Structure-activity studies on splitomicin derivatives as sirtuin inhibitors and computational prediction of binding mode

Article abstract of DOI:10.1021/jm700972e

NAD⁺-dependent histone deacetylases (sirtuins) are enzymes that cleave acetyl groups from lysines in histones and other proteins. Potent selective sirtuin inhibitors are interesting tools for the investigation of the biological functions of those enzymes and may be future drugs for the treatment of cancer. Splitomicin was among the first two inhibitors that were discovered for yeast sirtuins but showed rather weak inhibition on human enzymes. We present detailed structure-activity relationships on splitomicin derivatives and their inhibition of recombinant Sirt2. To rationalize our experimental results, ligand docking followed by molecular mechanics Poisson-Boltzmann/surface area (MM-PBSA) calculations were carried out. These analyses suggested a molecular basis for the interaction of the beta-phenylsplitomicins with human Sirt2. Protein-based virtual screening resulted in the identification of a novel Sirt2 inhibitor chemotype. Selected inhibitors showed antiproliferative properties and tubulin hyperacetylation in MCF7 breast cancer cells and are promising candidates for further optimization as potential anticancer drugs.

Full text of DOI:10.1021/jm700972e

J. Med. Chem. 2008, 51, 1203–1213
1203
Structure–Activity Studies on Splitomicin Derivatives as Sirtuin Inhibitors and Computational
Prediction of Binding Mode
Robert C. Neugebauer,† Urszula Uchiechowska,‡ Rene Meier,‡ Henning Hruby,† Vassil Valkov,† Eric Verdin,§
Wolfgang Sippl,‡ and Manfred Jung*,†
Institute of Pharmaceutical Sciences, Albert-Ludwigs-UniVersität Freiburg, Albertstr. 25, 79104 Freiburg, Germany, Department of
Pharmaceutical Chemistry, Martin-Luther UniVersität Halle-Wittenberg, Wolfgang-Langenbeckstr. 4, 06120 Halle/Saale, Germany, and
Gladstone Institute of Virology and Immunology, UniVersity of California, 1650 Owens Street, San Francisco, California 94158
ReceiVed August 6, 2007
NAD+-dependent histone deacetylases (sirtuins) are enzymes that cleave acetyl groups from lysines in histones
and other proteins. Potent selective sirtuin inhibitors are interesting tools for the investigation of the biological
functions of those enzymes and may be future drugs for the treatment of cancer. Splitomicin was among the
first two inhibitors that were discovered for yeast sirtuins but showed rather weak inhibition on human
enzymes. We present detailed structure–activity relationships on splitomicin derivatives and their inhibition
of recombinant Sirt2. To rationalize our experimental results, ligand docking followed by molecular mechanics
Poisson–Boltzmann/surface area (MM-PBSA) calculations were carried out. These analyses suggested a
molecular basis for the interaction of the beta-phenylsplitomicins with human Sirt2. Protein-based virtual
screening resulted in the identification of a novel Sirt2 inhibitor chemotype. Selected inhibitors showed
antiproliferative properties and tubulin hyperacetylation in MCF7 breast cancer cells and are promising
candidates for further optimization as potential anticancer drugs.
Chart 1. Known Sirtuin Inhibitors
Introduction
Histone deacetylases (HDACsa) are transcriptional regulators
that deacetylate histones and various nonhistone proteins. This
enzymatic activity is affecting the conformational state and the
activities of the substrate–proteins.1 Three classes of histone
deacetylases have been described in humans: class I and II have
been shown to be zinc-dependent amidohydrolases and 11
subtypes have been described (HDAC1-11). Class III enzymes
rely in their catalysis on NAD+ with the subsequent formation
of nicotinamide and O-acetyl ADP ribose (Chart 1) as a result
of the transacetylation. Based on the homology to the yeast
histone deacetylase Sir2p, the NAD+-dependent deacetylases
have been termed sirtuins and seven members (Sirt1–7) have
been identified in humans.2
Whereas class I and II histone HDACs have been identified
as bona fide anticancer targets and clinical studies of their
inhibitors as new anticancer agents are under way,3 much less
is known about the consequences of class III HDAC inhibition.2
Sirtuins have been linked to aging and overexpression of sirtuins
leads to a prolonged lifespan in yeast.4 Lately, sirtuins activity
has been tied to the pathogenesis of HIV5 and cancer6–8 and
also neurological diseases.9,10 Only a limited number of sirtuin
inhibitors is known and some of them do not inhibit human
subtypes.11,12 The first synthetic inhibitor that was discovered
is sirtinol (1;13 Chart 1), but in certain cases, it has been shown
that precipitation of the enzyme by the inhibitor contributes to
the in vitro inhibition.14 Structure–activity relationships of
sirtinol analogues have been reported very recently.15 Shortly
after sirtinol, splitomicin (2a) was published as an inhibitor of
yeast sirtuins16 that is not active on human subtypes. Initial
structure–activity relationships on 2a have been described,17,18
and for the first time selective inhibition of individual subtypes
(here yeast enzymes) could be demonstrated. Compound 2b
(HR73) was introduced as another splitomicin derivative by us
that was the first potent inhibitor (IC50 < 5 µM) of human
isoforms5 (see Chart 1). More inhibitors are available for Sirt2
that have been discovered using a virtual screening approach
but cellular activity and verification of protein hyperacetylation
has not been demonstrated.19,20 Lately, indoles like 3 with
activity down to around 0.1 µM have been presented.21 The
* To whom correspondence should be addressed. Tel.: +49-761-203-
4896. Fax: +49-761-203-6321. E-mail: manfred.jung@pharmazie.uni-
freiburg.de.
†
Albert-Ludwigs-Universität Freiburg.
Martin-Luther Universität Halle-Wittenberg.
University of California.
‡
§
a Abbreviations: HDAC, histone deacetylase; MM-PBSA, molecular
mechanics Poisson–Boltzmann/surface area; IPTG, isopropyl-ꢀ-D-thioga-
lactopyranosid; SAHA, suberoylanilide hydroxamic acid.
10.1021/jm700972e CCC: $40.75
2008 American Chemical Society
Published on Web 02/13/2008

1204 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5
Neugebauer et al.
Table 1. Inhibition of Sirtuins by Splitomicin Derivatives
Scheme 1. Synthesis of ꢀ-Phenylsplitomicins by Friedel–Crafts
Alkylation
Sirt2 (IC50 ( SE or inhibition
at concentration, µM)
cmpd
substitution
nicotinamide
1
5a
5b
5c
5d
5e
5f
5g
5h
5i
5j
5k
5l
5m
5n
5o
5p
32.3 ( 6.6
53.0 ( 15.8
5.2 ( 1.0
28.5 % @ 20
1.5 ( 0.5
5.4 ( 1.5
R1: Ph
R1: 4-ClPh
R1: 4-MePh
R1: 4-MeOPh
R1: 3,4-diClPh
R1: 4-FPh
42.3 % @ 20
7.9 ( 1.8
sirtuin inhibitor cambinol was discovered from random screen-
ing, and for the first time, anticancer activity in an animal model
could be demonstrated using this compound.22 Using a focused
library screening approach, we had identified kinase inhibitors
such as 4 (Ro-318220) as new lead structures for sirtuin
inhibitors that target the adenosine binding pocket. By applying
computational and competition experiments, we showed that
these inhibitors target the adenosine binding pocket.23 We set
out to perform structure–activity studies on splitomicin deriva-
tives and initial biological results led us to focus on ꢀ-aryl
derivatives and Sirt2 inhibition. Selected compounds were tested
for antiproliferative activity and for intracellular tubulin acety-
lation. To investigate the molecular interactions that are
important in determining the inhibitory activity and also to
clarify the spatial orientations that the splitomicins adopt within
the Sirt2 binding pocket, a molecular modeling study was carried
out. The newly synthesized ꢀ-phenylsplitomicins were first
analyzed by an automated molecular docking (GOLD) followed
by a molecular mechanics Poisson–Boltzmann/surface area
(MM-PBSA)24 analysis. Based on the docking and free energy
calculations, we observed a clear preference for one of the two
ꢀ-phenylsplitomicin stereoisomers. To experimentally verify the
proposed binding mode, the enantiomers of two potent ꢀ-phe-
nylsplitomicins were synthesized and tested. Furthermore,
introducing new analogues of splitomicins identified by virtual
screening, we were able to show that a lactam substitution of
the lactone ring is well tolerated, provided that other features
beneficial for strong sirtuin inhibition are retained. Because the
lactam ring, unlike the lactone ring in splitomicin, is not
expected to hydrolyze this suggests that splitomicins may act
as reversible inhibitors and not as acetylating agents as had been
postulated before.
