Development of potent adenosine monophosphate activated protein kinase (AMPK) activators

Article abstract of DOI:10.1002/cmdc.201500371

Previously, we reported the identification of a thiazolidinedione-based adenosine monophosphate activated protein kinase (AMPK) activator, compound 1 (N-[4-({3-[(1-methylcyclohexyl)methyl]-2,4-dioxothiazolidin-5-ylidene}methyl)phenyl]-4-nitro-3-(trifluoromethyl)benzenesulfonamide), which provided a proof of concept to delineate the intricate role of AMPK in regulating oncogenic signaling pathways associated with cell proliferation and epithelial-mesenchymal transition (EMT) in cancer cells. In this study, we used 1 as a scaffold to conduct lead optimization, which generated a series of derivatives. Analysis of the antiproliferative and AMPK-activating activities of individual derivatives revealed a distinct structure-activity relationship and identified 59 (N-(3-nitrophenyl)-N′-{4-[(3-{[3,5-bis(trifluoromethyl)phenyl]methyl}-2,4-dioxothiazolidin-5-ylidene)methyl]phenyl}urea) as the optimal agent. Relative to 1, compound 59 exhibits multifold higher potency in upregulating AMPK phosphorylation in various cell lines irrespective of their liver kinase B1 (LKB1) functional status, accompanied by parallel changes in the phosphorylation/expression levels of p70S6K, Akt, Foxo3a, and EMT-associated markers. Consistent with its predicted activity against tumors with activated Akt status, orally administered 59 was efficacious in suppressing the growth of phosphatase and tensin homologue (PTEN)-null PC-3 xenograft tumors in nude mice. Together, these findings suggest that 59 has clinical value in therapeutic strategies for PTEN-negative cancer and warrants continued investigation in this regard. Small-molecule AMPK activators: Metabolism reprogramming is a key feature of all cancer cells. Adenosine monophosphate activated protein kinase (AMPK) is an energy sensor; activation of this enzyme can restore normal cell metabolism. Starting with the previously reported AMPK activator OSU-53 (1), an optimization study was conducted, guided by docking studies. Compound 59 was identified as the optimal agent, with greater potency than OSU-53.

Full text of DOI:10.1002/cmdc.201500371

DOI: 10.1002/cmdc.201500371
Full Papers
Development of Potent Adenosine Monophosphate
Activated Protein Kinase (AMPK) Activators
Eman M. E. Dokla,[a, b] Chun-Sheng Fang,[a] Po-Ting Lai,[a] Samuel K. Kulp,[a]
Rabah A. T. Serya,[b] Nasser S. M. Ismail,[b] Khaled A. M. Abouzid,*[b] and Ching-Shih Chen*[a, c]
Previously, we reported the identification of a thiazolidine-
dione-based adenosine monophosphate activated protein
kinase (AMPK) activator, compound 1 (N-[4-({3-[(1-methylcyclo-
hexyl)methyl]-2,4-dioxothiazolidin-5-ylidene}methyl)phenyl]-4-
nitro-3-(trifluoromethyl)benzenesulfonamide), which provided
a proof of concept to delineate the intricate role of AMPK in
regulating oncogenic signaling pathways associated with cell
proliferation and epithelial–mesenchymal transition (EMT) in
cancer cells. In this study, we used 1 as a scaffold to conduct
lead optimization, which generated a series of derivatives.
Analysis of the antiproliferative and AMPK-activating activities
of individual derivatives revealed a distinct structure–activity
relationship and identified 59 (N-(3-nitrophenyl)-N’-{4-[(3-{[3,5-
bis(trifluoromethyl)phenyl]methyl}-2,4-dioxothiazolidin-5-yli-
dene)methyl]phenyl}urea) as the optimal agent. Relative to 1,
compound 59 exhibits multifold higher potency in upregulat-
ing AMPK phosphorylation in various cell lines irrespective of
their liver kinase B1 (LKB1) functional status, accompanied by
parallel changes in the phosphorylation/expression levels of
p70S6K, Akt, Foxo3a, and EMT-associated markers. Consistent
with its predicted activity against tumors with activated Akt
status, orally administered 59 was efficacious in suppressing
the growth of phosphatase and tensin homologue (PTEN)-null
PC-3 xenograft tumors in nude mice. Together, these findings
suggest that 59 has clinical value in therapeutic strategies for
PTEN-negative cancer and warrants continued investigation in
this regard.
Introduction
Adenosine monophosphate activated protein kinase (AMPK) is
a key player in maintaining energy homeostasis in response to
metabolic stress.[1–4] Beyond diabetes and metabolic syn-
drome,[5] there is a growing interest in the therapeutic exploi-
tation of the AMPK pathway in cancer treatment in light of its
unique ability to regulate cancer cell proliferation through the
reprogramming of cell metabolism.[6–8] This premise was corro-
borated by the ability of the antidiabetic drug metformin,
which acts by disrupting mitochondrial respiration, to activate
AMPK, thereby lowering the risk of cancer and improving out-
comes of many types of cancer cases.[9,10] In addition to cell
metabolism, recent evidence suggests that AMPK plays an in-
tricate role in regulating cancer cell proliferation, survival, and
invasive phenotype by targeting multiple oncogenic signaling
pathways, including those mediated by Akt, mTOR, and
MDM2.[11–13] Consequently, the past few years have witnessed
the development of a number of novel AMPK activators as dis-
cussed in several reviews,[14–16] including A-769662,[17] PT-1,[18]
OSU-53 (1; Figure 1),[19] (E)-3-((3-[(4-chlorophenyl)(phenyl)-
methylene]-2-oxoindolin-1-yl)methyl)benzoic
acid,[20]
and
991.[14,21] Mechanistically, these direct activators circumvent the
dependence on liver kinase B1 (LKB1), an upstream kinase of
AMPK under conditions of energy stress, which is often inacti-
vated in cancer cells through mutations in the course of tu-
morigenesis.[13]
Previously, based on the screening of a thiazolidinedione-
based focused compound library, we identified 1 as a direct
AMPK activator in LKB1-deficient MDA-MB-231 cells.[19] Com-
pound 1 served as a proof-of-concept compound to delineate
the intricate mechanisms by which AMPK regulates cell prolif-
eration and epithelial–mesenchymal transition (EMT) in cancer
cells,[11,12] among which the ability of 1 to target the Akt–
MDM2–Foxo3a signaling pathway is especially noteworthy.
