Structure-based design of pseudopeptidic inhibitors for SIRT1 and SIRT2

Article abstract of DOI:10.1021/jm200590k

The lack of substrate-bound crystal structures of SIRT1 and SIRT2 complicates the drug design for these targets. In this work, we aim to study whether SIRT3 could serve as a target structure in the design of substrate based pseudopeptidic inhibitors of SIRT1 and SIRT2. We created a binding hypothesis for pseudopeptidic inhibitors, synthesized a series of inhibitors, and studied how well the fulfillment of the binding criteria proposed by the hypothesis correlated with the in vitro inhibitory activities. The chosen approach was further validated by studying docking results between 12 different SIRT3, Sir2Tm, SIRT1 and SIRT2 X-ray structures and homology models in different conformational forms. It was concluded that the created binding hypothesis can be used in the design of the substrate based inhibitors of SIRT1 and SIRT2 although there are some reservations, and it is better to use the substrate-bound structure of SIRT3 instead of the available apo-SIRT2 as the target structure.

Full text of DOI:10.1021/jm200590k

ARTICLE
pubs.acs.org/jmc
Structure-Based Design of Pseudopeptidic Inhibitors for SIRT1 and
SIRT2
||
Tero Huhtiniemi,†,‡ Heikki S. Salo,*,†,‡ Tiina Suuronen,§ Antti Poso,† Antero Salminen,§,
Jukka Lepp€anen,† Elina Jarho,†,^ and Maija Lahtela-Kakkonen†
†School of Pharmacy and §Department of Neurology, Institute of Clinical Medicine, University of Eastern Finland, P.O. Box 1627,
70211 Kuopio, Finland
Department of Neurology, Kuopio University Hospital, P.O. Box 1777, 70211 Kuopio, Finland
^Department of Chemistry, Medicinal Chemistry, University of Gothenburg, 41296 Gothenburg, Sweden
S Supporting Information
b
ABSTRACT: The lack of substrate-bound crystal structures of SIRT1
and SIRT2 complicates the drug design for these targets. In this work,
we aim to study whether SIRT3 could serve as a target structure in the
design of substrate based pseudopeptidic inhibitors of SIRT1 and
SIRT2. We created a binding hypothesis for pseudopeptidic inhibi-
tors, synthesized a series of inhibitors, and studied how well the
fulfillment of the binding criteria proposed by the hypothesis corre-
lated with the in vitro inhibitory activities. The chosen approach was
further validated by studying docking results between 12 different
SIRT3, Sir2Tm, SIRT1 and SIRT2 X-ray structures and homology
models in different conformational forms. It was concluded that the created binding hypothesis can be used in the design of the
substrate based inhibitors of SIRT1 and SIRT2 although there are some reservations, and it is better to use the substrate-bound
structure of SIRT3 instead of the available apo-SIRT2 as the target structure.
’ INTRODUCTION
In general, peptidic inhibitors may show advantages over small
molecules in terms of specificity and affinity for different
targets,20 but unmodified peptides do not possess druglike
properties; they are poorly bioavailable and vulnerable to non-
specific enzymatic degradation, which drastically limits their use
as pharmacological tools. To move from peptidic toward small
molecule inhibitors Suzuki et al. have recently presented a
pseudopeptide backbone Cbz-Lys-NH-Ph with Nε-3-ethoxy-3-
oxopropanoyl21 and the Nε-thioacetyl15 substitutions at the
lysine side chain. Furthermore, the activity of the latter com-
pound was demonstrated in cellular studies. Pseudopeptides
have potential to overcome the aforementioned problems with
peptides.
In order to develop potent pseudopeptidic inhibitors it is
important to validate essential substrateꢀenzyme interactions,
which may also define affinity of inhibitors. The available crystal
structures are still limited, and for SIRT1 only a comparative
model has been reported.22 For SIRT2, the only reported crystal
structure is in the apo form.23 The fact that the substrate binding
is associated with significant domain movements24,25 compro-
mises the value of the available SIRT2 and SIRT1 structures in
the structure-based design of inhibitors.
The mammalian sirtuin family of class III deacetylases
(SIRT1-SIRT7) catalyzes the nicotinamide adenine dinucleotide
(NAD+) dependent deacetylation of Nε-acetyllysine residues in
multiple protein substrates to nicotinamide, O0-acetyl-ADP-
ribose (OAADPR) and Nε-deacetylated lysine. The reversibility
of the acetylation state of sirtuin target proteins by acetylases and
deacetylases regulates diverse biological processes which have
implications in devastating age-related diseases, such as cancer
and metabolic, cardiovascular and neurodegenerative diseases.1ꢀ4
Thus, it is important to discover small molecular regulators and to
study sirtuin functions in cells. The most relevant therapeutic
prospects for SIRT1 and SIRT2 inhibitors are for the treatment of
cancer, either by inducing cell death or by preventing angio-
genesis.1,5ꢀ9 SIRT2 inhibitors have also been demonstrated to
rescue proteotoxicity in neurodegenerative diseases.10,11
Several substrate based peptides have been reported to inhibit
sirtuins. The inhibition was gained by replacing the acetyllysine
with Nε-thioacetyl-,12ꢀ18 Nε-selenoacetyl-,17 Nε-isovaleryl-,17
Nε-3,3-dimethylacryl-,17 or Nε-trifluoroacetyllysine.14,19 How-
ever, apart from the acetyllysine analogue, the peptidic inhibitors
can have various sequences (Figure 1),16 which reflects the fact
that sirtuins can deacetylate various substrates.4 Recently, we
demonstrated that a tripeptide is sufficient length for peptidic
SIRT1 inhibitors and that an Nα-acetylated tripeptide has
sufficient length for SIRT2 inhibition.16,17
Received: December 21, 2010
Published: September 07, 2011
r
2011 American Chemical Society
6456
dx.doi.org/10.1021/jm200590k J. Med. Chem. 2011, 54, 6456–6468
|

Journal of Medicinal Chemistry
ARTICLE
resin with 1 M NaOH (aq) in dioxane gave a free carboxylic acid,
and cleavage with a mixture of MeOH, DIPEA and DMF gave a
methyl ester. Compounds 27ꢀ30 were synthesized in solution
phase using appropriate amine and TBTU as a coupling reagent.
Reference compounds 1, 2, 3 and 25 were synthesized as
described previously.16,17
’ RESULTS AND DISCUSSION
Peptidic Inhibitors. We have previously shown that pentapep-
tide HKK(thioAc)LM (1) and tripeptide KK(thioAc)L (2) give
similar inhibitory activity against SIRT1 while compound 1 is a 24
times more potent SIRT2 inhibitor than compound 2.16 We have
also shown that Nα-acetylated tripeptide Ac-AK(thioAc)A (3)
gives good SIRT1 and SIRT2 inhibition.17 In order to get a
complete picture about the structure activity relationship (SAR) of
peptidic inhibitors before the design of pseudopeptidic ones, we
synthesized compounds 8 and 13ꢀ17 (Table 1).
The free carboxyl terminal of compound 3 was blocked as
methyl amide (8) and as methyl ester (13). While SIRT1 slightly
favored the free carboxyl terminal (3), SIRT2 inhibition was
improved by the additional amide bond at the C-terminus (8).
Methyl ester (13) with the least hydrogen bonding sites gave the
weakest inhibition of SIRT1 and SIRT2.
Because Nα-acetylated tripeptide Ac-AK(thioAc)A (3) had
turned out to be a surprisingly good SIRT2 inhibitor compared
to tripeptide KK(thioAc)L (2),16,17 we examined the inhibitory
activity of tripeptide AK(thioAc)A (14). As expected, it was
significantly more potent than compound 2 against SIRT2. This
further reinforces the picture that no specific side chains are
required for binding but, quite the contrary, some side chains can
disrupt it. However, the Nα-acetylated 3 was more potent than
compound 14.
Figure 1. Nε-Thioacetyllysine containing sirtuin inhibitors with various
peptide sequences.
The KthioAc+1 residue is known to be important for inhibitory
activity.16 We synthesized compounds 15 and 16 in order to
study the role of the KthioAcꢀ1 residue in binding. Both
compounds showed only weak or no inhibitory activity against
SIRT1 and SIRT2 compared to compound 14. It can be
concluded that both residues flanking the thioacetyllysine are
needed for binding.
In this work, we aim to study whether SIRT3 could serve as a
target structure in design, and help to screen scaffolds for
pseudopeptidic inhibitors of SIRT1 and SIRT2. Based on the
SIRT3 crystal structure, we have created a binding hypothesis for
pseudopeptidic inhibitors. We have synthesized a series of
pseudopeptidic SIRT1 and SIRT2 inhibitors and studied how
they fulfill the binding hypothesis. The best inhibitors have
comparable potency to the peptidic ones.
In the available X-ray structures cocrystallized with a substrate
peptide, the peptide backbone has a change in the orientation at
the KAcꢀ1 position (see for example Figure 3a), which may help
to stabilize the substrateꢀprotein interaction at the KAcꢀ1 and
KAcꢀ2 sites. Compound 17 with a conformationally restricted
Ac-Pro moiety at the N-terminal was found to be nearly
equipotent with pentapeptide 1 and compound 3 for SIRT1,
but slightly less potent for SIRT2. The slight preference for
SIRT1 over SIRT2 makes the usage of proline a valid way to
restrict conformational orientation of SIRT1 inhibitors.
Based on the peptidic inhibitors, it was concluded that (1) the
amide bonds at both sides of the thioacetyllysine need to be
preserved to ensure hydrogen bonding and proper orientation of
the thioacetyllysine, (2) no side chains of amino acids need to be
mimicked, and (3) in order to obtain SIRT2 inhibition, the pseudo-
peptides should stretch over larger area than tripeptides cover.
Substrate Binding Area and the Creation of the Binding
Hypothesis. Although there are no substrate-bound structures
for SIRT1 or SIRT2, the key interactions of the substrateꢀ
enzyme complex in the vicinity of the acetylated lysine resi-
due are well-defined for example in the crystal structures for
SIRT3 (PDB code 3glr)24 and Sir2Tm (PDB code 2h2d)25
’ CHEMISTRY
Synthesis routes and reaction conditions are shown in
Scheme 1. To obtain compound 8, the amino terminal of Nε-
Cbz-Lys-OH was protected with a 9-fluorenylmethoxycarbonyl
(Fmoc) group to give 4. Compound 4 was taken to solid phase
peptide synthesis (SPPS) on Wang resin and cleavage as a methyl
ester with a mixture of MeOH, N,N-diisopropylethylamine
(DIPEA) and N,N-dimethylformamide (DMF) gave 5. The
methyl ester was treated with methylamine (6), and the Cbz
group was removed by hydrogenation on palladium (7). Finally,
Nε-thioacetylation of the free amine with ethyl dithioacetate gave
compound 8. Similarly, the Nε-thioacetyl group was introduced
to Nα-Fmoc-Lys-OH HCl and Nα-Cbz-Lys-OH to gain com-
3
pounds 12 and 26 respectively. Compounds 13ꢀ24 were
synthesized with SPPS on Wang resin using appropriate car-
boxylic acids, O-(benzotriazol-1-yl)-N,N,N0,N0-tetramethyluro-
nium tetrafluoroborate (TBTU) and DIPEA in the coupling
phase and piperidine for the deprotection. Cleavage from the
6457
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Scheme 1a
a Reagents and conditions: (a) 9-fluorenylmethyl chloroformate, 0 ꢀC to rt, 4 h; (b) solid phase peptide synthesis with Fmoc strategy [(coupling)
appropriate carboxylic acid, TBTU, DIPEA, DFM, rt, 1 h; (Fmoc deprotection) 20% piperidine/DMF; (cleavage as ꢀCOOH) 1 M NaOH (aq) /
dioxane (2:6), rt, 1 h, filtration, 3 M HCl (aq) or (cleavage as ꢀCOOCH3) DIPEA/MeOH/DMF (1:5:5), heat at reflux, overnight]; (c) 8 M
methylamine in EtOH, 100 ꢀC, 20 min; (d) 10% Pd on charcoal, MeOH, rt, overnight; (e) ethyl dithioacetate, EtOH/10% (w/v) Na2CO3 (aq) (3:1), rt,
12ꢀ48 h; (f) appropriate amine, TBTU, DMF/pyridine (1:1), 0 ꢀC to rt, 1ꢀ2 h.
main chain and the backbone of protein residues Gly295,
Glu296, Glu323 and Glu325 (SIRT3 numbering) (Figure 2a).