R1: 3,4-diFPh
R1: 3-BrPh
R1: 3-FPh
19.7 ( 5.5
10.1 ( 2.6
3.2 ( 0.8
R1: 3-MeOPh
R1: 4-tert.-BuPh
R1: 2-MeO
R1: 3-Cl-4-MePh
R1: 3-F-4-MePh
R1: 3-MeO-4-MePh
R1: H
2.3 ( 0.3
46.8 % @ 20
17.4 % @ 20
2.3 ( 0.6
2.0 ( 0.7
6.3 % @ 20
no inhib. @ 20
21.1 % @ 40
9.6 % @ 40
10.6 ( 1.6
5q
5r
8a
(R)-8a
(S)-8a
8b
R1: Bn
R1: Ph; 6-Pos: Br
R1: Ph
R1: Ph
3.4 ( 0.2
R1: Ph
41.8 % @ 100
26.7 % @ 20
1.5 ( 0.6
R1: 4-ClPh
8c
R1: 4-MePh
R1: 4-MePh
R1: 4-MePh
R1: 3-F-4-MePh
R1: 3-MeO-4-MePh
R1: 3-CNPh
R1: 3-CONH2
R1: 3-MePh
R1: 3-HOPh
R1: 4-PhPh
R1: 3-NHCOCH3Ph
R1: 3-BrPh
R: 8-Et
(R)-8c
(S)-8c
8d
8e
8f
8g
8h
8i
8j
8k
8l
9°
9b
9c
9d
9e
9f
9g
9h
9i
9j
9k
9l
9m
9n
9o
1.0 ( 0.3
35.1 % @ 100
3.4 ( 0.3
9.2 % @ 20
3.1 ( 0.5
20.0 ( 7.8
4.4 ( 0.6
15.4 % @ 20
28.3 % @ 20
17.1 % @ 20
9.8 ( 1.2
25.6 % @ 40
19.5 ( 9.7
25.9 % @ 40
26.2 % @ 40
no inhib. @ 40
no inhib. @ 40
18.4 % @ 20
37.5 % @ 20
22.2 % @ 20
20.6 % @ 20
15.1 % @ 20
no inhib. @ 40
21.8 % @ 20
24.0 % @ 20
49.1 % @ 40
R: 8-I
R: 8-CN
R: 8-Ph
R: 8-Bn
R: 8-CONH2
R: 8-(3-Py)
R: 8-(3-HOPh)
R: 8-(3-CH3CONHPh)
R: 8-(4-PhPh)
R: 8-(2-naphthyl)
R: 9-MeO
R: 8-HO
R: 8- CH3CO
R: 8-H
Results and Discussion
Chemistry. The 8-bromo substituent in 2b had been shown
to be crucial for the activity against human sirtuins as compared
to the unsubstituted congeners. Thus, we initially prepared a
series of ꢀ-aryl-8-bromosplitomicins 5a-p and 5r by the acid-
catalyzed cyclization of cinnamic acids 7a and 7c-e with
ꢀ-naphthols 6a-h (see Scheme 1 and Table 1). Similarly, the
use of acrylic acid (7b) instead of a cinnamic acid led to the
ꢀ-unsubstituted compound 5p. As an 8-methyl group proved
to be a good alternative to the 8-bromo-substituent (see below),
we also prepared a series of splitomicins 8 starting from
6-methyl-2-naphthol 6b or the R,ꢀ-unsaturated lactone 10b by
conjugate addition, respectively. Further use of 6-iodo-2-
naphthol (6c), 4,6-dibromo-2-naphthol (6d), 7-methoxy-2-
naphthol (6e), 6-benzyl-2-naphthol (6f), 2,6-dihydroxynaphtha-
lene (6g), and 6-acetyl-2-naphthol (6h) led to a series of
ꢀ-phenylsplitomicins 9 (see Scheme 1).
The 8-methyl-dehydrosplitomicin (10b) was synthesized using
propiolic acid as the alkylating agent, as reported before for
2-naphthol (see Scheme 2).17 The palladium-catalyzed conjugate
addition25 of arylboronic acids served as an alternative route to
ꢀ-arylsplitomicins. Addition of a Grignard reagent26 led to a
ꢀ-benzyl 8-bromosplitomicin 5q (see Scheme 2 and Table 1).
As all derivatives of ꢀ-phenylsplitomicins synthesized so far
were racemates, the respective enantiomers were prepared for

Splitomicin DeriVatiVes as Sirtuin Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5 1205
Scheme 2. ꢀ-Phenylsplitomicins via Conjugate Additions
Scheme 3. Asymmetric Synthesis of ꢀ-Phenylsplitomicins
also between 1 and 5 µM, whereas 5o is inactive. For both 8a
and 8c, the R-enantiomer is much more potent than the S-form.
Except for methyl, all other substituents in the 8-position
(9b-k) or the 9-methoxy group in 9l lead to a significant loss
of activity. Commercially available compounds 11a-d (see
Chart 2) with a lactam structure instead of the lactone ring of
splitomicin showed weak inhibitory potency. However, these
compounds lack the 8-substituent that was shown to be
necessary for good inhibition of human subtypes. They are
considerably more active than the corresponding 8-unsubstituted
lactone 9o and, therefore, 8-substituted lactams seem to be an
interesting target for further structure–activity-relationships. We
have prepared one such compound, 12, and it showed activity
in the low micromolar region (Table 2). Originally, splitomicins
were thought to be acylating inhibitors of sirtuins, and the
reactivity toward hydrolysis had been correlated with en-
zyme–inhibitory potency.17 While we cannot totally rule out
such a mode of inhibition, our docking studies and SAR,
especially with regard to the lactams, make this rather unlikely.
Competition analysis showed that the most potent inhibitor (R)-
8c is practically not competitive with respect to NAD+ (see
Supporting Information).
two potent ꢀ-phenyl-8-methyl-splitomicins applying an asym-
metric conjugate addition protocol using a chiral rhodium
catalyst27 (see Scheme 3). According to the literature, the
stereochemistry of the products can be assigned from the
configuration of the catalyst used,27 for example, the (S)-binap
generates the (S)-arylsplitomicin.
The bromo substituent in the 8-position led to active sirtuin
inhibitors, but also served as a useful starting point for structural
variations via Pd-catalyzed couplings employing the ꢀ-aryl
splitomicins as starting materials. Substituents introduced in the
8-position in this way were methyl-, ethyl-, phenyl-, 3-hydroxy-
phenyl-, biphenyl-, 2-naphthyl-, 3-pyridyl, 3-acetamidophenyl-,
and cyano (see Scheme 4 and Table 1). For alkylgroups,
alkylaluminium compounds served as the reagents,28 for aryl
groups, arylboronic acids under addition of fluoride29 served
as the reagents, and for the cyano group, zinc cyanide served
as the reagent.30 The cyano group in the 8- or the 3′-position
of arylsplitomicins could be converted to a carboxamide function
by mild palladium-catalyzed hydration, as described in the
literature31 (see Scheme 4).
Docking Analysis. All Sirt2 structures contain a conserved
270 amino acid catalytic domain with variable N- and C-termini.
The X-ray structures of several Sirt2 proteins have been
published in the past few years, whereas no 3D structure is
available for Sirt1.33,34 The structure of the catalytic domain
consists of a large classical Rossmann-fold and a small zinc
binding domain. The interface between the large and the small
subdomain is commonly subdivided into A, B, and C pocket.
This division is based on the interaction of adenine (A), ribose
(B) and nicotinamide (C), which are parts of the NAD+ cofactor.
Whereas the interaction of adenine and ribose is similar in all
available sirtuin X-ray structures, the interaction of the nicoti-
namide part is less clearly defined. Several so-called productive
and nonproductive conformations of nicotinamide have been
observed in the crystal structures, reflecting the high flexibility
of this part of the cofactor. In the X-ray structure of a Sirt2
homologue from Archeabacteria it was shown that the acetylated
peptide binds in a cleft between the two domains. The acetyl-
lysine residue inserts into a conserved hydrophobic pocket,
where NAD+ binds nearby. In the case of the human Sirt2 X-ray
structure, which was crystallized as trimer, no structural
information about the NAD+ or substrate binding is available.
However, due to the homology with bacterial sirtuins, it is
obvious that the interaction of NAD+ and substrate with Sirt2
is comparable with the bacterial X-ray structure complexes.
Using a protein-based virtual screening approach (described
in detail in Materials and Methods), we identified structurally
related lactams in the Chembridge database and purchased and
tested selected compounds 11a-d for sirtuin inhibition as well
(see Chart 2). As these compounds showed promising activity
(see below), we also prepared a derivative 12 with a bromo
substituent in the 8-position (see Scheme 5).
