Biogerontology
https://doi.org/10.1007/s10522-018-9744-x
RESEARCH ARTICLE
Repurposing existing drugs for new AMPK activators
as a strategy to extend lifespan: a computer-aided drug
discovery study
Sepideh Mofidifar . Farzin Sohraby . Milad Bagheri . Hassan Aryapour
Received: 7 October 2017 / Accepted: 4 January 2018
Ó Springer Science+Business Media B.V., part of Springer Nature 2018
Abstract Dietary restriction is one of the several
ways which could putatively extend organisms’ lifes-
pan, ranging from Saccharomyces cerevisiae to
rodents, by activating the AMP-activated protein
kinase (AMPK), an ATP/AMP sensor. Extensive
researches have shown that aging reduces sensibility
of AMPK and eventually causes energy imbalance in
cells. Research in mammals’ AMPK depicts that this
signaling molecule could control autophagy, improve
cellular stress resistance and suppress inflammatory
responses. Hence, in this study we performed a drug
repurposing of 1908 FDA-approved drugs in order to
discover putative safe activators of AMPK and to find
new applications for existing drugs. For this purpose,
FDA-approved drugs were screened by virtual screen-
ing and the ligand–protein interactions were carefully
inspected. Moreover, through MM/PBSA analysis, the
binding affinity of hit compounds in c and ab binding
sites were investigated. As Cangrelor, Nacitentan,
Levoleucovorin and Glisoxepide had lower binding
affinities; we predicted that they would probably prove
Electronic supplementary material The online version of
this article (https://doi.org/10.1007/s10522-018-9744-x) con-
tains supplementary material, which is available to authorized
users.
S. Mofidifar  F. Sohraby  M. Bagheri 
H. Aryapour (&)
Department of Biology, Faculty of Science, Golestan
University, Gorgan, Iran
e-mail: h.aryapour@gu.ac.ir
to be more potential activators than C2. However, hit-
compounds in ab binding site, exhibited higher
unfavorable binding affinity. Hence, present findings
can prove to be valuable for discovering new activa-
tors for AMPK.
Keywords AMPK  Drug repurposing  Virtual
screening  FDA-approved drugs
Introduction
Aging is a multifactorial process, which causes
deterioration of the physiological function. Studies
have shown that caloric restriction could potentially
decrease aging processes (Hall et al. 2013), which not
only has profound influence on life span, but also
delays some age-related pathologies, such as cancer
and neurodegenerative diseases (Maswood et al.
2004). DR is able to increase AMP/ATP and ADP/
ATP rations. Subsequently, the AMPK is allosterically
activated by high concentration of AMP (Salminen
and Kaarniranta 2012). AMPK is a conserved serine/
threonine protein kinase that appears to exert a variety
of biological functions. It is a putative regulator of
cell’s homeostasis (Hardie and Hawley 2001), which
sustains cellular life span by phosphorylating a large
number of downstream proteins. Therefore, in
response to energy balance, AMPK activates catabolic
processes, while inhibiting anabolic pathways (Rana
123

