Identification of small molecules that inhibit GSK-3β through virtual screening

Article abstract of DOI:10.1016/j.bmcl.2008.10.120

Glycogen synthase kinase-3β (GSK-3β) is involved in glycogen metabolism, neuronal cell development, osteoblast differentiation. Small molecule inhibitors of GSK-3β have various therapeutic potential for the treatment of diabetes type II, bipolar disorders

Full text of DOI:10.1016/j.bmcl.2008.10.120

Bioorganic & Medicinal Chemistry Letters 19 (2009) 533–537
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Identification of small molecules that inhibit GSK-3b through virtual screening
Nam Sook Kang a, Gil Nam Lee a, Chi Hyun Kim b, Myung Ae Bae b, Ikyon Kim c, Young Sik Cho b,
*
a Center for Drug Discovery Technologies, Korea Research Institute of Chemical Technology, PO Box 107, Yuseong-gu, Daejeon 305-600, Republic of Korea
b Center for Metabolic Syndrome Therapeutics, Korea Research Institute of Chemical Technology, PO Box 107, Yuseong-gu, Daejeon 305-600, Republic of Korea
c Center for Medicinal Chemistry, Korea Research Institute of Chemical Technology, PO Box 107, Yuseong-gu, Daejeon 305-600, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 4 June 2008
Revised 26 September 2008
Accepted 28 October 2008
Available online 5 November 2008
Glycogen synthase kinase-3b (GSK-3b) is involved in glycogen metabolism, neuronal cell development,
osteoblast differentiation. Small molecule inhibitors of GSK-3b have various therapeutic potential for
the treatment of diabetes type II, bipolar disorders, stroke and chronic inflammatory disease.
To identify GSK-3b inhibitors with novel scaffold from chemical library, we primarily screened out
putative inhibitors through computer modeling and subsequently evaluated the inhibitory activity of
selected compounds against GSK-3b by in vitro Z’-LYTETM assay. A series of compound KRMs strongly
inhibited phosphorylation of its substrate with IC50 value of approximately 0.5 lM. Also, we demon-
strated that KRM-189 and KRM-191 competed with ATP for GSK-3b, leading to decreased Vmax and con-
stant Km with increasing concentrations of ATP as determined from Lineweaver–Berk equation.
Moreover, they showed the selectivity for GSK-3b over other kinases with IC50 values of 2 to 10 lM or
Keywords:
Glycogen synthase kinase-3b
Inhibitor
Thiadiazolidinones
Virtual screening
more Incubation of cells with KRM-191 with highly selective and potent inhibitory activity caused accu-
mulation of b-catenin, downstream of GSK-3b signaling pathway, indicating that small molecule can pre-
vent degradation of b-catenin via GSK-3b inhibition. Our results suggest that modeling in combination
with in vitro assays can be used for the identification of selective and potent inhibitors.
Ó 2008 Elsevier Ltd. All rights reserved.
Glycogen synthase kinase-3 (GSK-3) is a cytoplasmic serine–
threonine kinase and exists in two highly homologous forms, GSK-
3
pathway transmitted by insulin or Wnt, which has been implicated
in glucose homeostasis, remodeling of bone mass or developmental
process of the embryo.2 GSK-3b is constitutively active in resting
cells and treatment of cells with agents, such as insulin and lithium
chloride (LiCl), is shown to cause GSK-3 inactivation through a PI 3-
kinase (PI3-K)-dependent mechanism. PI3-K-induced activation of
tively charged pocket of the kinase domain of GSK-3b are useful
for selectively inhibiting primed substrates. Therefore, intervention
of GSK-3b might be a useful target to the treatment and prevention
of diabetes, Alzheimer and osteoporosis. Up to date, a few com-
pounds are known to inhibit directly its enzyme. Lithium chloride
has a specific inhibitory activity in vitro and in intact cells although
millimolar concentration of IC50 is limited to therapeutic use.
Besides, small molecules such as bisindole or aniline maleimides,
kenpaullone, indirubin, or the marine natural product hymenialdi-
sine have been reported as GSK-3b inhibitors.3 All the small mole-
cules under development inhibit in a competitive manner with
ATP and as a result, show no selectivity over a wide variety of protein
kinases. In this presentation, we found out a scaffold structure of thi-
adiazolinone (TDZD) that might be a putative inhibitor of GSK-3b
through running computer modeling, and furthermore synthesized
TDZD derivatives to address moiety of which could be attributing to
their inhibitory activity. Addition of some groups to parent chemical
improved potently inhibitory activity against GSK-3b in enzyme
assay. It also inhibited enzyme by competing with ATP, but has
the selective inhibition for GSK-3b over other serine/threonine
kinases. In a bioassay measuring b-catenin accumulation as a result
of GSK-3b inhibition, it showed that b-catenin was still accumulated
in the total cell extracts as did insulin or lithium chloride in this
experiment. Taken together, computer modeling will provide a use-
ful tool for the primary identification of scaffolding structure fitted
into binding pocket of enzyme with known structure.
a
and GSK-3b.1 Especially, GSK-3b plays a key role in signaling
PKB/Akt results in phosphorylation of Ser21 on GSK-3
a and Ser9
on GSK-3b, therefore leading to the inhibition of GSK-3 activity.
The phosphorylated N-terminus becomes a primed pseudosub-
strate that occupies the positive binding pocket and the active site
of the enzyme and acts as a competitive inhibitor for true substrates.
Several known GSK-3 substrates participate in a wide network of
cellular processes, including glycogen metabolism, transcription,
translation, cytoskeletal regulation, intracellular vesicular trans-
port, cell cycle progression, and apoptosis. Phosphorylation of these
substrates by GSK-3b usually has an inhibitory effect. Arg96 is
shown to be a crucial component of the positive pocket that binds
primed substrates. Small molecule inhibitors that fit in the posi-
* Corresponding author at present address: Department of Pathology, University of
Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655-0002,
USA. Tel.: +82 42 860 7465.
E-mail address: yscho@krict.re.kr (Y.S. Cho).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.10.120

534
N. S. Kang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 533–537
O
R1
R2-NCO
Cl2
Cl
C
N
R1-NCS
R1
N
S Cl
R2
O
n-hexane
15 °C
N
n-hexane
rt
S
−
R1 = CH2CO2t-Bu
2 = CH2CO2Et
R
1) oxalyl chloride, cat. DMF
CH2Cl2, 0 °C - rt
O
O
CO2t-Bu
CO2H
HCO2H
rt
N
N
N
O
EtO2C
EtO2C
N
N
N
2) piperidine
CH2Cl2, 0 °C - rt
O
O
O
S
S
EtO2C
N
O
S
Scheme 1.
Figure 1. The residues within 5A around the bound-ligand are shown in 1R0E.pdb.
Figure 2. Inhibition curves of each compound were obtained from the incubation of
enzyme with various concentrations of it, respectively. Inhibitory potency was
compared with each other for % inhibition at 10 lM and IC50.
Table 1
Docking scores for potent GSK-3b inhibitors selected from chemical library.
Compound
Docking score
ate putative alignments of molecules or molecules fragments.
These putative poses are achieved using the Hammerhead scoring
function.6
KRM-191
KRM-296
KRM-192
KRM-195
KRM-189
6.81
6.8
6.77
6.57
3.97
Initially, in-house library including 170 compounds with TDZD
fragment as a small molecule library, synthesized from diverse iso-
cyanates and isothiocyanates via known method7 as shown in
Scheme 1,8 was docked into binding sites of GSK-3b. Before per-
forming docking, all compounds were minimized using Cerius29
suite of programs and the conserved water molecule (W82 as
shown in Fig. 1) was also specified near Thr138 side chain atom
and Asp200 backbone atom, which can play a crucial role in inhib-
itor binding.
To virtually screen out putative ligands that are capable of
acting as inhibitors of GSK-3b from in-house chemical library, we
employed SurflexDock4 method interfaced with Sybyl 7.3.1.5
SurflexDock4 is a new docking methodology that combines Ham-
merhead’s empirical scoring function6 with a molecular similarity
method to generate putative alignment of ligands. It4 employs an
idealized active site ligand (called a protomol) as a target to gener-
In an attempt to find out small molecules having an inhibition
activity against GSK-3b using SurflexDock,4 PDB ID 1R0E as a

N. S. Kang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 533–537
535
complex reference structure was adopted. Using active site
extracted from 1R0E as PDE ID, we generated a protomol as an ide-
alized active site ligand. SurflexDock’s protomols utilize CH4, C@O
and NH fragments. Protomol construction was based on protein
residues proximal to the native ligand and on parameter settings
to produce a small and buried docking target (SurflexDock param-
eters: proto_thresh 0.6 and proto_bloat 0). Each docking of puta-
tive ligands returned up to 50 scored poses, with the score
consisting of a nominal affinity score.6 Through SurflexDock meth-
od, we chose the pose with best score from multiple docking of the
same ligand. Out of 170 compounds, five compounds were finally
selected after considering a structural diversity and docking score.
The best pose of the chosen ligands and their resulting scores are
shown in Table 1.
To further investigate inhibitory potency of some derivatives
with thiadiazolidinones (TDZD) chosen through computer model-
ing for the inhibitors of GSK-3b, we performed Z’-LYTETM kinase as-
say,11 in which ratiometric fluorescence transfer between
substrate and product was reproducibly determined. Inhibition of
GSK-3b by each compound was shown in Figure 2 by plotting rel-
ative inhibition of enzyme at increasing doses of compound. IC50
values that are indicative of inhibitory activity of each compound
were summarized in Table 2. In vitro enzymatic data showed that
five compounds had their distinctive IC50 values with a similar
inhibitory potency at a fixed concentration. Scaffold with benzyl
and ethyl group as substituents exhibited low IC50 value, similar
to value reported previously.3 Derivatives with different combina-
tion of substituent groups had different IC50 values ranging from
0.5 to 2 lM.
To delineate the inhibitory mechanism of compounds with
most potent inhibitory activity under investigation (KRM-189
and KRM-191), kinetic experiment was carried out under different
concentrations of ATP and TDZD derivative. Double-reciprocal plot
of kinetic profiles was shown in Figure 3a and b. Lineweaver Burk
plot suggests that either KRM-189 or -191 was demonstrated to be
a typical competitive inhibition against substrate ATP, indicating
that chemical derivatives share binding pocket of enzyme with
ATP. With increasing concentrations of ATP, it displayed intersec-
tion of same Vmax and different Km values. A parent compound
(KRM-7777) has been reported to inhibit GSK-3b selectivity over
other protein kinases independent of ATP concentration. Therefore,
TDZD with some modification might shift inhibition mode from
noncompetitive to competitive while improving the inhibitory
competency.
Due to its involvement of GSK-3b in multiple pathways as de-
scribed above, selectivity of GSK-3b inhibition is one of a consider-
able important factor in the development of inhibitors for
therapeutic applications. Since GSK-3b is supposed to be phyloge-
netically most close to cyclin dependent kinases (CDKs),12 it is com-
pelling to achieve selectivity against CDKs. Moreover, structural
and functional similarities between GSK-3b and casein kinase 2
(CK2) were suggested on phylogenetic trees.13 PKA or PKC
Table 2
Summary for the inhibitory potency against GSK-3b and the selectivity of TDZD derivatives for GSK-3b over other kinases with reference to KRM-7777. Some other serine/
threonine kinases were tested with chemicals to inhibit potently GSK-3b. All the reaction conditions were almost similar with exception of substrates suitable for respective
enzyme, and were carried out according to the instructions provided by manufacturer.
Compound
Structure
% inhibition/IC50
PKA
GSK-3b
At 10a
CSNK2
At 100c
PKC,
a
CDK1
At 100c
IC50
b
d
d
d
IC50
IC50
At 100c
IC50
At 100c
IC50
O
N
N
KRM-7777
KRM-189
KRM-191
65
74
84
7149
548
30
17
21
30
88
84
89
13.8
5.8
15
0
S
O
O
S
Cl
N
35
87
N
O
O
O
O
467
2.2
6.9
5.3
18
O
N
O
N
O
S
O
O
O
O
KRM-192
75
80
1012
1863
36
31
81
20
78
51
48.3
12
20
O
N
N
S
O
O
F
N
KRM-195
KRM-296
38.7
4.1
N
O
S
O
N
N
65
71
539
25
21
42
75
6
17
93
0
15
36
N
O
S
O
O
O
LiCl
a
% inhibition at 10 lM concentration.
IC50 unit is nM.
b
c
% inhibition at 100
lM concentration.
d
IC50 unit is lM.

536
N. S. Kang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 533–537
Figure 3. The inhibitory potency of TDZD derivatives against GSK-3b. Compound selected through computer modeling was subjected to enzyme reaction to get enzyme
velocity with substrate and double-reciprocal plot with increasing concentrations of KRM-189 (a) and KRM-191 (b) at various doses of ATP.
insulin (Fig. 4). b-Catenin was continuously degraded under
normal differentiation, resulting in lower level. Well-known GSK-
3b inhibitor lithium chloride (LiCl) as well as insulin, which
activate insulin signaling to indirectly inactivate GSK-3b, has been
suggested to protect b-catenin from proteasome degradation. This
biological data shows that in vitro activity of KRM-191 is reflected
to b-catenin accumulation in cells left treated with KRM-191.
Taken together, this report exhibits one typical example of drug
development through docking virtual screening leading to potent
inhibitors. After well-defined target and optimized algorithm pack-
age are established, introduction of in silico modeling into drug
discovery and development pipeline will enrich the fraction of
drug-like compounds in library, resulting in the reduction of bio-
logical testing in lead discovery. This process has already been
included in drug discovery pipeline of pharmaceutical company.
Figure 4. Effect of GSK-3b inhibitors on accumulation of b-catenin in C2C12
myoblast. Myoblasts plated at a density of 5 Â 105 were exposed to chemicals at
2
l
M for 3 h. Cells harvested were lysed to obtain total proteins and then subjected
to SDS–PAGE for immunoblot with antibody raised against b-catenin. As an internal
standard, b-actin was detected by probing blotted membrane with anti b-actin.
belongs to be different from superfamily including GSK-3b on the
human kinome tree. Therefore, selective inhibition of TDZD deriv-
atives against GSK-3b was examined over other related kinases
using Z’-LYTE substrate. In our inhibition data, unexpectedly, nei-
ther CDK2 nor CK2 were affected at even high concentrations of
chemical derivatives whereas two serine/threonine kinases dis-
tinct from GSK-3b such as PKC and PKA were susceptible to inhibi-
tion by TDZD derivatives. Two potent compounds were 10-fold
more selective for GSK-3b over PKC and PKA and 2 order more
selective over CSNK2 and CDK1.
Acknowledgment
This research was supported by the Center for Biological, Mod-
ulators of the 21st Century Frontier R&D Program, Ministry of Sci-
ence and Technology, Korea.
References and notes
1. Woodgett, J. R. Sci. STKE 2001, 100, RE12.
2. (a) Eldar-Finkelman, H.; Kaidanovich, O. Expert Opin. Ther. Targets 2002, 6, 555;
(b) Westendor, J. J.; Kahler, R. A.; Schroeder, T. M. Gene 2004, 341, 19; (c)
Wodarz, A.; Nusse, R. Annu. Rev. Cell Dev. Biol. 1998, 14, 59.
3. Martinez, A.; Alonso, M.; Castro, A.; Perez, C.; Moreno, F. J. J. Med. Chem. 2002,
45, 1292.
To further examine downstream event of GSK inhibition in the
intact cells, b-catenin accumulation following GSK-3b inhibitor
treatment was followed with reference to lithium chloride and

N. S. Kang et al. / Bioorg. Med. Chem. Lett. 19 (2009) 533–537
537
4. Jain, A. N. J. Med. Chem. 2003, 46, 499.
5. SYBYL 7.3.1 Tripos Inc., St. Louis, MO 63144, USA.
4.54 (s, 2H), 4.39–4.12 (m, 4H), 1.36–1.16 (m, 6H); KRM-0195: 300 MHz 1H
NMR (CDCl3) 7.41–7.15 (m, 7H), 7.05–6.96 (m, 2H), 4.74 (s, 2H), 3.88
d
6. (a) Welch, W.; Ruppert, J.; Jain, A. N. Chem. Biol. 1996, 3, 449; (b) Jain, A. N. J.
Comput. Aided Mol. Des. 1996, 10, 427.
(t, J = 7.1 Hz, 2H), 2.93 (t, J = 7.2 Hz, 2H); KRM-0296: 300 MHz 1H NMR (CDCl3) d
4.48 (s, 2H), 4.36 (s, 2H), 4.26 (q, J = 7.2 Hz, 2H), 3.56 (t, J = 5.3 Hz, 2H), 3.40
(t, J = 4.4 Hz, 2H), 1.72–1.52 (m, 6H), 1.31 (t, J = 7.1 Hz, 3H).
9. Cerius2, Accelrys Inc., 9685 Scranton Road, San Diego, CA 92121-3752, USA.
11. Rodems, S. M.; Hamman, B. D.; Lin, C.; Zhao, J.; Shah, S.; Heidary, D.;
Makings, L.; Stack, J. H.; Pollok, B. A. Assay Drug Dev. Technol. 2002, 1,
9.
12. Frame, S.; Cohen, P. Biochem. J. 2001, 359, 1.
13. (a) Hanks, S.; Quinn, A. M. Meth. Enzyme 1991, 200, 36; (b) Allende, J. E.;
Allende, C. FASEB J. 1995, 9, 313.
7. Ottman, G.; Hooks, H. Angew. Chem., Int. Ed. Engl. 1966, 5, 672.
8. KRM-7777: 300 MHz 1H NMR (CDCl3) d 7.49–7.40 (m, 2H), 7.39–7.28 (m, 3H),
4.82 (s, 2H), 3.69 (q, J = 7.2 Hz, 2H), 1.26 (t, J = 7.2 Hz, 3H); KRM-0189: 300 MHz
1H NMR (DMSO-d6) d 7.55 (d, J = 8.4 Hz, 2H), 7.48 (t, J = 8.4 Hz, 2H), 7.32 (t,
J = 7.3 Hz, 1H), 3.99 (t, J = 5.5 Hz, 2H), 3.88 (t, J = 5.4 Hz, 2H); KRM-0191:
300 MHz 1H NMR (DMSO-d6) d 8.08–8.01 (m, 1H), 7.99–7.95 (m, 1H), 7.74–7.64
(m, 2H), 4.54 (s, 2H), 4.39–4.12 (m, 4H), 1.39–1.15 (m, 6H); KRM-0192:
300 MHz 1H NMR (DMSO-d6) d 8.09 (d, J = 8.6 Hz, 2H), 7.58 (d, J = 8.6 Hz, 2H),