Enzyme Inhibition. All splitomicin derivatives were tested
in initial studies for Sirt2 inhibition at concentrations of 20 and
40 µM in a homogeneous deacetylase assay using a fluorescent
lysine derivative14 that was developed in our group.32 Usually,
only if an inhibition of more than 50% at 40 µM or 20 µM was
observed, IC50 determinations were performed. Nicotinamide
and 1 were tested as reference compounds (see Table 1).
Depending on the substitution on the ꢀ-aryl ring, compounds
that contain either a bromo or a methyl substituent in the
8-position can be sirtuin inhibitors with a potency in the low
micromolar region, for example, a 4-methylphenyl group gives
good inhibition in both series (5c, 8c), whereas, for example, a
4-chlorophenyl substituent (5b, 8b) or sterically demanding
residues (5k, 8j) leads to decreased inhibition. Additionally,
polar moieties on the ꢀ-phenyl ring (8g, 8 h, 8k) diminish the
inhibitory potency, but the cyano derivative 8f shows good
inhibition. Compounds 5m-o were prepared as a consequence
of the good inhibition of the compounds 5c, 5i, and 5j, but no
improvement was achieved. Compounds 5m and 5n inhibit Sirt2
We have previously described the development of other series
of Sirt2 inhibitors, including adenosine mimetics, for example,
the bis-indolyl-maleimide 423 and suramin derivatives.35 Based
on docking studies that we carried out for human Sirt2 and
competition experiments with NAD+, we found that the
compounds interact with the adenine (for 4) and the nicotina-
mide subpocket (suramin), respectively. Due to the structural
dissimilarity between these inhibitors and the splitomicins, it
can be expected that they interact in a different way with sirtuin

1206 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5
Scheme 4. Pd-Catalyzed Reactions on Splitomicin Derivatives
Neugebauer et al.
Chart 2. Lactam Analogues of Splitomicins
molecules found in all Sirt2 monomers, which are located at a
narrow cavity nearby the active site, significantly improved the
docking results. The polar moieties of known inhibitors, such
as cambinol or 3, were found to interact with the polar residues
Gln167, Asn168, and the water molecules of the nicotinamide
subpocket (for details, see Supporting Information). The ꢀ-phe-
nylsplitomicins at first were prepared and tested as racemates,
and therefore, both stereoisomers (R and S) were considered in
the docking study. The obtained docking scores and the visual
analysis of the docking poses showed that only the (R)-isomer
is able to favorably interact with the nicotinamide subpocket
(Figure 1a,b). All docked ꢀ-phenylsplitomicins showed the same
binding mode, including a hydrogen bond to the water molecule
bonded to Gln167 (Figure 2). The ꢀ-phenyl substituent of all
(R)-isomers fits into a hydrophobic channel and is sandwiched
between Phe119 and His187. This channel represents the binding
site for the acetylated lysine residue of the substrate. For the
(S)-enantiomers, the docking showed that the lactone ring is
also facing toward the nicotinamide subpocket. However, no
direct hydrogen bond was observed between (S)-8a and the
residues of the active site (Figure 3a,b). Our proposed type of
interaction of (R)-8a and (R)-8c with Sirt2 is similar to docking
results recently published for other Sirt2 inhibitors.9,19 Also, in
these studies, polar moieties of the inhibitors were found to
interact with the nicotinamide binding pocket C (Phe96, Gln167,
Asn 168, Ile169).
Scheme 5. Synthesis of a Bromo-Substituted Lactam Analogue
Interaction possibilities at the binding pockets were analyzed
by calculating the contact preferences using the MOE program.
The purpose of this knowledge-based approach is to calculate
preferred locations for hydrophobic ligand atoms from the 3D
coordinates of the binding site. The visual analysis of the contact
preferences further supported the binding mode obtained by the
GOLD36 docking (Figure 4a,b). The most favorable interaction
with hydrophic ligand atoms was observed in the acetyl-lysine
channel and nearby Tyr104 and Val233. The favored polar
interactions were detected nearby the polar residues Gln167 and
Asn168, which are in agreement with the location of the water
molecules.
The docking study and the visual inspection of the interaction
possibilities clearly suggested that the R-enantiomers represent
the active form of the ꢀ-phenylsplitomicins, whereas for the
S-enantiomer, no clear binding mode was observed. The
experimental results showed that the R-enantiomers are more
than two log units more active than the S-enantiomers, which
is in perfect agreement with the docking results.
Table 2. Sirtuin Inhibition by the Compounds 11 (for Structures, see
Chart 2) and 12 (Scheme 5)
cmpd
Sirt2 (IC50 ( SE or inhibition at concentration, µM)
11a
11b
11c
11d
12
31.5% @ 40
16.9% @ 40
no inhib. @ 20
4.2% @ 40
6.4 µM ( 0.3
proteins. In the present study, the X-ray structure of human Sirt2
was used for an automated ligand docking using the program
GOLD.36 A series of docking simulations were carried out to
better rationalize the binding of the newly synthesized splito-
micin derivatives. The analysis of the X-ray structure of human
Sirt2 (which contains three monomers) and preliminary docking
simulations using the known inhibitors cambinol and the indole
3 showed that the compounds interact with the nicotinamide
subpocket (C) of Sirt2. The consideration of four water
In the next step, we analyzed the obtained docking scores
for all studied inhibitors. In the case of the enantiomers (R)-8a,
(S)-8a, (R)-8c, and (S)-8c, higher scores were observed for the
R-species. The docking results showed that the active inhibitors
can interact in similar ways with the adenine-binding pocket.

Splitomicin DeriVatiVes as Sirtuin Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5 1207
Figure 2. Docking solutions for the R-enantiomers of the active
ꢀ-phenylsplitomicins (orange). The substituted ꢀ-phenyl rings of the
inhibitors are all facing towards the acetyl-lysine binding channel. The
Conolly molecular surface of the binding pocket is colored according
to the electrostatic potential (blue ) positive potential, red ) negative
potential).
Figure 1. (a) Interaction of (R)-8c (magenta) with the amino acid
residues of human Sirt2. The four water molecules used within the
docking study are colored cyan. Hydrogen bonds are indicated by the
red dashed line. (b) Interaction of (R)-8c (magenta) at the nicotinamide
binding site. The Conolly molecular surface of the binding pocket is
colored according the electrostatic potential (blue ) positive potential,
red ) negative potential). (c) Schematic representation of the interaction
between (R)-8c and human Sirt2. Hydrophobic amino acids are colored
green, polar are colored purple. Hydrogen bonds are displayed as a
dashed line.
Figure 3. (a) Predicted binding mode of (S)-8c (green) at the
nicotinamide binding site of human Sirt2. The four water molecules
used within the docking study are colored cyan. The Conolly molecular
surface of the binding pocket is colored according the electrostatic
potential (blue ) positive potential, red ) negative potential). (b)
Schematic representation of the interaction between (S)-8c and human
Sirt2. Hydrophobic amino acids are colored green, polar are colored
purple.

1208 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5
Neugebauer et al.
Table 3. Cytotoxicity on MCF 7 Cells
cmpd
growth inhibition (GI50)
5c
8c
(R)-8c
(S)-8c
8g
43 µM
99 µM
84 µM
74% at 500 µM
64% at 500 µM
57% at 500 µM
9d
Table 4. Energy Contributions to the Free Energy of Binding as
Obtained by the MM-PBSA Approach
Sirt2/(R)-8c Sirt2/(S)-8c Sirt2/(R)-8a Sirt2/(S)-8a
∆Eelect
-7.81
-30.87
-38.68
-3.22
-31.54
-34.76
-2.09
-38.68
-40.77
-0.19
-31.44
-31.63
∆EvdW
∆Egas (∆Eelect
+
∆EvdW
∆Gsolv
T∆Stot
)
9.77
-21.50
-7.41
10.74
-19.29
-4.73
7.63
-25.79
-7.35
5.66
-23.27
-2.7
∆Gcalcd
8c were synthesized and tested as enantiomers, we focused our
free energy calculations on this set of inhibitors.
The binding free energies for the two pairs of stereoisomers
were estimated (Table 4) using the MM-PBSA method described
in the Materials and Methods section. The calculated mean
binding free energies were -7.41 kcal/mol for (R)-8c and -4.73
kcal/mol for (S)-8c, respectively. Also for 8a the MM-PBSA
calculation predicted the R-enantiomer to bind stronger to Sirt2
(R)-8a, -7.35 kcal/mol, vs (S)-8a, -2.70 kcal/mol. Thus,
binding of (R)-8c is 2.68 kcal/mol more favorable than for the
S-enantiomer, whereas in case of 8a the energy difference is
4.65 kcal/mol. The estimated binding free energies are in good
agreement with the biological data where the S-enantiomers are
more than two log units less-active than the R-enantiomers
(∆∆Gexpt > ∼3 kcal/mol, calculated on the basis of the IC50
values measured at 310 K). The detailed analysis of the
hydrogen bonding during the MD simulation showed that the
Sirt2/R-enantiomer complexes have more intermolecular hy-
drogen bonds stabilizing the complex than the complexes with
the S-enantiomers (see Materials and Methods section).
