in Coot46 and
refined with REFMAC.47 14d was generated with the Grade Web Server (Global Phasing Ltd.,
United Kingdom) and placed into 2FoꢀFc electron density maps using AFITTꢀCL (Version 2.1.0,
OpenEye Scientific Software, Inc., Santa Fe, NM, USA.). All residues except Pro99, Ser100 and
Thr101 were included in the model. The Nꢀterminal glycine, histidine and methionine originate
from the TEV cleavage site and the NdeIꢀrestriction site of the modified pET15bꢀ expression
vector. The structure was validated using the Molprobity server48 and PROCHECK.49 σꢀ
weighted iterative OMIT maps were generated with Phenix.50 RMSD values were determined
with SUPERPOSE51 and images were prepared with Pymol (The Pymol Molecular Graphics
System, Version 1.7, Schrödinger, LLC).
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 5. Data collection and refinement statistics
hSirt2-14d-NAD+
Data processing
PDB accession #
Spacegroup
5DY4
I2
a, b, c (Å)
84.30, 55.43, 96.18
90, 114.91, 90
46.78–1.77 (1.81–1.77)
38622 (2201)
260647 (15296)
98.2 (98.1)
α, β, γ (°)
Resolution (Å)
Unique observations
Observations
Completeness (%)
Multiplicity
6.7 (6.9)
[1]
Rmerge
0.095 (2.050)
10.3 (1.0)
I / σI
CC1/2
0.998 (0.554)
ACS Paragon Plus Environment
22

Page 23 of 47
Journal of Medicinal Chemistry
1
2
3
4
Refinement
5
6
Resolution (Å)
46.78–1.77 (1.81–1.77)
7
8
9
No. Amino acids
300
2649
2391
30
No. Atoms
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Protein
14d
NAD+
44
Waters
183
Zn2+
1
Rcryst/Rfree (%) [2]
B factors (Å2)
17.5/21.3
Protein
33.71
27.88
29.05
38.64
27.58
0.013
1.65
14d
NAD+
Waters
Zn2+
RMSD bond length (Å)
RMSD angles (°)
Ramachandran plot statistics
Most favoured region (%)
Additional allowed region (%)
Generously allowed region (%)
93.2
6.5
0.3
Disallowed region (%)
0.0
Values in parentheses represent the highest resolution shell.
[1]
R
R
= ∑hkl [(∑i |Ii ꢀ <I>|) / ∑iIi]
merge
[2]
= ∑hkl ||Fobs| ꢀ |Fcalc|| / ∑hkl |Fobs|
cryst
Rfree is the crossꢀvalidation R factor computed for a test set of 5 % of unique reflections
selected randomly.52 Ramachandran statistics as defined by PROCHECK.49
4. Protein crystallization: Crystallization assays were set up with the Oryx Nano pipetting robot
(Douglas Instruments, United Kingdom) using the vapor diffusion sitting drop method (Intelliꢀ
Plate 96ꢀ3 Low Profile, Art Robbins Instruments, USA) at 4 °C. Prior to crystallization, hSirt256ꢀ
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 47
1
2
(20 mg mLꢀ1) was incubated with 14d (100 mM stock solution in DMSO, 1% (v/v) final
3
4
356
5
6
7
8
DMSO concentration) and NAD+ (100 mM stock solution in 25 mM Tris/HCl, 150 mM NaCl,
pH 8.0, final concentration 10 mM, SigmaꢀAldrich, Germany). Crystals of the hSirt2ꢀ14dꢀ
complex were obtained after 2 days in a solution containing 27% (w/v) PEG 3350 in 0.05 M
HEPES buffer at pH 7.0 with a protein solution to reservoir ratio of 3:1. The crystal was
mounted on a nylon loop and cryoprotected by the addition of 20% (v/v) glycerol.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5. Cell Culture: HeLa (ATCCꢀ2) cells were cultured in Dulbecco’s Modified Eagle’s Medium
(high glucose) supplemented with 10% (v/v) fetal calf serum, 100 µg/ml kanamycin (all reagents
from SigmaꢀAldrich) in a humidified incubator at 37 °C with 5% CO2. For microscopic analysis,
0.5 x 104 cells were seeded on 12 mm diameter coverslips placed in 24ꢀwell plates and incubated
overnight before the experiment. For immunoblotting, 1.5 x 104 cells were seeded per well of 24ꢀ
well plates and incubated overnight. The drugs were added to cells for 1 hour from 10 mM stock
solutions in DMSO. Controls contained the corresponding amount of vehicle (DMSO).