Specifically, pharmacological activation of AMPK leads to pro-
tein phosphatase 2A-facilitated Akt dephosphorylation. This
Akt inactivation results in decreased phosphorylation of the E3
ligase murine double minute 2 (MDM2) and its substrate
Foxo3a, which facilitates the cytoplasmic sequestration and nu-
clear translocation of MDM2 and Foxo3a, respectively. The
physical separation of Foxo3a from MDM2 increases Foxo3a
protein stability and accumulation.
[a] E. M. E. Dokla, C.-S. Fang, P.-T. Lai, Dr. S. K. Kulp, Prof. C.-S. Chen
Division of Medicinal Chemistry & Pharmacognosy
College of Pharmacy, The Ohio State University
Room 336, Parks Hall, 500 West 12th Ave., Columbus, OH 43210 (USA)
E-mail: chen.844@osu.edu
[b] E. M. E. Dokla, Dr. R. A. T. Serya, Dr. N. S. M. Ismail, Prof. K. A. M. Abouzid
Pharmaceutical Chemistry Department, Faculty of Pharmacy
Ain Shams University, POB 11566, Abbassia, Cairo (Egypt)
E-mail: Khaled.abouzid@pharma.asu.edu.eg
[c] Prof. C.-S. Chen
Institute of Biological Chemistry, Academia Sinica
128 Academia Road Sec. 2, Nankang, Taipei (Taiwan)
These findings underscore the translational potential of tar-
geting AMPK activation as a therapeutic strategy for metastatic
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500371.
ChemMedChem 2015, 10, 1915 – 1923
1915
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 1. OSU-53 (1) as a scaffold to develop potent AMPK activators. A) Docking of 1 into the allosteric site formed at the interface between the kinase
domain and CBM of AMPK (PDB ID: 4CFE). B) Schematic diagram depicting the strategies for lead optimization of compound 1. C) General synthetic
procedures for compounds 2–87. Reagents and conditions: a) ArCHO, toluene, AcOH, piperidine, 808C, overnight; b) ArCH2Br, MeCN, K2CO3, RT, 2 days;
c) SnCl2·2H2O, EtOAc, reflux, 4–8 h; d) ArSO2Cl, pyridine, RT, 3 days; e) derivatives 2 or 3, BH3·DMS, THF, 708C, overnight; f) NaN3, NH4Cl, DMF, 1258C, 24 h;
g) morpholine carbonyl chloride, pyridine, RT, overnight; h) ArCOCl, pyridine, RT, overnight; i) phosgene, toluene, 1008C, 2 h; j) ArNH2, toluene, 1008C, 12 h;
k) ArNCS, DMF, RT, 12 h; l) 10% Pd/C, MeOH, 410 kPa, 16 h.
cancer treatment, thereby providing a rationale to use 1 as
a scaffold to conduct lead optimization. This effort has netted
a series of potent AMPK activators, among which 59 represent-
ed the optimal agent. Relative to 1, compound 59 was found
to exhibit multifold higher potency in facilitating AMPK activa-
tion, accompanied by parallel changes in Akt phosphorylation
and Foxo3a induction in MDA-MB-231 and other cancer cell
lines. Equally important, 59 is orally active in suppressing xeno-
graft tumor growth in nude mice without incurring acute tox-
icity.
a. Hydrophobic interactions between the cyclohexane/benzy-
lidine moieties and the large hydrophobic pocket formed
by residues Val11, Leu18, Ile46, and Leu47 (kinase domain)
and Val81, Val113, and Val118 (CBM domain);
b. Cation–p interactions between the thiazolidinedione ring
and Lys31 (kinase domain), as well as between the benzyli-
dine ring and Arg83 (CBM);
c. Hydrogen bonding between the -SO2NH- moiety and Lys29
and Asp88, and between the -NO2 substituent and Lys29.
Based on this docking analysis, we used the following strat-
egies to conduct the structural modification of 1 (Figure 1B).
General procedures for the synthesis of these compounds are
depicted in Figure 1C. The initial screening was based on the
activities of individual compounds, each at 5 mm, in suppress-
ing the viability of LKB1-deficient MDA-MB-231 cells by using
MTT assays.
Results
Structural optimization of 1
We carried out modeling analysis by docking 1 into the recent-
ly released X-ray crystal structure of human AMPK (PDB ID:
4CFE).[22] This analysis revealed that 1 binds key amino acids in
the allosteric site formed at the interface between the kinase
domain and carbohydrate binding module (CBM) through the
following forces, as depicted in Figure 1A.