et al. 2014). Consequently, mTOR proteins that are the
regulators of metabolism and cellular growth are
switched off (Dalle Pezze et al. 2016) and sirtuin is
switched on as an anti-aging protein (Cantó et al.
2009) (Fig. S1 in supporting information).
Structurally, AMPK is a heterotrimeric kinase
expressed in many cells. It is composed of one
catalytic subunit a and two regulatory subunits b and
c (Fig. S2 in supporting information) (Steinberg and
Kemp 2009). The a subunit consists of N-terminal
kinase domain (KD) and C-terminal subunit interact-
ing domain (SID). Phosphorylation of threonine 172 in
KD domain has crucial function in activating the
AMPK that is carried out by the signaling pathways of
calcium/calmodulin-dependent protein kinase kinase
(CaMKK) and liver kinase B1 (LKB1); two major
upstream activators of AMPK (Crute et al. 1998).
The b subunit of AMPK contains two main
domains: carbohydrate binding domain (CBD) and
subunit binding sequence at the C-terminal, which
interact with both a and c subunits (Townley and
Shapiro 2007). ‘‘ADaM’’ (allosteric drug and metabo-
lite) site is located in the interface between a and b
subunits, where a large number of AMPK activators
bind to it (Cameron and Kurumbail 2016). The
synthesis of activators in this site prevents the dephos-
phorylation of AMPK by phosphatases (Li et al. 2015).
The c subunit is sensitive to AMP: ATP ratio within
the cells. Each c subunit contains four repeats of
cystathionine beta-synthase (CBS) motifs. Although,
site 4 is a non-exchangeable AMP binding site, sites 1
and 3 binds to nucleotide in a reversible manner (Chen
et al. 2013). Studies have shown that AMP could
potentially change the proteins conformation in a way
that dephosphorylation is blocked (Sanders et al. 2007).
It has previously been investigated that drugs’ side-
effect is due to the fact that they may have more than
one target. Hence, drug repurposing has been devel-
oped, which identifies new targets for existing drugs
(Ma et al. 2013). Additionally, new target identifica-
tion for existing drugs may be a useful method for
saving money and time, because of complex and time-
consuming process of drug discovery (Gupta et al.
2013). In silico drug repurposing is one of the widely
used novel methods (Cavalla 2013). It also grabs aging
studies’ attention in recent years (Calvert et al. 2016).
Here, we performed the structure-based virtual
screening (SBVS) to predict valuable activators for
AMPK. Subsequently, molecular dynamics (MD)
123
Biogerontology
simulation, binding free energy calculations and
related analyses were carried out to investigate the
stability of resultant complexes. Finally, several drugs
were predicted to be the potential activators of AMPK.
Materials and methods
Protein preparation
The X-ray micrograph of the crystal structure of
human AMPK (PDB code 4ZHX) was obtained from
Protein Data Bank. Due to large number of missing
amino acids in 4ZHX (112 aa in a, 95 aa in b and 36 aa
in c subunit), Swiss-Model server was used to build a
reliable 3D model of the protein using amino acid
sequence (Bordoli et al. 2009). Then the model was
used as template in modeler for refining missing parts
of the experimental structure.
Molecular docking
In this study, a total of 1908 FDA-approved drugs were
retrieved from DrugBank database (on 18/01/2017).
These 2D drugs were converted to their 3D forms by
molconverter from Jchem toolkit (Version 16.8.15.0,
ChemAxon). After that, virtual screening was per-
formed by LeDock, which had 80.8% accuracy for the
best poses (Wang et al. 2016). Following which, the
protein was prepared and hydrogen atoms were added
to the crystal structure by LePro tool. LeDock program
is based on a combination of simulated annealing and
evolutionary optimization of the ligand pose and its
rotatable bonds (http://lephar.com). To investigate the
docking poses, LeDock score function was used. It uses
a physics/knowledge hybrid scoring scheme. Out of the
top 100 docked poses, the hit compounds were chosen
by visual inspection. In order to validate, the chosen hit
compounds, a different docking program with the
Lamarckian Genetic Algorithm (LGA), AutoDock
Vina was used (Trott and Olson 2010).
Molecular dynamics simulation
Among the various sampling algorithm for molecular
modeling, molecular dynamics is widely used in many
fields. In this algorithm, both ligand and protein are
more efficiently flexible than others. In molecular
dynamics simulations, almost all of the conditions
Biogerontology

Fig. 1 Validation of AMPK model: a Ramachandran plot, b verification of 3D analysis

Table 1 LeDock score and drug categorization of hit compounds with respect to c and ab binding sites
AMPK Generic name Ledock score Vina affinity Drug category
(binding (kcal/mol) (kcal/mol)
site)

c Cangrelor - 14.1 - 9.3 Receptor antagonist which inhibits ADP platelet aggregation
Nacitentan - 10.9 - 9.3 Blocks the stimulation of vasculature hypertrophy, inflammation,
fibrosis, proliferation, and vasoconstriction
Levoleucovorin - 10.2 - 10.9 Active form of folinic acid
Glisoxepide - 10.2 - 10 Increase secretion of (pro)insulin
ab Ceftolozane - 12 - 7.8 Antibiotic
Ceftriaxone - 8.84 - 8.1 Treatment of infections
Levomefolic - 8.35 - 7.3 Primary biologically active form of folate
acid

such as pressure, temperature, dielectric constant are
present (Ganesan et al. 2017).All the MD simulations
of this study, took advantage of the GROMACS 5.1.2
software (Pronk et al. 2013), using Amber99SB force
field (Hornak et al. 2006). Ligand topology files were
generated by ACPYPE tool. The protein–ligand
complexes were placed in the center of a triclinic
box setting a minimal distance of 1 nm from all edges,
which were solvated with TIP3P water molecules
(Jorgensen et al. 1983). Before minimization, the
systems were neutralized by adding Naþ and Cl at
the physiological concentration of 150 mM NaCl. The
steepest descent algorithm was used for energy
minimization of all atoms, in which the minimization
has stopped when the Fmax was smaller than
10 kJ/mol. Moreover, for equilibration of systems,
100 ps NVT and 300 ps NPT were performed while
the atomic position of protein and ligands were
restrained. After all, 10 ns the simulations were done
under constant temperature and pressure (300 K and
1 atm respectively). The trajectories information was
analyzed using GROMACS utilities and VMD.
Energy analysis via MM/GBSA approach
The molecular mechanics/generalized born surface
area (MM/GBSA) calculations were performed by
g_mmpbsa package (Kumari et al. 2014). In this
method, the binding free energy of hit compounds was
obtained from the following equation:


DGbind ¼ Gcomplex  Gligand þ Greceptor
In the present study, a total of 100 snapshots were
evenly taken in a time step of 200 frames. The binding
free energy, the Van der Waals (VDW) energy,
electrostatic energy, polar solvation energy and

123
 Biogerontology

eventually the solvent-accessible surface area (SASA)

energy were calculated in present study as described
previously (Kumari et al. 2014).
Additionally, to analyze the interactions between
residues and hit compounds, the MM/GBSA binding
energy decomposition analysis was performed (Tsui
and Case 2000). The binding interactions were
calculated by employing for the following compo-
nents: the van der Waals contribution (DGvdw), the
polar part of desolvation (DGGB), the electrostatic
contribution (DGele) and the non-polar part of desol-
vation (DGSA). Thus:

Fig. 2 RMSF values of the residues in the presence of hit

DG ¼ DGvdW þ DGele þ DGsol
compound during 10 ns of MD simulation at c binding site ¼ DGvdW þ DGele þ DGGB þ DGSA:

Fig. 3 Number of inter-

molecular hydrogen bonds
between hit compound and
the residues at c active site
of AMPK. a C2, b AMP,
c Cangrelor, d Nacitentan,
e Levoleucovorin and
f Glisoxepide

123
Biogerontology
Fig. 4 Number of inter-
molecular hydrogen bonds
between hit compounds and
the residues at ab active site
of AMPK. a A769662,
b Ceftolozane,
c Ceftriaxone and
d Levomefolic acid
Results and discussion
Model evaluation
After several rounds of energy minimization, the
refined model was evaluated by using Ramachandran
plot. According to our results, 92.7% residues were in
favored regions, 6.2% in the allowed regions and 1.1%
residues in the restricted regions. In general, well-built
stereochemical models implied a score close to 100%.
Therefore, the Ramachandran plot suggests that the
model was of great quality. For further confirmation,
the protein structure was also validated by Verify3D
program. The Verify3D analysis is compatible for the
model via the amino acid sequence. In current study,
all the amino acids showed an average 3D_ID score
above zero except 16 first amino acids of c subunit
(Fig. 1).
Virtual screening
Molecular docking can be used to predict the ligands
behavior in the binding site of protein. It consists of
two basic steps: sampling the ligand pose (position
and orientation) and ranking the binding affinity by
scoring function (Meng et al. 2011). Therefore,
Among 100 top scores, 3 ligands for ab binding site
and 4 ligands for c binding site were chosen by
checking the types and number of bonds in pocket.
According to LeDock score, the reference ligands, C2,
AMP and A769662 scored - 6.46, - 8.79 and
- 7.32 kcal/mol, respectively. However, the hit com-
pounds scored higher than the above-stated values for
the reference ligands (Table 1). Additionally, for
further validation of LeDock results, we redocked
and rescored top hit compounds by Vina. The
reference ligands (C2, AMP and A769662 scored
- 7.8, - 8.5 and - 7.5 kcal/mol, respectively) and
hit compounds scored approximately as same as
LeDock score which could indicate the accuracy of
our data.
Stability of MD simulations
Root mean square deviation (RMSD) is one of the
most important values to investigate the stability of a
system. The RMSD values of c and ab sites are shown
in Fig. S3. The backbone RMSDs for c site reached the
equilibrium state after about 2 ns simulation. Also, the
ab site RMSD plot illustrated the stability for all
systems after about 5 ns. The lower value of RMSD
shows more stable system. Therefore, all the presently
123
 Biogerontology

Table 2 The p-effects, hydrophobic (alkyl) bonds and the number of hydrogen bonds between the hit compounds and at c and ab
binding sites as well
Binding site Compounds p-effects Hydrophobic (Alkyl) H-bonds

(c) C2 – – –
AMP R304, V281, 7
V302, I245 (Pi-alkyl)
Cangrelor H303, H156 (Pi–Pi, T-shaped) R75-LIGC25 5
R75 (Pi-cation) LLE93-LIGC28
F249 (Pi-alkyl)
Nacitentan R75 (Pi-cation) AlA232-LIGC31 4
H303, F278 (Pi–Pi stack) I245-LIGC31
H156, F249 (Pi–Pi T-shaped)
Levolcovorin I155 (Pi-alkyl) 7
M90 (Pi-alkyl)
A76 (Pi-alkyl)
Glisoxepide H156 (Pi–Pi T-shaped) I155-LIGC5 2
K248 (Pi-cation)
R304, ALA232 (Pi-alkyl)
(ab) A769662 F103 (Pi–Pi, T-shaped) V81-LIG565 1
I59-LIG565
V24- LIG565
Ceftolozane I59 (Pi-alkyl) R83-LIGC3 2
V113-LIGC3
V24-LIGC26
L31-LIGC26
Ceftriaxone D33 (Pi-anion) 3
H109 (Pi–Pi stacked)
I59 (Pi-alkyl)
Levomefolic acid H109 (Pi–Pi T-shaped) 1

studied compounds, except Cangrelor showed the
most reliable stabilities in c binding site.
The RMSF value of the residues with 10 ns MD
simulation of each system was analyzed by ‘‘gmx
rmsf’’ tool. In this way, the flexibility of each residue
in the presence of hit compounds was compared. In
general, Fig. S4 depicts that the 4 activator-protein
complexes posed similar distribution at residues
75–325 of the active site, which fluctuated less than
0.4 nm. Although C2 showed the most significant
fluctuation at KD domain (at residues 21–291 of a
subunit), its RMSF difference with other ligands was
less than 0.1 nm. Besides, the AID part (residues
300–349) of a subunit illustrates this pattern as well.
Even though AMPK (with no ligands) and Glisox-
epide showed the most significant fluctuation at the
AID residues, they had an RMSF difference less than
0.1 nm. So, it seemed that the hit compounds and
reference ligands complexes had approximately the
same fluctuation for each residue in KD and AID
(Fig. 2).
The RMSF value for ab binding site is presented in
Fig. S6 in supporting information, which shows that
the C-terminals of c and b subunits are fluctuated at a
higher rate. At the active site (residues 21–60 and
80–115 for a and b subunits, respectively), the hit
compounds fluctuated approximately at the same rate
as AMPK. Among them Ceftriaxone had the most
similar RMSF value to that of AMPK.
Protein–Ligand interactions
The hydrogen bond between hit compounds and
protein has been illustrated in Figs. 3 and 4. The c

123
Biogerontology

Table 3 VDW, electrostatic, polar solvation, SASA and binding free energies of the hit compounds with respect to c and ab binding
sites
AMPK (Binding Generic name van der Waal Electrostatic Polar solvation SASA energy Binding energy
sites) energy energy energy

(c) C2 - 90.20 ± 2 - 678.8 ± 7.7 849.6 ± 4.9 - 12.0 ± 0.0 68.7 ± 6

AMP - 167.2 ± 1.5 - 117.6 ± 1.3 203.7 ± 1.7 - 16.2 ± 0.06 - 97.2 ± 1.5
Cangrelor - 272.3 ± 22.4 - 541.3 ± 76.9 862.8 ± 81.2 - 30.0 ± 0.9 19 ± 49.1
Nacitentan - 236.8 ± 1.8 - 40.80 ± 1.8 157.8 ± 4.2 - 22.1 ± 0.1 - 142 ± 2
Levoleucovorin - 195.3 ± 2.3 - 668.3 ± 8.4 918.8 ± 7.4 - 22.1 ± 0.1 32.7 ± 5.6
Glisoxepide - 198.8 ± 2.4 - 120.9 ± 2.2 197.4 ± 5.1 - 22.4 ± 0.1 - 144.8 ± 2.6
(ab) A769662 - 144.7 ± 1.8 - 44.50 ± 6.1 31.60 ± 3.2 - 17.2 ± 0.1 - 174.7 ± 4.5
Ceftolozane - 210.7 ± 2.4 - 328.0 ± 11.5 485.9 ± 12 - 23.9 ± 0.2 - 76.8 ± 4.3
Ceftriaxone - 196.2 ± 2.6 - 191.0 ± 6.3 398.0 ± 7.6 - 20.2 ± 0.2 - 9.4 ± 3.9
Levomefolic - 99.00 ± 3.9 - 346.3 ± 19.3 671.9 ± 22.8 - 16.1 ± 0.4 210.7 ± 10.2
acid
All values were calculated as kJ/mol

site analysis has revealed that Cangrelor and Levoleu-
covorin comprised about 10–11 hydrogen bonds the
same as C2 (reference ligand), while Nacitentan and
Glisoxepide showed less than 6 H-bonds. Addition-
ally, more information about hit compound-protein
interactions is presented in Table 2 and Fig. S6. We
have found that Arg304, Arg75, Arg157 and Ser321
were the key residues for H-bonds in C2complex.
Also, each of these residues could play a role in
binding of hit compounds via H-bond, p-effects or
hydrophobic interactions. It is obvious that Cangrelor
and Levoleucovorin have the most similarity in
residues with the reference compound.
Figure 4 provides information about the number of
H-bonds at ab binding site. Our results have shown
that Ceftolozane and Levomefolic acid had twice
hydrogen bonds than A769662; and Ceftriaxone
comprised approximately same number of hydrogen
bonds as A769662. Furthermore, Table 2 and Fig. S7
illustrates that there was a rare similarity in the
residues between hit compounds and A769662.
MM/PBSA and MM/GBSA data
In spite of continuous effort for improving scoring
functions, their precision for ranking the binding poses
and calculating binding free energies is still unsatis-
factory. Therefore, we recalculate the binding energy
of top hit compounds by MM/GBSA, which is more
rigorous than other available scoring function for
molecular docking (Hou et al. 2011).To further
explore the interactions between AMPK and the
activators, the binding free energy decomposition
was also estimated. The binding free energies of the hit
compounds in c docking site was encouraging because
all of them had shown higher affinity as compared to
the reference (C2) compound. Table 3 illustrates that
although the polar solvation energy was positively
unfavorable for interactions, their SASA energy
promoted the ligand binding capacity. Additionally,
it could be figured out that Van der Waals contribution
of hit compounds were much higher than C2, while
their electrostatic contributions were weaker.
MM/PBSA data revealed that binding free energies
of the hit compounds for the ab binding site were not
impressive compared to A769662 (Table 3). Among
them, Ceftolozane had the lowest binding free energy
(- 76.8 ± 4.3 kJ/mol) which is eventually a large
amount compared to the reference compound. While,
only Ceftolozane and Ceftriaxone had shown higher
Van der Waals energy than A769662, all of them
exhibited lower value for electrostatic energy.
The binding free energy decomposition has revealed
the atomistic details of each residue in protein-activator
interactions. As shown in Fig. 5, the common residues
between C2 and hit compounds that contributed in the
interactions were Arg75, Arg157, Lys175, Lys248,
Arg304 (where Arg75, Arg157 and Arg304 were also
mentioned in the Protein–Ligand interactions section).
It has been clarified from the present results that

123
 Biogerontology

Fig. 5 Decomposition of binding free energies on per-residue of c active site. a C2, b AMP, c Cangrelor, d Nacitentan,
e Levoleucovorin and f Glisoxepide

Levoleucovorin and Cangrelor interact with more Figure 6 provides information about free energy
similar residues compared to C2 than the other ligands. decomposition of ab binding site. According to the
Furthermore, the ligand-residue interaction energy was data, all the 3 hit compounds contributed to interaction
less for Nacitentan and Glisoxepide. with Gly41 and Val57 at subunit a and Arg108 at

123
Biogerontology
Fig. 6 Decomposition of binding free energies on per-residue of ab active site of AMPK. a A769662, b Ceftolozane, c Ceftriaxone and
d levomefolic acid
Fig. 7 Average distance between the center of mass of KD and
AID at a subunit during simulation time
subunit b. Arg108 is one of the most critical residues at
ab binding site (Xiao et al. 2013). Among these 3 hit
compounds, Ceftolozane has shown the most similar
common residues with the reference compound.
KD-AID distance analysis
Under energy stress, AMP binds to c subunit, which
provides a more active conformation by pulling AID
away from KD domain. In this way, KD’s structure
switches to closed state, Hence, Thr172 is protected
from dephosphorylation (Xin et al. 2013). In order to
monitor the distance between the center of mass of KD
and AID during simulation, we used distance analyzes
from Gromacs utilities. As we expected, when AMP
was bound to the binding site of the c subunit, the
distance between these two centers was at its maxi-
mum and in the presence of C2 in its binding site, the
distance was a bit lower than AMP (Fig. 7). All of the
123

hit compounds have indicated approximately the same
distance as active AMPK without any activators.
Conclusions
Regulation of the energy metabolism has crucial effect
on aging process, which is done by AMPK. Emerging
evidence indicates that the AMPK signaling is
decreased by aging. Therefore, in our study, some of
the FDA-approved drugs were suggested to increase its
potential activity. In this study, C2 and A769662 were
used as reference ligands at c and ab binding sites,
respectively. We have found 4 potential activators for c
and 3 activators for ab sites through virtual screening.
Considering RMSD and RMSF values, the MD simu-
lation stability was confirmed as RMSD value reached
plateau state after short range of fluctuation. In c
binding site, low binding free energy of hit compounds
was seen as compared to C2, indicating that these drugs
had strong affinity to this site. On the other hand, the
predicted ligands had higher binding free energy at ab
binding site, which indicates their low affinity to the
pocket. According to average distance between the
center of mass of KD and AID at a subunit and RMSF
results, the hit-compounds at c binding site showed the
same manner as reference ligands or active AMPK. So,
we have predicted that these hit compounds are putative
to be AMPK activators. Our study was based on drug
repurposing, which has shown that all of the presently
analyzed hit compounds are safe and could be evaluated
in phase 2 clinical trials for shortening aging process.
Acknowledgement The authors are grateful to Ms. Shady
Sheybani for her help to prepare some of the figures. This
investigation was supported by a grant number 96-1206 from
Golestan University, Gorgan, Iran.
References
Bordoli L, Kiefer F, Arnold K, Benkert P, Battey J, Schwede T
(2009) Protein structure homology modeling using
SWISS-MODEL workspace. Nat Protoc 4:1
Calvert S, Tacutu R, Sharifi S, Teixeira R, Ghosh P, Magalhães
JP (2016) A network pharmacology approach reveals new
candidate caloric restriction mimetics in C. elegans. Aging
Cell 15:256–266
Cameron KO, Kurumbail RG (2016) Recent progress in the
identification of adenosine monophosphate-activated pro-
tein kinase (AMPK) activators. Bioorganic Med Chem Lett
26:5139–5148
123
Biogerontology
Cantó C et al (2009) AMPK regulates energy expenditure by
modulating NAD ? metabolism and SIRT1 activity. Nat-
ure 458:1056
Cavalla D (2013) Predictive methods in drug repurposing: gold
mine or just a bigger haystack? Drug Discov Today
18:523–532
Chen L et al (2013) Conserved regulatory elements in AMPK.
Nature 498:E8
Crute BE, Seefeld K, Gamble J, Kemp BE, Witters LA (1998)
Functional domains of the a1 catalytic subunit of the AMP-
activated protein kinase. J Biol Chem 273:35347–35354
Dalle Pezze P et al (2016) A systems study reveals concurrent
activation of AMPK and mTOR by amino acids. Nat
Commun 7:13254
Ganesan A, Coote ML, Barakat K (2017) Molecular dynamics-
driven drug discovery: leaping forward with confidence.
Drug Discov Today 22:249–269
Gupta SC, Sung B, Prasad S, Webb LJ, Aggarwal BB (2013)
Cancer drug discovery by repurposing: teaching new tricks
to old dogs. Trends Pharmacol Sci 34:508–517
Hall JA, Dominy JE, Lee Y, Puigserver P (2013) The sirtuin
family’s role in aging and age-associated pathologies.
J Clin Investig 123:973–979. https://doi.org/10.1172/
JCI64094
Hardie DG, Hawley SA (2001) AMP-activated protein kinase:
the energy charge hypothesis revisited. BioEssays
23:1112–1119
Hornak V, Abel R, Okur A, Strockbine B, Roitberg A, Sim-
merling C (2006) Comparison of multiple Amber force
fields and development of improved protein backbone
parameters. Proteins 65:712–725
Hou T, Wang J, Li Y, Wang W (2011) Assessing the perfor-
mance of the molecular mechanics/Poisson Boltzmann
surface area and molecular mechanics/generalized born
surface area methods. II. The accuracy of ranking poses
generated from docking. J Comput Chem 32:866–877
Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein
ML (1983) Comparison of simple potential functions for
simulating liquid water. J Chem Phys 79:926–935
Kumari R, Kumar R, Consortium OSDD, Lynn A (2014)
g_mmpbsa A GROMACS tool for high-throughput MM-
PBSA calculations. J Chem Info Model 54:1951–1962
Li X et al (2015) Structural basis of AMPK regulation by ade-
nine nucleotides and glycogen. Cell Res 25:50
Ma D-L, Chan DS-H, Leung C-H (2013) Drug repositioning by
structure-based virtual screening. Chem Soc Rev
42:2130–2141
Maswood N et al (2004) Caloric restriction increases neu-
rotrophic factor levels and attenuates neurochemical and
behavioral deficits in a primate model of Parkinson’s dis-
ease. Proc Natl Acad Sci 101:18171–18176
Meng X-Y, Zhang H-X, Mezei M, Cui M (2011) Molecular
docking: a powerful approach for structure-based drug
discovery. Curr Comput-Aided Drug Des 7:146–157
Pronk S et al (2013) GROMACS 4.5: a high-throughput and
highly parallel open source molecular simulation toolkit.
Bioinformatics 29:845–854
Rana S, Blowers EC, Natarajan A (2014) Small molecule ade-
nosine 50 -monophosphate activated protein kinase
(AMPK) modulators and human diseases. J Med Chem
58:2–29
Biogerontology
Salminen A, Kaarniranta K (2012) AMP-activated protein
kinase (AMPK) controls the aging process via an inte-
grated signaling network. Ageing Res Rev 11:230–241
Sanders MJ, Ali ZS, Hegarty BD, Heath R, Snowden MA,
Carling D (2007) Defining the mechanism of activation of
AMP-activated protein kinase by the small molecule
A-769662, a member of the thienopyridone family. J Biol
Chem 282:32539–32548
Steinberg GR, Kemp BE (2009) AMPK in health and disease.
Physiol Rev 89:1025–1078
Townley R, Shapiro L (2007) Crystal structures of the adenylate
sensor from fission yeast AMP-activated protein kinase.
Science 315:1726–1729
Trott O, Olson AJ (2010) AutoDock Vina: improving the speed
and accuracy of docking with a new scoring function,
efficient optimization, and multithreading. J Comput Chem
31:455–461. https://doi.org/10.1002/jcc.21334
Tsui V, Case DA (2000) Theory and applications of the gener-
alized born solvation model in macromolecular simula-
tions. Biopolymers 56:275–291
Wang Z et al (2016) Comprehensive evaluation of ten docking
programs on a diverse set of protein–ligand complexes: the
prediction accuracy of sampling power and scoring power.
Phys Chem Chem Phys 18:12964–12975
Xiao B et al (2013) Structural basis of AMPK regulation by
small molecule activators. Nat Commun 4:3017
Xin F-J, Wang J, Zhao R-Q, Wang Z-X, Wu J-W (2013)
Coordinated regulation of AMPK activity by multiple
elements in the a-subunit. Cell Res 23:1237
123