The reactive acetyllysine is oriented toward the NAD+ binding
site by a hydrogen bond of the acetyl carbonyl with Val292. In
addition, at the KAcꢀ2 position, the substrate potentially has a
water molecule mediated hydrogen bond with Leu298.
Table 1. SIRT1 and SIRT2 Inhibitory Activities of the
Peptidic Inhibitors
compd
IC50a (μM)
SIRT1
SIRT2
113,16 HKK(thioAc)LM
0.31 (0.27ꢀ0.36) 6.3 (5.5ꢀ7.2)
0.57 (0.38ꢀ0.84) 151 (104ꢀ218)
0.24 (0.20ꢀ0.28) 9.8 (8.7ꢀ11)
Comparison of the SIRT3 structure to the crystal structure of
the Sir2Tm substrate complex (PDB code 2h2d)25 (Figure 2b)
indicates that small differences in the binding site sequences
(Figure 2c) do not lead to any significant changes in the substrate
binding orientations. The interaction pattern between substrates
and proteins is similar in SIRT3 and Sir2Tm structures, and
consists mainly of main chain interactions. It is assumed that this
can be generalized to the active sites of SIRT1 and SIRT2.
Substrate-bound SIRT3 (PDB code 3glr) was used as a target
structure to study the substrate binding area. Based on the
peptide modifications and substrate-bound crystal structures, it
was obvious that only main chain interactions were critical for the
substrate binding. However, we wanted to get a more detailed
insight on the substrate binding site in order to design potent
pseudopeptidic inhibitors for sirtuins. Two virtual libraries of
various possible pseudopeptides were created by attaching
various chemical structures to the terminals of a thioacetyllysine
amino acid. The libraries were docked in the crystal structure of
SIRT3 to study the chemical space of molecules that can be
docked to the substrate binding site allowing the thioacetyllysine
to be placed similarly as the acetyllysine in the X-ray structure. A
binding hypothesis for pseudopeptidic inhibitors was generated
based on the docking results (Figure 3aꢀe).
216
317
8
KK(thioAc)L
Ac-AK(thioAc)A
Ac-AK(thioAc)A-NHMe 0.40 (0.35ꢀ0.46) 2.9 (2.6ꢀ3.3)
Ac-AK(thioAc)A-OMe
AK(thioAc)A
13
14
15
16
17
1.0 (0.82ꢀ1.3)
13 (9.5ꢀ19)
0.66 (0.60ꢀ0.74) 36 (29ꢀ46)
K(thioAc)A
ndb
ndc
Ac-K(thioAc)A-OMe
51 (37ꢀ68)
862 (417ꢀ1782)
Ac-PK(thioAc)A
0.37 (0.35ꢀ0.39) 31 (28ꢀ33)
a Fluor de Lys based assay (repeated at least three times, 95% confidence
intervals in parentheses). b IC50 was not determined. Inhibition % at 200
μM ( standard deviation (n = 2ꢀ3) was 46.9 ( 3.0%. c IC50 was not
determined. Inhibition % at 200 μM ( standard deviation (n = 2ꢀ3)
was 2 ( 0.9%.
(Figures 2a and 2b, respectively). In the SIRT3 structure the
reactive acetyllysine of the substrate is located in a tunnel shaped
groove formed between the Rossman fold and the zinc binding
domains. The substrate binding has been claimed to induce
movement of the two domains relative to each other, which
significantly reduces the size of the cavity of the acetyllysine
binding site.25 The orientation of the substrate backbone is
dictated by a hydrogen bond network between the substrate
6458
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Interaction Criteria Proposed by the Binding Hypothesis.
Hydrogen bonds observed from the acetylated lysine in the SIRT3
crystal structure to Val292, Gly295, Glu296 and Glu325 were
assumed to be vital for the proper orientation of the acetylated
lysine (Figures 3a and 3b). A change in the substrate peptide
backbone orientation is seen at the KAcꢀ1 glycine residue position
(Figure 3a). This change in the orientation places the backbone to
a crevice between residues Pro297 and Pro326. The crevice
between residues Pro297 and Pro326 leads to a small cavity
formed by residues Pro297, Leu298, Gln300, Phe302 and
Leu303 (Figure 3c). It was hypothesized that binding affinity
could possibly be increased by occupying the cavity with an
appropriate structure and that Leu298 could offer an additional
hydrogen bonding site (Figure 3c), although the substrate peptide
in the crystal structure did not form a direct hydrogen bond at this
location. On the C-terminal side, it was noted that the location of
the amino acid Phe294 could offer a possibility for aromatic
interactions (Figure 3d,e). In addition, the backbone amine of
Glu325 could serve as an additional hydrogen bonding partner.
Analysis of the in Vitro Results in Light of Molecular
Docking. A series of pseudopeptides was designed (Table 2).
The first group of compounds has L-alanine at the C-terminal site
of the thioacetyllysine residue and various N-terminal modifica-
tions. The second group has the benzyloxycarbonyl group at the
N-terminal site of the thioacetyllysine residue and various
C-terminal modifications. The N- and C-terminal modifications
consist of non-natural peptidic fragments (18, 19), known
structures and their modifications (20, 21, 25, 27), structures
with different H-bonding properties (22, 23, 28, 29) and
structures from the virtual libraries selected based on the binding
hypothesis (24, 30).
The docking poses are presented in Figure 4 (see Figure S1 in
the Supporting Information for additional figures). The Glide
scores (Table 3) represent the overall docking score, while the
per-residue scores are listed to present hydrogen bond patterns
of the docking poses. The root-mean-square deviation (rmsd)
value represents the similarity of the thioacetylated lysine poses
of the docked compounds to the cocrystallized acetylated lysine
(PDB 3glr). The in vitro inhibitory activities are presented in
Table 2.
Figure 2. Crystal structures of (a) SIRT3 (PDB code 3glr)24 and (b)
Sir2Tm (PDB code 2h2d)25 showing the substrate binding site around
the acetylated lysine. (c) The sequence homology of SIRT1ꢀ3 and
Sir2Tm in the binding site area.
Compounds 18 and 24 have the Boc-L-alanine and Boc-
piperidine-3-carbonyl fragments at the N-terminus, respectively.
Figure 3. The binding hypothesis represented with wire spheres for pseudopeptidic inhibitors. (a) The molecular structure (green carbon atoms) represents the
binding mode of the peptide cocrystallized with SIRT3 (PDB 3glr). (b) The key hydrogen bonds formed by the acetylated lysine in the SIRT3 (blue and magenta
spheres). (c) N-Terminal modification from the virtual database screening that forms a hydrogen bond to Leu298 (blue sphere) and fills a cavity surrounded by
residues Pro297, Leu298, Phe302 and Leu303 (green sphere). (d, e) Possibility for aromatic interactions found from the virtual database screening (orange spheres).
6459
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Table 2. The Studied N- and C-Terminal Fragments and in
Vitro Inhibitory Activities of the Synthesized Compounds
Table 3. The Glide Docking Results of the Pseudopeptide
Compounds
Glide Glu325 Leu298 Glu296 Gly295 Val292 subset
compd gscore hbond hbond hbond hbond hbond rmsda
18 ꢀ8.85 ꢀ1.50
19 ꢀ8.06 ꢀ1.50
20 ꢀ8.47 ꢀ1.50
21 ꢀ8.41 ꢀ1.50
22 ꢀ8.66 ꢀ1.50
23 ꢀ8.81 ꢀ1.50
24 ꢀ9.71 ꢀ1.50
25 ꢀ8.00 ꢀ1.00
27 ꢀ8.00 ꢀ1.00
28 ꢀ8.28 ꢀ1.00
29 ꢀ8.60 ꢀ2.00
30 ꢀ8.97 ꢀ2.00
ꢀ0.72
0.00
0.00
0.00
0.00
0.00
ꢀ1.00
0.00
0.00
0.00
0.00
0.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ1.00
ꢀ0.94 ꢀ1.00
ꢀ0.96 ꢀ1.00
ꢀ1.00 ꢀ1.00
ꢀ0.85 ꢀ1.00
ꢀ0.96 ꢀ1.00
ꢀ2.00 ꢀ1.00
ꢀ1.00 ꢀ1.00
ꢀ1.00 ꢀ1.00
ꢀ0.96 ꢀ1.00
ꢀ1.00 ꢀ1.00
ꢀ1.00 ꢀ1.00
ꢀ1.00 ꢀ1.00
0.28
0.82
0.26
0.42
0.29
0.63
0.29
0.26
0.27
0.24
0.27
0.62
a Rmsd calculated between the common core substructure of the
compounds and the substrate peptide. The common core structure
was defined by the SMARTS pattern C(dO)NC(CCCCNCC)
(C(dO)N).
between the phenyl and carbonyl groups as has the Cbz group, a
known N-terminal fragment.15 In the docking calculations the free
amino terminal of the D-phenylalanine showed H-bonding to
Gly295but the phenyl groupwasnot placed in thecrevicebetween
Pro297 and Pro326 and no hydrogen bonding to Leu298 was
formed (see Figure S1 in the Supporting Information). This
compound did not fulfill any of the criteria proposed by the
binding hypothesis, and it is also the least active compound in the
series of N-terminal modifications. The additional H-bond to
Gly295 seen in the docking to SIRT3 does not correlate well with
the inhibitory activity toward SIRT1 or SIRT2.
In order to gain more insight into the effect of modifications in
the N-terminal, docking studies were carried out for compounds
20ꢀ23. The N-terminal Cbz group (20) has been presented by
Suzuki et al.15 In compound 21, the Cbz group was replaced by
the 3-phenylpropanoyl group. This modification was made in
order to study whether the removal of the oxygen in the chain
affects the activity. Docking calculations did not show any
significant differences in the binding of these compounds, and
the N-terminal structures were well superimposed (Figure 4b).