Step 1 (Series A): The methylcyclohexyl group of 1 was re-
placed with various substituted phenyl rings to exploit the hy-
drophobic pocket. Among 35 derivatives (2–36) examined
(Figure 2), 16 [3,5-di(trifluoromethyl)phenyl] and 26 [4-(methyl-
thio)phenyl] exhibited moderately higher antiproliferative po-
tencies than compound 1 (Figure 3A).
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1916
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Structures of compounds 2–86 in Series A–G.
Step 2 (Series B): The 3-trifluoromethyl-4-nitrophenyl ring of
16 and 26 were replaced with 4-chloro-3-nitro- or 4-chloro-2-
nitrophenyl moieties to generate compounds 37–39 (Figure 2).
As shown, substitution of the 3-trifluoromethyl-4-nitrophenyl
moiety of 16 and 26 with 4-chloro-3-nitro- substantially in-
creased the antiproliferative activities of the resulting com-
pounds (37 and 39; Figure 3A). Relative to 37, the antiprolifer-
ative potency of 38 was attenuated as a result of the different
configuration of the nitro group on the phenyl ring, indicating
its distinct role in interacting with the binding motif. Moreover,
40, a regioisomer of 16, was synthesized and assessed for its in
vitro antiproliferative activity, which was abolished by the
change in the regiochemical configuration of the terminal phe-
nylsulfonamide substructure.
68–72, 76, and 77) in conjunction with varying the terminal
substituted phenyl moiety (Figure 2). This strategy generated
an array of derivatives with the ability to suppress cell viability
by ꢀ75% at 5 mm, which included 43, 46–48, 50–54, 59, 62,
68, 69, 71, 74, and 75 (Figure 3A). A unique feature common
to many of these compounds is the 3-nitrophenyl moiety, with
the exception of 71, in which the 3-nitro substituent was re-
placed by a sulfonamide function.
Step 4 (Series F and G): Continued structural modifications of
the two optimal derivatives 51 and 59 generated Series F (79–
84) and G (85, 86), respectively (Figure 2). Substitution on the
benzylidene moiety of 51 with various functionalities, including
chloro, trifluoromethyl, and methoxy, at the C2’ and, to a great
extent, C3’ positions diminished antiproliferative efficacy rela-
tive to the parent compound (Figure 3A), indicating the dis-
tinct stereoelectronic effect of these substituents. In particular,
the addition of a chlorine atom to the C2’ position completely
abolished activity. Replacement of the phenyl ring of the ben-
zylidene moiety of 59 with a pyridine and/or replacing the ter-
Step 3 (Series C–E): As bioisosteric replacement represents
a common strategy for lead optimization,[23,24] we tested the
effects on antiproliferative efficacy of replacing the sulfona-
mide linker with amide (Series C, 41 and 42; Series D, 43–50),
urea (Series E, 51–67, 73–75, and 78), and thiourea (Series E,
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1917
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 3. Identification of optimal AMPK activators. A) Effects of 1–86, each at 5 mm, relative to DMSO control on the viability of MDA-MB-231 cells after 72 h
of incubation by MTT assays; data are the meanÆSD (n=6); *compounds with the ability to suppress ꢀ75% cell viability. Western blot analysis of B) dose-
dependent effects of metformin (left) and AICAR (right), and C) effects of selected derivatives, each at 2.5 mm, on AMPK phosphorylation in MDA-MB-231 cells
after 48 h of treatment.
minal 3,5-di(trifluoromethyl)phenyl with 4-trifluoromethylphen-
yl (85 and 86, respectively) did not show improved potency
relative to 59.
its close derivative 51, and 71 were selected for further evalua-
tions vis-à-vis compound 1.
MTT assays revealed a positive correlation between the anti-
proliferative efficacies of these compounds (IC50 values at 72 h:
59, 1.8 mm; 51, 3.5 mm; 71, 4 mm; 1, 6 mm; Figure 4A) and their
aforementioned abilities to upregulate AMPK phosphorylation
(Figure 3C). This observation was confirmed by the differences
in their dose-dependent effects on AMPK phosphorylation,
which was accompanied by parallel increases in the phosphor-
ylation of its downstream substrate acetyl-CoA carboxylase
(ACC; Figure 4B). Whereas 59 was effective in inducing AMPK
phosphorylation at ꢁ1 mm, other compounds examined would
require substantially higher concentrations (1, 7.5 mm; 51,
2.5 mm; 71, 2.5–5 mm) to achieve the same level of increase in
AMPK phosphorylation.