Antiproliferative Activity. Selected inhibitors 5c, 8c, and
both of its enantiomers, 8g and 9d, were tested for inhibition
of proliferation of MCF-7 breast cancer cells using the 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)
assay.43 Generally, the splitomicins are not very potent cytotoxic
agents on those cells. This may be due to their high lipophilicity.
But those compounds that inhibited the enzyme in the low
micromolar region (5c, 8c, (R)-8c) were also the most potent
antiproliferative agents. The weaker enyzme inhibitors showed
some inhibition of cell growth, but even at 500 µM, no complete
growth inhibition and, hence, no IC50 values could be obtained.
(see Table 3).
Protein Hyperacetylation. To verify whether the inhibitors
actually target the desired enzyme at concentrations where
antiproliferative activity was observed, we analyzed tubulin
acetylation by Western blot. Indeed, compounds 5c and (R)-8c
led to tubulin hyperacetylation consistent with Sirt2 inhibition.
The Sirt2 inhibitor 3 and the HDAC class I/II-inhibitor SAHA,
which is known to induce tubulin hyperacetylation via
HDAC644,45 inhibition, were used as reference compounds (see
Supporting Information). Taken together, the cellular data links
anticancer activity to Sirt2 inhibition. It is interesting to note
that the levels of tubulin hyperacetylation that can be reached
using SAHA or similar inhibitors of zinc-dependent HDACs
were never reached with any of the sirtuin inhibitors. This argues
Figure 4. Interaction possibilities at the binding pocket, as analyzed
by calculating the contact preferences using the MOE program. The
docked inhibitors (R-enantiomers) are colored atom-type coded. The
Conolly molecular surface of the binding pocket is colored according
the electrostatic potential (blue ) positive potential, red ) negative
potential). (a) The favourable hydrophobic contact preferences (contour
level 90%) are colored orange. (b) The favourable hydrophilic contact
preferences (contour level 90%) are colored magenta.
However, on the basis of the Goldscore values, no discrimination
could be derived between more- and less-active splitomicin
inhibitors. It is shown in many docking studies that often a low
correlation is observed between docking scores and biological
activities.37 Therefore, we focused on a more accurate method
for the analysis of key interactions necessary for strong
inhibitory activity.
MM-PBSA Approach. To make a more precise and quan-
titative analysis of the protein inhibitor interaction, several
Sirt2-splitomicin complexes were used as starting structures
for MD simulations. These simulations showed that both kinds
of complexes are stable during the 5 ns free MD simulation.
The inhibitor interactions observed in the starting structures
(obtained by means of GOLD) were generally maintained after
the MD simulations. Analyzing the root-mean-square deviation
(rmsd) from the complex structures of all of the heavy atoms
of the proteins, the rmsd remained approximately constant
around the range of 0.7–1.4 Å (for details, see Supporting
Information).
Successively, the MD trajectories were further analyzed
through the MM-PBSA method.38–42 In the MM-PBSA method,
the absolute free energy of a system is estimated from a
combination of molecular mechanics energy, a Poisson–Bolt-
zmann estimate of the electrostatic free energy, an estimate of
the solvation free energy determined from the exposed surface
area, and an estimate of the entropy of the molecule derived
from a normal modes calculation. Because the inhibitors 8a and

Splitomicin DeriVatiVes as Sirtuin Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5 1209
(1.0 mmol), Pd(OAc)2 (0.050 mmol), 2,2′-bipyridine (0.20 mmol),
acetic acid (1.0 mL), THF (0.5 mL), and water (0.3 mL) were added
under nitrogen. The mixture was stirred and heated at 40 °C for
three days and then neutralized with saturated sodium bicarbonate
solution and extracted with diethyl ether. The combined ether
solution was washed with brine, dried (sodium sulfate), and
concentrated. The residue was purified by column chromatography
to give splitomicins.
Asymmetric Conjugate Addition of Boronic Acids, Method
D. A solution of Rh(acac)(C2H4)2 (9.0 µmol), chiral ligand ((R)-
and (S)-binap, respectively; 9.9 µmol), R,ꢀ-unsaturated lactone 10b
(0.30 mmol), and boronic acid (3.0 mmol) in a mixture of 1,4-
dioxane (1.0 mL) and water (0.1 mL) were placed in a sealed tube
under nitrogen and stirred at 60 °C for 14 h. Then the solution was
passed through a short pad of silica gel with tert-butylmethylether
as eluent. The solvent was removed under reduced pressure, and
the resulting residue was purified by column chromatography,
furnishing the desired product.
Pd-Catalyzed Introduction of a Methyl Group, Method E.
The bromo-substituted splitomicin derivative (0.50 mmol), tri-
methylaluminium (0.50 mmol), and tetrakis(triphenylphospine)pal-
ladium (0.025 mmol) in dry 1,4-dioxane were refluxed for 2 h.
After cooling, the reaction mixture was diluted with water. The
product was extracted with dichloromethane and then purified by
column chromatography on silica gel.
Microwave-Accelerated Suzuki Coupling, Method F. To a
stirred mixture of arylboronic acid (0.90 mmol), bromo-substituted
splitomicin derivative (0.60 mmol), and powdered CsF (1.80 mmol)
in a mixture of dry methanol (1.5 mL) and dimethoxyethane (3.0
mL) was added tetrakis(triphenylphospine)palladium(0) (27 µmol).
The reaction mixture was flushed with nitrogen, sealed, and placed
in a microwave field (100 W) for 8 min. Subsequent filtration and
purification by column chromatography afforded the desired
product.
for a larger contribution of tubulin acetylation maintenance by
HDAC6 as compared to Sirt2.
Conclusion
Structure–activity relationships for ꢀ-aryl splitomicins led to
the identification of Sirt2 inhibitors that are active in the low
micromolar region. A link between increased enzyme inhibition
and anticancer activity could be established. This supports the
use of our fluorescent small molecule in sirtuin assays to identify
antiproliferative agents. Docking studies confirmed that in
the whole series of ꢀ-phenylsplitomicines the orientation of the
ꢀ-phenyl group is important for their Sirt2 activity. Only in the
case of the R-enantiomers of the chiral inhibitors the interaction
with the substrate binding channel is observed. The preparation
and biological testing of two pairs of enantiomers confirmed
the docking results. Furthermore, we reported calculations of
binding free energies between these two pairs of enantiomers
and Sirt2 using the recently developed MM-PBSA method. This
approach proved to be attractive for rationalizing at a quantita-
tive manner the interaction of the splitomicines and Sirt2. The
competition experiment carried out with NAD+ clearly showed
that the ꢀ-phenylsplitomicins are noncompetitve to the cofactor.
The comparison of the predicted binding mode of (R)-8a and
the observed interaction of NAD+ in the homologous bacterial
sirtuin X-ray structure complexes further supported the non-
competition between both molecules and is in agreement with
recently published structural data on sirtuin homologues.46
Further optimization of splitomicin-based sirtuin inhibitors and
the novel lactam analogues, therefore, can now be performed
in a more rational fashion.
Pd-Catalyzed Aromatic Cyanation, Method G. The bromo-
substituted splitomicin derivative (0.30 mmol) is combined with
zinc cyanide (0.18 mmol) and tetrakis(triphenylphospine)pal-
ladium(0) (7.2 µmol) in deoxygenated DMF (2.5 mL), and the
yellow slurry was heated to 80 °C under nitrogen. Then the mixture
was cooled down to room temperature, diluted with toluene (10
mL), and washed twice with 5% sodium bicarbonate solution. The
organic layer was collected, washed with brine, and concentrated
under reduced pressure to provide the product, which was purified
by column chromatography on silica gel.
Experimental Section
Materials and Methods. Standard chemicals were purchased
from Sigma, Aldrich, or Lancaster. 3-Methoxy-4-methylbenzalde-
hyde was purchased from Frinton Laboratories (Vineland, NJ) and
3-Fluoro-4-methylcinnamic acid from Matrix Scientific (Columbia,
SC). Compound 1 was purchased from Axxora (Lausen, Switzer-
land). Compound 2a,17 6-methyl-2-naphthol (6b),47 6-iodo-2-
naphthol (6c),48 and 4,6-dibromo-2-naphthol (6d)49 were prepared
according to published procedures. Cinnamic acids 7c-e were
prepared from the corresponding benzaldehydes according to
Pd-Catalyzed Hydration of Nitriles to Amides, Method H.