Both compounds bind to the crevice but do not reach all the way
to the cavity and do not form an H-bond to Leu298 thus fulfilling
one of the three binding criteria. The rmsd values illustrating the
similarity of the lysine orientation were small for both compounds
(Table 3). Compared to compound 20, compound 21 had
equipotent SIRT1 but slightly decreased SIRT2 inhibitory activity.
This decrease in the activity cannot be explained by docking
to SIRT3.
a Fluor de Lys based assay (repeated at least three times, 95% confidence
intervals in parentheses). b IC50 was not determined. Inhibition % at 200
c
μM ( standard deviation (n = 2ꢀ3) was 22.4 ( 3.0. IC50 was not
determined. Inhibition % at 200 μM ( standard deviation (n = 2ꢀ3)
was 57.2 ( 1.4.
The Boc-piperidine-3-carbonyl fragment was chosen among the
fragments in the previously created virtual libraries and was used
in the formation of the binding hypothesis (Figure 3c). Com-
pounds 18 and 24 were synthesized to test the hypothesis that
the shape of the receptor surface on the N-terminal side allows
the size of the N-terminal modification to be increased. The
docking poses of these compounds showed the N-terminal
fragment binding through the crevice, placing the tert-butyl
moiety in the small cavity and the carbamate carbonyl H-bonded
with Leu298 (Figure 4a and Table 3). Thus, these compounds
fulfilled all the N-terminal binding criteria proposed by the
binding hypothesis and are the two most potent compounds in
the series of the N-terminal modifications. Compared to Ac-
AK(thioAc)A peptide (3), compounds 18 and 24 showed
equipotent SIRT2 inhibitory activity and slightly reduced SIRT1
inhibitory activity. For complementary binders, the increase in
size usually increases the inhibitory activity; therefore, it may be
that the binding conformation of compounds 18 and 24 is not
optimal for SIRT1 and SIRT2.
Compound 21 was further modified by additional hydrogen
bonding groups (22 and 23). The position of the phenyl group of
compound 22 is slightly different compared to compound 21 but
is still positioned to the crevice (Figure 4c). The placement of
fluorine could allow H-bonding to Leu298, but the GlideScore
scoring function does not consider fluorine as a hydrogen bond
acceptor. The SIRT1 in vitro results do not support Leu298
H-bonding, but in the case of SIRT2 it cannot be excluded. The
docking pose of compound 23 to the SIRT3 is different
compared to compound 21 (Figure 4c). The phenyl group is
One D-amino acid was chosen into the set of N-terminal
modifications. D-Phenylalanine in compound 19 has two atoms
6460
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Figure 4. The docking poses of the pseudopeptides. The interactions were predicted with the Ligand Interactions tool available in MOE.26 (a) The
docking poses of compounds 18 and 24 showed binding to the crevice between Pro297 and Pro326, positioning of the tert-butyl groups in a pocket
surrounded by residues Pro297, Leu298, Gln300, Phe302 and Leu303 and hydrogen bonding to Leu298. (b) The docking poses of compounds 20
(cyan) and 21 (green). (c) The docking poses of compounds 22 (orange) and 23 (cyan) compared to compound 21 (green). Compound 23 was
hydrogen bonded to Gly295. (d) The docking pose of compound 30 showing aromatic interactions with Phe294 and hydrogen bonding to Glu325. The
best ranked pose (green) had the N-terminal amide in the cis-form while the second ranked pose (cyan) had the N-terminal amide in the trans-form.
However, the poses of the C-terminal fragment were identical. (e) The docking poses of compounds 25 (green) and 27 (cyan) placed the cyclohexyl and
phenyl moieties in the same location. (f) The docking poses of compounds 25 (green), 28 (cyan) and 29 (dark cyan) with the overlaid phenyl moieties.
The hydrogen bond acceptors hydroxyl and fluorine are toward the backbone amine of Glu325.
not positioned to the crevice, and the hydroxyl group forms an
H-bond to Gly295. However, the in vitro results are similar for
these compounds and, again, the additional hydrogen bond to
Gly295 found in the pose is not supported by the in vitro
activities. It is possible that the docking pose of 23 to SIRT3
may not represent the binding to SIRT1 or SIRT2.
more potent than compound 25 against both enzymes. The
Ligand Interaction tool of MOE did not detect aromatic inter-
actions between the phenyl ring of 25 and Phe294. However,
when the phenyl group was replaced by a cyclohexyl group (27),
the inhibitory activity against SIRT1 and SIRT2 decreased. This
proved the importance of the phenyl group even though the
docking scores and the per-residue hydrogen bonding scores
were nearly equal for these compounds (Table 3 and Figure 4e).
It may be that the receptor flexibility allows aromatic interactions
of the phenyl group with Phe294 although the interaction may
not be as optimal as for compound 30. However, compound 27
with the cyclohexyl group at the C-terminus does not fulfill the
criteria of the aromatic interactions or the H-bond to Glu325 and
is the least active compound in the series.
Compounds 28 and 29 are derivatives of compound 25 with
an additional hydrogen bonding group. The docking poses of
compounds 25, 28 and 29 were well superimposed (Figure 4f).
The C-terminal phenyl group of all three compounds had similar
position and orientation. The position of the hydroxyl oxygen of
compound 29 allowed hydrogen bond formation to residue
Glu325, and slight improvement is also seen in the inhibitory
activities. The fluorine of compound 28 is also in a position that
The C-terminal modification in compound 30 was chosen
among the fragments in the previously created virtual libraries
and was used in the creation of the binding hypothesis
(Figure 3e). The docking pose of compound 30 (Figure 4d)
positions the ketone carbonyl to form a hydrogen bond to
Glu325. When the pose was analyzed with the Ligand Interaction
tool of MOE (version 2009.10),26 the adjacent phenyl ring was
predicted to have aromatic interactions with Phe294. This
compound fulfills both binding criteria for the C-terminal bind-
ing proposed by the binding hypothesis and is the most potent
compound in the series of C-terminal modifications.
When the previously published compound 2515,17 is com-
pared to compound 30, it is seen that the phenyl group is
positioned differently (Figure 4e). Compound 25 also lacks the
H-bond to Glu325. The differences in the docking poses are
reflected by the in vitro inhibitory activities; compound 30 was
6461
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Table 4. The Number of Successfully Docked Compounds on
Various SIRT3, SIRT2 and Sir2Tm Crystal Structures and
SIRT1 and SIRT2 Modelsa
compounds (excluding compound 2) were docked again to several
sirtuin X-ray structures and two homology models representing
different conformational forms. These structures included four
different SIRT3 X-ray structures, one SIRT2 X-ray structure, one
SIRT2 homology model, one SIRT1 homology model and five
Sir2Tm structures. The homology models of SIRT1 and SIRT2
were based on the substrate binding X-ray structures of SIRT3. The
target set included one apo structure from each protein.
PDB
code
cocrystallized
substrate
no. of successfully
docked compounds
protein
3glr
SIRT3
SIRT3
SIRT3
SIRT3
SIRT2
Sir2Tm
Sir2Tm
Sir2Tm
Sir2Tm
Sir2Tm
SIRT1
SIRT2
AceCS2-KAc
18/20
0/20
3gls
apo
The resulting docking poses were visually analyzed to see
whether the synthesized compounds fulfill the hydrogen bonding
network of the thioacetyllysine (see Figures S2ꢀS13 in the
Supporting Information for the docking poses). The docking
results are presented in Table 4. The results of SIRT3, Sir2Tm
and SIRT1 calculations indicated that the obtained docking
poses are not sensitive to small sequence differences. The
compounds could be docked into correct docking poses with a
high success rate on all of these proteins. However, the protein
conformation seemed to have a bigger effect on the docking
results. Dependence of the success rate on the conformation
variation was observed on SIRT3 and Sir2Tm proteins. When
the apo conformations of SIRT3, SIRT2, and Sir2Tm were used
as the docking targets, none of the docked compounds resulted in
a correct pose. This can be explained by the significant con-
formation difference and enlargement of the binding site cavity,
caused by the movement of the two large domains.25 The success
rate was increased on SIRT2 when the homology model was
used. However, a lower success rate was obtained with the SIRT2
model than with the SIRT1 model. The results do not explain
whether the lower success rate is caused by sequence differences
or by not optimally placed side chains of the SIRT2 model.
The results of this analysis gave further justification for our
choice to use the SIRT3 crystal structure complexed with
AceCS2-KAc substrate. As the small sequence differences in the
substrate binding site did not seem to affect significantly the
docking poses, the results can be postulated to give a somewhat
reasonable prediction on the binding poses also on SIRT1 and
SIRT2. The in vitro results supported these assumptions as
several new active inhibitors were discovered in the process.
Inhibition of SIRT1 Activity in Cell Culture Models. To
examine if the compounds are also active at the cellular level, we
used Western blot analysis to detect the changes in acetylation
level of p53 after etoposide-induced DNA damage in three
different cell types. SIRT1 is known to be involved in deacetyla-
tion of p53 at those conditions.15,27 The effect of compounds 20,
24 and 30 on p53 acetylation rate in cells (Figure 5) is in
accordance with the in vitro inhibition profile of SIRT1. Those
compounds were not toxic to cells at concentrations used (data
not shown).
3glt
AceCS2-KAc-ADPR
13/20
18/20
0/20
3glu
1j8f
AceCS2
apo
2h2d
2h2f
2h2g
2h2h
2h2i
model
model
p53-K382Ac
p53-K382
H3-K115Ac
H4-K79Ac
apo
19/20
19/20
18/20
18/20
0/20
18/20
8/20
a In order to be considered as a successful docking pose, the thioace-
tyllysine had to form the three hydrogen bonds seen in substrate-bound
sirtuin structures (e.g. PDB code 3glr).
allows hydrogen bonding to the backbone amine of Glu325, but
no effect on the inhibitory activities was observed. No hydrogen
bonding scores were seen, either, in the per-residue scores
(Table 3), but this is because the GlideScore scoring function
does not consider fluorine as a hydrogen bond acceptor.
The binding hypothesis created by using SIRT3 structure
seems to predict quite well binding to SIRT1 and SIRT2
although there are some reservations. The most active N- and
C-terminal modifications (18, 24 and 30) fulfilled all the binding
criteria proposed by the hypothesis while the least active
modifications (19 and 27) failed to fulfill them. On the N-term-
inal side, the occupation of the cavity seemed to be important for
SIRT2 while SIRT1 can be inhibited with shorter compounds.
This is well in line with the SAR of the peptidic inhibitors.
Hydrogen bonding to Leu298 in the SIRT3 structure does not
seem to correlate with SIRT1 in vitro activity, while some
correlation in the case of SIRT2 was seen. The proposed
hydrogen bonding of compounds 19 and 23 to Gly295 was
not supported by the in vitro results. On the C-terminal side,
both the hydrogen bonding to Glu325 and the aromatic inter-
actions with Phe294 in the SIRT3 structure predicted good
inhibitory activities for SIRT1 and SIRT2 inhibitors. The pro-
posed binding hypothesis can be used in the design of SIRT1 and
SIRT2 inhibitors although all affinity differences between these
two targets cannot be explained. This would require detailed
structural data on the SIRT1 and SIRT2 substrate binding sites in
the substrate binding conformation, but because of the lack of
crystal structures, this data is not available.