6. Immunoblotting: For the detection of acetylated tubulin, total tubulin and glyceraldehydeꢀ3ꢀ
phosphate dehydrogenase (GAPDH) levels in cellular samples, we kept the cells at 37 °C and
washed them with prewarmed PBS. Next, the cells were lysed in 100 µL 1x reducing sample
buffer containing protein inhibitor mix and 2 mM EDTA (SigmaꢀAldrich). Samples were
centrifuged at 10,000 g at 4°C for 5 min and the supernatants were stored at ꢀ70 °C. Samples
were analyzed by SDSꢀPAGE and blotted onto polyvinylidene difluoride membrane (Millipore).
The blot was developed sequentally using a monoclonal mouse antibody against acetylated
alphaꢀtubulin at Lysꢀ40 (1:5000, clone 6ꢀ11Bꢀ1), than using a monoclonal mouse antibody
against alphaꢀtubulin (1:5000, clone DM1A), next using a monoclonal mouse antibody against
GAPDH (1 µg/mL, CB1001, clone 6C5, Calbiochem). Antibodies were detected by antiꢀmouse
ACS Paragon Plus Environment
24

Page 25 of 47
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
IgGꢀperoxidase conjugate (Fcꢀspecific), (1:5000, SigmaꢀAldrich). Peroxidase reaction detected
using Immobilon Western substrate (Millipore) by a BioꢀRad ChemiDoc MP Imaging system
and its ImageLab 4.1 software. Intensity of spots was analysed by ImageJ 1.49 using Measure
command and subtracting background values. The sample values were normalized by the
average control value on the corresponding blot.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7. Immunofluorescence microscopy: Immunofluorescence microscopy was performed as
previously reported.30
8. Computational Methods: 3D structures of all compounds in this study were generated from
SMILES strings, and a subsequent energy minimization was carried out using the MMFF94x
force field implemented in Molecular Operating Environment System (MOE) 2012.10
(Chemical Computing Group, Montreal, Canada). All compounds were used in the protonation
state at physiological level. A maximum of 100 conformations were generated for each ligand
using the Conformational Search module implemented in MOE. All protein structures were
prepared by using the Structure Preparation module in MOE. Hydrogen atoms were added and
the protonation state for titratable amino acids was calculated using the Protonate 3D module in
MOE. Protein structures were energy minimized using the AMBER99 force field53 with a
tethering force constant of (3/2) kT / 2 (σ = 0.5 Å) for all atoms during the minimization. AM1ꢀ
BCC charges were used for ligands.54 All molecules except the zinc ion were removed from the
structures. Proteinꢀligand docking was performed using program GLIDE (Suite 2012ꢀ5.8,
Schrödinger LLC, New York, USA). The position of the inhibitor 10 in its crystal structure with
hSirt2/AcꢀLysꢀH3 peptide (PDB ID 4RMH) was used to define the size of the grid box (10 Å
radius). Docking was performed using GLIDEꢀExtra Precision (XP). The ligand was treated as
flexible and 20 docking poses were calculated for each inhibitor. All other options were left at
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 47
1
2
3
4
5
6
7
8
9
their default values. The topꢀranked pose from each docking run was included in the final
analysis and viewed graphically together with the protein structure using the program MOE
2010.13 (Chemical Computing Group, Montreal, Canada). Using the docking setup, 10 could be
correctly docked into its crystal structure with RMSD values below 0.5 Å (PDB ID 4RMG and
4RMH).30 Also, other active aminothiazole derivatives were predicted to adopt a similar binding
mode as was observed for 10. Binding free energies for the inhibitors in this study were
calculated using the topꢀranked docking poses. Structurally conserved water molecules included
for docking studies were maintained during the geometry optimization of the complexes. The
proteinꢀinhibitor complexes were energy minimized using the AMBER PFROSST force field and
the GBSA solvation model implemented in MOE 2012.10. Using the resulting ꢀEGBSA value as
descriptor, a significant correlation to the pIC50 values was observed (r2=0.67, RMSE=0.60) for
the 30 compounds for which an IC50 and exact stereochemistry was determined (Figure S1).