Identification of optimal AMPK activators by cell-based
assays
Of the 85 derivatives examined, 21 compounds were found to
be effective in suppressing the viability of MDA-MB-231 cells
by ꢀ75% at 5 mm after 72 h of exposure, including 37, 39, 43,
46–48, 50–54, 59, 62, 68, 69, 71, 74, 75, 80, 85, and 86 (Fig-
ure 3A, indicated by ). We further assessed the effects of
*
these compounds vis-à-vis 1 on the phosphorylation status of
AMPK in MDA-MB-231 cells by western blot analysis. Due to
a lack of LKB1, MDA-MB-231 cells were resistant to the effect
of metformin-induced metabolic stress on AMPK activation
(Figure 3B, left), reminiscent to that previously reported.[25] In
addition, these cells were relatively insensitive to 5-aminoimi-
dazole-4-carboxamide ribonucleotide (AICAR)-facilitated AMPK
phosphorylation (right). In contrast, these derivatives exhibited
differential effects on AMPK phosphorylation, with 59 showing
the highest potency, among those with high activity, including
51, 53, 69, 71, 74, and 86 (Figure 3C). Based on these data, 59,
Mechanistic interrogation of the effects of 51, 59, and 71 on
AMPK signaling
It is well recognized that AMPK acts as an energy sensor by
mediating the phosphorylating inactivation of ACC and other
key lipogenic enzymes in response to cellular AMP or ADP ele-
vation.[1] Equally important, we previously used 1 as a proof-of-
concept compound to demonstrate the suppressive effect of
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1918
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
(PTEN)-null PC-3 prostate cancer
cells. MTT assays indicate that
the antiproliferative efficacies of
1 and 59 in MDA-MB-468 and
PC-3 cells are similar to those in
MDA-MB-231 cells with the re-
spective IC50 values as follows: 1,
5.8 and 6.2 mm; 59, 1.5 and
1.6 mm (data not shown). More
important, 1 and 59 affected
AMPK phosphorylation and vari-
ous AMPK downstream markers
in these two cell lines in
a manner similar to that in MDA-
MB-231 cells (Figure 5A,B), indi-
cating these effects are not
a cell-line-specific phenomenon.
Moreover, consistent with our
previous results demonstrating
the role of AMPK in suppressing
EMT via Foxo3a upregulation,[12]
exposure of MDA-MB-231 and
PC-3 cells to 59 dose-depend-
ently increased Foxo3a expres-
sion, leading to the reversal of
their mesenchymal phenotype,
as manifested by increases in
the epithelial markers claudin-
1 in MDA-MB-231 cells, which
are deficient in E-cadherin, and
E-cadherin in PC-3 cells, in con-
junction with parallel decreases
~
&
*
Figure 4. A) Time [24 ( ), 48 ( ), and 72 h ( )] and dose-dependent suppressive effects of 51, 59, and 71 versus
1 on the viability of MDA-MB-231 cells as determined by MTT assays. Data are the meanÆSD (n=6). B) Western
blot analysis of the dose-dependent effects of individual agents on the phosphorylation of AMPK and down-
stream targets, including ACC, p70S6K, and Akt in MDA-MB-231 cells after 48 h of exposure. C) Dose-dependent
effects of 60 on the phosphorylation of AMPK in MDA-MB-231 cells after 48 h of exposure.
AMPK activation on multiple oncogenic signaling pathways, in-
cluding those mediated by Akt and mTOR/p70S6K.[11,12] Conse-
quently, as part of the mechanistic interrogation of 59, 51, and
71, we compared their effects versus those of 1 on these
aforementioned signaling markers. As shown, 59 (0.3–2.5 mm)
and, to a slightly lesser extent, 51 (0.5–5 mm) were effective in
increasing the phosphorylation of AMPK and its target ACC, ac-
companied by parallel decreases in the phosphorylation of
p70S6K, a marker of mTOR activity, and Akt in a manner remi-
niscent of that of 1 (2.5–10 mm) in MDA-MB-231 cells (Fig-
ure 4B). In contrast, 71, unlike 59 and 51, was ineffective in
downregulating the phosphorylation of p70S6K and Akt, sug-
gesting a clear structure–activity relationship (SAR) among
these three AMPK-activating agents in modulating AMPK
downstream effectors (see the following section for further dis-
cussion). This distinct SAR was further manifested by 60, which
differs from 59 in the substitution position of the nitro group
(C4 versus C3). Despite largely shared structural motifs, 60 ex-
hibited substantially lower efficacy than 59 in suppressing via-
bility (IC50: 6 mm versus 1.8 mm for 59, data not shown) and
eliciting AMPK activation (Figure 4C) in MDA-MB-231 cells.
To refute the possibility that the capacity of 59 to activate
AMPK signaling is unique to MDA-MB-231 cells, we expanded
our investigation to two additional cell lines, MDA-MB-468
breast cancer cells and phosphatase and tensin homologue
in the mesenchymal markers vimentin and snail in both cell
lines (Figure 5C).
59 is not a PAINS compound
It has recently been reported that so-called pan-assay interfer-
ence (PAINS) compounds—those with certain classes of chemi-
cally reactive substructures—could give rise to false-positive
results by interfering with assays through covalent modifica-
tions, metal chelation, photometric interference, redox activity,
and/or aggregate formation.[26–28] These types of interference
might lead to false leads that could not be optimized. Families
of these PAINS compounds include alkylidine-containing het-
erocycles, hydroquinones, catechols, hydroxyphenylhydro-
zones, Mannich bases of phenols, and 2-amino-3-carbonyl thio-
phenes. Although 59 was not recognized by the PAINS filters
(http://cbligand.org/PAINS/login.php),[27] it contains an alkyli-
dene thiazolidinedione substructure, which could serve as a Mi-
chael acceptor to covalently modify the target protein AMPK
upon binding. This possibility was refuted by the distinct SAR
displayed by various close derivatives of 59 in activating
AMPK, as demonstrated in Figures 3 and 4. Moreover, we ob-
tained two additional lines of evidence to support the position
that 59 is not a PAINS compound. First, compound 87, a satu-
rated counterpart of compound 59, retained the ability to sup-
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1919
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 5. Western blot analysis of the effect of compound 59 versus 1 on the phosphorylation of AMPK, ACC, p70S6K, and Akt in A) MDA-MB-468 and B) PC-
3 cells after 48 h of treatment. C) Suppressive effect of 59 on EMT in MDA-MB-231 and PC-3 cells, as manifested by increases in the expression of Foxo3a and
the epithelial marker claudin-1 or E-cadherin, accompanied by parallel decreases in the expression of the mesenchymal markers vimentin and snail.