’ CONCLUSIONS
Structure-based computational design approach was success-
fully applied in the design of Nε-thioacetyllysine containing
pseudopeptidic inhibitors of SIRT1 and SIRT2. Several designed
pseudopeptidic inhibitors maintained the potency comparable to
the peptidic inhibitors. Those N- and C-terminal fragments
which gave the best inhibitory activities were the same fragments
that fulfilled all the interaction criteria proposed by the binding
hypothesis. On the other hand, those fragments which did not
fulfill the set criteria gave the least active compounds. The
pseudopeptides were shown to inhibit class III HDACs
(SIRT1 and SIRT2), however no high inhibition responses were
There are 11 globular isoforms of human histone deacetylases
(HDACs) with conical pockets that fit acetylated lysine residues.
The pseudopeptidic compounds were tested on a mixture of class
I and class II HDACs to study inhibition among HDACs. The
compounds did not show high inhibition activities in the HDAC
assay (Table 2). Compound 27 was the only compound that
showed some response.
Sensitivity of the Modeling Approach To Target Protein
Differences. In order to test how sensitive the used molecular
docking approach is to sequence differences, the synthesized
6462
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Figure 5. Inhibition of SIRT1 activity increases p53 acetylation after DNA damage. (A) Effect of 10 μM compound 30 on p53 acetylation after 100 μM
etoposide (eto) treatment for 5 h in three different cell lines. (B) Effect of 10 μM compound 20 and 10 or 100 μM compound 24 on p53 acetylation after
100 μM etoposide treatment for 5 h in SH-SY5Y cells. + present; ꢀ absent.
observed in an assay that contained a mixture of class I and class II
HDACs. Compounds 20, 24 and 30 demonstrated SIRT1 inhibi-
tion also in cultured cells. It was concluded that the homologous
SIRT3 structure can be used for designing substrate based inhibitors
of SIRT1 and SIRT2. On the other hand, the available experimental
SIRT2 structure in the apo conformation is not suitable for the
design of substrate based modulators as the sirtuins undergo
significant conformational changes upon substrate binding.25
of 11 for the standard deviation setting. The Leadlike filter was left on,
and no exclusion for the reactive compounds was done. The RECAP
synthesis was set to continue until 10000 compounds or 1000000
repeated attempts had been performed. The thioacetyllysine structure
was set as a root structure by placing the structure in a MOE database
and by forcing the root structure to be always selected from this database.
The weight setting, which defines the probability that a fragment is
selected from this database, was set to 0 for the root database and 1 for
the small size fragment database, forcing the fragments to be selected
from the latter. Overall 10000 structures were generated. Hydrogens
were added to the created molecules using the Database Wash function
in MOE. Prior to subjecting the ligands to docking, the compounds were
preprocessed with Ligprep (version 2.3).28 The molecules were im-
ported to Ligprep in sdf format. The ionization and tautomeric states
were determined with EPIK29 (version 2.0) in pH of 7 ( 2. OPLS2005
force field was used for the geometric optimization.
’ EXPERIMENTAL SECTION
The Virtual Databases. The first virtual database was created
using the RECAP synthesis tool of MOE (Molecular Operating
Environment) version 2008.10.26 The thioacetyllysine structure was
used as a root structure in the RECAP synthesis process, and molecular
fragments were attached to both terminals of the thioacetyllysine. The
fragments were taken from a small-size fragment subset of the RECAP
fragment database provided with MOE. The subset contained all
fragments of size 12 or less heavy atoms from the MOE RECAP
database. The RECAP synthesis process was run with default settings
except the size limit was increased to 35 heavy atoms with default value
The second virtual database was created using CombiGlide (version
2.5).30 The fragment ideas were obtained from a list of selected structure
sets available from the chemical vendor Sigma-Aldrich. The carboxylic
acid containing fragments were downloaded in sdf format from the
Sigma-Aldrich web page (http://www.sigmaaldrich.com/chemistry/
chemistry-services/selected-structure.html, accessed January 27th, 2010).
6463
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
The carboxylic acid fragments were prepared with the Reagent Prepara-
tion process of CombiGlide. The carboxylic acid group was defined as the
reactive functional group. No changes to the ionization, tautomeriza-
tion and steroisomer states of the fragments were made at this point.
Thioacetyllysineꢀalanine structure, used as a core structure in the
CombiGlide process, was built and energy minimized in MOE (version
2009.10)26 and later prepared with Ligprep (version 2.3)28 using
OPLS2005 forcefield for geometric optimization. The fragments were
attached to the N-terminal of the core structure using the Combina-
torial Library Enumeration tool of CombiGlide. Overall 6748 struc-
tures were generated. Prior to subjecting the structures to docking they
were prepared with Ligprep. OPLS2005 force field was used for
geometric optimization, all possible protonation and tautomeric states
were generated with EPIK at pH 7 ( 2 and chiralities if defined were
taken from the 3D structure.
The Docking Calculations. The docking calculations were per-
formed using Glide (version 5.5)31 molecular docking program. The
X-ray crystal structure of human SIRT3 protein complexed with an
acetylated lysine structure (PDB code 3glr)24 was used as a target in the
docking calculations performed on the virtual databases. The protein
structure was prepared using the Protein Preparation Wizard available in
the Schr€odinger Suite 2009.32 The PDB structure was downloaded from
www.rcsb.org using the Import command of the Protein Preparation
Wizard (PDB download date December 18th, 2009). In the preprocess
phase of the Protein Preparation Wizard the bond orders were assigned,
hydrogens were added, distant water molecules were removed, metals
were treated and overlaps were detected. Later all water molecules and
HET groups except the zinc were removed. Exhaustive sampling method
was used to optimize the hydrogen bonding and determine the orienta-
tions of hydroxyl groups, amide groups of Asn and Gln amino acids, and
the proper state and orientations of the histidine imidazole rings. Finally
a restrained minimization was performed using OPLS2001 force field,
with a 0.3 Å rmsd atom displacement limit.
The prepared protein structure was used in the receptor grid
generation process. The acetyllysine containing peptide was defined as
a ligand to be excluded, and the peptide defined the grid box location and
the size of the outer box. Default values were used for all the other
available options and settings for the grid calculations.
The same receptor grid files were used in the docking calculations of
both virtual databases. The dockings were performed using the SP
(standard precision) precision option. The compounds were docked
flexibly using the options that allow the algorithm to sample ring
conformations but penalize the nonplanar amide bond torsion angles.
50 poses of each docked compound were subjected to post docking
minimization. The other options and settings were left at their default
values.
The synthesized and tested pseudopeptides were subjected to docking
calculations. The molecular structures were exported in sdf format from a
ChemBioOffice33 database and prepared using Ligprep (version 2.3).28
OPLS2005 force field was used for geometric optimization, and all pos-
sible ionization and tautomeric forms were created at pH 7 ( 2 using
EPIK.29 The chiralities specified in the input structures were retained. The
same grid files were used in the dockings as in the virtual library dockings.
The dockings were performed using the SP precision settings treating the
compounds as flexible. The ring conformations were allowed to be
sampled and nonplanar amide bond torsions penalized to favor planar
conformations. 50 poses of each ligand were subjected to postdocking
minimization, and an option for adding a strain correction term if the
strain crossed the default strain limit was set on. The best ranked pose of
each docked ligand was included in the analysis.
X-ray structures (PDB codes: 3glr, 3gls, 3glt and 3glu),24 to 5 Sir2Tm X-ray
structures (PDB codes: 2h2d, 2h2f, 2h2g, 2h2h and 2h2i),25 to the only
available SIRT2 X-ray crystal structure (PDB code 1j8f)23 (the PDB files
were downloaded between June 1 and June 3, 2010) and to homology
models of SIRT1 and SIRT2. The homology models were built using
MODELER in Discover Studio.34 The templates used for the homology
models were the crystal structure of SIRT3 with substrate (PDB codes: 3glr,
3glt and 3glu). Several models were generated, and side chains were
optimized. The model with the smallest rmsd compared to 3glr was chosen
for dockings. The protein structures were prepared with Schr€odinger Suite
2009 (Update 2) Protein Preparation Wizard32 in a similar automated
fashion that was used for preparing the protein used in the virtual database
dockings. The ligands were treated with Ligprep (version 2.3)28 as
described earlier. The grid box center for SIRT3 proteins was defined
based on the amino acids Pro297, Pro326 and Phe294 for SIRT3 structures.
Also the dockings for the 3glr crystal structure were repeated using the same
grid box center in order to improve the comparability of the docking results.
The corresponding amino acids were also used for defining the grid box
center on the Sir2Tm, SIRT1 and SIRT2 proteins. The dockings were
performed with the same Glide settings as the earlier dockings.
Result Analysis and Figure Preparation. The figures of the
crystal structures 3glr and 2h2d (Figure 2) were prepared using MOE
(version 2009.10).26 Prior to the figure preparation, the structures were
treated using the Protein Preparation Wizard (Schr€odinger Suite 2009
Update 2).32 This included bond order assignment, addition of hydro-
gens, removal of water molecules and heteroatomic groups. For the
SIRT3 structure a water molecule close to amino acid Leu298 was not
removed, and the zinc atoms for both structures were left untouched.
The Ligand Interactions tool in MOE was used for predicting the
interactions between the substrate peptides and their protein counter-
parts. The protein sequences for the sequence alignment were obtained
from the UniprotKB35 database (access date July 8, 2010). The sequence
alignment was made with the online Clustalw2 (version 2.0.12)36
alignment tool available at the UniprotKB web page. The alignment
figure was prepared using Jalview (version 2.5.1).37 The graphical effects
on the final figure were prepared with Inkscape (version 0.47).
The binding mode hypothesis figure (Figure 3) was prepared using
MOE (version 2009.10).26 The protein structure in the figure is the one
used in the virtual database screening. MOE pharmacophore annotations
were used for describing the hypothesized interactions. The graphical
effects on the figure were prepared using Gimp version 2.2.13 and
ImageMagick version 6.5.1ꢀ0.
The figures describing the docking poses (Figure 4) were prepared
using MOE (version 2009.10).26 The interactions in the figures includ-
ing the hydrogen bond possibilities and aromatic interactions were
analyzed and visualized using the Ligand Interaction module available in
MOE. The score limit for presenting a hydrogen bond was set to 10%.
The figure labels were added with ImageMagick version 6.5.1ꢀ0.
Hydrogen bond scoring of Table 3 is based on the per-residue
interaction scoring output of Glide. The rmsd value of Table 3 is a rmsd
between a common substructure of the peptide substrate of the crystal
structure 3glr and the synthesized ligands, defined with a SMARTS
pattern C(dO)NC(CCCCNCC)(C(dO)N). The rmsd calculation
was performed with a Schr€odinger script rmsd.py (version 1.14.2.6)38
downloaded from the Schrodinger Script Center Web site (http://www.
schrodinger.com/scriptcenter, access date June 23th, 2010).