Besides the correlation coefficient r2, the model was also tested using leave one out (LOO) crossꢀ
validation (q2LOO=0.62). Furthermore, we investigated the effect of other descriptors like ligand
charge, diameter, or polar surface descriptors for establishing a QSAR model. We successfully
applied this approach recently for establishing predictive models with high robustness.55 Among
192 tested 2D descriptors in MOE, the “diameter” and the topological surface descriptor
“PEOE_VSA+4” showed the highest correlation. A QSAR model based on three descriptors,
namely “ꢀEGBSA score”, “diameter”, and “PEO_VSA+4” using MOE PLS methodology was
generated and validated. The final PLS model yielded a correlation coefficient of r2=0.81
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(RMSE=0.45) and a q2
of 0.76 demonstrating the robustness of the model (Figure S2, Table
LOO
S1). Since LOO crossꢀvalidation is sometimes misleading and resulting in too optimistic models,
a more demanding crossꢀvalidation procedure was applied for the QSAR model. 10ꢀfold cross
ACS Paragon Plus Environment
26

Page 27 of 47
Journal of Medicinal Chemistry
1
2
3
4
validation was carried out using randomly selected compounds (20%) as test set. Repeating this
5
6
random splitting 10 times, resulted in a mean q2
value of 0.74, demonstrating the
L20%O
7
8
9
robustness of the model. All results are summarized in Table S2 in the Supplementary
Information.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9. Chemistry: Starting materials and reagents were obtained from commercial suppliers and
used without further purification. Thinꢀlayer chromatography (TLC) for reaction monitoring was
performed with alumina plates coated with Merck silica gel 60 F254 (layer thickness: 0.2 mm)
and analyzed under UVꢀlight (254 nm). A mixture of ethyl acetate and cyclohexane (2:1) was
used as mobile phase. If the purity of the synthesized compounds was not adequate, we
performed flash column chromatography with TELOS FlashꢀLL Silica Columns 60M (0.040ꢀ
0.063 mm, 230ꢀ400 mesh) as a stationary phase on a Biotage Isolera One automated flash
purification system with UVꢀVis detector. Cyclohexane and ethyl acetate was used as mobile
phase and gradient was adjusted based on TLC results. Yields were not optimized. 1HꢀNMR and
13CꢀNMR spectra were recorded on Bruker Avance III HD spectrometer at 400 MHz and 100
1
MHz. The spectra are referenced against the NMR solvents and are reported as follows: H:
chemical shift δ (ppm), multiplicity (s = singlet, d = doublet, dd = doublet of doublets, t = triplet,
q = quartet, quint = quintet, sex = sextet, m = multiplet, b = broad), integration, coupling
constant (J in Hz). 13C: chemical shift δ (ppm), abbreviations: (q) = quaternary carbons,
quaternary carbons that could not be found in 13C spectra but in HMBC or HSQC are
additionally marked with an asterisk (*). The assignment resulted from HMBC and HSQC
experiments. Purity was determined for all tested compounds by HPLC and UV detection (λ =
210 nm) and was > 95%. HPLC analysis was performed using the following conditions: Eluent
A: H2O containing 0.05% TFA, Eluent B: acetonitrile containing 0.05% TFA, Eluent C: nꢀ
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 47
1
2
hexane, Eluent D: propanꢀ2ꢀol, flow rate 1 mL minꢀ1. Method 1.1 (M1.1), analytical column,
Phenomenex SynergiTM 4 µm MAXꢀRP 80 Å, 150 x 4.6 mm, isocratic conditions (A = 55%, B =
45%). Method 1.2 (M1.2), analytical column, Phenomenex SynergiTM 4 µm HYDROꢀRP 80 Å,
250 x 4.6 mm. Method 1.3 (M1.3), analytical column, Phenomenex SynergiTM 4 µm POLARꢀRP
80 Å, 150 x 4.6 mm. Method 2.1 (M2.1), analytical column, Phenomenex SynergiTM 4 µm
MAXꢀRP 80 Å, 150 x 4.6 mm, isocratic conditions (A = 40%, B = 60%). Method 2.2 (M2.2),
analytical column, Phenomenex SynergiTM 4 µm POLARꢀRP 80 Å, 150 x 4.6 mm. Method 3
(M3), analytical column, Phenomenex SynergiTM 4 µm HYDROꢀRP 80 Å, 250 x 4.6 mm,
isocratic conditions (A = 30%, B = 70%). Method 4 (M4), analytical column, Phenomenex
SynergiTM 4 µm HYDROꢀRP 80 Å, 250 x 4.6 mm, isocratic conditions (A = 65%, B = 35%).
Method 5 (M5), analytical column, Phenomenex SynergiTM 4 µm HYDROꢀRP 80 Å, 250 x 4.6
mm, isocratic conditions (A = 5%, B = 95%). Method 6.1 (M6.1), analytical column,
Phenomenex SynergiTM 4 µm HYDROꢀRP 80 Å, 250 x 4.6 mm, linear gradient conditions (0−4
min, A = 90%, B = 10%; 4−29 min, linear increase to 100 % of B; 29−31 min, B = 100%; 31−40
min, A = 10%, B = 90%). Method 6.2 (M6.2), analytical column, Phenomenex SynergiTM 4 µm
MAXꢀRP 80 Å, 150 x 4.6 mm. Method 6.3 (M6.3), analytical column, Phenomenex SynergiTM
4 µm HYDROꢀRP 80 Å, 250 x 4.6 mm. Method 7 (M7), analytical column, Phenomenex LuxTM
5 µm CELLULOSEꢀ1, 250 x 4.6 mm, linear gradient conditions (0−4 min, C = 90%, D = 10%;
4−30 min, linear increase to C = 50%, D = 50%; 30−36 min, linear gradient to C=90%, D=10%,
36–40 min C=90%, D=10%. Melting temperatures were determined in glass capillary tubes with
the Stuart Melting Point Apparatus SMP2. Optical rotation was measured with a PerkinElmer
Model 341 Polarimeter. Mass spectra with electrospray ionization (ESI) were recorded on an
Advion expression CMS spectrometer, with electron ionization (EI) on an Agilent Technologies
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
28

Page 29 of 47
Journal of Medicinal Chemistry
1
2
6890 N Network GCꢀMS system. The synthesis of compounds 16e,35a 22b,36 23b,37 33,56 has
already been reported. Due to changes in the experimental procedure and to show unpublished
characterization data, we have outlined the synthesis and the characterization data for those
compound as well.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
General Procedure for the Synthesis of 1,3,4ꢀThiadiazolꢀ2ꢀamines (11a-e):31 A wellꢀstirred
mixture of thiosemicarbazide (1 equiv, 25 mmol), the carboxylic acid (1.2 equiv), and 8 mL of
concentrated sulphuric acid was slowly heated to 80 – 90 °C and maintained at this temperature
for 7 hours. After cooling, the reaction mixture was poured into ice water and was treated with
concentrated ammonia to pH = 12. The crude product, which precipitated upon the addition of
the ammonia, was filtered off and washed with water.
5ꢀ(1ꢀNaphthylmethyl)ꢀ1,3,4ꢀthiadiazolꢀ2ꢀamine (11b):57 From thiosemicarbazide and 1ꢀ
1
naphthylacetic acid. Yield: 3% of a beige solid. Rf : 0.15; HꢀNMR (DMSOꢀD6, δ [ppm]):
8.16ꢀ8.09 (m, 1H, naphthyl Hꢀ8), 7.98ꢀ7.92 (m, 1H, naphthyl Hꢀ5), 7.90ꢀ7.82 (m, 1H, naphthyl
Hꢀ4), 7.60ꢀ7.43 (m, 4H, naphthyl Hꢀ2,3,6,7), 6.97 (bs, 2H, ꢀNH2), 4.62 (s, 2H, ArꢀCH2ꢀAr);
13CꢀNMR (DMSOꢀD6, δ [ppm]): 169.01 q (aminothiadiazole Cꢀ2), 158.20 q (aminothiadiazole
Cꢀ5), 134.54 q (naphthyl Cꢀ1), 133.93 q (naphthyl Cꢀ4a), 131.66 q (naphthyl Cꢀ8a), 128.99
(naphthyl Cꢀ5), 128.18 (naphthyl Cꢀ4), 127.53 (naphthyl Cꢀ2), 126.74 (naphthyl Cꢀ7), 126.32
(naphthyl Cꢀ6), 126.11 (naphthyl Cꢀ3), 124.30 (naphthyl Cꢀ8), 33.80 (ArꢀCH2ꢀAr); ESIꢀMS(+):
242.1 [M+H]+
General Procedure for Acylation of Thiazolꢀ2ꢀamines and 1,3,4ꢀThiadiazolꢀ2ꢀamines to
generate amides (12a-g, 17a-j, 18a-i, 24a-b, 31, 37, 43, 44):32 To a solution of the amine
(1 equiv, 2 mmol), dissolved in 10ꢀ30 mL acetonitrile, N,Nꢀdiisopropylethylamine (1.5 equiv)
was added. The mixture was stirred and cooled to 0 °C. The acyl chloride (1.5 equiv) was
ACS Paragon Plus Environment
29