lent adducts, the course of enzyme activation would
not be adversely affected by drug removal. However,
this washout experiment indicated that AMPK phos-
phorylation in drug-treated MDA-MB-231 cells re-
turned to the basal level following the removal of 59
from the medium (Figure 6B). This rapid restoration
of AMPK phosphorylation suggests
nature of this ligand–protein interaction.
a reversible
In vivo efficacy of 59 versus 1 in suppressing PC-3
xenograft tumor growth
Based on the above findings, we selected 59 as the
lead AMPK activator vis-à-vis 1 for the evaluation of
in vivo efficacy. In light of the ability of 59 to target
the AMPK–Akt signaling axis, we examined 59 versus
1 as a single agent against the growth of subcutane-
ous xenograft tumors established from PTEN-null PC-
3 cells in athymic nude mice (n=7 for each group).
As shown, daily administration of 59 at 50 mgkgÀ1
Figure 6. Evidence that 59 is not a PAINS compound. A) Structure of 87, a saturated
derivative of 59 (left); western blot analysis of the dose-dependent effects of 87 on the
phosphorylation of AMPK in MDA-MB-231 cells after 48 h of exposure (right). B) Time-
dependent effects of 59, via continuous exposure or removal (washout) after 12 h of
drug exposure, on the phosphorylation of AMPK in MDA-MB-231 cells.
press cell viability (IC50: 9 mm, data not shown) and to activate
AMPK, although with lower potency, in MDA-MB-231 cells (Fig-
ure 6A).
via oral gavage significantly suppressed PC-3 xenograft tumor
growth after 28 days of treatment (65% inhibition, p=0.0003
relative to vehicle-treated control; Figure 7A, left).
The lower efficacy of 87 in eliciting AMPK activation might
arise from loss of the rigid conformation associated with the
removal of the double bond, resulting in weaker interactions
with the binding motif. Second, we conducted a washout ex-
periment, in which 59 was removed from the culture medium
after 12 h of drug exposure. In principle, if the mode of AMPK
activation by 59 was mediated through the formation of cova-
In contrast, 1 gave rise to modest, but statistically insignifi-
cant inhibition (36% inhibition; p=0.069). Equally important,
59 was well tolerated, as no significant loss of body weight
was noted (Figure 7A, right). Moreover, western blot analysis
of tumor lysates showed that this tumor-suppressive activity
correlated with the ability of 59 to activate AMPK signaling, as
manifested by parallel changes in the phosphorylation levels
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1920
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
This role of the nitro substituent in mediating AMPK activation
was also manifested by the differential activity between 51
and 59. Relative to 59, compound 51 showed lower activities
in antiproliferation and AMPK activation, which might be due
to the stereochemical effect of the additional chlorine atom on
protein binding. Together, these data suggest intricate inter-
play between the nitro substituent on the terminal phenyl ring
and neighboring functionalities in interacting with the binding
pocket of AMPK. Compound 59 exhibited multifold higher po-
tency than 1 in upregulating the phosphorylation of AMPK
and its downstream target ACC in multiple cell lines, irrespec-
tive of their LKB1 functional status. Equally important, this
AMPK activation was accompanied by parallel changes in the
phosphorylation/expression levels of p70S6K, Akt, Foxo3a, and
various EMT-associated markers, confirming its ability to target
the AMPK–Akt–Foxo3a signaling axis.
Our previous study demonstrated the in vivo efficacy of 1 in
suppressing PTEN-functional MDA-MB-231 xenograft tumor
growth.[11] In light of its predicted activity against tumors with
activated Akt status, we examined the tumor-suppressive activ-
ity of 59 versus 1 (50 mgkgÀ1 daily via oral gavage) in PTEN-
null PC-3 tumor-bearing athymic nude mice. Consistent with
the relative in vitro antiproliferative potency, 59 showed higher
in vivo efficacy than 1 in inhibiting PC-3 xenograft tumor
growth, which correlated with their respective activities in facil-
itating AMPK activation in tumors. Together, these findings
suggest that 59 has clinical value in therapeutic strategies for
PTEN-negative cancer and warrants continued investigation in
this regard.
~
*
Figure 7. A) In vivo efficacy of 59 ( ) versus 1 ( ) in suppressing PC-3 xeno-
&
graft tumor growth; vehicle control ( ) is shown for comparison. Athymic
nude mice bearing established subcutaneous PC-3 xenograft tumors were
treated once daily with vehicle, 1, or 59 via oral gavage at 50 mgkgÀ1 for
28 days. Tumor volume (left) and body weight (right) data are presented as
meansÆSEM (n=7). B) Western blot analysis of the phosphorylation of
AMPK, ACC, and Akt in lysates from four representative tumors from each
group of mice after 28 days of treatment.
of AMPK, ACC, and Akt (Figure 7B). As compared with 1, com-
pound 59 elicited a greater extent of AMPK activation in
tumors.
