SDS-PAGE sodium dodecyl sulfate-Polyacrylamide gel electrophoresis
Sir2Tm Thermotoga maritima Sirtuin
SEC size exclusion chromatography
Sir silent information regulator
STACs Sirtuin activating compounds
TEV tobacco etch virus
TFA trifluoroacetic acid
Tm temperature of melting
TNF tumor necrosis factors
Trx-tag thioredoxin tag
TSA thermal shift assay
Vmax maximum reaction velocity
XDS X-ray detector software
yHst2 yeast Hst2
ySirt2 yeast Sirt2
zSirt5 zebrafish Sirt5
Sirtuins are a highly conserved family of protein deacylases that are important regulators
of metabolism, longevity and aged-related diseases. There are seven sirtuin isoforms in mammals
with different subcellular localization, substrates and biological functions. Three sirtuin
isoforms, Sirt3-5, are located in the mitochondria and play significant roles in all mitochondrial
processes. This study investigates the modulations of small molecule compounds on
mitochondrial sirtuins Sirt3 and Sirt5 using structural characterization as well as biochemical and
Resveratrol, a polyphenol found in red wine, was reported to activate Sirt1. Testing
resveratrol and its related compounds including piceatannol, polydatin, 4’-bromo-resveratrol
against Sirt3 showed an inhibitory effect. Among these compounds, brRESV is the most potent
Sirt3 inhibitor. Crystal structures of Sirt3 in complex with FdL-1 peptide and
piceatannol/polydatin showed a direct interaction between the compounds and the peptide
coumarin ring to induce non-productive substrate binding, thus inhibit the enzyme activity.
Crystal structures of the complex Sirt3/FdL-1/4’-bromo-resveratrol and Sirt3/ACS2/4’-bromo-
resveratrol revealed two different compound binding sites. Biochemical and binding data
indicated that the binding site of 4’-bromo-resveratrol in the FdL-1 complex involved in
inhibition mechanism whereas the compound binding site in the ACS2 complex might imply the
site for the activation mechanism of resveratrol on Sirt1.
Resveratrol unrelated compounds including SRT1720, a potent synthetic Sirt1 activator,
and Ex-527, a potent Sirt1 inhibitor, also inhibited Sirt3. In the crystal structure of the complex
Sirt3/carba-NAD+/SRT1720, the compound showed competition with substrate peptide by
occupying the binding region of acetyl lysine. SRT1720 interacts with NAD+ and the binding
analysis indicated that the NAM moiety of NAD+ is essential for SRT1720 binding. The crystal
structure of Sirt3 in the presence of NAD+ and SRT1720 only showed the ADP-ribose part
implying the hydrolysis of NAD+ and the importance of NAM moiety in SRT1720 binding. In
contrast to the SRT1720 inhibition mechanism, Ex-527 stabilized a closed sirtuin conformation
and prevented the product 2'-O-acetyl-ADP-ribose release. Crystal structure of the complex of
Sirt3/native O-alkylamidate intermediate provided more information about the deacetylation
Sirtuine sind eine Familie hochkonservierter Proteindeacylasen, die wichtige Regulatoren
des Metabolismus, der Lebensdauer und alters-assoziierter Krankheiten sind. Säugetiere besitzen
sieben Sirtuinisoformen, die sich in ihrer subzellulären Lokalisation, ihren Substraten und
biologischen Funktionen unterscheiden. Drei Sirtuinisoformen, Sirt3-5, sind mitochondrial
lokalisiert und spielen eine entscheidende Rolle in allen mitochondrialen Prozessen. Diese Arbeit
untersucht die Modulation von niedermolekularen Wirkstoffen auf die mitochondrialen Sirtuine
Sirt3 und Sirt5 mittels struktureller Charakterisierung als auch anhand biochemischer und
Über Resveratrol, ein in Rotwein vorkommendes Polyphenol, ist bekannt, dass es Sirt1
aktiviert. Die Untersuchung von Resveratrol und dessen verwandten Wirkstoffen wie
Piceatannol, Polydatin, 4´-Bromoresveratrol gegen Sirt3 zeigte eine inhibitorische Wirkung. Von
diesen Wirkstoffen ist 4´-Bromoresveratrol der stärkste Sirt3-Inhibitor. Kristallstrukturen von
Sirt3 komplexiert mit dem FdL-1-Peptid und Piceatannol/Polydatin zeigten eine direkte
Interaktion dieser Wirkstoffe mit dem Coumarinring des Peptids. Dies induziert eine
unproduktive Substratbindung, die dadurch die Enzymaktivität inhibiert. Kristallstrukturen der
Sirt3/FdL-1/4’-bromo-resveratrol- und Sirt3/ACS2/4’-bromo-resveratrol-Komplexe offenbarten
zwei unterschiedliche Wirkstoffbindestellen. Biochemische Daten und Bindungsdaten deuteten
an, dass die 4’-bromo-resveratrol-Bindestelle im FdL-1-Komplex im Inhibitionsmechanismus
involviert ist. Die Wirkstoffbindestelle im ACS2-Komplex hingegen könnte die Bindestelle für
den Aktivierungsmechanismus von Resveratrol gegenüber Sirt1 implizieren.
Wirkstoffe wie SRT1720, ein potenter synthetischer Sirt1-Aktivator, und Ex-527, ein
potenter Sirt1-Inhibitor, die beide Resveratrol nicht ähneln, inhibierten auch Sirt3. In der
Komplexstruktur von Sirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 kompetierte der Wirkstoff mit dem
Substratpeptid, indem er die Bindestelle des Acetyllysins besetzt. SRT1720 interagiert mit NAD+
und Bindestudien deuten an, dass die Nikotinamidgruppe des NAD+ für die SRT1720-Bindung
essenziell ist. Im Gegensatz zum Inhibitionsmechanismus von SRT1720 stabilisiert Ex-527 die
geschlossene Sirtuinkonformation und verhinderte dadurch die Freisetzung des Produkts 2´-O-
acetyl-ADP-Ribose. Die Kristallstruktur des Komplexes aus Sirt3/nativem O-alkylamidat-
Intermediats lieferte mehr Information über die Deacetylierungsreaktion.
1. Introduction 9
1.1. Caloric restriction (CR) and aging
Caloric restriction (CR) is a dietary regimen based on low calorie intake up to 40 percent.
Over 75 years ago, Clive McCay and colleagues first found that rats fed a caloric restricted diet
live longer than ad libitum (freely fed) (McCay, et al., 1989). Since then, similar observations
were reported in a variety of species including yeast, fruit flies, worms, etc. (Lin, et al., 2002;
Masoro, 2005). In addition, many studies found that CR without malnutrition can prevent or
delay a wide number of chronic diseases, such as cancer, diabetes, autoimmune, respiratory
disease, Alzheimer disease and Parkinson disease (Cohen, et al., 2009; Longo and Fontana,
2010; Masoro, 1990) indicating that CR retards aging processes.
1.2. Sirtuins and their roles in CR, aging and human age-related diseases
Sirtuins are a family of NAD+-dependent protein deacetylases conserved throughout
evolution from archaebacteria to eukaryotes. They are homologs to the yeast Sir2 (silent
information regulator 2) (Lin, et al., 2000).
Sir2 was proven to be required for lifespan extension in yeast by CR (Kaeberlein, et al.,
1999; Lin, et al., 2000). The ability of glucose restriction to extend lifespan was blocked in yeast
deleting Sir2 gene (Lin, et al., 2000). Since then, many studies focus on Sir2 homologs and their
relation to aging. Sir2 homolog induces lifespan extension in worms (Wang and Tissenbaum,
2006) and flies (Rogina and Helfand, 2004). Sirt1 was described as a key role in regulating the
metabolic response to CR (Cantó and Auwerx, 2009). Sirt3 mediates CR to age-related hearing
loss, the hallmark of mammalian aging and required for the reduction of oxidative damage
(Someya, et al., 2010). Sirt6 prolongs lifespan in male mice (Kanfi, et al., 2012) and can be act
as a tumor suppressor (Lombard and Miller, 2012).
Many reports suggested the relation of sirtuins to various age-related diseases such as
metabolic abnormalities, cancer, neurodegenerative diseases, cardiovascular, etc. (Sebastian, et
1. Introduction 10
al., 2012). Sirtuins can control tumorigenesis due to their ability in regulation of genomic
stability such as Sirt1 modulates cellular stress responses and DNA repair, deacetylates the
proto-oncogene Myc to prevent transformation (Martinez-Pastor and Mostoslavsky, 2012; Yuan,
et al., 2009). Some reports suggested that Sirt3 and Sirt6 seem to be tumor suppressors due to
their ability to destabilize HIF-1α (hypoxia-inducible factor 1-α) through down regulation of
ROS (reactive oxygen species) and induce apoptosis in cancer cell lines (Sebastian, et al., 2012).
Sirt1 was described as a protector against neurodegenerative diseases such as Alzheimer disease,
Parkinson disease and Huntington disease (Arima, et al., 1998; Haass and Selkoe, 2007; Jiang, et
al., 2012). Sirt2 supports differentiation and migration of some brain cells through deacetylating
α-tubulin and Par-3 (protease activated receptor 3) (Beirowski, et al., 2011; Li, et al., 2007).
Moreover, Sirt1 and Sirt7 possess cardiovascular protective properties by deacetylating p53 or
regulating fatty acid oxidation (Sebastian, et al., 2012). Sirt3 is a regulator of cardiac function by
reducing cellular ROS or suppressing Akt phosphorylation via AMPK (AMP-activated kinase)
(Pillai, et al., 2010; Sack, 2011). Sirt6 protect against cardiac hypertrophy by inhibiting NF-κB
(nuclear factor-κB) or IGF (insulin-like growth factor)-Akt signaling (Sundaresan, et al., 2012;
Yu, et al., 2012).
1.3. The mammalian sirtuin enzyme family
1.3.1. Overview of the mammalian sirtuin family: classification, localization and function
In mammals, there are seven members of the sirtuin family, Sirt1–7 that differ in their
cellular localization and function (Haigis and Sinclair, 2010; Michan and Sinclair, 2007). The
seven mammalian sirtuins share a highly conserved catalytic core domain but have differences in
their N- and C-terminal ends (Frye, 2000) (Figure 1.1).
1. Introduction 11
Figure 1.1. Schematic illustration of seven mammalian sirtuins.
Based on phylogenetic analysis, mammalian sirtuins can be divided into four classes.
Sirt1-3 belong to class I, Sirt4 to class II, Sirt5 to class III and Sirt6, Sirt7 to class IV (Frye,
2000). Sirt1 is mainly found in the nucleus but also present in the cytosol (Michan and Sinclair,
2007), Sirt2 in the cytoplasm (North, et al., 2003), Sirt3-5 in mitochondria (Gertz and Steegborn,
2010), and Sirt6 and Sirt7 in the nucleus (Michishita, et al., 2005).
Sirt1-3 have robust deacetylation activity, whereas Sirt4 and Sirt6 were reported to be
ADP-ribosyltransferases (Haigis, et al., 2006; Mao, et al., 2011). Sirt6 can also act as a
deacetylase (Michishita, et al., 2008). Sirt7 was recently confirmed as a deacetylase due to its
important role in deacetylation of H3K18Ac (acetylated lysine 18 of histone H3) (Barber, et al.,
2012). Sirt5 was initially reported to deacetylate CPS1 (carbamoyl phosphate synthase 1)
(Nakagawa, et al., 2009) but recently described as a protein desuccinylase and demalonylase
(Du, et al., 2011) indicating that sirtuins are a family of deacylases.
Sirt1 is the most studied mammalian sirtuin isoform which was first described as a
histone deacetylase (Haigis and Sinclair, 2010) but also has other protein targets such as p53
which is deacetylated upon DNA damage or oxidative stress (Vaziri, et al., 2001) and forkhead
transcription factors (FOXO) in lipid and glucose metabolism (Motta, et al., 2004). Sirt2 is a
tubulin deacetylase (North, et al., 2003). Sirt3-5 play important roles in metabolism, apoptosis
and intracellular signaling (Verdin, et al., 2010). More details of Sirt3-5 are in the Mitochondrial
1. Introduction 12
sirtuins section (see below). Sirt6 is involved in DNA repair (Lombard, 2009), regulates immune
response that relates to NF-κB targets (Kawahara, et al., 2009; Michishita, et al., 2008) and
controls TNF (tumor necrosis factors) production (Van Gool, et al., 2009). Sirt7 is H3K18Ac
deacetylase that functions in chromatin regulation, cellular transformation programs and tumour
formation (Barber, et al., 2012).
1.3.2. Mitochondrial sirtuins
NAD+-dependent protein deacylase is a major enzymatic activity of sirtuins. Proteomic
studies implied that a large number of mitochondrial proteins are acetylated (Verdin, et al.,
2010). NAD+ is an essential electron carrier in various metabolic processes such as energy
production, fatty acid metabolism, urea cycle, etc., which are integrated by mitochondria. Since
sirtuins use NAD+ as a cosubstrate, Sirt3-5 which have been identified in the mitochondrial
matrix are directly linked to mitochondrial processes and influence mitochondrial functions
Figure 1.2. Mitochondrial sirtuins and their links to mitochondrial processes. Figure is
reused with permission from Elsevier Limited (Verdin, et al., 2010).
1. Introduction 13
Sirt3 seems to be the major mitochondrial deacetylase since mice lacking Sirt3 showed a
hyperacetylation of mitochondrial proteins (Lombard, et al., 2007). The mitochondrial protein
acetylcoenzyme A synthase 2 (ACS2) which converts acetate to acetyl-CoA in the presence of
ATP is the first identified Sirt3 substrate (Hallows, et al., 2006). Glutamate dehydrogenase
(GDH) is also Sirt3 substrate since Sirt3 deacetylates and activates GDH activity by 10%
(Schlicker, et al., 2008). Sirt3 interacts with Complex I of the electron transport chain and
deacetylates various proteins in this complex (Ahn, et al., 2008). The mitochondrial matrix
enzyme CPS1 which plays an important role in the rate-limiting step of the urea cycle was
identified as a Sirt5 substrate (Nakagawa, et al., 2009). Recently, Sirt5 has been determined as a
major demalonylase and desuccinylase since mice lacking Sirt5 showed hypermalonylation and
hypersuccinylation (Peng, et al., 2011). Two Sirt5 residues in the catalytic pocket, Tyr102 and
Arg105, are mandatory for the demalonylase and desuccinylase activities (Du, et al., 2011).
These two residues are conserved in sirtuin class III of different species (Du, et al., 2011). The
enzymatic function of Sirt4 remains unclear. So far, Sirt4 has no detectable deacetylase activity
and weak ADP-ribosyltransferase activity (Ahuja, et al., 2007; Haigis, et al., 2006).
1.3.3. Structure of sirtuins
Crystal structures of different sirtuin homologs including apo protein or in complex with
substrates or small molecules have been published. The structure of yeast Hst2 (PDB ID 1Q14)
(Zhao, et al., 2003) contains the full-length protein whereas the remaining structures show only
the core domain. The full length yHst2 structure implies the role of N- and C-terminal region of
sirtuins in the regulation of substrate binding (Zhao, et al., 2003).
The conserved catalytic core consists of two domains, a large Rossmann-fold domain for
NAD+ binding and a variant small zinc-binding domain that may be involved in substrate
binding (Sanders, et al., 2010; Sauve, et al., 2006) (Figure 1.3). The Rossmann-fold domain
consists of six parallel β strands forming a β sheet which is sandwiched between several α
helices on each side (Min, et al., 2001). The two modules of the zinc-binding domain include
three β strands forming an antiparallel β sheet and a helical region with three or four helices
1. Introduction 14
dependent on the sirtuin member. Four cysteine residues coordinate the zinc ion in a tetrahedral
conformation to stabilize the structure (Min, et al., 2001). A cleft where catalysis takes place is
formed between two domains by four linking loops. This region is the most conserved with high
sequence homology among sirtuin family.
The largest of these four linking loops, called cosubstrate binding loop, is very dynamic.
The cosubstrate binding loop is highly flexible when NAD+ is not bound and becomes well-
ordered through NAD+ binding (Zhao, et al., 2004) indicating that its conformation is dependent
on the presence of NAD+. The NAD+ binding site can be divided in three pockets: an adenine
binding pocket (pocket A), a nicotinamide (NAM) ribose binding pocket (pocket B) and a NAM
binding pocket (pocket C). Different ligands in NAD+ site induce slightly conformational
changes of the cosubstrate binding loop (Avalos, et al., 2005; Sanders, et al., 2007).
The acetyl lysine of peptide substrate inserts into a hydrophobic tunnel of the cleft
between two domains. When comparing the protein conformation with and without peptide
substrate, peptide binding induces a shift in the linking loop between two domains and brings
two domains closer together (Cosgrove, et al., 2006). The substrate peptide orientation and
interaction with protein residues described in different crystal structures strengthen one
mechanism that different sirtuins discriminate among substrates.
The NAD+ cosubstrate inserts from opposite site with the acetylated substrate into the
cleft between two domains (Sanders, et al., 2010). The conformation of NAM ribose is variable
dependent on NAD+ analogs, substrate peptides and sirtuin homologs. The density of the ADPR
part including the adenine ribose and NAM ribose is well defined whereas the NAM moiety is
almost invisible in different sirtuin structures in the presence of NAD+ (Chang, et al., 2002;
Nguyen, et al., 2013; Pan, et al., 2011) indicating the flexibility of this part or the hydrolysis of
NAD+ during crystallization.
1. Introduction 15
Figure 1.3. Overall structure of sirtuins. The catalytic core of human Sirt3 (ribbon) in complex
with carba-NAD(CAS# 112345-60-5)+ (stick, orange) and ACS2 peptide (acetyl lysine in stick, yellow) (PDB ID
4FVT) (Szczepankiewicz, et al., 2012) is shown as a representative. The large Rossmann-fold
domain is in purple. The small zinc-binding domain that contains Zn2+ ion (sphere, yellow) is in
blue. The loops connecting two domains are in green. The cosubstrate binding loop is
highlighted in red.
1.3.4. Enzymatic activity of sirtuins
Although the initial activity of sirtuins was reported as NAD+-dependent ADP-
ribosylation (Tanny, et al., 1999), protein deacylation is the most prevalent reaction that sirtuin
enzymes catalyze. The deacetylation reaction occurs in two continuous stages to generate
deacetylated protein, NAM and 2’-O-acetyl-ADP-ribose (2’-OAADPr) (Sauve, et al., 2006;
Tanner, et al., 2000) (Figure 1.4). In the first stage, sirtuins cleave NAD+ to produce NAM and
the nucleophilic addition of the acetyl oxygen to C1’ of the ADP-ribose moiety to form C1’-O-
alkylamidate intermediate (Sauve, 2010). The nucleophilic attack mechanism has been subject to
debate between SN1 and SN2 type for the cleavage of the glycosidic bond between NAM and the
1. Introduction 16
rest of NAD+ (Smith and Denu, 2007). NAM can inhibit sirtuins by rebinding to reverse the
reaction through the base-exchange mechanism (Sauve, et al., 2006). In the second stage, the
C1’-O-alkylamidate intermediate converts to the bicyclic intermediate by using the conserved
Histidine as a general base to induce nucleophilic attack of the 2’-OH group of the ribose onto
the iminium carbon of the O-alkylamidate intermediate. The crystal structure of the bicyclic
intermediate between thiosuccinyl H3 peptide and NAD+ on Sirt5 was recently solved to provide
an evidence for the mechanism (Zhou, et al., 2012). The bicyclic intermediate is disrupted by a
base activated water molecule to form deacetylated protein and 2’-O-acetyl-ADP-ribose (Sauve
and Youn, 2012). Both 2’-O-acetyl-ADP-ribose and 3’-O-acetyl-ADP-ribose exist in equilibrium
as solution products of sirtuins (Jackson and Denu, 2002).
Figure 1.4. Mechanism of sirtuin-catalyzed deacetylation. Protein deacetylation reaction with
two continuous stages catalyzed by sirtuins. Figure is reused with permission from American
Society for Biochemistry and Molecular Biology (Feldman, et al., 2012).
1. Introduction 17
1.4. Sirtuin modulators
CR extends lifespan of a variety of species and delays or prevents many age-related
diseases. The role of sirtuins in CR-mediated longevity has been proven in several studies (Cantó
and Auwerx, 2009; Kanfi, et al., 2012; Someya, et al., 2010). Due to the fact that not many
people would be willing to keep a CR lifestyle, recent studies focus on mimicking CR’s effects,
especially enhancing the activity of sirtuins by small molecules providing a foundation for drug
Howitz and colleagues screened small molecule compounds to identify several activators
and inhibitors of Sirt1. They reported that two polyphenols, quercetin and piceatannol (Figure
1.5), can activate Sirt1 activity 5- and 8-fold, respectively. The sirtuin activating compounds
were called STACs. Subsequently, they screened the polyphenol family and found resveratrol
(3,4’,5-trihydroxystilbene) (Figure 1.5) as the most potent activator candidate with ~13-fold
increase in substrate deacetylation of Sirt1 and lower Km of the enzyme for the substrate and
NAD+ (Howitz, et al., 2003). Resveratrol, a natural polyphenol found in red wine and other
plant-based foods, is able to mimic CR in anti-aging and possess many other benefits such as
antivirus, anti-inflammation, anti-diabete and cardioprotective effects (S. Mohar, 2012).
Resveratrol was reported to extend lifespan of different organisms including yeast, worm and fly
dependent on sirtuin activity (Howitz, et al., 2003; Wood, et al., 2004). This compound can also
induce lifespan extension in fish but the relation of its effect to sirtuins is unclear (Valenzano, et
1. Introduction 18
Figure 1.5. Resveratrol and its related compounds. Figures are adapted with permission from
Blum et al. (Blum, et al., 2011). Copyright (2011) American Chemical Society.
However, the role of resveratrol as a Sirt1 activator has been questioned in some reports.
Reveratrol effect was initially found with a fluorophore labeled substrate (Howitz, et al., 2003).
It was later shown that for this substrate, the enhancement of Sirt1 activity by resveratrol
depends on the fluorophore labeling (Borra, et al., 2005; Kaeberlein, et al., 2005). Resveratrol
did not activate Sirt1 when using fluorophore-free peptide substrates (Beher, et al., 2009;
Pacholec, et al., 2010). In addition, the crystal structure of Sirt5 in complex with resveratrol and
of Sirt3 in complex with piceatannol revealed the direct interaction of the compounds to the
coumarin tag of the Fluor-de-lys peptide (Gertz, et al., 2012), indicating the fluorophore
influence in compound binding. In another study, resveratrol induced metabolic changes
mediated via AMPK rather than Sirt1 (Um, et al., 2010).
Subsequently, a study found that Sirt1 fluorophore-free substrates containing a
hydrophobic amino acid residue (Trp, Tyr or Phe) at position +1 or +6 were selectively activated
by STACs (Hubbard, et al., 2013). In the same report, Glu230 of Sirt1, a conserved residue from
flies to humans, was identified as the critical residue for Sirt1 activation by STACs. The
compounds stimulate Sirt1 via allosteric activation mechanism mediated by Glu230 containing
N-terminal domain of the enzyme. In addition, another study testing many physiological
deacetylation sites in parallel using peptide arrays showed that substrate sequence determines
1. Introduction 19
resveratrol effects (Lakshminarasimhan, et al., 2013). Moreover, there is no evidence of Sirt1
independent AMPK phosphorylation in STAC-treated cells that goes against a report about
resveratrol mediated AMPK pathway (Park, et al., 2012). Therefore, sirtuins could be directly
activated by STACs.
220.127.116.11. Other activators
Resveratrol is a natural compound which might have benefits in prevention or treatment
of age related diseases via sirtuin activation. However, the compound has low bioavailability and
might not specific for sirtuin target (Alcaín and Villalba, 2009; Pirola and Frojdo, 2008). New
synthetic sirtuin activators that are much more effective than resveratrol have been developed.
Milne and colleagues identified resveratrol unrelated activators including SRT1460, SRT1720,
and SRT2183 (Figure 1.6) that work up to 1000 fold more potently than resveratrol (Milne, et al.,
2007). Among these compounds, SRT1720 is the most effective with EC1.5 = 0.16 µM and
maximum activation = 781%. This small molecule compound was demonstrated as therapeutics
for treatment of many diseases such as type 2 diabetes, metabolic disorders, inflammation, etc.
(Villalba and Alcaín, 2012).
Figure 1.6. Resveratrol unrelated activators. Figures are adapted with permission from Blum
et al. (Blum, et al., 2011). Copyright (2011) American Chemical Society.
1. Introduction 20
While sirtuin activators have mainly been developed for Sirt1, sirtuin inhibitors have
been studied on different sirtuin members including ySir2 (yeast Sir2), Sir2Tm (Thermotoga
maritima sirtuin), and mammalian Sirt1, Sirt2, Sirt3 and Sirt5.
The first sirtuin inhibitors are mostly based on substrates and products of the
deacetylation reaction. NAM inhibits Sir2 activity by rebinding to attack the O-alkylamidate
intermediate (Sauve and Schramm, 2003). Carba-NAD(CAS# 112345-60-5)+ is a weak inhibitor of sirtuins (Landry,
et al., 2000). Thioacetyllysine derived peptides have been described as sirtuin inhibitors by
hindering the reaction via formation of a stable S-alkylamidate intermediate instead of the native,
transient O-alkylamidate intermediate (Smith and Denu, 2007). Moreover, the replacement of N-
acetyl group with other groups and chemical modifications of peptide substrate have also
reported as an approach to develop sirtuin inhibitors (Chen, 2011).
Besides substrate and product based inhibitors, a variety of small molecule compounds
have been studied (Blum, et al., 2011; Cen, 2010). Some of these compounds such as
splitomicin, sirtinol, cambinol (Figure 1.7) have week effects on sirtuins with micromolar range
of IC50 value. Their derivatives showed improved potency but still moderate isoform selectivity.
Splitomicin showed a moderate inhibition on Sir2 with an IC50 value of 60 µM (Bedalov, et al.,
2001) but a weak inhibition on Sirt1. HR73, a splitomicin derivative, inhibited Sirt1 with an IC50
value of less than 5 µM (Pagans, et al., 2005). Sirtinol inhibited both Sir2 and Sirt2 (Grozinger,
et al., 2001) but its analogs, meta- and para-sirtinol, are more potent (Mai, et al., 2005).
Cambinol inhibited Sirt1 and Sirt2 with IC50 values of 56 and 59 µM, respectively but showed
weak inhibition against Sirt5 and no inhibition against Sirt3 (Heltweg, et al., 2006). Phenyl ring
modifications of cambinol improved potency and selectivity of this compound on Sirt1 and Sirt2
(Medda, et al., 2009). Other compounds such as Ex-527 and suramin (Figure 1.7) have higher
effects on sirtuins with nanomolar range of IC50 value. Ex-527, an indole derivative, has high
selectivity for Sirt1 with an IC50 value of 0.098 µM and lower potency against Sirt2 (IC50 19.6
µM) and Sirt3 (IC50 48.7 µM) (Solomon, et al., 2006). Suramin, a symmetric polyanionic
naphthylurea, is a potent inhibitor of many sirtuin isoform including Sirt5 (IC50 22 µM)
(Schuetz, et al., 2007), Sirt1 (IC50 0.297 µM) and Sirt2 (IC50 1.15 µM) (Trapp, et al., 2007).
1. Introduction 21
Among these compounds, the inhibition mechanisms of suramin and Ex-527 on sirtuins
were revealed based on crystal structures (Gertz, et al., 2013; Schuetz, et al., 2007; Zhao, et al.,
2013). In Sirt5-suramin complex structure, one suramin molecule links two monomers of Sirt5
together. Suramin inhibits Sirt5 by binding into the B- and C-pockets of the NAD+ binding site
and the substrate binding site (Schuetz, et al., 2007). Ex-527 stabilizes the closed sirtuin
conformation to prevent product release (Gertz, et al., 2013).
Figure 1.7. Representative sirtuin inhibitors. Figures are adapted with permission from Blum
et al. (Blum, et al., 2011). Copyright (2011) American Chemical Society.
Sirtuins modulation by small molecule compounds could have benefits to treat many
human age-related diseases such as cancer and neurodegenerative diseases. Resveratrol is a
natural polyphenol that can mimic CR to activate Sirt1 and has important role in delaying and
preventing some diseases. SRT1720, a resveratrol unrelated compound, is the most potent
activator of Sirt1. The molecular mechanisms of these activators on Sirt1 are unclear, especially
their structural information. So far, all activators have been described only for Sirt1 whereas
inhibitors have been identified for different sirtuin isoforms. In this study, the mitochondrial
1. Introduction 22
sirtuins Sirt3 and Sirt5 are used as models to investigate the regulation of small molecule
compounds on the sirtuin family. The small molecule compounds used in this study include
resveratrol and its related compounds with improved solubility due to an additional hydroxyl
group (piceatannol (3,5,3',4'-tetrahydroxy-trans-stilbene)) or glucose group (polydatin
(reservatrol-3-β-D-glucoside)) or bromide group (4’-bromo-resveratrol (5-(2-(4-hydroxyphenyl)vinyl)-1,3-benzenediol)) (Figure 1.5) and resveratrol unrelated compounds including SRT1720 (N-(2-(3-(piperazin-1-ylmethyl)imidazo[2,1-b]thiazol-6-yl)phenyl)quinoxaline-2-carboxamide) and Ex-527 (6-chloro-2,3,4,9-tetrahydro-1H-carbazole-1-
carboxamide). To obtain the purpose of this study, crystal structures of Sirt3 and Sirt5 in
complex with different substrate peptides in the presence of different small molecule compounds
were solved. In addition, biochemical and biophysical studies were conducted to support the
regulation mechanisms of small molecule compounds on sirtuins implied by crystal structures.
2. Materials and Methods 23
2. Materials and Methods
2.1.1. Chemicals, peptides and compounds
All chemicals were from Sigma, Roth and Applichem if not stated differently. Fluor-de-
lys 1 (FdL-1) peptide was from Enzo Life Science (New York, USA). The fluorophore-free
peptides listed in Table 2.1 were synthesized and HPLC (high performance liquid
chromatography) purified by GL Biochem (Shanghai, China). 4’-bromo-resveratrol was from
Matrix Scientific (Columbia, USA). SRT1720 was from Cayman Chemical (Ann Arbor, USA).
Table 2.1. List of peptides used in this study.
2.1.2. Plasmid vectors
The catalytic core domain gene of zebrafish Sirt5 (zSirt5) (residues 30-298) was cloned
into the vector pET151/D-TOPO (Life Technologies, USA) coding for His-tag (hexahistidine
tag) and carrying the resistance marker to ampicillin. The catalytic core domain gene of human
Sirt3 (hSirt3) (residues 118-399) was cloned into the vector pVFT3S (Sungkyunkwan university,
2. Materials and Methods 24
South Korea) coding for His-Trx-tag (hexahistidine-thioredoxin tag) and carrying the resistance
marker to kanamycin.
2.1.3. Oligonucleotide primers
All primers using for cloning or site-directed mutagenesis listed in Table 2.2 were HPLC
purified or HPSF (high purity salt free) from Sigma, USA or Eurofins MWG Operon, Germany.
Table 2.2. List of primers used in this study. Restriction sites or stop codon are underlined.
Bio-Rad Sub-cell horizontal gel electrophoresis system (Bio-Rad, USA) was used to perform nucleic acid electrophoreses with 1x TAE (Tris-acetate-EDTA (ethylenediaminetetraacetic acid)) as the running buffer. Samples and DNA maker (New
England Biolabs, USA) were mixed with loading buffer (6 mM EDTA, 6 % glycerol and 0.015
% bromophenol blue) before loading on a 1 % (w/v) agarose matrix (in 1x TAE buffer)
containing 1 µg/ml ethidium bromide. After electrophoresis, the gel was placed under UV light
for DNA visualization.
The catalytic core domain genes of zSirt5 and hSirt3 were amplified using PCR
(polymerase chain reaction). 50 µl of a PCR reaction contains the following: 10-50 ng of
template DNA, 0.5 µM of each primer, 2 Units of DNA polymerase (Thermo Scientific, USA)
and 1x HF buffer, 0.2 µM of each deoxynucleotide. The PCR program included 1) initial
denaturation at 95 ºC for 2 minutes; 2) denaturation at 95 ºC for 1 minute, annealing at 60 ºC for
1 minute, extension at 72 ºC for 1 minute and 3) final elongation at 72 ºC for 10 minutes. Step 2
was repeated 30 times. The PCR products were visualized and purified using agarose gel
electrophoresis and gel extraction kit (Qiagen, USA).
zSirt5 gene was directly mixed with the vector pET151/D-TOPO without using
restriction enzymes. hSirt3 gene and the vector pVFT3S were treated with restriction enzymes
2. Materials and Methods 26
BamHI and XhoI (Thermo Scientific, USA). After visualized and purified using agarose gel
electrophoresis, hSirt3 gene was ligated into the vector using a molar ratio of 3:1 (gene : vector)
in the presence of T4 DNA ligase (New England Biolabs, USA) and incubation at 20 ºC
overnight. 3 µl of the ligated products was used for the transformation of the recombinant
plasmids into 50 µl of TOP10 competent cells to amplify the plasmids. The mixture was placed
on ice for 30 minutes, heat shock at 42 ºC for 30 seconds, and then put on ice for 5 minutes. 450
µl of LB media was added to recover the cells at 37 ºC for 1 hour followed by plating on LB agar
plates containing appropriate antibiotics and incubated at 37 ºC overnight. Subsequently, the
plasmids were extracted using plasmid extraction kit (Qiagen, USA).
2.2.3. Site-directed mutagenesis
50 µl of the PCR reaction for site-directed mutagenesis contains the following: 5-50 ng of
template DNA, 125 ng of each forward and reverse primers, 0.2 mM deoxynucleotide mix, 1.25
Units of Pfu Turbo DNA polymerase (Agilent Technologies, USA), 1x cloned Pfu DNA
polymerase reaction buffer. The PCR program for site-directed mutagenesis was: 1) initial
denaturation at 95 ºC for 5 minutes; 2) denaturation at 95 ºC for 1 minute, annealing at 55 ºC for
1 minute, extension at 68 ºC for 10 minutes; 3) final elongation at 68 ºC for 10 minutes. Step 2
was repeated 18 times. Subsequently, the PCR product was treated with 5 Units of DpnI
restriction enzyme at 37 ºC for 1 hour to digest the template plasmid vector and 1 µl of the
reaction mixture was transformed into 50 µl of TOP10 competent cells using the transformation
protocol as in the Cloning section.
The recombinant plasmids were transformed into 50 µl of E. coli Rosetta (DE3)
competent cells for expression. The cells were placed on a 1 millimeter electroporation cuvette
(Serva, Germany) and pulsed with a voltage of 2.5 kV using the Bio-Rad Gene Pulser
electroporation system (Bio-Rad, USA). 450 µl of LB media was added to recover the cells at 37
ºC for 1 hour followed by transferring to LB media containing appropriate antibiotics and
2. Materials and Methods 27
incubation at 37 ºC by shaking. When the OD600 reached 0.6 – 0.8, the temperature was reduced
to 15 ºC. IPTG (isopropyl β-D-thiogalactopyranoside) was added into media to induce protein
expression. The cells were grown at 15 ºC overnight and harvested by centrifugation at 5,000
RPM for 20 minutes at 4 ºC and stored at -80 ºC.
2.2.5. Cell disruption
Frozen cells were resuspended in an appropriate lysis buffer and disrupted using
Microfluidizer (Microfluidics, USA) at 4 ºC. The lysed cells were then centrifuged at 18,000
RPM for 45 minutes in a refrigerated Beckman Coulter Avanti J-26XP centrifuge fitted with a
JA-30.50 Ti rotor (Beckman Coulter, USA) to remove cell debris.
2.2.6. SDS-Polyacrylamide gel electrophoresis (SDS-PAGE)
SDS-PAGE was performed to analyze the purity and size of proteins. The stacking gel,
running gel (15% acrylamide), buffers were prepared by following a published protocol
(Sambrook and Russell, 2001). Protein samples were denatured for 5 minutes at 95 ºC before
loading on the gel. The Mini-PROTEAN Tetra Cell vertical electrophoresis system (Bio-Rad,
USA) was used for electrophoresis. After electrophoresis, the gel was rinsed in water, followed
by a quick soaking in hot Coomassie blue solution (0.025 % (w/v) Coomassie-Briliant Blue R-
250 (Applichem, Germany), 50 % (v/v) methanol, 10 % (v/v) acetic acid). Subsequently, the gel
was transferred to a destaining solution containing 20 % (v/v) methanol and 12 % (v/v) acetic
The catalytic core domain of zSirt5 and hSirt3 were purified using similar protocols. The
fusion proteins were purified by affinity chromatography with TALON resin followed by
removing the His-tag of zSirt5 or His-Trx-tag of hSirt3 using tobacco etch virus (TEV) protease.
2. Materials and Methods 28
To separate tag and protease, the tag-cleaved hSirt3 was resubjected to a TALON column while
the digested zSirt5 was loaded into a HiTrap SP column. Finally, zSirt5 and hSirt3 were
subjected to Superose-12 gel filtration column (GE Healthcare, Waukesha, USA), and the
purified proteins were shock frozen and stored at -80 °C.
18.104.22.168. Affinity chromatography (AC)
The first step to purify His-tagged hSirt3 and zSirt5 was affinity chromatography. 1 ml
bed volume of TALON resin was used for every liter of E. coli culture. The resin was washed
with water followed by equilibration in lysis buffer. The supernatant containing the recombinant
protein in lysis buffer was incubated with the equilibrated resin at 4 °C for 1 hour. After the
incubation, the flow through was collected by gravity flow using a glass column (Bio-Rad,
USA). The column was washed twice with 20 bed volumes of two wash buffers and the protein
was then eluted with elution buffer. The purity and size of the proteins were analyzed using
Lysis buffer: 50 mM Tris, pH 7.8 for hSirt3 and 8.5 for zSirt5, 200 mM NaCl
The first wash buffer: 50 mM Tris, pH 7.8 for hSirt3 and 8.5 for zSirt5, 500 mM NaCl
The second wash buffer: 50 mM Tris, pH 7.8 for hSirt3 and 8.5 for zSirt5, 200 mM NaCl, 5 mM
Elution buffer: 50 mM Tris, pH 7.8 for hSirt3 and 8.5 for zSirt5, 200 mM NaCl, 250 mM
In the second AC of hSirt3 purification, the protein was eluted using gel filtration buffer.
22.214.171.124. Tag cleavage
The His-tag of zSirt5 and the His-Trx-tag of hSirt3 were cleaved using TEV protease.
The purified proteins after AC step were dialyzed in the buffer containing 30 mM HEPES, pH
2. Materials and Methods 29
6.5, 50 mM NaCl for zSirt5 and gel filtration buffer for hSirt3 at 4 °C. The ratio of protease :
protein is 1:20 mg and incubated at 4 °C overnight.
126.96.36.199. Ion exchange chromatography (IEC)
IEC was performed using a 1 ml HiTrap SP cation exchange column (GE Healthcare,
USA) that was equilibrated with buffer A (30 mM HEPES, pH 6.5). After tag cleavage, zSirt5
was applied on the column. The column was washed with 3 column volumes of buffer A
followed by elution of the protein using a linear gradient against buffer B (30 mM HEPES, pH
6.5, 1 M NaCl). Protein fractions were analyzed using SDS-PAGE and then pooled.
188.8.131.52. Size exclusion chromatography (SEC)
Elution samples from the second AC of hSirt3 or from the IEC of zSirt5 purification were
pooled and concentrated to 1 ml using an Amicon centrifugal concentrator (Millipore, USA) and
injected to an equilibrated Superose-12 size exclusion column (GE Healthcare, USA) and eluted
with gel filtration buffer containing 20 mM Tris, pH 7.8 for hSirt3 and 8.5 for zSirt5, 150 mM
NaCl. Subsequently, the purity of the eluted fractions was assessed using SDS-PAGE before
appropriate fractions were pooled and concentrated.
2.2.8. Fluorescence-based Flour-de-Lys assay
Deacetylase activity of sirtuins was tested using a commercial FdL assay kit (Enzo Life
Sciences, USA) containing the p53-derived FdL-1 substrate peptide RHK(acK) with a C-
terminally attached fluorophore. 50 µl of a reaction mixture consisting of 1.5 µg of sirtuin, 100
µM FdL-1, 2.5 mM NAD+ in the appropriate protein buffer was incubated at 37 °C for 30
minutes. Subsequently, a developer mixture containing 2 mM NAM and 10 mg/ml trypsin was
added to the reaction mixture and incubated at room temperature for 45 minutes. Trypsin cleaves
the coumarin tag from deacetylated FdL-1. Fluorescence was determined at an excitation
wavelength of 360 nm and an emission wavelength of 460 nm using a FluoDiaT70 microplate
2. Materials and Methods 30
reader (Photal Otsuka Electronics, Japan). A blank containing all the components of the assay
except the enzyme was subtracted.
2.2.9. Enzyme-coupled continuous assay
The continuous assay was performed using a published protocol (Smith, et al., 2009).
NAM, one of the products of the deacetylation reaction, is first converted to nicotinic acid and
ammonia by nicotinamidase. The ammonia is then transferred to α-ketoglutarate via glutamate
dehydrogenase yielding glutamate, under consumption of NADPH which is measured
spectrophotometrically at 340 nm and thus proportional to sirtuin activity. 100 µl of a reaction
mixture contains 2 µM of hSirt3 or 10 µM of zSirt5, 500 µM substrate peptide, 640 µM NAD+, 1
mM DTT, 3.3 mM α-ketoglutarate, 2 µM nicotinamidase, 2 units of bovine GDH and 0.2 mM
NADPH in a buffer containing 20 mM Na-PO4, pH 7.5. The reaction was performed at room
temperature for 1 hour and continuously measured using a spectrophotometer (Cary 50, Agilent
2.2.10. Mass spectrometry (MS)
50 µl of a reaction mixture consisting of 10 μM hSirt3 (in 20 mM Tris pH 7.8, 150 mM
NaCl) or zSirt5 (in 20 mM Tris pH 8.5, 150 mM NaCl), 0.5 mM ACS2 peptide and 2.5 mM
NAD+ in the presence of different compound concentrations in 2% (v/v) DMSO, or with 2%
(v/v) DMSO as a control was incubated at 37 °C. The reaction was stopped after different time
points by adding 0.25% (v/v) trifluoroacetic acid (TFA) followed by dilution to 1 µM peptide in
0.1% (v/v) formic acid (FA). Subsequently, the solution was filtered to separate the substrate
peptide from the reaction mixture using 10 kDa cutoff concentrators (Pall Life Sciences, USA).
Finally, 5 µl of each sample containing the filtered substrate peptide was subjected to nano-LC-
MS/MS analysis as described before (Fischer, et al., 2012). Specific deacetylation activity was
determined by linear fitting of the time-series experiments. The results were analyzed using
Xcalibur (Thermo Scientific, USA).
2. Materials and Methods 31
2.2.11. Thermal denaturation shift assay
Protein thermal denaturation assay measures the thermal stability of a target protein and a
subsequent increase in protein melting temperature due to the binding of a ligand to the protein
based on the fluorescence change of the dye SYPRO Orange (Life Technologies, USA). 50 µl of
a sample mixture contains 0.1 mg/ml of protein, 1 µl of 1:10 diluted SYPRO dye, 500 µM
NAD+, compounds or 2% (v/v) DMSO as a control followed by adding 15 µl of mineral oil. The
temperature was gradually increased from 25 °C to 73 °C using 2 °C intervals. The change in
fluorescence was measured at an excitation wavelength of 465 nm and an emission wavelength
of 580 nm using a FluoDiaT70 microplate reader (Photal Otsuka Electronics, Japan).
2.2.12. Binding analysis by microscale thermophoresis (MST)
Binding affinities were measured by microscale thermophoresis (Wienken, et al., 2010)
with 1 μM hSirt3 in 20 mM Tris pH 7.8, 150 mM NaCl in the presence or absence of different
concentrations of compounds or ACS2 peptide. Protein and ligands were mixed at room
temperature and transferred to capillaries for scanning before thermophoresis analysis at 25 ºC
using the NanoTemper Monolith NT.label-free instrument (NanoTemper Technologies,
Germany) with the intrinsic protein fluorescence signal (excitation at 280 nm, emission at 360
nm). The excitation UV-LED power was set to 25% and IR-laser power to 20; 40 and 80%. The
Kd values were determined through non-linear fitting (1-site equation) of the measured
thermophoresis values using Prism (Graphpad Software, CA, USA). Each experiment was
repeated at least twice.
2.2.13. Crystallization and structure determination
Crystallization trials were performed using a Phoenix robot (Art Robbins, USA) for
initial screening with a mixture of 0.15 μl of protein and 0.15 μl of reservoir solution on a 96
well sitting drop plate (Corning, Intelli, Greiner etc. plates) and incubated at 20 °C in a
Formulatrix imager (Formulatrix Inc., USA). The selected conditions were further optimized
2. Materials and Methods 32
manually by mixing 1 μl of protein and 1 μl of reservoir solutions on a 24 well sitting drop
corning plates and incubation at 20 °C.
The X-ray diffraction data were collected at 100 K with an MX-225 CCD detector
(Rayonix, Evanston, IL, USA) at beam line MX14.1 of the BESSY II electron storage ring
(Berlin, Germany) (Mueller, et al., 2012). The wavelength was 0.92 Å allowing to observe the
anomalous diffraction of the Br atom. Diffraction data were processed using XDS (Kabsch,
2010). Crystal structures were solved by Patterson searches with the program MolRep (Vagin
and Isupov, 2001) using chain A of the complex hSirt3/FdL-1/PCT (PDB ID 4HD8) (Gertz, et
al., 2012) as a search model for hSirt3 structures and the complex hSirt5/suramin (PDB ID
2NYR) (Schuetz, et al., 2007) as a search model for zSirt5 structures. Structure refinement was
performed using Refmac (Murshudov, et al., 1997), and manual rebuilding was done in Coot
(Emsley and Cowtan, 2004). Parameter files for 4’-bromo-resveratrol, polydatin and SRT1720
were generated using ProDrg (Schuttelkopf and van Aalten, 2004). The quality of the refined
structures was evaluated using Coot and MolProbity (Chen, et al., 2010). The structure figures
were prepared using Pymol (The PyMOL Molecular Graphics System, Schrödinger, LLC).
3. Results 33
3.1. Sirt3 studies
3.1.1. Sirt3 purification
hSirt3 purification was first performed using the published protocol with the vector
pET21b coding for a His-tag (Jin, et al., 2009) but the solubility and purity of the expressed
protein were very low. pVFT3S, a vector coding for a His-Trx-tag, was then used to improve
protein solubility. The hSirt3 purification protocol in this study was illustrated by a diagram in
figure 3.1A. After the first AC step, the highly expressed proteins with the size of ~60 and ~45
kDa were collected. The hSirt3 construct (118-399) is ~31 kDa and the His-Trx-tag is ~14 kDa
indicating that the second band (~45 kDa) is the target protein. MS analysis was used to identify
the protein shown in the first band (~60 kDa) and found that it was an E. coli chaperon. After
TEV protease incubation and the second AC, the His-Trx-tag was removed but the E. coli
chaperon had not yet separated. After SEC, the last step of purification, the purified hSirt3 was
obtained in the fractions B3 – B10 shown in the second peak of SEC profile with the purity ≥
95%, the E. coli chaperon was eluted in the fractions A4 – A8 shown in the first peak (Figure
3.1B). The yield of purification was 8-10 mg of the purified protein per 12 liters of the expressed
3. Results 34
Figure 3.1. Human Sirt3 purification. (A) Four steps of hSirt3 purification. (B) Step 4: SEC,
hSirt3 and E. Coli chaperon were separated. Fractions B3-B10 (green box) were pooled for
further studies. B.L, before loading.
3. Results 35
3.1.2. Resveratrol and its related compounds
184.108.40.206. Resveratrol and its related compounds are hSirt3 inhibitors
The FdL assay was performed to investigate the effects of resveratrol (RESV) and its
related compounds including piceatannol (PCT), polydatin (PD) and 4’-bromo-resveratrol
(brRESV) on hSirt3. The protein is weakly inhibited by RESV, PCT and PD (Figure 3.2A).
brRESV showed a much more potent effect and inhibited hSirt3 activity almost completely at 0.2
mM compound concentration (Figure 3.2A). In comparison to human Sirt1 (hSirt1), the same
inhibition effect of 0.2 mM brRESV but ~17-fold activation by 0.2 mM RESV was observed
Figure 3.2. FdL assay. (A) Inhibition of hSirt3 activity by resveratrol-related compounds at 0.2
mM and 1 mM compound concentration. (B) 0.2 mM RESV activates whereas brRESV inhibits
hSirt1. Activities were normalized to the control in the absence of compound. Error bars
represent standard errors of two independent measurements.
The FdL substrate peptide is attached with a fluorophore that can interact with small
molecules, potentially leading to artificial effects on sirtuin activity (Borra, et al., 2004; Gertz, et
al., 2012; Pacholec, et al., 2010). Therefore, a MS-based assay and ACS2 peptide, the acetylated
3. Results 36
peptide derived from a physiological Sirt3 substrate, were used to confirm and quantify hSirt3
inhibition by brRESV. A dose-response experiment at 500 µM ACS2 peptide yielded an IC50
value of 143.0 ± 3.6 µM (Figure 3.3). This result confirms that inhibition by brRESV applies to
Figure 3.3. IC50 determination of brRESV on hSirt3 against 500 µM ACS2 peptide using
MS. Error bars represent standard errors of linear fits to time series experiments.
220.127.116.11. Crystallization trials of hSirt3 in complex with resveratrol related compounds
To determine the inhibition mechanisms of resveratrol related compounds on hSirt3,
different complexes of the protein with FdL-1 or ACS2 peptide and compounds in the presence
or absence of NAD+ were crystallized (Table 3.1) and crystals were obtained in different
morphologies. The complex hSirt3/FdL-1/PCT, hSirt3/FdL-1/PD and hSirt3/FdL-1/brRESV
form rod-shaped crystals, the crystals of the complex hSirt3/ACS2 and hSirt3/ACS2/ NAD+/
brRESV have plate shapes and crystals of the complex hSirt3/ACS2/brRESV are in diamond
shapes (Figure 3.4). Especially, Sirt3 crystals in the presence of ACS2 peptide and brRESV were
3. Results 37
obtained in 30 different conditions of the screening core suite JCSG I – IV with the same
morphology (Figure 3.5).
Table 3.1. Crystallization trials of hSirt3 complexes
Crystallization trials Number of crystal conditions
hS3 + FdL-1 + PCT/PD/brRESV 1
hS3 + NAD+ + brRESV 0
hS3 + ACS2 + brRESV 30
hS3 + ACS2 + NAD+ + brRESV 1
Figure 3.4. Crystals of different hSirt3 complexes with FdL-1 or ACS2 peptide and
resveratrol related compounds in the presence or absence of NAD+. The compound labeled
as italic was not found in the structures.
3. Results 38
Figure 3.5. hSirt3 crystals in the presence of ACS2 peptide and brRESV were obtained in
30 different conditions with the same morphology. Three conditions were zoomed in as
18.104.22.168. Crystal structures and inhibition mechanisms of hSirt3 in complex with resveratrol
22.214.171.124.1. hSirt3 in complex with FdL-1 peptide and piceatannol/polydatin
In the crystal structures of the complex hSirt3/FdL-1/PCT and hSirt3/FdL-1/PD, the
compounds locate next to the coumarin ring of FdL-1 peptide (Figure 3.6A). The crystal contact
was formed by the interaction between two fluorophores of FdL-1 peptide molecules that belong
to two symmetry-related monomers (Figure 3.6B,C). PCT and PD bind to the protein at the same
site and directly interact with FdL-1 peptide to induce non-productive substrate binding, thus
inhibits hSirt3 activity (Gertz, et al., 2012).
3. Results 39
Figure 3.6. Crystal structure of hSirt3 in complex with PCT/PD and FdL-1 peptide. (A)
Overall structures of the complex hSirt3/FdL-1/PCT and hSirt3/FdL-1/PD. PCT is shown in
stick-yellow, PD in stick-green. FdL-1 peptides of two complexes are in stick and in the same
color as the corresponding compound. (B, C) The interface with the neighboring symmetry-
related monomer: Two FdL-1 peptides form π-stacking interactions and two PCT/PD molecules
overlay each other. Omit Fo-Fc difference density is contoured at 3.0σ. The symmetry-related
monomer is in grey.
126.96.36.199.2. hSirt3 in complex with FdL-1 peptide and 4’-bromo-resveratrol
In the hSirt3/FdL-1/brRESV complex structure, the compound was found in the active
site (Figure 3.7) and different from the PCT/PD binding site.
3. Results 40
Figure 3.7. Crystal structure of hSirt3 in complex with brRESV and FdL-1 peptide. (A)
Overall structure of hSirt3/FdL-1/brRESV complex. The dashed line indicates a loop not defined
by electron density. (B) FdL-1 peptide and brRESV ligands of hSirt3, overlaid with omit Fo-Fc
difference density (2.5 σ; green) and anomalous density (5 σ; magenta) showing the positions of
Br and Zn2+. (C) 2Fo-Fc electron density (1 σ; blue) of FdL-1 peptide and brRESV in hSirt3/
A closer look at the compound binding site (Figure 3.8) shows that the A-ring hydroxyl
groups of brRESV form hydrogen bonds with Asn229 and Asp231 of hSirt3. Furthermore,
residues Ile230, Leu199, and Ile154 form a hydrophobic patch for A-ring binding, and Phe157,
Leu195, and Phe180 a hydrophobic cleft for accommodating the B-ring. This cleft extends in a
hydrophobic pocket (formed by Ile179, Leu173, and Tyr171) for binding the bromine atom, and
Arg158 and Pro176 form a lid shielding this pocket from solvent.
3. Results 41
Figure 3.8. Closer view on the brRESV binding site showing interacting residues. Hydrogen
bonds are indicated by dashed lines.
Superposition of the crystal structure of the complex hSirt3/FdL-1/brRESV and Sir2Tm
in complex with p53 peptide and NAD+ (Hoff, et al., 2006) reveals that brRESV occupies part of
the NAD+ binding pocket (Figure 3.9), thus prevents the C-pocket insertion of the NAD+
nicotinamide moiety necessary for catalysis.
Figure 3.9. Superposition of the hSirt3/FdL-1/brRESV structure with a Sir2Tm/p53/NAD+
complex (PDB ID 2H4F) (Hoff, et al., 2006). FdL-1 peptide and brRESV are in pink, p53
peptide and NAD+ are in green. The protein part of the Sir2Tm/p53/NAD+ complex is omitted
3. Results 42
The overall hSirt3 and FdL1-peptide conformations in the superposition of the brRESV
complex with hSirt3 bound to FdL-1 and other resveratrol-related compounds, PCT/PD, are
identical (Figure 3.10). However, FdL-1 peptide conformations are different, in particular the
fluorophore orientation. PCT/PD interact extensively with the FdL-1 fluorophore to induce non-
productive peptide binding (Gertz, et al., 2012) whereas brRESV does not directly contact this
substrate peptide but blocks productive NAD+ binding.
Figure 3.10. Superposition of the hSirt3/FdL-1/brRESV with the hSirt3/FdL-1/PCT
complex. PCT is shown in stick-yellow, brRESV in stick-pink. FdL-1 peptides of two complexes
are in stick and in the same color as the corresponding compound.
To test for competition between brRESV and FdL-1 peptide, IC50 values for brRESV
inhibition of hSirt3 were determined at three different concentrations of FdL-1 peptide (50, 100,
3. Results 43
and 200 µM). The IC50 value obtained were ~100 µM for all three peptide concentrations (Figure
3.11), indicating that the inhibitor brRESV does not compete against FdL-1.
Figure 3.11. IC50 determination for brRESV inhibition of hSirt3 at 50, 100, and 200 M
FdL-1 substrate peptide.
We also investigated the competition between brRESV and NAD+ for hSirt3 binding.
The binding affinity of brRESV to the apo protein was 7.6 ± 0.9 µM, and the Kd increased to
higher than 50 µM in the presence of 2 mM NAD+ (Figure 3.12). The results reveal that brRESV
competes with NAD+ for binding to hSirt3, supporting the conclusion that the internal brRESV
binding site is the one relevant for inhibition of hSirt3 activity.
Figure 3.12. Binding affinity of brRESV to hSirt3 in the presence or absence of 2 mM
3. Results 44
188.8.131.52.3. hSirt3 in complex with ACS2 peptide and 4’-bromo-resveratrol
In the complex structure hSirt3/ACS2/brRESV, the compound molecule was found at the
bottom of the Rossmann-fold domain, interacting with Arg139, Met331, and Arg335 (Figure
3.13). In this exposed position, the compound interacts only through its A-ring with this shallow
hSirt3 pocket, and the bromo-containing aromatic ring points towards the symmetry-related
monomer in the crystal lattice.
Figure 3.13. Crystal structure of hSirt3 in complex with brRESV and ACS2 substrate
peptide. (A) Overall structure of hSirt3/ACS2/brRESV complex. A missing loop is indicated by
a dashed line. (B) 2Fo-Fc electron density (1 σ; blue) of ACS2 peptide in hSirt3/ACS2/brRESV
complex. (C) brRESV ligand of hSirt3, overlaid with omit Fo-Fc difference density (2.5 σ; green)
and anomalous density (5 σ; magenta) showing the position of Br and surrounding residues
important for compound binding. Residues from the symmetry-related monomer are labeled with
a star. (D) 2Fo-Fc electron density (1 σ; blue) of brRESV in hSirt3/ACS2/brRESV complex.
3. Results 45
The superposition of the hSirt3/brRESV complexes with FdL-1 and ACS2 peptide,
respectively, reveals that the inhibitor cannot bind at the catalytic pocket when the ACS2 peptide
is bound, since it would clash with the C-terminal part of this substrate peptide (Figure 3.14).
Figure 3.14. Superposition of the hSirt3/ACS2/brRESV structure with the hSirt3/FdL-
1/brRESV complex. FdL-1 peptide and brRESV are in pink, ACS2 peptide is in cyan. Phe157
of the hSirt3/ACS2/brRESV and hSirt3/FdL-1/brRESV complex are shown in cyan and pink,
brRESV in the complex structure with hSirt3/ACS2-peptide does not show many
interactions with hSirt3, rendering it a less likely inhibition site. Figure 3.15 illustrates the
hydrogen bonds between brRESV with Arg139, Met331 (backbone), and Arg335 and
additionally with Arg384 of the symmetry-related monomer.
3. Results 46
Figure 3. 15. Closer view on the brRESV binding site in the hSirt3/ACS2/brRESV complex
showing interacting residues. Hydrogen bonds are indicated by dashed lines and residues of the
next symmetry-related monomer are labeled with a star.
To test whether the surface site occupied by brRESV in its hSirt3/ACS2 complex is
relevant for inhibition, the role of the interacting residues R139, M331, R335, and R384 were
tested by site-directed mutagenesis to alanine. Microscale thermophoresis (MST) results indicate
that the hSirt3-R335A variant has a slightly reduced binding affinity for the compound compared
to wildtype hSirt3 (Figure 3.16A), but the change was not statistically significant. The hSirt3
variants R384A, R139A, and M331A showed no change in the affinity for the compound.
Moreover, we performed activity assays with 500 µM ACS2 as peptide substrate and in presence
of brRESV (at its IC50 concentration, 140 µM) to examine the effects on the activity of the
mutant proteins. The screening indicated that the activity of hSirt3 R335A is slightly higher than
for the other mutants and the wildtype protein (data not shown). The IC50 value of brRESV on
the R335A variant activity was then determined. Figure 3.16B shows only a small shift for the
brRESV inhibition curves between wildtype protein and the R335A variant, resulting in IC50
values of 143.0 ± 3.6 and 179.2 ± 12.3 µM, respectively. Therefore, the position of brRESV in
the complex structure hSirt3/ACS2/brRESV seems not to be primarily responsible for the
inhibitory effect of the compound on hSirt3.
Figure 3.16. (A) Affinity measurements for binding of brRESV to hSirt3 wildtype as well as
R335A and R384A variants. Error bars represent standard errors of two independent
measurements. (B) IC50 determination of brRESV on wildtype hSirt3 or the R335A variant
against 500 µM ACS2 peptide using MS. Error bars represent standard errors of linear fits to
time series experiments.
To test for brRESV competition with ACS2 peptide, MS assays were performed to
determine Km and Vmax values for the peptide at different compound concentrations. The result
shows that the higher the brRESV concentration, the more the Km for substrate peptide increases,
or affinity for substrate decreases, while the Vmax is not altered, indicating competitive inhibition
3. Results 48
Figure 3.17. brRESV inhibition of hSirt3 with ACS2 peptide as a substrate. brRESV
concentrations of 0, 50, and 150 µM resulted in Km values for substrate peptide of 31.3 ± 9.0,
48.6 ± 7.5, and 253.3 ± 52.9 µM, respectively, while the Vmax is roughly constant at ~50 nmol
mg-1 min-1. Error bars represent standard errors of linear fits to time series experiments.
This competition was confirmed through binding data from MST measurements. ACS2
peptide bound to apo hSirt3 with a Kd value of 64.4 ± 9.1 µM. In presence of 50 µM brRESV,
the Kd value increased to more than 200 µM (Figure 3.18A). Consistently, binding affinity of
brRESV to the apo protein was 7.6 ± 0.9 µM, and the Kd increased to higher than 37 µM in the
presence of 500 µM ACS2 peptide (Figure 3.18B). Thus, brRESV inhibits hSirt3 through
binding competition with ACS2 peptide substrate.
Figure 3.18. (A) Binding affinity of ACS2 peptide to hSirt3 in the presence or absence of 50 µM
brRESV. (B) Binding affinity of brRESV to hSirt3 in the presence or absence of 500 µM ACS2
peptide. Error bars represent standard errors of two independent measurements.
3. Results 49
3.1.3. Resveratrol unrelated compounds
The effect of SRT1720 on hSirt3 was investigated using FdL-1 and ACS2 peptide. The
FdL assay indicates that 20 µM SRT1720 inhibits 92% hSirt3 activity but activates hSirt1
activity to 1.6 fold and has no effect on human Sirt2 (hSirt2) (Figure 3.19).
Figure 3.19. Effects of 20 µM SRT1720 on hSirt1, hSirt2 and hSirt3 in the FdL assay.
Activities were normalized to the control in the absence of compounds. Error bars represent
standard errors of two independent measurements.
MS was used to examine the effect of the compound on ACS2, the fluorophore-free
peptide. The IC50 value of SRT1720 on hSirt3 is 11 ± 1 M for ACS2 peptide (Figure 3.20).
Figure 3.20. IC50 determination of SRT1720 on hSirt3 against 500 µM ACS2 peptide using
MS. Error bars represent standard errors of linear fits to time series experiments.
3. Results 50
Since SRT1720 was reported as a noncompetitive inhibitor of hSirt3 against NAD+ (Jin,
et al., 2009), hSirt3 in complex with the compound was crystallized in the presence or absence of
NAD+. Crystals were observed in many conditions with different morphologies (Figure 3.21).
Crystals of the complex hSirt3/NAD+ were further soaked with SRT1720 at different time points.
Figure 3.21. Crystals of hSirt3 in complex with NAD+ in the presence or absence of SRT1720.
Since the compound was not found in the structures, it was labeled as italic.
3. Results 51
In both cases, hSirt3/NAD+ soaking or co-crystallizing with SRT1720, the structures only
showed the ADPR moiety of NAD+ and no density for the compound (Figure 3.22). This might
indicate that NAD+ was hydrolyzed during crystallization that was mentioned in the Introduction
Figure 3.22. Active site of the hSirt3/ADPR complex. 2Fo-Fc electron density of ADPR is
contoured at 1.0σ. The cosubstrate binding loop is shown in purple.
To prevent the hydrolysis of NAD+ during crystallization, we used inert carba-NAD(CAS# 112345-60-5)+
(Szczepankiewicz, et al., 2012). This NAD+ analog is stable in crystallization conditions due to
its ability to prevent the NAM displacement reactions. The replacement indeed resulted the
crystal structure of a Sirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 complex. The quinoxaline ring of SRT1720
interacts with NAM moiety of carba-NAD(CAS# 112345-60-5)+ and Phe157 of the cobsubstrate binding loop through
π-stacking interaction (Figure 3.23). This observation explains the absence of the compound in
Sirt3/NAD+ complex structure when using unstable NAD+ and indicates the role of NAD+, in
particular NAM moiety, for the compound binding. Superposition of the complex hSirt3/carba-
NAD+/SRT1720 with hSirt3/ACS2/carba-NAD(CAS# 112345-60-5)+ (PDB ID 4FVT) (Szczepankiewicz, et al., 2012)
reveals that the piperazine group and part of the connected imidazothiazole system of SRT1720
occupy the binding region of acetyl lysine (Figure 3.24); thus confirm the inhibition mechanism
competitive with substrate peptide (Jin, et al., 2009). Due to the compound binding, the
3. Results 52
cosubstrate binding loop was shifted and Phe157 was re-orientated whereas the carba-NAD(CAS# 112345-60-5)+
conformation was almost unchanged (Figure 3.24).
Figure 3.23. Active site of the hSirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 complex. 2Fo-Fc electron density
of carba-NAD(CAS# 112345-60-5)+ and SRT1720 is contoured at 1.0σ. The cosubstrate binding loop is shown in red,
SRT1720 in orange and carba-NAD(CAS# 112345-60-5)+ in yellow.
Figure 3.24. Superposition of the complex hSirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 with
hSirt3/ACS2/carba-NAD(CAS# 112345-60-5)+ (green, PDB ID 4FVT) (Szczepankiewicz, et al., 2012).
3. Results 53
To confirm the critical role of the NAM moiety of NAD+ for the compound binding, the
stability of hSirt3 towards thermal denaturation was measured using thermal shift assay (TSA)
(Figure 3.25). In the presence of 50 µM SRT1720, the half-point of the hSirt3 melting transition
(Tm) value increased from less than 324 K of the DMSO control to more than 326 K. The Tm
value increased to more than 327 K when adding 500 µM NAD+ in the presence of 50 µM
SRT1720. The measurements imply that SRT1720 supports the hSirt3 stabilization and NAD+
supports the compound binding.
Figure 3.25. hSirt3 stability TSA test in presence or absence of 50 µM SRT1720 and/or 500
µM NAD+. (A) Transition curves. The lines shown are nonlinear fits for a two state transition.
(B) Melting temperatures. Error bars represent standard errors of nonlinear fits.
The stability measurements of SRT1720 to hSirt3 are consistent with the binding data
using microscale thermophoresis (MST) (Figure 3.26). The Kd value for the affinity of SRT1720
to hSirt3 is 7.5 ± 1.3 µM. In the presence of 500 µM NAD+, the Kd value decreased to 2.6 ± 0.3
µM. When adding 500 µM ADPR instead, the Kd value is 7.8 ± 1.3 µM indicating that ADPR
does not support the compound binding but NAD+ with additional NAM moiety in comparison
to ADPR provides a positive effect. The results of binding and stability measurements strengthen
the conclusion of the contribution of NAD+, in particular its NAM moiety, in compound binding.
3. Results 54
Figure 3.26. Binding affinity of SRT1720 to hSirt3 in the presence or absence of 500 µM
NAD+ or ADPR. Error bars represent standard errors of two independent measurements.
Ex527 was reported as a potent inhibitor of hSirt1 with the IC50 value of approximately
0.1 M (Solomon, et al., 2006). It has much lower potency against hSirt3 with IC50 about 50 M.
In this study, hSirt3 was used as a model to understand the inhibition of this compound on
sirtuins. In the crystal structure of the complex hSirt3/NAD+/Ex-527, Ex-527 binds to hSirt3 at C
pocket (Figure 3.27). When using ADPR instead, the compound can also bind to hSirt3 at the
same site (Gertz, et al., 2013) indicating that the compound binds to the protein when the
cosubstrate binding pocket is occupied either by NAD+ or the product 2’-O-acetyl-ADP-ribose.
3. Results 55
Figure 3.27. Active site of the hSirt3/ NAD+/Ex-527 complex. 2Fo-Fc electron density of
NAD+ and Ex-527 is contoured at 1.0σ.
To examine whether Ex-527 can bind to hSirt3 during the step of forming O-
alkylamidate intermediate, the crystal of hSirt3/ACS2 was soaked with NAD+ and Ex-527 in 80
minutes. The obtained structure showed very clear native O-alkylamidate intermediate state of
the reaction but density for Ex-527 was not found (Figure 3.28). In comparison to the S-
alkylamidate intermediate, the ribose moiety of the native O-alkylamidate intermediate has a
different conformation (Figure 3.29). In combination with inhibition kinetics and binding
analysis, the inhibition mechanism of Ex-527 on sirtuins was revealed to be that the compound
stabilizes the closed enzyme conformation of the complex with 2’-O-acetyl-ADP-ribose, thus
prevents product release (Gertz, et al., 2013).
3. Results 56
Figure 3.28. Active site of the hSirt3/ O-alkylamidate intermediate complex. 2Fo-Fc electron
density of the intermediate is contoured at 1.0σ.
Figure 3.29. Superposition of the hSirt3/ O-alkylamidate intermediate complex with hSirt3/S-
alkylamidate intermediate complex (pink, PDB ID 3GLT) (Jin, et al., 2009).
3. Results 57
3.2. Sirt5 studies
3.2.1. Sirt5 purification
In a previous study, human Sirt5 (hSirt5) purification yielded high amount of the purified
protein (Gertz, et al., 2012). However, crystallization trials resulted in twinned crystals and low
occupancy of ligands. Therefore, zSirt5, an orthologue of hSirt5 was used in this study to
overcome the issue. The zSirt5 purification protocol in this study was illustrated by a diagram in
figure 3.30A. After the AC step, the highly expressed protein with the size of ~32 kDa was
collected. The zSirt5 construct (30-298) is ~31 kDa and the His-tag is ~1 kDa. After TEV
protease incubation and cation exchange, the His-tag was removed and zSirt5 was eluted in the
fractions of the first peak with a small remaining contamination. After SEC, the last step of
purification, the purified zSirt5 was obtained in the fractions B5 – B12 with the purity ≥ 95%
(Figure 3.30B). The yield of purification was 6 mg of the purified protein per 12 liters of the
3. Results 58
Figure 3.30. Zebrafish Sirt5 purification. (A) Four steps of zSirt5 purification. (B) Step 4:
SEC, contamination was separated, purified zSirt5 including fractions B5-B12 (green box) was
pooled for further studies. B.L, before loading.
3. Results 59
3.2.2. Resveratrol and its related compounds are zSirt5 activators on FdL-1 peptide
Similar to Sirt3 study, the FdL assay was performed to investigate the effects of RESV
and its related compounds including PCT, PD and brRESV on zSirt5. Among these compounds,
1 mM RESV can activates zSirt5 up to 12-fold, thus becomes the most potent activator of the
enzyme (Figue 3.31). PCT, PD and brRESV slightly activate zSirt5 with nearly 3-fold in the
presence of 0.2 mM and 6-fold in the presence of 1 mM compound concentration, except
Figure 3.31. Activation of zSirt5 activity on FdL-1 peptide by resveratrol-related
compounds at 0.2 mM and 1 mM compound concentration. Activities were normalized to the
control in the absence of compound. Error bars represent standard errors of two independent
To prevent the artificial effect of FdL-1 peptide as mentioned in the Sirt3 study section,
several fluorophore-free acetylated peptides were screened to find potent peptide substrates for
zSirt5 using continuous assay. Among chosen peptides, p53 peptide is the most potent substrate
of zSirt5 but not CPS1, the acetylated peptide derived from the physiological mammalian Sirt5
substrate (Figure 3.32). Lamin_B2 peptide is also a potent zSirt5 peptide substrate but has low
solubility due to its rich hydrophobic residues. ME peptide has the similar linear of NADPH
3. Results 60
consumption as CPS1 peptide indicating that it is also a zSirt5 substrate (Figure 3.32). ME and
p53 peptide were chosen for further study. The regulations of resveratrol related compounds on
these peptides were examined using MS. However, these compounds did not show significant
effects (weak or unclear activation or inhibition) on the peptides (Figure 3.33).
Figure 3.32. zSirt5 substrates were identified using continuous assays
Figure 3.33. Insignificant effects of resveratrol related compounds on zSirt5 using MS. (A)
p53 peptide. (B) ME peptide with 0.2 mM of each compound. Error bars represent standard
errors of linear fits to time series experiments.
3. Results 61
3.2.3. Crystallization trials and crystal structures of zSirt5 in complex with peptide
substrates in the presence of resveratrol
Since RESV is the most potent activator of zSirt5 in comparison to its related
compounds, it was used for crystallization study to determine the activation mechanism on the
enzyme. Different crystallization trials including zSirt5 in the presence or absence of peptide
substrate and RESV were setup. The diamond crystals were obtained from the condition
containing the mixture of zSirt5 and RESV in comparison with no crystal when using DMSO as
a control (Figure 3.34). The compound thus seems to be important for crystal growing. The
crystals of the complex zSirt5/FdL-1 in the presence of RESV are in rod shape with a nice
packing (Figure 3.34). However, the diffractions of these crystals are quite weak and the best
data set is only 3.2 Å. The complex zSirt5/ME in the presence of RESV has very big rod crystals
(Figure 3.34) and their diffractions are up to 2 Å. The complex zSirt5/p53 in the presence of
RESV formed long stick crystals (Figure 3.34).
Figure 3.34. Crystals of different zSirt5 complexes with FdL-1 or p53 or ME peptide in the
presence of RESV. Since the compound was not found in the structures, it was labeled as italic.
3. Results 62
Similar to other sirtuin/peptide complexes, ME or p53 peptide binds to the cleft between
the Rossmann-fold and zinc-binding domain of zSirt5 and the acetyl lysine binds into a
hydrophobic tunnel pointing toward the catalytic residue His158 (Figure 3.35). No density fits to
RESV implying the compound could not bind or bound with very low occupancy that could not
be observed, consistent with the lack of an effect in activity assays with these peptides.
Superposition of the complex zSirt5/ME/RESV and zSirt5/p53/RESV indicates that the protein
has the same conformation in both structures (data not shown).
Figure 3.35. Active site of complex structures. (A) zSirt5/p53/RESV complex and (B)
zSirt5/ME/RESV complex. Omit Fo-Fc difference density is contoured at 3.0σ.
zSirt5 was crystallized in the presence of RESV but the obtained structure only showed
the apo enzyme. The asymmetric unit of the apo-zSirt5 has four monomers whereas the complex
zSirt5/FdL-1/RESV contains five zSirt5 monomers and the electron density of FdL-1 was found
in only one monomer (Figure 3.36). RESV was included in the solution but not present in the
structures. The protein packing may be caused by the unspecific binding between monomers
including disulfide bonds formed by the residues Cys278.
3. Results 63
Figure 3.36. (A) Superposition of apo-zSirt5 (salmon pink) and the zSirt5/FdL-1/RESV complex
(cyan), the density of the compound was not found. The asymmetric unit of apo structure is
tetramer and of the zSirt5/FdL-1/RESV is pentamer but only one monomer contains FdL-1
peptide. zSirt5 is shown in cartoon, Zn2+ as a sphere and the peptide is in stick representation.
(B) Active site of the monomer containing FdL-1 peptide. 2Fo-Fc electron density of FdL-1
peptide is contoured at 1.0σ.
Superposition of the FdL-1-containing monomer and apo monomer reveals two positions
of conformation change: the loop 250 – 260 and the loop 277 – 284 (Figure 3.37). The loop 250
– 260 is peptide-binding loop, thus it moves closer to the peptide when the peptides bind to the
protein. The loop 277 – 284 is on the surface of monomer contacts. Figure 3.38 shows the crystal
contact of FdL-1 containing monomer of the zSirt5/FdL-1/RESV complex. The peptide is close
to the loop 277 – 284 of the next symmetry-related monomer indicating that the loop
conformation is also influenced by the presence of the peptide. The overall conformations of the
complex zSirt5/FdL-1/RESV and zSirt5/ME/RESV are almost identical except a slight difference
in the loop 277-284 (Figure 3.39). This may cause by different sequences and lengths of the
peptides. Therefore, the loop 277-284 conformation is very flexible and depends on the presence,
sequence and length of peptides.
3. Results 64
Figure 3.37. Superposition of apo-zSirt5 (salmon pink) and the FdL-1 containing zSirt5
monomer (cyan) with the positions of conformation changes showed in black boxes: loop
250-260 and loop 277-284.
Figure 3.38. Crystal contact of the zSirt5/FdL-1/RESV complex. The symmetry-related
monomer is shown in grey. The FdL-1 peptide and the loop 277-284 of the complex is in dark
blue and of the symmetry-related monomer is in grey-black.
3. Results 65
Figure 3.39. Superposition of the complex zSirt5/FdL-1/RESV (cyan) and zSirt5/ME/RESV
(yellow). zSirt5 is shown in cartoon, Zn2+ as a sphere and the peptide is in stick representation.
4. Discussion 66
4.1. Sirt3 studies
4.1.1. Resveratrol and its related compounds
Resveratrol and its related compounds are hSirt3 inhibitors with brRESV as the most
potent candidate. Unlike PCT or PD which directly interact with the FdL-1 fluorophore for the
inhibition effect, brRESV in the FdL-1 complex occupies a part of the C pocket where the NAM
moiety of NAD+ binds to initialize the reaction and where NAM can bind to an alkylimidate
complex (Sauve, et al., 2006) for the reverse reaction. In comparison to the Sir2Tm/NAM
complex structure (Avalos, et al., 2005), the 1-OH group of the brRESV A-ring plays a role like
the NAM carboxamide group, which also interacts with the conserved Asp101 in Sir2Tm
(Asp231 in hSirt3) (Figure 4.1). A novel binding site for the 4’-bromo-reveratrol B-ring in
hSirt3/FdL-1 complex provides a base to develop a new class of sirtuin inhibitors.
Figure 4.1. Superposition of the complex hSirt3/FdL-1/brRESV with Sir2Tm/p53/NAM
(PDB ID 1YC5) (Avalos, et al., 2005). Dashed lines indicate hydrogen bonds of the brRESV A-
ring hydroxyl groups and of the NAM carboxamide group to protein residues. brRESV is shown
in pink and NAM in yellow. The protein part of the Sir2Tm/p53/NAM complex is omitted for
clarity. Sir2Tm residues are labeled with a star.
4. Discussion 67
Superposition of the complex hSirt3/ACS2/brRESV and hSirt3/FdL-1/brRESV reveals
that a loop of a symmetry related monomer in the FdL-1 complex prevents the compound to bind
to this allosteric site (Figure 4.2). RESV was reported as the most potent natural small molecule
activator on Sirt1 and its orthologs (Howitz, et al., 2003). An activation mechanism of this
compound on Sirt1 has not been successfully investigated due to the lacking of a Sirt1/RESV
complex structure. Superposition of the hSirt3/ACS2/brRESV complex with a hSirt1 model
indicates that the N-terminal extension could contribute to the external brRESV binding site. The
residue Glu230 that is essential for Sirt1 activation (Hubbard, et al., 2013) is located next to this
binding site, thus it might be the allosteric activation site in hSirt1 (Figure 4.3).
Figure 4.2. Analysis of crystal contacts. Superposition of the complex hSirt3/ACS2/brRESV
(blue) and hSirt3/FdL-1/brRESV (pink and grey). brRESV of the ACS2 complex (cyan) clashes
with the substrate binding loop of the symmetry related monomer (grey) of the FdL-1 complex.
brRESV of the symmetry related FdL-1 complex is shown in pink and labeled with a star.
4. Discussion 68
Figure 4.3. Superposition of the complex hSirt3/ACS2/brRESV with a homology model of
hSirt1. brRESV (cyan) of hSirt3 (blue) is close to the putative position of Glu230 of hSirt1
(magenta and orange, respectively) suggesting the involvement of the compound binding site to
an allosteric hSirt1 activation mechanism. hSirt1 N- and C-terminus and Glu230 are labeled with
a star. A missing hSirt3 loop is indicated by a dashed line.
4.1.2. Resveratrol unrelated compounds
SRT1720 was known as a potent synthetic activator on Sirt1 with a much higher effect
than RESV (Milne, et al., 2007). SRT1720 inhibits hSirt3 with high potency and isoform
selectivity. Although it showed competitive inhibition with substrate peptide and uncompetitive
with NAD+ using kinetic analysis (Jin, et al., 2009), the lacking of a complex structure with
details about the binding site prevents full understanding of the inhibition mechanism. The
hSirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 complex structure in this study reveals the molecular inhibition
4. Discussion 69
mechanism of the compound. Consistent with the kinetic study (Jin, et al., 2009), the compound
interacts with NAM moiety of NAD+ and Phe157 of the cosubstrate binding loop to form π-
stacking sandwich and interferes substrate peptide by occupying a part of acetyl lysine. Further
improvements to obtain more effective inhibitor can be developed from the novel binding site of
SRT1720. It can be strengthening either the π-stacking interaction or substrate peptide
competition. Based on the inhibition mechanism of Ex-527 (Gertz, et al., 2013) and the binding
site of SRT1720 in hSirt3 complex structure, an activation mechanism of SRT1720 on Sirt1 can
be speculated. SRT1720 would clash with the acetyl ribose moiety of the product (Figure 4.4)
indicating that the compound might bind to Sirt1 after product formation and support product
Figure 4.4. Superposition of the complex hSirt3/carba-NAD(CAS# 112345-60-5)+/SRT1720 with hSirt3/OAcADPR/Ex-527 (PDB ID 4BVH) (Gertz, et al., 2013). SRT1720 (orange) would
clash with the product 2’-O-acetyl-ADP-risbose (green). Carba-NAD(CAS# 112345-60-5)+ and Ex-527 were omitted
The inhibitor Ex-527 has high selectivity for Sirt1 and lower potency on other isoforms
(Solomon, et al., 2006). Ex-527 binds to sirtuins at the C pocket but extends in a different
direction in comparison to the internal binding site of brRESV in the FdL-1 complex (Figure
4.5). The Ex-527 molecule extends perpendicular to its carbamide in a hydrophobic pocket close
4. Discussion 70
to the acetyl-Lys binding site whereas brRESV extends toward a pocket close to the enzyme’s
surface. The crystal structure of the complex hSirt3/NAD+/Ex-527 and hSirt1/NAD+/Ex-527
(Zhao, et al., 2013) indicate that the compound can interact with the ribose and NAM moiety of
NAD+, force NAD+ to bind in non-productive conformation and thus affects the peptide binding.
However, a previous study indicated that the compound is an uncompetitive inhibitor of Sirt1
against both peptide substrate and NAD+ (Napper, et al., 2005). Based on competition
experiments on the base exchange activity observed after intermediate formation, it was
suggested to bind after binding of both substrates, and possibly to act by preventing the release
of one or both of the products, 2’-O-acetyl-ADP-ribose and the acetylated peptide (Napper, et al.,
2005). Moreover, the uncompetitive inhibition behavior in kinetic experiments was confirmed in
our lab; thus the hSirt3/NAD+/Ex-527 complex structure appears not to be relevant for the
inhibition mechanism. Our study reveals that Ex-527 inhibits sirtuins by binding to the product
2'-O-acetyl-ADP-ribose complex to prevent product release. The complex hSirt3/OAADPr/Ex-
527 explains the kinetics mentioned above. In addition, we confirmed that the compound binds
to the enzyme after intermediate formation and one product molecule per enzyme molecule is
formed before inhibition.
Figure 4.5. Superposition of the complex hSirt3/FdL-1/brRESV with hSirt3/OAcADPR/Ex-
527 (PDB ID 4BVH) (Gertz, et al., 2013). Both brRESV (pink) and Ex-527 (green) occupy the
C-site but extend in different directions.
4. Discussion 71
4.2. Sirt5 studies
An effort to investigate the activation mechanism of RESV on zSirt5 has not been
successful. The compound was not found in crystal structures of zSirt5 complex that might due
to its low solubility. However, the study obtained different zSirt5 structures including the apo
structure and the protein in complex with different peptide substrates. Therefore, it provides
structural models for further study on screening and investigating the regulation of other small
molecule compounds with higher solubility, potency and selectivity on zSirt5.
Ahn, B.H., Kim, H.S., Song, S., Lee, I.H., Liu, J., Vassilopoulos, A., Deng, C.X., and Finkel, T.
(2008). A role for the mitochondrial deacetylase Sirt3 in regulating energy homeostasis. Proc