ARTICLE

Flavones with inhibitory effects on glycogen synthase kinase 3b

Yearam Jung1 . Soon Young Shin2 . Young Han Lee2 .

Yoongho Lim1

Received: 9 March 2017 / Accepted: 22 March 2017 / Published online: 30 March 2017

� The Korean Society for Applied Biological Chemistry 2017

Abstract Because glycogen synthase kinase-3 (GSK-3)

activity is linked to various human diseases, it has been

targeted in new drug development. Flavonoids, including

luteolin, apigenin, and quercetin, inhibit GSK-3b; however,the relationships between their structural properties and

inhibitory effects are unclear. We measured the inhibitory

effects of 34 flavonoid derivatives on GSK-3b and calcu-

lated hologram quantitative structure–activity relationships

to provide information on pharmacophores for designing

novel compounds with better activities. The in vitro bind-

ing effect of flavonoids was confirmed using Western

blotting for myricetin, which showed the best inhibitory

activity, and the binding mode between myricetin and

GSK-3b was elucidated using in silico docking.

Keywords Flavone � Glycogen synthase kinase 3b �Hologram quantitative structure–activity relationships �In silico docking � Myricetin

Introduction

Since the identification of glycogen synthase kinase-3

(GSK-3), its cellular multi-functions, including cell pro-

liferation, apoptosis, and development, have been identified

[1]. Because the activity of GSK-3 is linked to various

human diseases such as neurodegenerative disorders,

inflammation, and diabetes, it has been targeted in the

development of new drugs [2]. The first GSK-3b inhibitor

was lithium ion, which enhances serine phosphorylation

and competes with magnesium ion [3–5]. Its half-maximal

inhibitory concentration (IC50) is 1–2 mM [4]. Stau-

rosporine isolated from Streptomyces staurosporeus is a

potent GSK-3b inhibitor (IC50 = 15 nM) [6]. Flavonoids

have been reported to have inhibitory effects on GSK-3b[7]. Although the inhibitory effects of several flavonoids

against GSK-3b have been tested [8], the relationships

between their structural properties and inhibitory effects

have not been reported. Quantitative structure–activity

relationships (QSARs) can provide information on phar-

macophores for designing novel compounds with better

activities. Even the data reported previously contained the

IC50 values of several citrus flavonoids [8]; we tested the

inhibitory effects of 34 flavonoid derivatives on GSK-3bhere. The hologram QSARs (HQSARs) could be calculated

in this work. The aim of this study is to identify the

structural conditions of flavonoids that can enhance the

inhibitory effects on GSK-3b. To confirm the in vitro

binding effect of flavonoids, Western blot analysis was

carried out for myricetin, which showed the best inhibitory

activity in this research, and in silico docking was per-

formed to elucidate the binding mode between myricetin

and GSK-3b.

Materials and methods

The 34 flavones tested here included 13 hydroxyflavones,

two methoxyflavone, five hydroxymethoxyflavones, and 14

glycosylated flavones (Table S1). All compounds were

Electronic supplementary material The online version of thisarticle (doi:10.1007/s13765-017-0271-2) contains supplementarymaterial, which is available to authorized users.

& Yoongho Lim

yoongho@konkuk.ac.kr

1 Division of Bioscience and Biotechnology, BMIC,

Konkuk University, Seoul 05029, Republic of Korea

2 Department of Biological Sciences, Konkuk University,

Seoul 05029, Republic of Korea

123

Appl Biol Chem (2017) 60(3):227–232 Online ISSN 2468-0842

DOI 10.1007/s13765-017-0271-2 Print ISSN 2468-0834

used without further purification, and their purities were

determined using high-performance liquid chromatography

[9].

The GSK-3b kinase assay was carried out using the EMD

Millipore KinaseProfiler service assay protocol (EMD Mil-

lipore Corp., Billerica, MA, USA). GSK-3bwas supplied byEMD Millipore Corp. GSK-3b was incubated with 8 mM

3-(N-morpholino)propanesulfonic acid buffer (pH 7.0),

0.2 mM ethylenediaminetetraacetic acid, 20 lM substrate

for phosphorylation (YRRAAVPPSPSLSRHSSPHQS),

10 mM Mg acetate, and 10 lM ATP. The specific activity

of 33P-ATP was adjusted to approximately 500 cpm/pM.

The magnesium–ATP mixture was added for the initiation

of the enzyme reaction. The incubation was performed for

40 min at room temperature. And, 3% phosphoric acid was

added for the stop of the reaction. Ten lL of the reaction

solution was spotted on a P30 filtermat (PerkinElmer,

London, UK). It was washed with 50 mM phosphoric acid

three times and with methanol once. It was dried for

scintillation counting [10, 11]. All experiments were

repeated three times at five different concentrations (0, 5,

10, 20, and 40 lM). The IC50 values were obtained using

SigmaPlot software (Systat, Chicago, IL, USA). The sig-

moid curve fit method was used. The negative logarithmic

values (pIC50) were used as the biological data for the

HQSAR calculations (Table S1) [12]. The procedures for

HQSAR calculations followed the method reported previ-

ously [9, 13].

HCT116 cells were either left untreated or treated with

30 mM LiCl in the absence or presence of different con-

centrations of myricetin. After the indicated time periods,

cell lysates containing 20 lg protein were separated by

10% SDS-PAGE and transferred to nitrocellulose mem-

branes. The blots were probed with an antibody against

phospho-GSK-3b (Ser9; Cell Signaling Technology, Dan-

vers, MA, USA; 1:1000) for 4 h, followed by incubation

with a secondary antibody conjugated to horseradish per-

oxidase for 2 h, as described previously [13]. Glyceralde-

hyde 3-phosphate dehydrogenase (GAPDH) was used as an

internal control. The results were visualized using an

enhanced chemiluminescence detection system (GE

Healthcare, Piscataway, NJ, USA). The relative band

intensities were measured with a quantitative scanning

densitometer and image analysis software, Bio-1D ver

97.04 (Froebel, Wasserburg, Germany).

The three-dimensional (3D) structure of myricetin was

obtained from the crystal structure deposited in the protein

data bank (PDB) as a complex with human pancreatic

alpha-amylase (4gqr.pdb) [14]. Human GSK-3b consists of

420 amino acids. Sixty-nine 3D structures of human GSK-

3b are deposited in the PDB. All of these structures were

determined using X-ray crystallography. Because stau-

rosporine isolated from Streptomyces staurosporeus is

known to be a representative GSK-3b inhibitor, the crys-

tallographic structure (1q3d.pdb) containing staurosporine

as a ligand was chosen [15]. In silico docking experiments

followed the method reported previously [11]. The struc-

ture of 1q3d.pdb consists of two chains, A and B. Chain A

was used because its structure includes two more residues

than chain B. The 3D structure of 1q3d-A is composed of

35–119, 126–286, and 293–385 residues. The apo-protein

of 1q3d-A was prepared using the Sybyl program. To

obtain the solution structure of the apo-GSK-3b protein,

energy minimization was performed according to a method

reported previously [12]. The superposed result between

apo-1q3d-A and the crystallographic structure deposited in

the PDB produced a root-mean-square standard value of

0.29 A. The binding site of the substrate was determined

using the LigPlot program and consisted of the following:

Ile62, Gly63, Gly65, Val70, Ala83, Leu132, Asp133,

Tyr134, Val135, Gln185, Asn186, Leu188, Cys199, and

Asp200. The flexible docking method used the Sybyl

program, and the details of this method have been reported

previously [12]. To confirm that this docking method was

working well, an original ligand, staurosporine, was

docked to apo-1q3d-A. Of the 30 complexes of stau-

rosporine and 1q3d-A generated from the 30 iterations, the

5th complex showed the best docking pose and its binding

energy was -16.67 kcal/mol. Ten hydrophobic interac-

tions (Ile62, Ala83, Lys85, Leu132, Tyr134, Gln185,

Asn186, Leu188, Cys199, and Asp200) and two hydrogen

bonds (H-bonds, Asp133 and Val135) were observed in the

complex based on the LigPlot analysis (Fig. S5). Statistical

significance was analyzed using Student’s t test. A p value

of less than 0.05 was considered statistically significant.

All experiments were performed in triplicate [16].

Results and discussion

The hologram QSAR (HQSAR) studies generate fragment-

based structure–activity relationships using molecular

holograms [17]. The IC50 values on GSK-3b of the 34

flavonoid derivatives ranged from 0.90 lM (8, myricetin)

to 98.73 lM (20, apigetrin) (Fig. S1). The flavonoid

derivatives used here consist of 14 glycosylated flavones

and 20 aglycons (Table S1). The biological data for the

HQSAR calculations were the negative logarithmic IC50

(pIC50) values, as listed in Table S1. The derivatives were

divided into a training set and a test set. The former was

prepared to build up the HQSAR models, and the latter was

used to validate the models. For the test set, six derivatives,

3, 6, 7, 11, 12, and 26, were selected arbitrarily. This

selection was validated using hierarchical clustering anal-

ysis (HCA) [18]. The HCA results obtained from the Sybyl

7.3 program (Tripos, St. Louis, MO, USA) indicated that

228 Appl Biol Chem (2017) 60(3):227–232

123

the six derivatives are not structurally similar derivatives,

as shown in Fig. S2.

HQSAR calculations were carried out using the QSAR/

HQSAR module provided by Sybyl 7.3. The hologram

lengths were set to 53, 59, 61, 71, 83, 97, 151, 199, 257,

307, 353, and 401, which were provided as default values

in the Sybyl 7.3 program. The fragment sizes ranged

between 4 and 7. Fifteen HQSAR models were generated

based on the influence of the fragment distinction in terms

of atom (A), bond (B), connections (C), hydrogen atoms

(H), chirality (Ch), and donor, and acceptor (DA), as listed

in Table S2. Because the fragment distinction including A,

B, and DA and the fragment distinction including A, B, Ch,

and DA showed good cross-validated correlation coeffi-

cient (q2) values, 0.519 and 0.446, respectively, 12 models

were generated for each fragment distinction using six

different fragment sizes, 2–5, 3–6, 4–7, 5–8, 6–9, and 7–10,

as listed in Table S3. Although for the fragment distinction

including A, B, and DA the variation of the fragment size

did not improve the q2 value, for the fragment distinction

including A, B, Ch, and DA, the model with a fragment

size of 5–8 or 7–10 increased the q2 value to 0.556 or

0.661, respectively. The residual values between the

experimental data and the values predicted by the HQSAR

model using a fragment size of 5–8 showed better results

than those using a fragment size of 7–10. Therefore, the

HQSAR model with the fragment distinction of A, B, Ch,

and DA and the fragment size of 5–8 was chosen. Its sta-

tistical parameters are as follows: cross-validated correla-

tion coefficient (q2), 0.566; non-cross-validated correlation

coefficient (r2), 0.942; standard error of estimate, 0.163;

hologram length, 61; and optimal number of components

(N), 6 [19]. The pIC50 values calculated using this HQSAR

model are listed in Table S4. The experimental data were

plotted against the values predicted by this HQSAR model

(Fig. S3). Its Pearson R correlation coefficient was 0.9403.

The residuals between the experimental pIC50 values and

the predicted pIC50 values were obtained. The residuals for

the training set ranged from 0.12% to 21.26%, and for the

test set, they ranged from 4.07% to 21.21%. As a result,

this HQSAR model is reliable and can be used for the

development of novel GSK-3b inhibitors.

To visualize the results obtained from the HQSAR cal-

culations, contribution maps were generated using the

Sybyl program. Positive, neutral, and negative contributions

are expressed in green, white, and red colors, respectively.

In the case of 3,30,40,5,50,7-hexahydroxyflavone (8, myr-

icetin), which showed the best activity, hydroxyl groups at

the meta and para positions of the B-ring are shown in green

(Fig. 1A). This figure demonstrates that the electronegative

Fig. 1 Contribution maps generated from the HQSAR calculations.

Positive, negative, and neutral contributions are colored in green, red,

and white, respectively. (A) 3,30,40,5,50,7-hexahydroxyflavone (8,

myricetin), (B) apigenin-7-O-glucoside (20, apigetrin), (C) 5-hy-

droxy-30,40,50,6,7,8-hexamethoxyflavone (16, gardenin), and

(D) myricetin-3-O-rhamnoside (30, myricitrin)

Appl Biol Chem (2017) 60(3):227–232 229

123

fragments of the m-, p-positions of the B-ring increase the

inhibitory effects on GSK-3b. The bulky fragment at the

7-position of apigenin-7-O-glucoside (20, apigetrin), which

showed the worst activity, negatively affects the inhibitory

activity (Fig. 1B). The bulky fragments at the 6-position of

the A-ring of 5-hydroxy-30,40,50,6,7,8-hexamethoxyflavone

(16, gardenin) and at the 3-position of the B-ring of myr-

icetin-3-O-rhamnoside (30, myricitrin) also negatively

affect the inhibitory activity (Figs. 1C, D). These findings

provide information about the structural conditions of fla-

vonoids that can enhance the inhibitory effects on GSK-3b.Myricetin, which had the best inhibitory effect on GSK-

3b, is found in various plants such as tea, vegetables, fruits,

and berries [20]. Its antioxidative and anticancer activities

have been studied thoroughly [20, 21]. However, its inhi-

bitory effect on GSK-3b has not been reported. Unlike

most protein kinases, GSK-3b is active in resting condi-

tions and is inactivated by cellular responses. The activity

of GSK-3b is negatively regulated by phosphorylation at

the serine-9 residue in its regulatory N-terminal region

[22]. Thus, increased phosphorylation of this serine-9

residue in GSK-3b is functionally associated with the

inhibition of its kinase activity. To confirm the result

obtained from the in vitro GSK-3b kinase assay, HCT116

human colon cancer cells were treated with different con-

centrations of myricetin, and the phosphorylation status at

serine-9 was then assessed by Western blot analysis using a

phosphospecific antibody. Lithium, a known GSK-3binhibitor, was used as a reference compound. The level of

serine-9 phosphorylation of GSK-3b increased rapidly

within 30 min after LiCl treatment (Fig. 2A), indicating

the rapid inactivation of GSK-3b by lithium. HCT116 cells

were then treated with different concentrations (20, 40, and

80 lM) of myricetin for 30 min, and we measured the

serine-9 phosphorylation of GSK-3b (Fig. 2B). Densito-

metric analysis demonstrated that myricetin caused an

increase in the level of serine-9 phosphorylation of GSK-

3b in a dose-dependent manner. As a result, the data

obtained here suggest that the binding of myricetin to

GSK-3b is associated with the inhibition of GSK-3bactivity.

To elucidate the molecular binding mode between

myricetin and GSK-3b, in silico docking analysis was

carried out. Among sixty-nine 3D structures of human

GSK-3b deposited in the PDB, the crystallographic struc-

ture (1q3d.pdb) containing staurosporine (Fig. S4) as a

Fig. 2 Effect of myricetin on

the serine-9 phosphorylation of

GSK-3b. (A) HCT116 cells

were serum-starved for 24 h,

and 30 mM LiCl was then

added for various time periods.

Whole lysates were prepared,

and western blot analysis was

performed using a

phosphospecific antibody

against serine-9 of GSK-3b.GAPDH was used as an internal

control. (B) HCT116 cells were

serum-starved for 24 h, and

30 mM LiCl or different

concentrations (20, 40, and

80 lM) of myricetin were

added for 30 min. Whole lysates

were prepared, and western blot

analysis was performed as in

(A). The phosphorylated levels

at serine-9 of GSK-3b were

quantified using densitometry,

and the results are shown as the

fold change compared to

unstimulated control cells

(Cont). GAPDH was used as an

internal control

230 Appl Biol Chem (2017) 60(3):227–232

123

ligand was chosen [15]. This structure consists of two

chains, A and B. Chain A was used because its structure

includes two more residues than chain B. First, the ligand

contained in 1q3d.pdb, staurosporine, was docked into the

apo-GSK-3b protein (1q3d-A). The 5th complex showed

the best binding pose, and its binding energy was

-16.67 kcal/mol. Ten hydrophobic interactions and two

hydrogen bonds (H-bonds) were observed in the LigPlot

analysis (Fig. S5). Next, Myricetin was docked to apo-

1q3d-A. Among the 30 complexes of myricetin and 1q3d-

A generated from the 30 iterations, the second complex

showed the best docking pose and lowest binding energy

(-17.87 kcal/mol). The LigPlot analysis revealed five

hydrophobic interactions (Ile62, Val70, Ala83, Leu132,

and Leu188) and four H-bonds (Asp133, Tyr134, and

Val135) (Fig. S6). To compare the binding conditions of

myricetin to GSK-3b with those of staurosporine, 3D

images of the binding sites were generated using the

PyMOL program (The PyMOL Molecular Graphics Sys-

tem, Version 1.3, Schrodinger, LLC). In the complex of

myricetin docked to GSK-3b, four H-bonds were observed,which are expressed as yellow dot circles (Fig. 3A). The

distances between the oxygen in the ketone of Asp133 and

the hydrogen in the 5-hydroxyl group of myricetin, the

oxygen of the hydroxyl group of Tyr134 and the hydrogen

of the 30-hydroxyl group of myricetin, the hydrogen of the

amide of Val135 and the oxygen of the ketone in myricetin,

and the oxygen of the ketone of Val135 and the hydrogen

of the 3-hydroxyl group of myricetin were 2.93, 3.09, 2.81,

and 2.03 A, respectively. In the case of staurosporine, two

H-bonds were observed (Fig. 3B). The distances between

the oxygen of the ketone of Asp133 and the hydrogen of

the nitrogen of the pyrrolone of staurosporine and between

the hydrogen of the amide of Val135 and the oxygen of the

ketone of the pyrrolone of staurosporine were 2.62 and

3.01 A, respectively. The interactions between myricetin

and the binding site of GSK-3b were better than those

between staurosporine and GSK-3b. These results may

explain why the binding energy of myricetin is lower than

that of staurosporine.

Luteolin abolishes GSK-3 activation [23]. Apigenin

induces cell death by inhibition of GSK-3b in BxPC-3 and

PANC-1 human pancreatic cancer cells [24]. Pretreatment

of BxPC-3 human pancreatic cancer cells with luteolin and

apigenin inhibits cell proliferation by reducing the

expression of GSK-3b [24]. The inhibitory effects of 22

flavonoids were measured using the Kinase-Glo Lumines-

cent Kinase Assay. Of these flavonoids, luteolin, apigenin,

and quercetin showed good IC50 values of 1.5, 1.9, and

2.0 lM, respectively [8]. In the paper published previously

[8], the IC50 values of several flavonoids showed several

tens of micromolar level and others were over hundreds of

micromolar level, so that they could not obtain the

structure–activity relationships (SAR) data quantitatively.

Since the IC50 values of 34 flavonoids tested here ranged

between 0.9 and 98 lM, however, we could obtain the

SAR data quantitatively. While the electronegative frag-

ments of the m-, p-positions of the B-ring increase the

inhibitory effects on GSK-3b, the bulky fragments at the 6-

and 7-positions of the A-ring and at the 3-position of the

B-ring negatively affect the inhibitory activity. While the

paper published previously focused on in silico experi-

mental results, the current results showed in vitro biolog-

ical data including Western blot analysis as well as in silico

Fig. 3 (A) 3D image of the binding site surrounding myricetin

docked to GSK-3b generated using the PyMOL program and (B) thatsurrounding staurosporine docked to GSK-3b

Appl Biol Chem (2017) 60(3):227–232 231

123

experimental results. In addition, the structural conditions

of flavonoids that enhance the inhibitory effects on GSK-

3b were identified using hologram QSAR calculations.

These findings will help in the development of novel GSK-

3b inhibitors.

Acknowledgment This paper was supported by Konkuk University

in 2014.

References

1. Takahashi-Yanaga F (2013) Activator or inhibitor? GSK-3 as a

new drug target. Biochem Pharmacol 86:191–199

2. Hur EM, Zhou FQ (2010) GSK3 signalling in neural develop-

ment. Nat Rev Neurosci 11:539–551

3. Kirshenboim N, Plotkin B, Shlomo SB, Kaidanovich-Beilin O,

Eldar-Finkelman H (2004) Lithium-mediated phosphorylation of

glycogen synthase kinase-3beta involves PI3 kinase-dependent

activation of protein kinase C-alpha. J Mol Neurosci 24:237–245

4. Klein PS, Melton DA (1996) A molecular mechanism for the

effect of lithium on development. Proc Natl Acad Sci U S A

93:8455–8459

5. Ryves WJ, Harwood AJ (2001) Lithium inhibits glycogen syn-

thase kinase-3 by competition for magnesium. Biochem Biophys

Res Commun 280:720–725

6. Leclerc S, Garnier M, Hoessel R, Marko D, Bibb JA, Snyder GL,

Greengard P, Biernat J, Wu YZ, Mandelkow EM, Eisenbrand G,

Meijer L (2001) Indirubins inhibit glycogen synthase kinase-3

beta and CDK5/p25, two protein kinases involved in abnormal

tau phosphorylation in Alzheimer’s disease. A property common

to most cyclin-dependent kinase inhibitors? J Biol Chem 276:

251–260

7. Bapteste E (2014) The origins of microbial adaptations: how

introgressive descent, egalitarian evolutionary transitions and

expanded kin selection shape the network of life. Front Microbiol

5:83

8. Johnson JL, Rupasinghe SG, Stefani F, Schuler MA, Gonzalez de

Mejia E (2011) Citrus flavonoids luteolin, apigenin, and quercetin

inhibit glycogen synthase kinase-3beta enzymatic activity by

lowering the interaction energy within the binding cavity. J Med

Food 14:325–333

9. Cho M, Yoon H, Park M, Kim YH, Lim Y (2014) Flavonoids

promoting HaCaT migration: I. Hologram quantitative structure-

activity relationships. Phytomedicine 21:560–569

10. Sharma SK, Kumar P, Narasimhan B, Ramasamy K, Mani V,

Mishra RK, Majeed AB (2012) Synthesis, antimicrobial, anti-

cancer evaluation and QSAR studies of 6-methyl-4-[1-(2-substi-

tuted-phenylamino-acetyl)-1H-indol-3-yl]-2-oxo/thioxo-1,2,3,4-

tetrahydropyrimidine-5-carboxylic acid ethyl esters. Eur J Med

Chem 48:16–25

11. Shin SY, Yoon H, Ahn S, Kim DW, Kim SH, Koh D, Lee YH,

Lim Y (2013) Chromenylchalcones showing cytotoxicity on

human colon cancer cell lines and in silico docking with aurora

kinases. Bioorg Med Chem 21:4250–4258

12. Jung H, Shin SY, Jung Y, Tran TA, Lee HO, Jung KY, Koh D,

Cho SK, Lim Y (2015) Quantitative relationships between the

cytotoxicity of flavonoids on the human breast cancer stem-like

cells mcf7-sc and their structural properties. Chem Biol Drug Des

86:496–508

13. Shin SY, Jung H, Ahn S, Hwang D, Yoon H, Hyun J, Yong Y,

Cho HJ, Koh D, Lee YH, Lim Y (2014) Polyphenols bearing

cinnamaldehyde scaffold showing cell growth inhibitory effects

on the cisplatin-resistant A2780/Cis ovarian cancer cells. Bioorg

Med Chem 22:1809–1820

14. Williams LK, Li C, Withers SG, Brayer GD (2012) Order and

disorder: differential structural impacts of myricetin and ethyl

caffeate on human amylase, an antidiabetic target. J Med Chem

55:10177–10186

15. Bertrand JA, Thieffine S, Vulpetti A, Cristiani C, Valsasina B,

Knapp S, Kalisz HM, Flocco M (2003) Structural characteriza-

tion of the GSK-3beta active site using selective and non-selec-

tive ATP-mimetic inhibitors. J Mol Biol 333:393–407

16. Jung Y, Shin SY, Yong Y, Jung H, Ahn S, Lee YH, Lim Y (2015)

Plant-derived flavones as inhibitors of aurora B kinase and their

quantitative structure-activity relationships. Chem Biol Drug Des

85:574–585

17. Waller CL (2004) A comparative QSAR study using CoMFA,

HQSAR, and FRED/SKEYS paradigms for estrogen receptor

binding affinities of structurally diverse compounds. J Chem Inf

Comput Sci 44:758–765

18. Hwang D, Park HJ, Seo EK, Oh JY, Ji SY, Park DK, Lim Y

(2013) Effects of flavone derivatives on antigen-stimulated

degranulation in RBL-2H3 cells. Chem Biol Drug Des

81:228–237

19. Shin SY, Woo Y, Hyun J, Yong Y, Koh D, Lee YH, Lim Y

(2011) Relationship between the structures of flavonoids and their

NF-kappaB-dependent transcriptional activities. Bioorg Med

Chem Lett 21:6036–6041

20. Ross JA, Kasum CM (2002) Dietary flavonoids: bioavailability,

metabolic effects, and safety. Annu Rev Nutr 22:19–34

21. Ong KC, Khoo HE (1997) Biological effects of myricetin. Gen

Pharmacol 29:121–126

22. Sutherland C, Leighton IA, Cohen P (1993) Inactivation of

glycogen synthase kinase-3 beta by phosphorylation: new kinase

connections in insulin and growth-factor signalling. Biochem J

296(Pt 1):15–19

23. Sawmiller D, Li S, Shahaduzzaman M, Smith AJ, Obregon D,

Giunta B, Borlongan CV, Sanberg PR, Tan J (2014) Luteolin

reduces Alzheimer’s disease pathologies induced by traumatic

brain injury. Int J Mol Sci 15:895–904

24. Johnson JL, Gonzalez de Mejia E (2013) Interactions between

dietary flavonoids apigenin or luteolin and chemotherapeutic

drugs to potentiate anti-proliferative effect on human pancreatic

cancer cells, in vitro. Food Chem Toxicol 60:83–91

232 Appl Biol Chem (2017) 60(3):227–232

123