PAINS have become a major issue in drug discovery because
they give rise to false-positive results in enzyme-based assays
due to their confounding properties, which would lead to
problematically flat SAR if the chemically reactive substructure
remains unchanged.[26–28] In our study, there was a clear SAR in
the course of optimization, and it is notable that disruption of
the conjugated system by removing the double bond did not
abrogate the ability of the resulting compound, i.e., 87, to acti-
vate AMPK, although with lower potency. In addition, the
washout experiment indicates the ligand–protein binding is re-
versible, which, in conjunction with the improved in vitro and
in vivo antitumor efficacy of 59 relative to 1 by targeting the
AMPK–Akt signaling axis, refutes the possibility that 59 be-
haves like a PAINS compound.
Discussion
Using 1 as a proof of concept, we previously demonstrated
the relevance of AMPK activation as a therapeutic strategy to
suppress tumorigenesis and invasive phenotype by targeting
multiple oncogenic pathways.[11,12] However,
1 exhibited
modest potency in facilitating AMPK activation, which under-
lies the impetus of lead optimization of 1 to develop more
potent AMPK activators. Based on modeling analysis, we modi-
fied different substructures of compound 1 in a stepwise
manner, as outlined in Figure 1B, to generate a total of 85 de-
rivatives, among which 59 represented the lead agent. Analysis
of the antiproliferative and/or AMPK-activating activities of in-
dividual derivatives revealed a distinct SAR. Among various
functionalities that were modified, the linker between the two
aromatic systems and the nitro substituent on the phenyl ring
had the most profound effects on the aforementioned activi-
ties. Among the 21 derivatives selected for western blot analy-
sis (Figure 3C), compounds with a urea linker (51–54, 59, 62,
68, 69, 71, 74, 80, 85, 86) generally showed superior potency
in facilitating AMPK phosphorylation relative to those contain-
ing sulfonamide (37, 39) or amide (43, 46–48, 50) groups (Fig-
ure 3C). Moreover, shifting the orientation of the nitro function
of 59 (C3!C4) substantially attenuated the antiproliferative
and AMPK-activating activities of the resulting compound 60.
Conclusions
In light of the demonstrated ability of compound 1 to target
the Akt–Foxo3a signaling pathway via AMPK activation, this
study described the lead optimization of 1 to develop a series
of potent AMPK activators. This effort netted the lead com-
pound 59, which displayed high in vitro and in vivo efficacy in
activating AMPK and suppressing PTEN-negative tumor
growth, which might have translational potential in fostering
novel therapeutic strategies for cancer therapy.
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1921
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Alexa Fluor 555- and 488-conjugated goat anti-rabbit and anti-
mouse IgG were purchased from Invitrogen (Carlsbad, CA, USA),
and anti-mouse and anti-rabbit secondary antibodies were ob-
tained from Jackson ImmunoResearch Laboratories (West Grove,
PA, USA).
Experimental Section
Chemistry
Detailed information on the syntheses of compounds 2–87 is
given in the Supporting Information. Compound 1 was synthesized
as previously reported (compound 53 therein).[19] All commercially
available reagents were used without further purification unless
otherwise stated. Routine 1H and 13C NMR spectra were recorded
on Bruker AV300 or Bruker Ascend 400 instruments. Samples were
dissolved in CDCl3 or deuterated dimethyl sulfoxide ([D6]DMSO),
and tetramethylsilane (TMS) was used as a reference. Electrospray
ionization mass spectrometry analyses were performed with a Mi-
cromass Q-Tof II high-resolution electrospray mass spectrometer at
The Ohio State University Campus Chemical Instrument Center. All
compounds used for bioassays were characterized by 1H NMR,
13C NMR, and HRMS, and were found to have purity >95%. The
purity of all compounds was confirmed by a Hitachi Elite LaChrom
HPLC system (including a Versa Grad Prep 36 pump, an L-2400 UV
detector, an L-2200 autosampler, and a 150”4.6 mm Agilent
ZORBAX Eclipse XDB-C18 5m column; detection: l 360 nm). A linear
solvent gradient with a mobile phase of 40% H2O in CH3CN to
100% CH3CN in 30 min was used.
Cell viability assay: Cell viability was assessed by using the MTT
assay in six replicates. Cells were seeded and incubated in 96-well
plates in the respective medium with 10% FBS for 24 h, and then
exposed to test agents at the indicated concentrations dissolved in
DMSO in 5% FBS-supplemented medium. After 72 h of treatment,
the medium was removed and replaced by 200 mL of 0.5 mgmLÀ1
MTT in 10% FBS-containing medium, and cells were incubated at
378C for 2 h. Supernatants were removed, and the MTT dye was
solubilized in DMSO (120 mL per well). Absorbance at l 570 nm
was determined on a plate reader. Cell viabilities are expressed as
percentages of that in the corresponding vehicle-treated control
group.
Cell lysis and immunoblotting: Cells were exposed to the test
agents in 10 cm dishes and then collected by scraping. The cell
pellets were washed once with PBS, and then lysed in SDS lysis
buffer containing 50 mm Tris·HCl (pH 8), 10 mm EDTA, 1% SDS, and
a commercial protease inhibitor cocktail (2 mm AEBSF, 1 mm EDTA,
130 mm bestatin, 14 mm E-64, 1 mm leupeptin, 0.3 mm aprotinin).