The cyano-substituted splitomicin derivative (0.5 mmol) was
dissolved in a mixture of THF (0.8 mL) and water (0.2 mL).
Acetamide (2.0 mmol) and Pd(OAc)2 (0.05 mmol) were added, and
the mixture was strirred at room temperature for 6 h. Then the
solvents were removed under reduced pressure and the residue was
subjected to column chromatography to yield the amide.
Synthesis of Inhibitors. The compounds were synthesized using
the above-mentioned methods and general methods for organic
syntheses. They were analyzed for identity and purity using IR,
1
standard procedures. H NMR spectra were obtained on a Bruker
1
Avance DRX 400 MHz spectrometer, and H chemical shifts are
reported in ppm (δ). Merck Kieselgel 60 was used for flash
chromatography with either cyclohexane/ethyl acetate mixtures or
cyclohexane/dichloromethane mixtures as eluents so that the Rf
value of the desired product was about 0.3. Absolute configurations
of enantiomers were assigned from the stereochemistry of the
catalysts according to the literature.27 A CEM Discover was used
as the microwave reactor.
Cyclization of Naphthols with Cinnamic and Acrylic Acids
with Amberlyst 15, Method A. The ꢀ-naphthol (3 mmol), the
cinnamic acid (6 mmol), and Amberlyst 15 ion-exchange resin (300
mg) were refluxed in toluene (15 mL) overnight. The resin was
filtered off and washed with toluene. The solvent was removed
under reduced pressure. Separation by column chromatography led
to the isolation of the splitomicins.
Method B for Deactivated Cinnamic Acids. The naphthol (3
mmol), the deactivated cinnamic acid (6 mmol), and 100 µL of
sulfuric acid were refluxed in toluene (15 mL) overnight. After
cooling, the reaction mixture was washed twice with aqueous
sodium carbonate solution. The organic phases were collected and
dried over sodium sulfate, and the solvent was removed under
reduced pressure. Separation by column chromatography yielded
the desired products.
1
MS, H and 13C NMR, and elemental analyses or HPLC. Enan-
tiomeric excess was determined using HPLC. The details are given
in the Supporting Information (13C NMR data not shown).
Analytical data for three selected compounds is listed below.
8-Bromo-1-phenyl-1,2-dihydro-benzo[f]chromen-3-one (5a).
1
By method A, yield ) 201 mg (19%) as a white solid. H NMR
(CDCl3): 8.02 (d, J ) 2.0 Hz, 1H, 7-H), 7.77 (d, J ) 9.0 Hz, 1H,
10-H), 7.65 (d, J ) 9.0 Hz, 1H, 6-H), 7.52 (dd, J1 ) 9.0 Hz, J2 )
2.0 Hz, 1H, 9-H), 7.29 (d, J ) 9.0 Hz, 1H, 5-H), 7.29–7.20 (m,
3H, Ph-H), 7.10–7.08 (m, 2H, Ph-H), 4.91 (dd, J1 ) 6.7 Hz, J2 )
2.4 Hz, 1H, 1-H), 3.22 (dd, J1 ) 15.7 Hz, J2 ) 6.7 Hz, 1H, 2-H),
3.15 (dd, J1 ) 15.7 Hz, J2 ) 2.4 Hz, 1H, 2-H).
8-Methyl-1-(4-methylphenyl)-1,2-dihydro-benzo[f]chromen-
3-one (8c). By method E, using 5c as the bromo-substituted
splitomicin derivative, yield ) 113 mg (75%) as a white solid. 1H
NMR (CDCl3): 7.79 (d, J ) 8.9 Hz, 1H, 10-H), 7.71 (d, J ) 8.6
Hz, 1H, 6-H), 7.65 (s, 1H, 7-H), 7.33 (d, J ) 8.9 Hz, 1H, 9-H),
Conjugate Addition of Boronic Acids, Method C. To a Schlenk
tube, aryl boronic acid (3.0 mmol), R,ꢀ-unsaturated lactone 10b

1210 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5
Neugebauer et al.
7.32 (d, J ) 8.6 Hz, 1H, 5-H), 7.09 (d, J ) 8.2 Hz, 2H, Ph-H),
7.02 (d, J ) 8.2 Hz, 2H, Ph-H), 4.92 (dd, J1 ) 2.2 Hz, J2 ) 6.6
Hz, 1H, 1-H), 3.20 (dd, J1 ) 6.6 Hz, J2 ) 15.7 Hz, 1H, 2-H), 3.15
(dd, J1 ) 15.7 Hz, J2 ) 2.2 Hz, 1H, 2-H), 2.48 (s, 3H, CH3-Np),
2.28 (s, 3H, CH3-Ph).
structures were energy minimized using the MMFF94s force field
and the conjugate gradient method, until the default derivative
convergence criterion of 0.01 kcal/(mol × Å) was met. The
crystal structure of human Sirt2 (pdb code 1J8F) was taken from
the Protein Data Bank. Sirt2 is a monomer in solution and,
therefore, only the chain B was chosen from the trimeric Sirt2
structure of 1J8F. As in our recent docking study,23 monomer
B of Sirt2 was selected, as it showed the best stereochemical
quality examined with the program PROCHECK. Because the
X-ray structures might contain a strain from the crystallization
process, the Sirt2 structure was first energy minimized. After
adding all hydrogen atoms, including all water molecules, a
descent minimization was carried out on the Sirt2 structure using
the MMF94 force field and the GB/SA continuum51 solvent
model for water. During the minimization, a tethering constant
of 100 kcal/(mol × Å) on the backbone atoms was applied, after
a stepwise reduction of the tethering to 1 kcal/(mol × Å).
Inhibitor Docking. Docking of the newly synthesized inhibitors
was carried out using program GOLD 3.036 and the minimized
native human Sirt2 structure. This approach has been shown to be
successful for ligand docking by us35,52 and others.20 All torsion
angles in each inhibitor were allowed to rotate freely. The binding
site was defined on Ile169 with an radius of 15 Å. Goldscore was
chosen as fitness function due to the success in former docking
studies on sirtuins.23 For each molecule, 10 docking runs were
performed. The resulting solutions were clustered on the basis of
the heavy atom rmsd values (1 Å). The top-ranked poses for each
ligand were retained and analyzed graphically within MOE 2006.08
(Chemical Computing Group).53 Interaction possibilities at the
binding pockets were analyzed by calculating the contact prefer-
ences using the MOE program. The purpose of this knowledge-
based approach is to calculate, from the 3D coordinates of the
binding site, preferred locations for hydrophobic ligand atoms. The
calculation was carried out on the solvent stripped X-ray strucutre
of Sirt2 (1j8f, chain B). The calculated polar and hydrophobic
contact preferences were then viewed superimposed on the crystal
structure and homology models using the MOE program. The
Conolly molecular surface of the binding pocket was calculated
using the MOE program and colored according the molecular
electrostatic potential (using MMF94 charges within MOE). The
conformation of the binding mode of NAD+ was taken from our
former docking studies of sirtuin ligands.23,35
Virtual Screening. We screened the Chembridge database for
inhibitors structurally related to the ꢀ-phenylsplitomicins. The
compounds of the Chembridge database were transformed into
3D molecular structures using the Omega module from OpenEye
Software.54 The ∼328000 molecules were stored in a MOE
database (Chemical Computing Group) and ESshape3D MOE
fingerprints were calculated. The ESShape3D is an eigenvalue
spectrum shape fingerprint. Each fingerprint is a fixed length
and allows for comparison of 3D shapes made from the heavy
atoms of a molecule. Compounds similar to 8a (with an inverse
distance > 0.9) were retrieved and further analyzed. Among the
retrieved compounds 17 lactam analogs of 8a were identified.
The lactams were docked into the human Sirt2 protein structure
using the GOLD program and the same settings as described
above. GoldScores were calculated for all docking poses that
were subsequently visually analyzed within MOE. To confirm
the virtual screening results, four lactam analogs of 8a that
showed a similar binding mode as the ꢀ-phenylsplitomicins were
selected and purchased from Chembridge.
(R)-8-Methyl-1-(4-methylphenyl)-1,2-dihydro-benzo[f]chromen-
3-one ((R)-8c). By method D, using 10b as R,ꢀ-unsaturated
splitomicin derivative, (R)-binap as chiral ligand, and p-tolyl boronic
acid, yield ) 16 mg (18%) as a white solid; ee ) 98.9%, as
determined by chiral HPLC. [R]589 ) 11.3° (c ) 1.0 mg/mL,
1
CH2Cl2, T ) 20 °C); for H NMR data, see 8c.