Chemistry. All reagents and solvents were commercial high purity
quality. Purification of products 8, 13, 16ꢀ18, 20ꢀ24 and 27ꢀ30 was
performed by CombiFlash column chromatography on normal phase
silica (average particle size, 35 to 70 μm; mesh, 230 to 400; average pore
size, 60 Å) or by crystallization methods. Compounds 14, 15 and 19
were purified by preparative HPLC (Shimadzu LC-10Avp (Shimadzu,
Kyoto, Japan)) on a reverse phase C18 column (Supelcogel ODP-50,
25 cm ꢁ 21.2 mm, 5 μm) with a linear gradient of 5ꢀ90% solvent
In order to test whether the docking results can be somewhat
generalized to other sirtuin structures with differences in the conforma-
tions and in the amino acid sequences, the 20 different tested com-
pounds (1, 3, 8, 13ꢀ25 and 27ꢀ30) were docked to 4 human SIRT3
6464
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
B (0.05%acetic acid/acetonitrile) in solvent A (0.05 acetic acid/H2O) in
30 min with the flow rate 10 mL/min. The peptide was detected by UV
at λ = 215 nm. NMR spectra were recorded on a Bruker Avance 500 AV
1.83 (s, 3 H), 2.37 (s, 3 H), 2.57 (d, 3 H), 3.43 (m, 2 H), 4.13ꢀ4.28 (m, 3 H),
7.73 (m, 1 H), 7.82 (d, 1 H), 7.92 (d, 1 H), 8.07 (d, 1 H), 9.93 (m, 1 H). 13
C
NMR ((CD3)2SO): δ = 17.92, 18.29, 22.45, 22.81, 25.52, 26.81, 31.33, 32.77,
45.32, 48.08, 48.34, 52.36, 169.23, 170.98, 172.33, 172.55, 198.76. ESI-MS
(m/z): 402.2 [M + H]+, 424.3 [M + Na]+. Anal. (C17H31N5O4S 0.3hexane
1
(Bruker Biospin, Switzerland) 500.1 MHz for H and 125.8 MHz for
13C. The chemical shifts are expressed in ppm relative to the shift of used
solvent as an internal standard (1H NMR, DMSO at 2.50 ppm, CH3OH
at 3.31 ppm and CHCl3 at 7.26 ppm; 13C NMR, (CD3)2SO at 39.52
ppm, CD3OD at 49.00 ppm and CDCl3 at 77.16 ppm). Positive ion mass
spectra were acquired with a quadrupole ion trap mass spectrometer
(Finnigan MAT, San Jose, CA) equipped with an electrospray ionization
source (ESI-MS). The purity of compound 14 was determined using
Agilent 1100 HPLC (Agilent Technologies Inc., Waldbronn, Karlsruhe,
Germany) with diode array detection, using reversed phase column
(Zorbax Eclipse XDB C 18, 4.6 ꢁ 50 mm, 1.8 μm, Agilent Technologies,
Palo Alto, CA, USA), 5% solvent B (0.05% AcOH in CH3CN) in solvent
A (0.05% AcOH in H2O) for 5 min, then linear gradient 5ꢀ80% B in A
in 15 min with the flow rate 1 mL/min. The purity of compound 15 was
determined using Shimadzu LC-10Avp (Shimadzu, Kyoto, Japan), using
a reverse phase column (Supelcogel ODP-50 C 18, 4 ꢁ 150 mm, 5 μm),
with an isocratic eluent (0.05% AcOH, 2.5% MeOH, 97.45% H20) in 30
min with the flow rate 1 mL/min. The purity of the rest of the
compounds was determined by combustion analysis for CHN by
Thermo Quest CE Instruments EA 1110 CHNS-O elemental analyzer.
Peptide 14 showed 93% purity, and all other compounds had purity of
g95%.
3
3
0.2H2O) C, H, N.
(S)-Methyl 2-((S)-2-((S)-2-Acetamidopropanamido)-6-etha-
nethioamidohexanamido)propanoate (13). 1H NMR (MeOD):
δ = 1.34 (d, 3 H), 1.39 (d, 3 H), 1.47 (m, 2 H), 1.62ꢀ1.92 (m, 4 H), 1.98
(s, 3 H), 2.45 (s, 3 H), 3.59 (m, 2 H), 3.71 (s, 3 H), 4.28ꢀ4.45 (m, 3 H).
13C NMR (MeOD): δ = 17.28, 17.80, 22.39, 24.06, 28.27, 32.86, 33.12,
46.96, 49.46, 50.63, 52.76, 54.12, 173.28, 173.92, 174.52, 175.18, 201.57.
ESI-MS (m/z): 403.2 [M + H]+, 425.3 [M + Na]+. Anal. (C17H30N4O5S)
C, H, N.
(S)-2-((S)-2-((S)-2-Aminopropanamido)-6-ethanethioami-
dohexanamido)propanoic Acid (14). 1H NMR ((CD3)2SO): δ =
1.25 (d, 3 H), 1.28ꢀ1.41 (m, 2 H), 1.32 (d, 3 H), 1.48ꢀ1.74 (m, 4 H),
2.37 (s, 3 H), 3.42 (m, 2 H), 3.83 (m, 1 H), 4.09 (m, 1 H), 4.27 (m, 1 H),
8.12 (d, 1 H), 8.48 (d, 1 H), 9.99 (m, 1 H). 13C NMR ((CD3)2SO): δ =
17.50, 17.55, 22.78, 26.95, 31.87, 32.81, 45.37, 47.95, 48.18, 52.51,
169.79, 170.62, 174.18, 198.83. ESI-MS (m/z): 347.14 [M + H]+.
HPLC: tR 9.31 min, area percent 93% at 260 nm.
(S)-2-((S)-2-Amino-6-ethanethioamidohexanamido)propanoic
Acid (15). 1H NMR (D2O): δ = 1.44 (d, 3 H), 1.50 (m, 2 H), 1.73 (m,
2 H), 1.96 (m, 2 H), 2.51 (s, 3 H), 3.63 (t, 2 H), 4.03 (t, 1 H), 4.36 (q,
1 H). 13C NMR (D2O): δ = 17.82, 22.71, 27.86, 31.87, 33.71, 47.11,
50.92, 54.36, 170.49, 178.16, 201.83. ESI-MS (m/z): 276.19 [M + H]+.
HPLC: tR 3.18 min, area percent 99% at 260 nm.
(S)-Methyl 2-((S)-2-Acetamido-6-ethanethioamidohexa-
namido)propanoate (16). 1H NMR (CDCl3): δ = 1.43 (d, 3 H),
1.46 (m, 2 H), 1.66ꢀ1.88 (m, 4 H), 2.01 (s, 3 H), 2.56 (s, 3 H), 3.65 (m,
2 H), 3.76 (s, 3 H), 4.52 (m, 1 H), 4.57 (m, 1 H), 6.63 (m, 1 H), 7.12 (m,
1 H), 8.21 (m, 1 H). 13C NMR (CDCl3): δ = 17.47, 22.30, 23.13, 26.81,
32.62, 33.71, 45.84, 48.15, 52.30, 52.52, 170.45, 171.57, 173.31, 200.75.
ESI-MS (m/z): 332.22 [M + H]+, 354.25 [M + Na]+. Anal.
Manual Solid Phase Peptide Synthesis (SPPS). Wang resin
(polymer-bound p-alkoxybenzyl alcohol) was used as solid support for
the peptide synthesis of compounds 5 and 13ꢀ24. Solid phase synthesis
was performed in a 10 mL syringe equipped with a frit. Fmoc-Ala-Wang
resin (loading 0.4ꢀ0.8 mmol/g) was swelled for 1 h in 3 mL of DMF. In
the deprotection phase, the Nα-Fmoc protection group was removed
with 5 mL of 20% (V/V) piperidine in DMF for 15 min and the resin was
rinsed 5 times with DMF. In the coupling phase, the following Nα-Fmoc-
amino acid or acetic acid (2ꢀ4 equiv) was preactivated (1ꢀ3 min) with
the coupling reagent TBTU (2ꢀ4 equiv) and DIPEA (5ꢀ10 equiv) in
3ꢀ5 mL of DMF. This solution was added on the resin, and the syringe
was shaken at rt for 60 min. Then the resin was rinsed 5 times with DMF.
The cycle of Fmoc-deprotection and coupling was repeated until the
desired resin-bound peptide was completed. Before cleavage from the
resin, the resin was washed once with AcOH, five times with dichlor-
omethane (DCM), and once with MeOH to remove the excess solvents
and then dried under vacuum.
Cleavage of Peptide-Resin as Carboxylic Acid. The dried
peptide-resin was preswelled in dioxane for 15 min. The excess of
dioxane was removed, and 8 mL of cool (0 ꢀC) cleavage mixture of 1 M
NaOH (aq)/dioxane (2 and 6 mL) was added on the peptide-resin. The
reaction mixture was shaken for 1 h at rt. The mixture was filtrated, the
filtrate was collected and the resin was washed with H2O/dioxane (1:3),
dioxane and H2O. The filtrate was neutralized with 3 M HCl (aq), and
the solvents were evaporated. Acetone was added on the residue, and the
mixture was filtrated to remove inorganic salts. The solvent of the filtrate
was evaporated under reduced pressure to yield the crude product,
which was purified by column chromatography or by crystallization.
Cleavage of Peptide-Resin as Methyl Ester. The dry peptide-
resin was placed in a flask. Cleavage mixture DIPEA/MeOH/DMF (1:5:5,
11ꢀ22 mL) was added, and the reaction mixture was refluxed under argon
atmosphere overnight. The mixture was filtrated, filtrate was collected and
the resin was rinsed 5 times with MeOH/DMF (1:1). Solvent was
evaporated from the combined filtrates under reduced pressure to yield
the crude product, which was purified by column chromatography or by
crystallization.
(C14H25N3O4S 0.3Et2O) C, H, N.
3
(S)-2-((S)-2-((S)-1-Acetylpyrrolidine-2-carboxamido)-6-
ethanethioamidohexanamido)propanoic Acid (17). 1H NMR
((CD3)2SO): δ = 1.22ꢀ1.37 (m, 5 H), 1.46ꢀ1.79 (m, 5 H), 1.79ꢀ1.89
(m, 3 H), 1.96ꢀ2.22 (m, 3 H), 2.34ꢀ2.40 (m, 3 H), 3.37ꢀ3.59 (m,
4 H), 4.11ꢀ4.41 (m, 3 H), 7.85ꢀ7.94 (m, 1 H), 8.12ꢀ8.22 (m, 1 H),
9.86ꢀ9.97 (m, 1 H), 12.01ꢀ12.53 (br, 1 H). 13C NMR ((CD3)2SO):
δ = 16.97, 17.05, 21.03, 22.01, 22.38, 22.50, 22.76, 22.86, 24.31, 26.77,
26.79, 29.44, 31.28, 31.58, 31.73, 32.78, 45.27, 45.36, 46.29, 47.41, 47.47,
47.65, 51.71, 51.94, 59.34, 60.19, 168.51, 168.93, 171.27, 171.55, 171.80,
171.97, 173.91, 173.96, 198.77, 217.03. ESI-MS (m/z): 415.23 [M +
H]+. Anal. (C18H30N4O5S 0.8AcOH) C, H, N.
3
(6S,9S,12S)-9-(4-Ethanethioamidobutyl)-2,2,6,12-tetramethyl-
4,7,10-trioxo-3-oxa-5,8,11-triazatridecan-13-oic Acid (18). 1H
NMR ((CD3)2SO): δ = 1.16 (d, 3 H), 1.26 (d, 3 H), 1.37 (s, 9 H),
1.22ꢀ1.41 (m, 2 H), 1.45ꢀ1.71 (m, 4 H), 2.37 (s, 3 H), 3.41 (m, 2 H), 3.96
(m, 1H), 4.17(m,1H),4.28(m, 1H),6.96(d, 1H), 7.69(d,1H),8.17(d, 1
H), 9.90(m, 1 H), 12.17 (br, 1H). 13CNMR((CD3)2SO): δ= 17.09, 17.99,
21.04, 26.89, 28.17, 32.11, 32.79, 45.38, 47.45, 49.73, 51.75, 78.10, 155.07,
171.16, 171.98, 173.89, 198.75. ESI-MS (m/z): 447.11 [M + H]+. Anal.
(C17H30N4O5S 0.6AcOH 0.15hexane) C, H, N.
3
3
(S)-2-((S)-2-((R)-2-Amino-3-phenylpropanamido)-6-ethane-
thioamidohexanamido)propanoic Acid (19). 1HNMR((CD3)2SO):
δ=1.15(m,2H),1.25(d,3H),1.38ꢀ1.61(m,4H),2.37(s,3H),2.66ꢀ3.00
(m, 2 H), 3.39 (m, 2 H), 3.68 (m, 1 H), 4.09 (m, 1 H), 4.25 (m, 1 H),
7.18ꢀ7.32 (m, 5 H), 8.19 (m, 2 H), 9.97 (m, 1 H). 13C NMR ((CD3)2SO):
δ = 17.43, 22.41, 26.86, 32.07, 32.76, 45.31, 47.82, 51.95, 55.15, 126.39, 128.18,
129.29, 137.49, 170.79, 171.88, 173.91, 198.74. ESI-MS (m/z): 423.27
(S)-2-((S)-2-Acetamidopropanamido)-6-ethanethioamido-N-
((S)-1-(methylamino)-1-oxopropan-2-yl)hexanamide (8). 1H NMR
((CD3)2SO): δ = 1.15ꢀ1.21 (m, 6 H), 1.29 (m, 2 H), 1.45ꢀ1.73 (m, 4 H),
[M + H]+. Anal. (C20H30N4O4S 0.5H2O) C, H, N.
3
6465
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
(S)-2-((S)-2-(Benzyloxycarbonylamino)-6-ethanethioamido-
hexanamido)propanoic Acid (20). 1H NMR ((CD3)2SO): δ = 1.27
(d, 3 H), 1.34 (m, 2 H), 1.45ꢀ1.69 (m, 4 H), 2.37 (s, 3 H), 3.43 (m, 2 H),
4.00 (m, 1 H), 4.19 (m, 1 H), 5.02 (s, 2 H), 7.27ꢀ7.42 (m, 6 H), 8.15 (d,
1 H), 9.92 (m, 1 H), 12.47 (br, 1 H). 13C NMR ((CD3)2SO): δ = 17.11,
22.97, 26.89, 31.66, 32.79, 45.31, 47.38, 54.21, 65.31, 127.64, 127.74, 128.31,
137.05, 155.89, 171.67, 173.99, 198.74. ESI-MS (m/z): 410.26 [M + H]+,
156.77, 170.04, 201.24. ESI-MS (m/z): 432.26 [M + H]+. Anal.
(C22H26FN3O3S 0.2H2O) C, H, N.
3
(S)-Benzyl 6-Ethanethioamido-1-(2-hydroxyphenylamino)-
1-oxohexan-2-ylcarbamate (29). 1H NMR (CDCl3): δ = 1.47 (m,
2 H), 1.64ꢀ2.02 (m, 4 H), 2.51 (s, 3 H), 3.66 (s, 2 H), 4.41 (br, 1 H), 5.13
(s, 2 H), 5.52 (d, 1 H), 6.86 (dd, 1 H), 6.98 (d, 1 H), 7.11 (dd, 2 H),
7.28ꢀ7.40 (m, 5 H), 7.52 (br, 1 H), 8.33 (s, 1 H), 8.55 (br, 1 H). 13C NMR
(CDCl3):δ= 22.71, 27.23, 31.84, 34.31, 45.67, 55.12, 67.78, 119.45, 120.85,
122.74, 125.19, 127.47, 128.28, 128.61, 128.81, 135.88, 148.52, 156.83,
171.58, 201.38. ESI-MS (m/z): 430.23 [M + H]+. Anal. (C22H27N3O4S)
C, H, N.
432.27 [M + Na]+. Anal. (C19H27N3O5S 0.1hexane) C, H, N.
3
(S)-2-((S)-6-Ethanethioamido-2-(3-phenylpropanamido)
hexanamido)propanoic Acid (21). H NMR ((CD3)2SO): δ =
1
1.24 (m, 2 H), 1.27 (d, 3 H), 1.39ꢀ1.67 (m, 4 H), 2.37 (s, 3 H), 2.45 (m,
2 H), 2.80 (m, 2 H), 3.41 (m, 2 H), 4.18 (m, 1 H), 4.29 (m, 1 H),
7.13ꢀ7.29 (m, 5 H), 7.93 (d, 1 H), 8.17 (d, 1 H), 9.90 (m, 1 H), 12.47
(br, 1 H). 13C NMR ((CD3)2SO): δ = 17.03, 22.73, 26.92, 31.07, 31.93,
32.78, 36.68, 45.37, 47.35, 51.84, 125.81, 128.18, 128.19, 141.27, 171.21,
171.49, 173.96, 198.72. ESI-MS (m/z): 408.23 [M + H]+. Anal.
(C20H29N3O4S 0.1H2O 0.1hexane) C, H, N.
(S)-Benzyl 6-Ethanethioamido-1-oxo-1-(2-oxo-2-phenylethy-
lamino)hexan-2-ylcarbamate (30). 1H NMR (CDCl3): δ = 1.50 (m,
2 H), 1.65ꢀ1.98 (m, 4 H), 2.53 (s, 3 H), 3.64 (m, 2 H), 4.35 (m, 1 H), 4.76
(m, 2 H), 5.14 (s, 2 H), 5.54 (d, 1 H), 6.95 (br, 1 H), 7.28ꢀ7.39 (m, 5 H),
7.51 (dd, 2 H), 7.64 (dd, 1 H), 7.74 (br, 1 H), 7.96 (d, 2 H). 13C NMR
(CDCl3): δ = 22.72, 27.11, 32.83, 34.21, 46.07, 46.45, 54.49, 67.37, 128.09,
128.18, 128.45, 128.75, 129.16, 134.39, 134.48, 136.18, 156.55, 171.86,
194.05, 201.13. ESI-MS (m/z): 456.26 [M + H]+. Anal. (C24H29N3O4S)
C, H, N.
In Vitro Assay for SIRT1 and SIRT2 Activities. The Fluor de Lys
fluorescence assays were based on the method described in the BioMol
product sheet using BioMol KI177 substrate for SIRT1 and KI179
substrate for SIRT2. Determined Km for SIRT1 substrate was 58 μM
and for SIRT2 substrate 198 μM.16
Briefly, assays were carried out using Fluor de Lys acetylated 40 μM
SIRT1- or 138 μM SIRT2-peptide substrate (concentrations were 70%
of Km values), 500 μM NAD+ (N6522, Sigma), recombinant GST-
SIRT1/2-enzyme and SIRT assay buffer (HDAC assay buffer, KI143,
supplemented with 1 mg/mL bovine serum albumin (BSA), A3803, Sigma).
GST-SIRT1-enzyme and GST-SIRT2-enzyme were produced as described
recently.39,40 The buffer, SIRT1/2-peptide substrate, NAD+ and dimethyl
sulfoxide (DMSO)/compounds in DMSO (2.5 μL in 50 μL total volume of
reaction mixture; DMSO from Sigma, D2650) for testing were preincubated
for 5 min at rt. The reaction was started by adding the SIRT1- or SIRT2-
enzyme. The reaction mixture was incubated for 1 h at 37 ꢀC. After that Fluor
de Lys developer (KI176) plus 2 mM nicotinamide in 50 μL were added and
incubation was continued for 45 min at 37 ꢀC. Fluorescence readings were
obtained using the Victor 1420 Multilabel Counter (Wallac, Finland) with
excitation wavelength 355 nm and emission 460 nm.
The IC50 values were based on 9-point doseꢀresponse determination
(2000 μM, 1000 μM, 100 μM, 10 μM, 1 μM, 0.1 μM, 0.01 μM, 0.001 μM
and 0.0001 μM) where more necessary dose points were added between
the critical concentrations depending on the compound. Each experiment
was repeated at least three times and calculated using Graph Pad Prism
Software version 4.03 ( 19922005 GraphPad Software, Inc.). The SIRT1
and SIRT2 assays differ in their active enzyme concentrations, and
consequently, SIRT1 and SIRT2 IC50 values cannot bedirectly compared.
In Vitro Assay for HDAC Activities. The Fluor de Lys fluores-
cence assay was based on the method described in the product sheet for
HDAC Fluorimetric Assay/Drug Discovery Kit of Enzo Life Sciences
(BML-AK500).
3
3
(S)-2-((S)-6-Ethanethioamido-2-(3-(2-fluorophenyl)propa-
namido)hexanamido)propanoic Acid (22). 1H NMR ((CD3)2SO):
δ = 1.19ꢀ1.31 (m, 2 H), 1.26 (d, 3 H), 1.41ꢀ1.67 (m, 4 H), 2.37 (s, 3 H),
2.44 (m, 2 H), 2.82 (dd, 2 H), 3.40 (m, 2 H), 4.18 (m, 1 H), 4.28 (m, 1 H),
7.06ꢀ7.16 (m, 2 H), 7.19ꢀ7.32 (m, 2 H), 7.97 (d, 1 H), 8.18 (d, 1 H), 9.90
(m, 1 H), 12.33 (br, 1 H). 13C NMR ((CD3)2SO): δ = 17.03, 22.74, 24.17
(JCF = 2.49 Hz), 26.91, 31.90, 32.78, 35.06, 45.36, 47.36, 51.89, 114.98 (d,
JCF = 21.88 Hz), 124.24 (d, JCF = 3.32 Hz), 127.77 (d, JCF = 15.72 Hz),
127.96 (d, JCF = 8.09 Hz), 130.58 (JCF = 4.89 Hz), 160.43 (JCF = 243.12
Hz), 170.87, 171.47, 173.98, 198.73. ESI-MS (m/z): 426.23 [M + H]+.
Anal. (C20H28FN3O4S 0.4AcOH 0.05hexane) C, H, N.
3
3
(S)-2-((S)-6-Ethanethioamido-2-(3-(2-hydroxyphenyl)propa-
namido)hexanamido)propanoic Acid (23). 1HNMR((CD3)2SO):
δ = 1.21ꢀ1.33 (m, 2 H), 1.27 (d, 3 H), 1.42ꢀ1.68 (m, 4 H), 2.37 (s, 3 H),
2.39 (m, 2 H), 2.72 (dd, 2 H), 3.41 (m, 2 H), 4.18 (m, 1 H), 4.29 (m, 1 H),
6.68(m, 1H), 6.76(d, 1H), 6.98(m, 1H),7.04(d, 1H), 7.91(d, 1H), 8.17
(d, 1 H), 9.28 (br, 1 H), 9.91 (m, 1 H), 12.29 (br, 1 H). 13C NMR
((CD3)2SO): δ = 17.05, 22.79, 25.62, 26.92, 31.89, 32.79, 35.09, 45.36,
47.39, 51.91, 114.87, 118.83, 126.88, 127.41, 129.54, 155.02, 171.54, 171.75,
173.98, 198.73. ESI-MS (m/z): 424.20 [M + H]+. Anal. (C20H29N3O5S
3
0.1H2O 0.7AcOH) C, H, N.
3
(S)-2-((S)-2-((R)-1-(tert-Butoxycarbonyl)piperidine-3-carbo-
xamido)-6-ethanethioamidohexanamido)propanoic Acid (24).
1H NMR ((CD3)2SO): δ = 1.26 (d, 3 H), 1.21ꢀ1.37 (m, 4 H), 1.39 (s,
9 H), 1.43ꢀ1.86 (m, 6 H), 2,32 (m, 1 H), 2.37 (s, 3 H), 2.59ꢀ2.85 (m, 2 H),
3.43 (m, 2 H), 3.78ꢀ3.97 (m, 2 H), 4.16 (m, 1 H), 4.24 (m, 1 H), 7.98 (d,
1H), 8.09(m, 1H), 9.92(m, 1H), 12.36(br, 1H). 13CNMR((CD3)2SO):
δ = 17.11, 21.03, 22.79, 22.84, 24.10 (br), 26.88, 27.44, 28.04, 31.73, 32.78,
41.86, 45.33, 47.38, 51.79, 78.65, 153.80, 171.39, 172.68, 173.99, 198.75. ESI-
MS (m/z): 487.12 [M + H]+, 509.32 [M + Na]+. Anal. (C22H38N4O6S
3
0.4AcOH 0.2hexane) C, H, N.
(S)-Benzyl 1-(Cyclohexylamino)-6-ethanethioamido-1-ox-
3
1
ohexan-2-ylcarbamate (27). H NMR (CDCl3): δ = 1.04ꢀ1.95
(m, 16H), 2.54 (s, 3 H), 3.55ꢀ3.78(m, 3 H), 4.09 (m, 1 H), 5.11 (s, 2 H),
5.45 (d, 1 H), 5.86 (d, 1 H), 7.29ꢀ7.39 (m, 5 H), 7.72 (m, 1 H). 13C
NMR (CDCl3): δ = 22.67, 24.88, 25.54, 26.99, 32.56, 33.01, 33.12, 34.22,
45.89, 48.64, 54.51, 67.28, 128.15, 128.45, 128.74, 136.23, 156.59, 170.52,
201.08. ESI-MS (m/z): 420.30 [M + H]+, 442.32 [M + Na]+. Anal.
(C22H33N3O3S) C, H, N.
Briefly, assay was carried out using Fluor de Lys acetylated 50 μM HDAC-
peptide substrate, human cervical cancer cell line (HeLa) nuclear extract
dilution as recommended in the kit and 200 μM test compounds diluted in
HDAC assay buffer. The reaction was started by adding the substrate as
recommended in the kit protocol. The reaction mixture was incubated for 1 h
at 37 ꢀC. After that Fluor de Lys developer plus 1 μM nicotinamide was added
and incubation was continued for 15 min at 25 ꢀC. Fluorescence readings
were obtained using the Victor 1420 Multilabel Counter (Wallac, Finland)
with excitation wavelength 355 nm and emission 460 nm.
(S)-Benzyl 6-Ethanethioamido-1-(2-fluorophenylamino)-
1-oxohexan-2-ylcarbamate (28). H NMR (CDCl3): δ = 1.50
1
(m, 2 H), 1.66ꢀ2.05 (m, 4 H), 2.53 (s, 3 H), 3.66 (m, 2 H), 4.36 (m,
1 H), 5.15 (s, 2 H), 5.43 (d, 1 H), 7.03ꢀ7.16 (m, 3 H), 7.28ꢀ7.40 (m,
5 H), 7.53 (br, 1 H), 8.11 (br, 1 H), 8.19 (dd, 1 H). 13C NMR (CDCl3):
δ = 22.82, 27.23, 31.89, 34.30, 45.81, 55.42, 67.65, 115.18 (d, J = 19.17
Hz), 122.21, 124.70 (d, J = 3.65 Hz), 125.25 (d, J = 7.63 Hz), 125.77
(J = 10.05 Hz), 128.25, 128.54, 128.77, 135.98, 152.94 (J = 244.37),
The experiment was repeated twice. Means and standard deviations
were calculated using SPSS Software version 14.0 for Windows (IBM
Corporation, USA).
6466
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Cell Culture. Clonetics normal human astrocytes (NHA) were
purchased from Lonza (Basel, Switzerland). Cells were cultured in
Dulbecco’s modified Eagle medium (DMEM) (Sigma, St. Louis, MO,
USA) with the astrocyte growth medium (AGM) supplement (Lonza,
Basel, Switzerland), 10% fetal bovine serum (Lonza), 100 units/mL
penicillin (Lonza), 100 μg/mL streptomycin (Lonza), and 2 mM
glutamine (Lonza). Human retinal pigment epithelial cells (ARPE-19)
were obtained from American Type Culture Collection (ATCC). The
cells were grown in DMEM/Nut MIX F-12 medium (Lonza) including
10% fetal bovine serum (Hyclone, Logan, UT, USA), 100 units/mL
penicillin, 100 μg/mL streptomycin, and 2 mM glutamine. SH-SY5Y
neuroblastomas (DSMZ, Braunschweig, Germany) were cultured in 50%
DMEMꢀ50% optimized modified Eagle medium (OPTI-MEM) (Gibco,
Rockville, MD, USA) containing 5% fetal bovine serum, 100 units/mL
penicillin (Lonza), 100 μg/mL streptomycin (Lonza), and 2 mM gluta-
mine (Lonza). All celltypes were platedto12-well plates (Nunc, Roskilde,
Denmark) at density 105 cells/well, and the experiments were initiated
after 24 h. Test compounds and etoposide (Sigma, St. Louis, MO, USA)
were added at the same time and incubated for 5 h before harvesting.
Western Blotting. For the Western blot analysis, the cells were
lysed into the M-PER Mammalian Protein Extraction Reagent (Thermo
Fisher Scientific, Waltham, MA, USA). Before estimation of protein
levels with the DC protein assay kit (Bio-Rad Laboratories, Hercules,
CA, USA) the lysates were centrifuged at 16000g for 20 min. Equal
amounts of total protein were electrophoretically separated in 4ꢀ12%
SDSꢀPAGE gel (Invitrogen, Carlsbad, CA, USA). Subsequently, proteins
were transferred onto Hybond-ECL nitrocellulose transfer membrane
(Amersham, GE Healthcare, London, U.K.) using electroblotting (Trans-
Blot Cell wet blotting apparatus, Bio-Rad Laboratories, Hercules, CA,
USA). The specific protein was detected with rabbit monoclonal acetyl-p53
(K382) antibody (Abcam, Cambridge, U.K.). Actin (H-196, sc-7210, Santa
Cruz, CA, USA) was used as loading control. The staining of specific
proteins was carried out using horseradish peroxidase (HRP)-linked
secondary antibody (donkey-anti-rabbit, NA934, GE Healthcare, London,
U.K.) and chemiluminescent substrate (Immobilon Western Chemi-
luminescent HRP Substrate, Millipore, Billerica, MA, USA). Filters were
exposed to a medical X-ray film (SuperRX-film, Fujifilm Corporation,
Tokyo, Japan), developed with Kodak X-OMAT 2000 developing machine
(Eastman Kodak Company, Rochester, NY, USA) and scanned with the
Epson Perfection V750 Pro scanner (Epson America Inc., Long Beach, CA,
USA) using Jasc Paint Shop Pro 8 software (Jasc Software Inc., Eden Prairie,
MN, USA). Images were processed with the use of UN-SCAN-IT gel
software (Silk Scientific Corporation, Orem, UT, USA).
’ ACKNOWLEDGMENT
We thank Prof. Kristina Luthman for helpful discussions
during this work and M.Sc. Marko Lehtonen, Tanja Bruijn, Miia
Reponen, Tiina Koivunen and Anneleena Holopainen for their
skillfull assistance. We thank the Graduate School of Drug
Discovery, Academy of Finland (Grants 127062 and 132780),
Orion Farmos Research Foundation, Finnish Cultural Founda-
tion, Emil Aaltonen Foundation and the Saastamoinen Founda-
tion for financial support. We thank Biocenter Kuopio for the
facilities and the CSC-IT center of Science Limited for providing
software licenses for the Shcr€odinger software package. This
work is part of COST Action TD09056: “Epigenetics: Bench to
Bedside”.
’ ABBREVIATIONS USED
SIRT1ꢀ7, human silent information regulator type 1ꢀ7;
Sir2Tm, silent information regulator 2 Thermotoga maritima;
NAD+, nicotinamide adenine dinucleotide; cbz, carbonyloxy-
benzyl; fmoc, 9-fluorenylmethoxycarbonyl; SPPS, solid phase
peptide synthesis; DIPEA, N,N-diisopropylethylamine; DMF,
N,N-dimethylformamide; TBTU, O-(benzotriazol-1-yl)-N,
N,N0,N0-tetramethyluronium tetrafluoroborate; SAR, structure
activity relationship; rmsd, root-mean-square deviation; Boc,
tert-butoxycarbonyl; HDAC, histone deacetylase; SH-SY5Y, hu-
man derived neuroblastoma cell line; DMSO, dimethyl sulfoxide;
DCM, dichloromethane; GST, glutathione-S-transferase; BSA,
bovine serum albumin; DMEM, Dulbecco’s modified Eagle
medium; AGM, astrocyte growth medium; OPTI-MEM, opti-
mized modified Eagle medium; SDSꢀPAGE, sodium dodecyl
sulfate polyacrylamide gel electrophoresis; HRP, horseradish
peroxidase; LDH, lactate dehydrogenase
’ REFERENCES
(1) Haigis, M. C.; Sinclair, D. A. Mammalian sirtuins: biological
insights and disease relevance. Annu. Rev. Pathol. 2010, 5, 253–295.
(2) Lavu, S.; Boss, O.; Elliott, P. J.; Lambert, P. D. Sirtuins—novel
therapeutic targets to treat age-associated diseases. Nat. Rev. Drug
Discovery 2008, 7, 841–853.
(3) Milne, J. C.; Lambert, P. D.; Schenk, S.; Carney, D. P.; Smith, J. J.;
Gagne, D. J.; Jin, L.; Boss, O.; Perni, R. B.; Vu, C. B.; Bemis, J. E.; Xie, R.;
Disch, J. S.; Ng, P. Y.; Nunes, J. J.; Lynch, A. V.; Yang, H.; Galonek, H.;
Israelian, K.; Choy, W.; Iffland, A.; Lavu, S.; Medvedik, O.; Sinclair,
D. A.; Olefsky, J. M.; Jirousek, M. R.; Elliott, P. J.; Westphal, C. H. Small
molecule activators of SIRT1 as therapeutics for the treatment of type 2
diabetes. Nature 2007, 450, 712–716.
Lactate Dehydrogenase Assay. Lactate dehydrogenase (LDH)
leakage from the cells to medium was used as a marker for cytotoxicity.
LDH was measured from cell culture medium with the cytotoxicity kit
obtained from Promega (G1780).
(4) Taylor, D. M.; Maxwell, M. M.; Luthi-Carter, R.; Kazantsev,
A. G. Biological and potential therapeutic roles of sirtuin deacetylases.
Cell. Mol. Life Sci. 2008, 65, 4000–4018.
’ ASSOCIATED CONTENT
(5) Peck, B.; Chen, C. Y.; Ho, K. K.; Di Fruscia, P.; Myatt, S. S.;
Coombes, R. C.; Fuchter, M. J.; Hsiao, C. D.; Lam, E. W. SIRT inhibitors
induce cell death and p53 acetylation through targeting both SIRT1 and
SIRT2. Mol. Cancer Ther. 2010, 9, 844–855.
(6) Neugebauer, R. C.; Sippl, W.; Jung, M. Inhibitors of NAD+
dependent histone deacetylases (sirtuins). Curr. Pharm. Des. 2008,
14, 562–573.
S
Supporting Information. Detailed experimental proce-
b
dures for the syntheses and NMR spectra for the compounds
4ꢀ24 and 26ꢀ30, ESI-MS results, elemental analysis data and
additional docking pose figures. This material is available free of
charge via the Internet at http://pubs.acs.org.
(7) Mai, A.;Altucci, L. Epi-drugstofight cancer: from chemistry to cancer
treatment, the road ahead. Int. J. Biochem. Cell Biol. 2009, 41, 199–213.
(8) Guarani, V.; Potente, M. SIRT1—a metabolic sensor that
controls blood vessel growth. Curr. Opin. Pharmacol. 2010, 10, 139–145.
(9) Alcain, F. J.; Villalba, J. M. Sirtuin inhibitors. Expert Opin. Ther.
Pat. 2009, 19, 283–294.
’ AUTHOR INFORMATION
Corresponding Author
*Tel: +358 40 3553693. Fax: +358 17 162424. E-mail: Heikki.Salo@
uef.fi.
Author Contributions
‡These authors contributed equally to this work.
(10) Luthi-Carter, R.; Taylor, D. M.; Pallos, J.; Lambert, E.; Amore,
A.; Parker, A.; Moffitt, H.; Smith, D. L.; Runne, H.; Gokce, O.; Kuhn, A.;
6467
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468

Journal of Medicinal Chemistry
ARTICLE
Xiang, Z.; Maxwell, M. M.; Reeves, S. A.; Bates, G. P.; Neri, C.;
Thompson, L. M.; Marsh, J. L.; Kazantsev, A. G. SIRT2 inhibition
achieves neuroprotection by decreasing sterol biosynthesis. Proc. Natl.
Acad. Sci. U.S.A. 2010, 107, 7927–7932.
(11) Outeiro, T. F.; Kontopoulos, E.; Altmann, S. M.; Kufareva, I.;
Strathearn, K. E.; Amore, A. M.; Volk, C. B.; Maxwell, M. M.; Rochet,
J. C.; McLean, P. J.; Young, A. B.; Abagyan, R.; Feany, M. B.; Hyman,
B. T.; Kazantsev, A. G. Sirtuin 2 inhibitors rescue alpha-synuclein-
mediated toxicity in models of Parkinson’s disease. Science 2007,
317, 516–519.
(30) CombiGlide, version 2.5; Schr€odinger, LLC: New York, NY,
2009.
(31) Glide, version 5.5; Schr€odinger, LLC: New York, NY, 2009.
(32) Schr€odinger Suite 2009 Protein Preparation Wizard, Epik version
2.0; Impact version 5.5; Prime version 2.1; Schr€odinger, LLC: New York,
NY, 2009.
(33) Kerwin, S. M. ChemBioOffice Ultra 2010 suite. J. Am. Chem.
Soc. 2010, 132, 2466–2467.
(34) Discovery Studio Modeling Environment, Release 2.5; Accelrys
Software Inc.: San Diego, 2009.
(12) Fatkins, D. G.; Monnot, A. D.; Zheng, W. Nepsilon-thioacetyl-
lysine: a multi-facet functional probe for enzymatic protein lysine
Nepsilon-deacetylation. Bioorg. Med. Chem. Lett. 2006, 16, 3651–3656.
(13) Fatkins, D. G.; Zheng, W. Substituting N-thioacetyl-lysine for
N-acetyl-lysine in peptide substrates as a general approach to inhibiting
human NAD-dependent protein deacetylases. Int. J. Mol. Sci. 2008,
9, 1–11.
(14) Smith, B. C.; Denu, J. M. Acetyl-lysine analog peptides as
mechanistic probes of protein deacetylases. J. Biol. Chem. 2007, 282,
37256–37265.
(15) Suzuki, T.; Asaba, T.; Imai, E.; Tsumoto, H.; Nakagawa, H.;
Miyata, N. Identification of a cell-active non-peptide sirtuin inhibitor
containing N-thioacetyl lysine. Bioorg. Med. Chem. Lett. 2009, 19,
5670–5672.
(16) Kiviranta, P. H.; Suuronen, T.; Wallꢀen, E. A. A.; Lepp€anen, J.;
Tervonen, J.; Kyrylenko, S.; Salminen, A.; Poso, A.; Jarho, E. M.
N(epsilon)-thioacetyl-lysine-containing tri-, tetra-, and pentapeptides
as SIRT1 and SIRT2 inhibitors. J. Med. Chem. 2009, 52, 2153–2156.
(17) Huhtiniemi, T.; Suuronen, T.; Lahtela-Kakkonen, M.; Bruijn,
T.; J€a€askel€ainen, S.; Poso, A.; Salminen, A.; Lepp€anen, J.; Jarho, E.
N(epsilon)-Modified lysine containing inhibitors for SIRT1 and SIRT2.
Bioorg. Med. Chem. 2010, 18, 5616–5625.
(35) The Universal Protein Resource (UniProt) in 2010. Nucleic
Acids Res. 2010, 38, D142ꢀ148.
(36) Larkin, M. A.; Blackshields, G.; Brown, N. P.; Chenna, R.;
McGettigan, P. A.; McWilliam, H.; Valentin, F.; Wallace, I. M.; Wilm, A.;
Lopez, R.; Thompson, J. D.; Gibson, T. J.; Higgins, D. G. Clustal W and
Clustal X. Bioinformatics 2007, 23, 2947–2948.
(37) Waterhouse, A. M.; Procter, J. B.; Martin, D. M.; Clamp, M.;
Barton, G. J. Jalview Version 2—a multiple sequence alignment editor
and analysis workbench. Bioinformatics 2009, 25, 1189–1191.
(38) rmsd.py version 1.14.2.6, Maestro version 9.1; Schrodinger, LLC:
New York, NY, 2010.
(39) Tervo, A. J.; Kyrylenko, S.; Niskanen, P.; Salminen, A.;
Lepp€anen, J.; Nyr€onen, T. H.; J€arvinen, T.; Poso, A. An in silico
approach to discovering novel inhibitors of human sirtuin type 2.
J. Med. Chem. 2004, 47, 6292–6298.
(40) Kiviranta, P. H.; Lepp€anen, J.; Rinne, V. M.; Suuronen, T.;
Kyrylenko, O.; Kyrylenko, S.; Kuusisto, E.; Tervo, A. J.; J€arvinen, T.;
Salminen, A.; Poso, A.; Wallꢀen, E. A. A. N-(3-(4-Hydroxyphenyl)-
propenoyl)-amino acid tryptamides as SIRT2 inhibitors. Bioorg. Med.
Chem. Lett. 2007, 17, 2448–2451.
(18) Smith, B. C.; Denu, J. M. Mechanism-based inhibition of Sir2
deacetylases by thioacetyl-lysine peptide. Biochemistry 2007, 46,
14478–14486.
(19) Smith, B. C.; Denu, J. M. Sir2 deacetylases exhibit nucleophilic
participation of acetyl-lysine in NAD+ cleavage. J. Am. Chem. Soc. 2007,
129, 5802–5803.
(20) McGregor, D. P. Discovering and improving novel peptide
therapeutics. Curr. Opin. Pharmacol. 2008, 8, 616–619.
(21) Asaba, T.; Suzuki, T.; Ueda, R.; Tsumoto, H.; Nakagawa, H.;
Miyata, N. Inhibition of human sirtuins by in situ generation of an
acetylated lysine-ADP-ribose conjugate. J. Am. Chem. Soc. 2009, 131,
6989–6996.
(22) Huhtiniemi, T.; Wittekindt, C.; Laitinen, T.; Lepp€anen, J.;
Salminen, A.; Poso, A.; Lahtela-Kakkonen, M. Comparative and phar-
macophore model for deacetylase SIRT1. J. Comput.-Aided. Mol. Des.
2006, 20, 589–599.
(23) Finnin, M. S.; Donigian, J. R.; Pavletich, N. P. Structure of the
histone deacetylase SIRT2. Nat. Struct. Biol. 2001, 8, 621–625.
(24) Jin, L.; Wei, W.; Jiang, Y.; Peng, H.; Cai, J.; Mao, C.; Dai, H.;
Choy, W.; Bemis, J. E.; Jirousek, M. R.; Milne, J. C.; Westphal, C. H.;
Perni, R. B. Crystal structures of human SIRT3 displaying substrate-
induced conformational changes. J. Biol. Chem. 2009, 284, 24394–
24405.
(25) Cosgrove, M. S.; Bever, K.; Avalos, J. L.; Muhammad, S.; Zhang,
X.; Wolberger, C. The structural basis of sirtuin substrate affinity.
Biochemistry 2006, 45, 7511–7521.
(26) MOE (The Molecular Operating Environment), Version 2008.10
and 2009.10; software available from Chemical Computing Group Inc.,
1010 Sherbrooke Street West, Suite 910, Montreal, Canada H3A 2R7.
http://www.chemcomp.com.
(27) Solomon, J. M.; Pasupuleti, R.; Xu, L.; McDonagh, T.; Curtis,
R.; DiStefano, P. S.; Huber, L. J. Inhibition of SIRT1 catalytic activity
increases p53 acetylation but does not alter cell survival following DNA
damage. Mol. Cell. Biol. 2006, 26, 28–38.
(28) LigPrep, version 2.3; Schr€odinger, LLC: New York, NY, 2009.
(29) Epik, version 2.0; Schr€odinger, LLC: New York, NY, 2009.
6468
dx.doi.org/10.1021/jm200590k |J. Med. Chem. 2011, 54, 6456–6468