After centrifugation of lysates for 20 min at 14000”g, the superna-
tants were collected. A sample of each supernatant (1 mL) was
used for determination of protein concentration using a colorimet-
ric bicinchoninic acid assay (Pierce, Rockford, IL, USA), and to the
remaining sample was added an equal volume of 2” SDS-PAGE
sample loading buffer (62.5 mm Tris·HCl, pH 6.8, 4% SDS, 5% b-
mercaptoethanol, 20% glycerol, 0.1% bromophenol blue), followed
by incubation in boiling water for 5 min. Equivalent amounts of
proteins were resolved by SDS-PAGE, and then transferred to nitro-
cellulose membranes using a semidry transfer cell. The transblotted
membrane was washed twice with Tris-buffered saline containing
0.1% Tween 20 (TBST). After blocking with TBST containing 5%
nonfat milk for 40 min, the membrane was incubated with the ap-
propriate primary antibody (1:1000) in TBST-1% nonfat milk at 48C
overnight. The membrane was then washed three times with TBST
for a total of 15 min, followed by incubation with goat anti-rabbit
or anti-mouse IgG–horseradish peroxidase conjugates (1:2000) for
1 h at room temperature and four washes with TBST for a total of
1 h. The immunoblots were visualized by enhanced chemilumines-
cence.
Docking analysis
Docking was performed with the Discovery Studio CDocker proto-
col (version 2.5, Accelrys Inc., San Diego, CA, USA). The X-ray crys-
tal structure of human AMPK was obtained from the RCSB Protein
Data Bank server (PDB ID: 4CFE). The protein structure was pre-
pared according to the standard protein preparation procedure in-
tegrated in Accelrys Discovery Studio 2.5, which involved adding
hydrogen atoms, completing the missing loops, and assigning
force field parameters. CHARMM force field was used for simulation
studies, and then the protein structure was minimized using 1000
iterations of steepest descent minimization algorithm. The ligands
were drawn using the sketching tools of Accelrys Discovery Studio
and were prepared for docking by adding hydrogen atoms and
partial charges using the Momany–Rone method. The CDocker
protocol was used for docking. The binding site was defined by
the residues within 10 Š distance from the co-crystallized ligand.
The default values of CDocker were used. Ten different docking sol-
utions were generated and visually inspected for selection of the
best binding mode. The docking mode was represented using Chi-
mera software version 1.8.1.[29]
Testing the in vivo efficacy of 1 and 59 in PC-3 tumor-bearing nude
mice: Athymic nude mice (Hsd:Athymic Nude-Foxn1nu; 5–6 weeks
of age; Harlan, Indianapolis, IN, USA) were group-housed under
conditions of constant photoperiod (12 h light, 12 h dark) with ad
libitum access to sterilized food and water. All experimental proce-
dures using mice were done in accordance with protocols ap-
proved by the Institutional Animal Care and Use Committee of The
Ohio State University (US Office of Laboratory Animal Welfare, As-
surance #A3261-01). Ectopic tumors were established in athymic
nude mice by subcutaneous injection of PC-3 cells (1”106 cells per
mouse). The establishment and growth of tumors were monitored
by direct measurement with calipers. Mice with established tumors
(mean starting tumor volume ~50 mm3) were randomized to three
treatment groups (n=7). Mice were treated once daily for 28 days
by oral gavage with a) compound 1 at 50 mgkgÀ1 body weight,
b) compound 59 at 50 mgkgÀ1 body weight, or c) vehicle (10%
DMSO/0.5% methylcellulose/0.1% Tween 80 [v/v] in sterile water).
Tumor burdens were determined weekly using calipers [tumor
volume=(width)2 ”(length)”0.52].
Biological methods
Cell lines, culture, reagents, and antibodies: The MDA-MB-231 and
MDA-MB-468 breast cancer and PC-3 prostate cancer cell lines
were purchased from American Type Culture Collection (Manassas,
VA, USA). Media used for the maintenance of these cells were as
follows: MDA-MB-231 and MDA-MB-468, DMEM; PC-3, RPMI 1640,
all of which were supplemented with 10% fetal bovine serum
(FBS). Cells were incubated at 378C in a humidified incubator con-
taining 5% CO2. MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-
tetrazolium bromide] was obtained from TCI America (Portland,
OR, USA), and the ECL Western Blotting System was from GE
Healthcare Life Sciences (Pittsburgh, PA, USA). Antibodies specific
for the following protein targets were used: p-Thr172-AMPK,
AMPK, p-Thr389-p70S6K, p70S6K, Ser473-Akt, Akt, Foxo3a, claudin-
1, vimentin, snail, and b-actin (Cell Signaling Technology Inc., Bev-
erly, MA, USA); E-cadherin, BD Biosciences (San Diego, CA, USA).
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1922
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[10] W. W. Wheaton, S. E. Weinberg, R. B. Hamanaka, S. Soberanes, L. B. Sulli-
van, E. Anso, A. Glasauer, E. Dufour, G. M. Mutlu, G. S. Budigner, N. S.
Chandel, Elife 2014, 3, e02242.
[11] K. H. Lee, E. C. Hsu, J. H. Guh, H. C. Yang, D. Wang, S. K. Kulp, C. L. Sha-
piro, C. S. Chen, J. Biol. Chem. 2011, 286, 39247–39258.
Statistical analysis: In vitro experiments were performed using 3–6
replicates in each group. All in vitro experiments were carried out
at least twice on different occasions. Differences between group
means of tumor burden were analyzed for statistical significance
using two-sided Student t tests. Differences were considered signif-
icant at p<0.05.