Recombinant Proteins. Human Sirt2 (N-terminally tagged with
6 His) was prepared as described previously,50 with minor
modifications. In brief, the plasmid pEV1440, containing the full-
length human Sirt2 cDNA, was transformed in E. coli strain BL21
for expression. The culture is grown in LB media to an optical
density of 0.6 (A600) at 37 °C, induced with 0.1 mM IPTG for 2 h,
and pelleted. Lysis is performed using a French press. The soluble
overexpressed recombinant protein was purified using Ni-NTA
resin. The identity of the produced His-Sirt2 was verified using
SDS electrophoresis. Deacetylase activity of the produced Sirt2 was
dependent on NAD+ and could be inhibited with sirtinol and
nicotinamide.
Fluorescent Deacetylase Assay. All compounds were evaluated
for their ability to inhibit recombinant sirtuins using a homogeneous
fluorescent deacetylase assay.32 Stock solutions of inhibitors were
prepared in DMSO, and 3 µL or less of a suited DMSO inhibitor
solution were added to the incubation mixture. The assay was
carried out in 96-well plates: 60 µL reaction volume contained the
fluorescent histone deacetylase substrate ZMAL (10.5 µM), NAD+
(500 µM), and Sirt2 (3 µL). After 4 h of incubation at 37 °C, the
deacetylation reaction was stopped and the formed metabolite ZML
(deacetylated form of ZMAL) was developed using a tryptic digest
to form a fluorophor. Then, fluorescence was measured in a plate
reader (BMG Polarstar) with a coumarin filter (excitation 390 nm
and emission 460 nm). The amount of remaining substrate in the
positive control with inhibitor versus negative control without
inhibitor was employed to calculate inhibition. All determinations
were at least carried out in duplicate. IC50 data were analyzed using
GraphPad Prism Software.
Cytotoxicity Assays. MCF-7 human breast tumor cells were
obtained from Max-Delbrück-Centre of Molecular Medicine (Ber-
lin) and maintained in RPMI-1640 supplemented with 10% fetal
bovine serum, 1% 2 mM L-glutamine, and 1% penicillin/strepto-
mycin (Invitrogen). The cells were incubated at 37 °C in an
atmosphere of 5% CO2. Cells were seeded at 5000 per well in 100
µL of growing medium in 96-well tissue culture plates. At 24 h
after seeding, diluted compounds or DMSO vehicle control were
added to each well and incubated for 72 h at 37 °C. Growth
inhibition was determined using the MTT assay. Data was plotted
as the percentage of DMSO-treated control against compound
concentration using GraphPad Prism 4.0. The 50% growth inhibition
(GI50) was calculated as the compound concentration required to
reduce cell number by 50% compared with control.
Western Blot. Cells were seeded at 500000 per well in 6-well
tissue culture plates. At 24 h after seeding, cells were treated with
diluted compounds or DMSO and incubated for 8 h at 37 °C. The
cells were washed with PBS and lysed in 50 mM Tris-HCl, pH
7.5, 0.5 mM EDTA, 150 mM NaCl, 0.5% NP-40, 1× complete
protease inhibitors (Roche, Germany). Protein concentration was
determined with BCA protein assay kit (Pierce). For Western blot
analysis, protein samples were separated in 30% SDS-PAGE gels
and blotted onto ImmobilonP membrane (Millipore). Membranes
were blocked with Roti-Block (Roth) and probed with antiacetylated
R-tubulin (6-11B1; Sigma) and anti R-tubulin (B-5-1-2; Sigma).
ECL detection (Pierce) was performed according to manufacturer’s
instructions.
MD Simulations. Molecular dynamics (MD) and thermodynamic
computations were carried out using Amber 9.038 and the
Amber1999SB force field.55 We focused our MD simulations on
the two pairs of separated stereoisomers 8a and 8c. The initial
structure of the four Sirt2-inhibitor complexes were taken from the
GOLD docking study. The ligand force field parameters were taken
from the general Amber force field (GAFF),56 whereas AM1 ESP
atomic partial charges were assigned to the inhibitors.57 In the
crystal structure of human Sirt2, each domain is associated with a
divalent zinc ion coordinated to four anionic cystein side groups.
Computational Details. All calculations were performed on
a Pentium IV 1.8/2.2 GHz based Linux cluster (20 CPUs). The
molecular structures of the inhibitors were generated using
the MOE modeling package (Chemical Computing Group). The

Splitomicin DeriVatiVes as Sirtuin Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5 1211
The divalent zinc ions were represented by a cationic dummy atom
(CaDA) approach of Pang et al.,24,58 which treats the zinc ion as
a tetrahedron-shaped divalent cation with dummy atoms filling the
tetrahedral corners. Parameters and libraries for the tetrahedral-
zinc groups and anionic cystein residues were obtained from Pang
et al.24,58,59 An explicit solvent model TIP3P water was used, and
the complexes were solvated with a 10 Å water cap. A total of
eight sodium ions were added as counterions to neutralize the
system. Prior to the free MD simulations, three steps of relaxation
were carried out; in the first step, we kept the protein fixed with a
constraint of 500 kcal mol-1Å-1, and we relaxed the position of
the tetrahedron-shaped zinc cations (0.1 ps MD). In the second step,
the zinc divalent cations and the inhibitor structures were relaxed
for 0.5 ps, during which the protein atoms were restrained to the
X-ray coordinates with a force constant of 500 kcal mol-1Å-1. In
the final step, all restraints were removed and the complexes were
relaxed for 1 ps. The temperature of the relaxed system was then
equilibrated at 300 K through 20 ps of MD using 2 fs time steps. A
constant volume periodic boundary was set to equilibrate the temper-
ature of the system by the Langevin dynamics39 using a collision
frequency of 10 ps-1 and a velocity limit of 5 temperature units. During
the temperature equilibration routine, the complex in the sovent box
was restrained to the initial coordinates with a weak force constant of
10 kcal mol-1Å-1. The final coordinates of the temperature equilibra-
tion routine (after 20 ps) was then used to complete a 1 ns molecular
dynamics routine using 2 fs time steps, during which the temperature
was kept at 300 K by the Langevin dynamics39 using a collision
frequency of 1 ps-1 and a velocity limit of 20 temperature units, and
the pressure of the solvated system was equilibrated at 1 bar at a certain
density in a constant pressure periodic boundary by an isotropic
pressure scaling method employing a pressure relaxation time of 2
ps. The time step of the free MD simulations was 2 fs, with a cutoff
of 9 Å for the nonbonded interaction, and SHAKE55 was employed
to keep all bonds involving hydrogen atoms rigid. Electrostatic
interactions were computed using the Particle Mesh Ewald method.40
The MD simulations of the Sirt2-inhibitor complexes were performed
in total for 6 ns.
computed by a linear Poisson–Boltzmann (PB) equation,38 which
is a three-dimensional vector differential equation as in eq 6
ꢀ(r) Ø(r) ) -4π × F(r)
(6)
in which ꢀ(r) is the dielectric constant (ꢀ ) 1 for the solute interior
and ꢀ ) 80 for implicit PB water) and F(r) is the charge density.
The grid point potentials were then summed up for each atom i to
yield atomic potentials Øi. The PB implicit solvent molecules at
the solute–solvent boundary were allowed to energetically converge
over 1000 iterations before the single-point Poisson computations
were implemented by PBSA for each snapshot. A spherical solvent
probe (radii) of 1.4 Å and atomic radii provided by the Amber
force field were used for the implicit solvent molecules and solute
atoms, respectively, during the PBSA computations.
The absolute entropy was computed for each solute species by
normal-mode analysis60 integrated in the NMODE module of
Amber 9.0. The total entropy (Stot), as formulated in eq 7, arose
from changes in the degree
Stot ) Strans + Srot + Svib
(7)
of freedom [translational (Strans), rotational (Srot), and vibrational
(Svib)] of each species.42
Considering all absolute energy terms as given in eq 2, the
binding free energy ∆G takes the following form
∆Gbinding ) [∆Hgas + ∆Gsolv] - T∆Stot
(8)
Parameter/topology files used in MM-PBSA computations were
prepared for the complex, the protein, and the inhibitors using the
LEAP module. Snapshots extracted from trajectories were premi-
nimized in the gas phase by the SANDER module using a conjugate
gradient method until the root-mean-square-deviation of the ele-
ments of the gradient vector was less than 10-4 kcal/mol-1Å-1
.
Frequencies of the vibrational modes were computed at 300 K for
these minimized structures, including all snapshot atoms and using
a harmonic approximation of the energies.42
Hydrogen Bond Analyses. A comprehensive hydrogen bond
analyses was carried out on the trajectory of the splitomicin-Sirt2
complexes. Watson–Crick base pair formation and intermolecular
protein-inhibitor interactions contributing to the specificity of
binding was characterized by analyzing hydrogen bond formations
through the trajectory of 6 ns MD simulation. A solvation pattern
for bound isomeres was also characterized by hydrogen bond
formation for the last 5 ns of the 6 ns simulation, during which all
water molecules are thought to be pre-equilibrated. A cutoff distance
of 3 Å between the heavy atoms of donor and acceptor groups and
a hydrogen bonding angle cutoff of 120° was used to analyze
hydrogen bond formation. Intermolecular hydrogen bond occurrence
>1% was only found for the R-isomer (Gln167 and Ile169, 2.85%
and 4.23%). A solvation pattern (water molecules) for this stereo-
isomer was found with 17.44% occurrence. For the S-isomer, only
a solvation pattern with an occurrence of 11.25% was observed
throughout the 5 ns period.
MM-PBSA. The net change in binding free energy accompany-
ing the formation of the protein–ligand complex is approximated
by the following equation
∆G ) ∆H - T∆S
(1)
in which T is the temperature of the system at 300 Kelvin. The
binding free energy (∆G) of the protein–ligand complex is
computed as:
∆G ) Gcomplex -[Gprotein + Gligand
]
(2)
where Gcomplex is the absolute free energy of the complex, Gprotein
is the absolute free energy of the protein, and Gligand is the absolute
free energy of the ligand. We extracted 100 snapshots (at time
intervals of 2 ps) for each species (complex, protein, and ligand)
from the last 200 ps of the MD simulations of the complexes. The
enthalpy term in eq 1 is dissected into subenergy terms
Acknowledgment. Funding by the Deutsche Forschungs-
gemeinschaft (Ju295-4/1 and -4/2, Si868/1-1) is gratefully
acknowledged.
Htot ) Hgas + Gsolv
(3)
(4)
Hgas ) Eel + EvdW + Eint
where Hgas is the potential energy of the solute, which is determined
as the sum of van der Waals (Evdw), electrostatic (Eel), and internal
Supporting Information Available: Spectral data for newly
synthesized compounds, inhibitor competition analysis, Western blot
and data on compound purity, and data on the binding mode
analysis. This material is available free of charge via the Internet
at http://pubs.acs.org.
energies (Eint) in gas phase by using the SANDER module of
38
Amber.
G
solv
is the solvation free energy for transferring the solute
from vacuum into solvent and is a sum of electrostatic (Gel) and
nonelectrostatic (hydrophobic) contributions (Gnonel), as shown in
eq 5
References
Gsolv ) Gel + Gnonel
(5)
(1) Grozinger, C. M.; Schreiber, S. L. Deacetylase enzymes: Biological
functions and the use of small-molecule inhibitors. Chem. Biol. 2002,
9, 3–16.
(2) North, B. J.; Verdin, E. Sirtuins: Sir2-related NAD-dependent protein
deacetylases. Genome Biol. 2004, 5, 224.
(3) Johnstone, R. W. Histone-deacetylase inhibitors: novel drugs for the
treatment of cancer. Nat. ReV. Drug DiscoVery 2002, 1, 287–299.
Gel in eq 4 was computed at 0.15 M salt concentration by the
PBSA module of Amber 9.0 by dividing implicitely solvated solute
species into 0.4 Å cubic grid points and summing up the electrostatic
potentials computed at each grid point. Electrostatic potential Ø(r)
at a grid point r, which is not at the solvent–solute boundary, was

1212 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5
Neugebauer et al.
(4) Imai, S.; Armstrong, C. M.; Kaeberlein, M.; Guarente, L. Transcrip-
tional silencing and longevity protein Sir2 is an NAD- dependent
histone deacetylase. Nature 2000, 403, 795–800.
(27) Hayashi, T. Rhodium-catalyzed asymmetric addition of aryl- and
alkenylboron reagents to electron-deficient olefins. Pure Appl. Chem.
2004, 76, 465–475.
(28) Ohta, A.; Inoue, A.; Watanabe, T. Introduction of the methyl group
into the pyrazine ring. Heterocycles 1984, 22, 2317–2321.
(29) Wright, S. W.; Hageman, D. L.; Mcclure, L. D. Fluoride-mediated
boronic acid coupling reactions. J. Org. Chem. 1994, 59, 6095–6097.
(30) Tschaen, D. M.; Desmond, R.; King, A. O.; Fortin, M. C.; Pipik, B.;
King, S.; Verhoeven, T. R. An improved procedure of aromatic
cyanation. Synth. Commun. 1994, 24, 887–890.
(31) Maffioli, S. I.; Marzorati, E.; Marazzi, A. Mild and reversible
dehydration of primary amides with PdCl2 in aqueous acetonitrile.
Org. Lett. 2005, 7, 5237–5239.
(32) Heltweg, B.; Trapp, J.; Jung, M. In vitro assays for the determination
of histone deacetylase activity. Methods 2005, 36, 332–337.
(33) Avalos, J. L.; Bever, K. M.; Wolberger, C. Mechanism of sirtuin
inhibition by nicotinamide: Altering the NAD(+) cosubstrate specific-
ity of a Sir2 enzyme. Mol. Cell 2005, 17, 855–868.
(34) Finnin, M. S.; Donigian, J. R.; Pavletich, N. P. Structure of the histone
deacetylase SIRT2. Nat. Struct. Biol. 2001, 8, 621–625.
(35) Trapp, J.; Meier, R.; Hongwiset, D.; Kassack, M. U.; Sippl, W.; Jung,
M. Structure-activity studies on suramin analogues as inhibitors of
NAD(+)-dependent histone deacetylases (sirtuins). ChemMedChem
2007, 2, 1419–1431.
(36) Jones, G.; Willet, P.; Glen, R. C.; Leach, A. R.; Taylor, R.
Development and validation of a genetic algorithm for flexible docking.
J. Mol. Biol. 1997, 267, 727–748.
(37) Tame, J. R. Scoring functionssThe first 100 years. J. Comput.-Aided
Mol. Des. 2005, 19, 445–451.
(38) Case, D. A.; Cheatham, T. E. I.; Darden, T.; Gohlke, H.; Luo, R.;
Merz, K. M. J.; Onufriev, A.; Simmerling, C.; Wang, W.; Woods, R.
The Amber biomolecular simulation programs. J. Comput. Chem.
2005, 26, 1668–1688.
(39) Pastor, R. W.; Brooks, B. R.; Szabo, A. An analysis of the accuracy
of Langevin and molecular dynamics algorithms. Mol. Phys. 1988,
65, 1409–1419.
(40) Darden, T.; Pedersen, L. Particle mesh Ewald: An Nlog(N) method
for Ewald sums in large systems. J. Chem. Phys. 1993, 103, 8577–
8593.
(41) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. Numerical integration of
the cartesian equations of motions of a system with constraints: Molecular
dynamics of n-alkanes. J. Comput. Phys. 1977, 23, 327–341.
(42) Gohlke, H.; Case, D. A. Converging free energy estimates: MM-
PB(GB)SA studies on the protein-protein complex Ras-Raf. J. Com-
put. Chem. 2004, 25, 238–250.
(43) Mossman, T. Rapid colorimetric assay for cellular growth and survival:
application to proliferation and cytotoxic assays. J. Immunol. Methods
1983, 65, 55–63.
(44) Hubbert, C.; Guardiola, A. R.; Shao, R.; Kawaguchi, Y.; Ito, A.; Nixon,
A.; Yoshida, M.; Wang, X. F.; Yao, T. P. HDAC6 is a microtubule-
associated deacetylase. Nature 2002, 417, 455–458.
(45) Zhang, Y.; Li, N.; Caron, C.; Matthias, G.; Hess, D.; Khochbin, S.;
Matthias, P. HDAC-6 interacts with and deacetylates tubulin and
microtubules in vivo. EMBO J. 2003, 22, 1168–1179.
(46) Sanders, B. D.; Zhao, K.; Slama, J. T.; Marmorstein, R. Structural
basis for nicotinamide inhibition and base exchange in Sir2 enzymes.
Mol. Cell 2007, 25, 463–472.
(47) Utley, J. H. P.; Rozenberg, G. G. Electroorganic reactions. Part 56:
Anodic oxidation of 2-methyl- and 2-benzylnaphthalenes: Factors
influencing competing pathways. Tetrahedron 2002, 58, 5251–5265.
(48) Smith, W. B. Some Observations on the iodination of 2-naphthol and
its methyl ether. J. Org. Chem. 1985, 50, 3649–3651.
(5) Pagans, S.; Pedal, A.; North, B. J.; Kaehlcke, K.; Marshall, B. L.;
Dorr, A.; Hetzer-Egger, C.; Henklein, P.; Frye, R.; McBurney, M. W.;
Hruby, H.; Jung, M.; Verdin, E.; Ott, M. SIRT1 regulates HIV
transcription via Tat deacetylation. PLoS Biol. 2005, 3, e41.
(6) Vaziri, H.; Dessain, S. K.; Ng Eaton, E.; Imai, S. I.; Frye, R. A.;
Pandita, T. K.; Guarente, L.; Weinberg, R. A. hSIR2(SIRT1) functions
as an NAD-dependent p53 deacetylase. Cell 2001, 107, 149–159.
(7) Bereshchenko, O. R.; Gu, W.; Dalla-Favera, R. Acetylation inactivates
the transcriptional repressor BCL6. Nat. Genet. 2002, 32, 606–613.
(8) Ota, H.; Tokunaga, E.; Chang, K.; Hikasa, M.; Iijima, K.; Eto, M.;
Kozaki, K.; Akishita, M.; Ouchi, Y.; Kaneki, M. Sirt1 inhibitor,
Sirtinol, induces senescence-like growth arrest with attenuated Ras-
MAPK signaling in human cancer cells. Oncogene 2005.
(9) Outeiro, T. F.; Kontopoulos, E.; Altman, S.; 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. Sirtuin 2 inhibitors rescue {alpha}-synuclein-
mediated toxicity in models of Parkinson’s disease. Science 2007, 317,
516–519.
(10) Garske, A. L.; Smith, B. C.; Denu, J. M. Linking SIRT2 to Parkinson’s
disease. ACS Chem. Biol. 2007, 2, 529–532.
(11) Schäfer, S.; Jung, M. Chromatin modifications as targets for new
anticancer drugs. Arch. Pharm. Chem. Life Sci. 2005, 338, 347–357.
(12) Biel, M.; Wascholowski, V.; Giannis, A. Epigenetics-An epicenter
of gene regulation: Histones and histone-modifying enzymes. Angew.
Chem., Int. Ed. 2005, 44, 3186–3216.
(13) Grozinger, C. M.; Chao, E. D.; Blackwell, H. E.; Moazed, D.;
Schreiber, S. L. Identification of a class of small molecule inhibitors
of the sirtuin family of NAD-dependent deacetylases by phenotypic
screening. J. Biol. Chem. 2001, 276, 38837–38843.
(14) Heltweg, B.; Dequiedt, F.; Verdin, E.; Jung, M. A nonisotopic substrate
for assaying both human zinc and NAD+-dependent histone deacety-
lases. Anal. Biochem. 2003, 319, 42–48.
(15) Mai, A.; Massa, S.; Lavu, S.; Pezzi, R.; Simeoni, S.; Ragno, R.;
Mariotti, F. R.; Chiani, F.; Camilloni, G.; Sinclair, D. A. Design,
synthesis, and biological evaluation of sirtinol analogues as class III
histone/protein deacetylase (sirtuin) inhibitors. J. Med. Chem. 2005,
48, 7789–7795.
(16) Bedalov, A.; Gatbonton, T.; Irvine, W. P.; Gottschling, D. E.; Simon,
J. A. Identification of a small molecule inhibitor of Sir2p. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 15113–15118.
(17) Posakony, J.; Hirao, M.; Stevens, S.; Simon, J. A.; Bedalov, A.
Inhibitors of Sir2: Evaluation of splitomicin analogues. J. Med. Chem.
2004, 47, 2635–2644.
(18) Hirao, M.; Posakony, J.; Nelson, M.; Hruby, H.; Jung, M.; Simon,
J. A.; Bedalov, A. Identification of selective inhibitors of NAD+-
dependent deacetylases using phenotypic screens in yeast. J. Biol.
Chem. 2003, 278, 52773–52782.
(19) Tervo, A. J.; Kyrylenko, S.; Niskanen, P.; Salminen, A.; Leppanen,
J.; Nyronen, T. H.; Jarvinen, T.; Poso, A. An in silico approach to
discovering novel inhibitors of human sirtuin type 2. J. Med. Chem.
2004, 47, 6292–6298.
(20) Tervo, A. J.; Suuronen, T.; Kyrylenko, S.; Kuusisto, E.; Kiviranta,
P. H.; Salminen, A.; Leppanen, J.; Poso, A. Discovering inhibitors of
human sirtuin type 2: Novel structural scaffolds. J. Med. Chem. 2006,
49, 7239–7241.
(21) Napper, A. D.; Hixon, J.; McDonagh, T.; Keavey, K.; Pons, J. F.;
Barker, J.; Yau, W. T.; Amouzegh, P.; Flegg, A.; Hamelin, E.; Thomas,
R. J.; Kates, M.; Jones, S.; Navia, M. A.; Saunders, J. O.; DiStefano,
P. S.; Curtis, R. Discovery of indoles as potent and selective inhibitors
of the deacetylase SIRT1. J. Med. Chem. 2005, 48, 8045–8054.
(22) Heltweg, B.; Gatbonton, T.; Schuler, A. D.; Posakony, J.; Li, H.;
Goehle, S.; Kollipara, R.; Depinho, R. A.; Gu, Y.; Simon, J. A.;
Bedalov, A. Antitumor activity of a small-molecule inhibitor of human
silent information regulator 2 enzymes. Cancer Res. 2006, 66, 4368–
4377.
(23) Trapp, J.; Jochum, A.; Meier, R.; Saunders, L.; Marshall, B.; Kunick,
C.; Verdin, E.; Goekjian, P. G.; Sippl, W.; Jung, M. Adenosine
mimetics as inhibitors of NAD+-dependent histone deacetylases, from
kinase to sirtuin inhibition. J. Med. Chem. 2006, 49, 7307–7316.
(24) Pang, Y. P. Novel zinc protein molecular dynamics simulations: Steps
toward antiangiogenesis for cancer treatment. J. Mol. Model. 1999,
5, 196–202.
(49) Fries, H.; Schimmelschmidt, I. Darstellung von dibromnaphthol aus
tribromnaphthol. Justus Liebigs Ann. Chem. 1930, 484, 245–271.
(50) North, B. J.; Schwer, B.; Ahuja, N.; Marshall, B.; Verdin, E.
Preparation of enzymatically active recombinant class III protein
deacetylases. Methods 2005, 36, 338–345.
(51) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. Semi-
analytical treatment of solvation for molecular mechanics and dynam-
ics. J. Am. Chem. Soc. 1990, 112, 6127–6129.
(52) Trapp, J.; Jung, M. The role of NAD+ dependent histone deacetylases
(sirtuins) in aging. Curr. Drug Targets 2006, 7, 1553–1560.
(53) Molecular Operating EnVironment (MOE) 2006.8; Chemical Comput-
ing Group Inc.: Montreal, Quebec, Canada, 2006.
(54) Omega; OpenEye Scientific Software: Santa Fe, NM, http://www.
eyesopen.com.
(55) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.;
Simmerling, C. Comparison of multiple Amber force fields and
development of improved protein backbone parameters. Proteins 2006,
65, 712–725.
(25) Lu, X. Y.; Lin, S. H. Pd(II)-bipyridine catalyzed conjugate addition
of arylboronic acid to R,ꢀ-unsaturated carbonyl compounds. J. Org.
Chem. 2005, 70, 9651–9653.
(26) Abouassali, M.; Decoret, C.; Royer, J.; Dreux, J. Study on reactivity
of coumarins. 1. Regioselectivity of Grignard reagents with coumarin.
Tetrahedron 1976, 32, 1655–1659.
(56) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A.
Development and testing of a general amber force field. J. Comput.
Chem. 2004, 25, 1157–1174.

Splitomicin DeriVatiVes as Sirtuin Inhibitors
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 5 1213
(57) Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A. Automatic atom
type and bond type perception in molecular mechanical calculations.
J. Mol. Graph. Model. 2006, 25, 247–260.
(58) Pang, Y. P.; Xu, K.; Yazal, J. E.; Prendergas, F. G. Successful
molecular dynamics simulation of the zinc-bound farnesyltransferase
using the cationic dummy atom approach. Protein Sci. 2000, 9, 1857–
1865.
(59) Pang, Y. P. http://mayoresearch.mayo.edu/mayo/research/camdl/
zinc_protein.cfm.
(60) Case, D. A. Molecular dynamics and normal mode analysis of
biomolecular rigidity. Rigidity Theory and Applications; Plenum: New
York, 1999; pp 329–344.
JM700972E