[12] C. C. Chou, K. H. Lee, I. L. Lai, D. Wang, X. Mo, S. K. Kulp, C. L. Shapiro,
C. S. Chen, Cancer Res. 2014, 74, 4783–4795.
[13] D. B. Shackelford, R. J. Shaw, Nat. Rev. Cancer 2009, 9, 563–575.
[14] F. Giordanetto, D. Karis, Expert Opin. Ther. Pat. 2012, 22, 1467–1477.
[15] Z. Wang, J. Huo, L. Sun, Y. Wang, H. Jin, H. Yu, L. Zhang, L. Zhou, Curr.
Comput.-Aided Drug Des. 2011, 7, 214–227.
[16] S. Rana, E. C. Blowers, A. Natarajan, J. Med. Chem. 2015, 58, 2–29.
[17] B. Cool, B. Zinker, W. Chiou, L. Kifle, N. Cao, M. Perham, R. Dickinson, A.
Adler, G. Gagne, R. Iyengar, G. Zhao, K. Marsh, P. Kym, P. Jung, H. S.
Camp, E. Frevert, Cell Metab. 2006, 3, 403–416.
[18] T. Pang, Z. S. Zhang, M. Gu, B. Y. Qiu, L. F. Yu, P. R. Cao, W. Shao, M. B. Su,
J. Y. Li, F. J. Nan, J. Li, J. Biol. Chem. 2008, 283, 16051–16060.
[19] J. H. Guh, W. L. Chang, J. Yang, S. L. Lee, S. Wei, D. Wang, S. K. Kulp, C. S.
Chen, J. Med. Chem. 2010, 53, 2552–2561.
Abbreviations
ACC: acetyl-CoA carboxylase; AICAR: 5-aminoimidazole-4-carboxa-
mide ribonucleotide; AMPK: adenosine monophosphate activated
protein kinase; CBM: carbohydrate binding module; EMT: epithe-
lial–mesenchymal transition; LKB1: liver kinase B1; MDM2: murine
double minute 2; MTT: 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-
tetrazolium bromide; PTEN: phosphatase and tensin homologue.
[20] L. F. Yu, Y. Y. Li, M. B. Su, M. Zhang, W. Zhang, L. N. Zhang, T. Pang, R. T.
Zhang, B. Liu, J. Y. Li, J. Li, F. J. Nan, ACS Med. Chem. Lett. 2013, 4, 475–
480.
[21] B. C. Bookser, Q. Dang, T. S. Gibson, H. Jiang, D. M. Chung, J. Bao, J.
Jiang, A. Kassick, A. Kekec, P. Lan, H. Lu, G. M. Makara, F. A. Romero, I.
Sebhat, D. Wilson, D. Wodka (Merck Sharp & Dohme Corp., Metabasis
Therapeutics Inc., Schering Corp.), Pat. Appl. No. CA 2741125 A1, 2010.
[22] B. Xiao, M. J. Sanders, D. Carmena, N. J. Bright, L. F. Haire, E. Underwood,
B. R. Patel, R. B. Heath, P. A. Walker, S. Hallen, F. Giordanetto, S. R. Martin,
D. Carling, S. J. Gamblin, Nat. Commun. 2013, 4, 3017.
[23] L. M. Lima, E. J. Barreiro, Curr. Med. Chem. 2005, 12, 23–49.
[24] N. A. Meanwell, J. Med. Chem. 2011, 54, 2529–2591.
[25] R. J. Dowling, M. Zakikhani, I. G. Fantus, M. Pollak, N. Sonenberg, Cancer
Res. 2007, 67, 10804–10812.
[26] J. B. Baell, L. Ferrins, H. Falk, G. Nikolakopoulos, Aust. J. Chem. 2013, 66,
1483–1494.
[27] J. B. Baell, G. A. Holloway, J. Med. Chem. 2010, 53, 2719–2740.
[28] J. Baell, M. A. Walters, Nature 2014, 513, 481–483.
Acknowledgements
This work was supported by the US National Institutes of Health
grant R01CA112250 (C.S.C.) and a joint supervision fellowship
from the Egyptian government (E.M.E.D.).
Keywords: AMPK · activators · antitumor agents · drug
discovery · lead optimization
[1] B. B. Kahn, T. Alquier, D. Carling, D. G. Hardie, Cell Metab. 2005, 1, 15–
25.
[2] Y. C. Long, J. R. Zierath, J. Clin. Invest. 2006, 116, 1776–1783.
[3] C. T. Lim, B. Kola, M. Korbonits, J. Mol. Endocrinol. 2010, 44, 87–97.
[4] D. G. Hardie, Genes Dev. 2011, 25, 1895–1908.
[5] B. B. Zhang, G. Zhou, C. Li, Cell Metab. 2009, 9, 407–416.
[6] D. G. Hardie, Annu. Rev. Pharmacol. Toxicol. 2007, 47, 185–210.
[7] D. G. Hardie, Cold Spring Harbor Symp. Quant. Biol. 2011, 76, 155–164.
[8] H. C. Chuang, C. C. Chou, S. K. Kulp, C. S. Chen, Curr. Pharm. Des. 2014,
20, 2607–2618.
[29] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt,
E. C. Meng, T. E. Ferrin, J. Comput. Chem. 2004, 25, 1605–1612.
[9] A. Leone, E. Di Gennaro, F. Bruzzese, A. Avallone, A. Budillon, Cancer
Treat. Res. 2014, 159, 355–376.
Received: August 15, 2015
Published online on September 9, 2015
ChemMedChem 2015, 10, 1915 – 1923
www.chemmedchem.org
1923
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim