Pharmacophore identification, virtual screening and biological evaluation of prenylated flavonoids derivatives as PKB/Akt1 inhibitors

Article abstract of DOI:10.1016/j.ejmech.2011.10.006

A total of 24 well-defined PKB/Akt1 inhibitors were used to generate pharmacophore models applying Catalyst/HypoGen program. The best ranked model (Hypo-1) was then validated by cost analysis, prediction capability, Cat-Scramble and receiver operating cha

Full text of DOI:10.1016/j.ejmech.2011.10.006

European Journal of Medicinal Chemistry 46 (2011) 5949e5958
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Pharmacophore identification, virtual screening and biological evaluation of
prenylated flavonoids derivatives as PKB/Akt1 inhibitors
Xiaowu Dong a, Xinglu Zhou b, Hui Jing b, Jianzhong Chen a, Tao Liu a, Bo Yang b, Qiaojun He b,
,
Yongzhou Hu a
*
a ZJU-ENS Joint Laboratory of Medicinal Chemistry, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou 310058, China
b Institute of Pharmacology & Toxicology, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou 310058, China
a r t i c l e i n f o
a b s t r a c t
Article history:
A total of 24 well-defined PKB/Akt1 inhibitors were used to generate pharmacophore models applying
Catalyst/HypoGen program. The best ranked model (Hypo_1) was then validated by cost analysis,
prediction capability, Cat-Scramble and receiver operating characteristic (ROC) studies. Then,
pharmacophore-based virtual screening combined with docking study was performed to search an in-
house compound database. Nine preferable hits 75e80, HTS-02143, BTB-14740 and HTS-08006 were
prepared and biologically evaluated. Several compounds were identified as good PKB/Akt1 inhibitors,
suggesting that Hypo_1 would be reliable and useful in virtual screening. Flow cytometric and western
blotting analysis on compounds 79 and 80 further demonstrated that the inhibition of phosphorylation
Received 14 May 2011
Received in revised form
25 September 2011
Accepted 2 October 2011
Available online 8 October 2011
Keywords:
PKB/Akt1
Pharmacophore model
HypoGen
of PKB/Akt1 and its substrates (such as GSK3
b
) was responsible for their cytotoxic activities.
Ó 2011 Elsevier Masson SAS. All rights reserved.
Molecular docking
Virtual screening
1. Introduction
pyridineepyrazolopyridine [17] and oxindoleepyridine [18].
However, drawbacks of this kind of inhibitors as a clinical agent lie in
their short half-life in animals and poor oral bioavailability [17].
Therefore, it is still greatly interested and valued in searching novel
chemical scaffolds with PKB/Akt inhibitory activity.
PKB/Akt, the serine/threonine proteases, belongs to the AGC
family of kinases. It is a central node of the PI3K/Akt signaling
pathway which is one of the most frequently mutated or over
expressed signaling abnormality in cancers [1e4]. PKB/Akt can be
classified into three subfamilies: PKB/Akt1, PKB/Akt2 and PKB/Akt3
[5]. PKB/Akt have a high degree of overall homology being approx-
imately 80% identical [6], and share similar downstream targets, but
differ in the levels of expression and activation in various tumors [7].
Inhibition of PKB/Akt alone or in combination with other cancer
chemotherapeutics can result in increasing apoptosis of cancer cells,
decreasing tumor growth and reversing tumor resistance [8e10]. All
of these findings led to extensive studies of PKB/Akt inhibitors as
anticancer agents. In general, there are three strategies applied to
development of novel PKB/Akt inhibitors, targeting toward pleck-
strin homology (PH) domain [11], ATP-binding kinase domain [12],
and hinge-region domain [13]. Among these approaches, the most
highly explored strategy is to target ATP-binding kinase domain,
leading to discover lots of potent PKB/Akt inhibitors, such as bis-
pyridinylethylenes [14], isoquinoline (indazole)epyridines [15,16],
Pharmacophore-based virtual screening is a rational strategy for
identification of novel hits or leads with diverse chemical scaffolds
[19,20]. There are several methods for generation of pharmaco-
phore models, such as HypoGen [21], MOE [22], Phase [23] and
GALAHAD [24]. HypoGen is one of the most widely used
approaches in pharmacophore-based drug design. Böhm et al.
identified several DNA gyrase inhibitors with novel scaffolds using
Catalyst and LUDI programs to screen Available Chemical Dictio-
nary (ACD) and Roche compound inventory (RCI) database [25].
Besides, Catalyst/HypoGen was applied by Yu et al. to establish
a pharmacophore model of KDR kinase inhibitors and successfully
discovered a novel hit in screening the Traditional Chinese Medi-
cine Database (TCMD) [26]. Nevertheless, few reports about
pharmacophore-based virtual screening for discovering novel PKB/
Akt inhibitors are available.
In the present study, a novel pharmacophore model was
proposed based on a set of pyridinyl-bridged PKB/Akt1 inhibitors
using Catalyst/HypoGen program (Catalyst 4.1, Molecular Simula-
tions Inc., San Diego, CA). The pharmacophoric features in Hypo_1
was then demonstrated to be consistent with the interaction mode
* Corresponding author. Tel./fax: þ86 571 888208460.
E-mail addresses: huyz@zju.edu.cn, huyz@zjuem.zju.edu.cn (Y. Hu).
0223-5234/$ e see front matter Ó 2011 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2011.10.006

5950
X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
proposed in molecular docking, which was carried out to evaluate
the molecular interactions between the hits and PKB/Akt1. Finally,
the preferably screened hits were picked up and biologically eval-
uated to validate robustness and predictive ability of the virtual
screening model.
the default parameters in the Catalyst molecular modeling package
(Catalyst 4.10 documentation).
An initial analysis of the “show function mapping” tools indi-
cated that five kinds of chemical features, including H-bond
acceptor (HBA), H-bond donor (HBD), hydrophobic (HY), aromatic
(RA) and positive ionizable (NI) features, effectively map all critical
pharmacophoric features of all molecules in both training and test
sets. Based on all of these features of 24 compounds in the training
set, 10 pharmacophore models were generated using Catalyst/
HypoGen program (Catalyst 4.1, Molecular Simulations Inc., San
Diego, CA) with the default uncertainty value of 3.0. The uncer-
tainty value represents a ratio range of uncertainty in the activity
value based on the expected statistical irregularities of biological
data collection.
2. Computational methods
2.1. Generation of pharmacophore models
For the pharmacophore modeling studies,
a set of 74
compounds were selected from the literatures [14e16,27]. PKB/
Akt1 inhibitory activity data (IC50, Table 1) of them span over 6
orders of magnitude (from 0.16 to 32250 nM). The 2D structures of
these compounds were illustrated in Fig. 1. The dataset was split
into a training set of 24 compounds and a test set of 50 compounds.
Besides, the 1940 negatives compounds used in receiver operating
characteristic (ROC) were retrieved from Available Chemical
Directory (ACD) database (Symyx Technologies, Santa Clara, CA)
using “Random Percent Filter protocol” by Pipeline Pilot software
(SciTegic, Inc., San Diego, CA). All the compounds were optimized in
Discovery Studio 2.0 software (Accelrys, Inc. San Diego, CA) using
the CHARMm-like force field [28]. Then, a representative family of
conformations was generated for each compound using the Poling
Algorithm [29] and the ‘best conformational analysis’ method with
2.2. Pharmacophore model quality examination
In order to evaluate the reliability and accuracy of the generated
3D pharmacophore models, cost analysis, prediction capability,
Cat-Scramble validation and receiver operating characteristic (ROC)
studies were performed.
2.2.1. Cost analysis
The quality of HypoGen-generated pharmacophore models can
be described in terms of fixed cost, null cost and total cost, all of
which are well-defined by Debnath [30]. The fixed cost represents
the simplest model that fits the data. The null cost represents the
cost of a hypothesis with no features which estimates every activity
to be the average activity. If a returned cost (total cost) differs from
the null cost by 40e60 bits, it is highly probable that the hypothesis
has 75e90% chance representing the true correlation of the data. If
the difference becomes more than 60 bits, the probability of rep-
resenting the true correlation of the data are even higher.
Table 1
The experimental data and Hypo_1-predicted IC50 values of compounds 1e74.
Cmpd.
PKB/Akt1
IC50(nM)
Error
factor
Cmpd.
PKB/Akt1
IC50(nM)
Error
factor
Exp.
Pred.
Exp.
Pred.
1a,b
2a,b
3a,b
4a,b
5b
0.16
1.8
2.9
2.6
1.5
3.9
18
13.3
18
3.6
49
39
13
7.7
57
12.1
3.3
126
14
6.8
0.42
5.6
13
0.91
5.5
10
34
7.6
38
6.2
4
7.6
1.2
8.6
13
2.6
3.1
4.6
38a
39
40
41
42
14170
5290
4040
6760
8020
4040
875
488
312
473
685
237
1232
92
1117
215
1659
4022
1503
331
147
7.4
2800
8200
1800
760
1400
1800
690
310
660
440
490
1600
190
230
430
540
540
410
590
940
20
ꢀ5.1
1.5
2.2.2. Prediction capability of the generated pharmacophore model
The pharmacophore model was also used to predict the PKB/Akt
inhibitory activities of other compounds beyond in the training set.
For such a purpose, 50 PKB/Akt1 inhibitors in the test set were
estimated by the best pharmacophore model which has the highest
correlation coefficient (r), lowest total cost, and lowest RMSD value.
ꢀ2.2
ꢀ8.8
ꢀ5.7
ꢀ2.2
ꢀ1.3
ꢀ1.6
2.1
ꢀ1.1
ꢀ1.4
6.8
ꢀ2.9
3.7
6b
2.6
1.9
43
44
7b
8a,b
9a,b
10
ꢀ1.7
45a
46a
47a
48a
49
2.1
1.7
11
ꢀ12
ꢀ5.1
ꢀ11
1.1
ꢀ4.4
ꢀ1.2
ꢀ1.6
1.7
2.2.3. Cat-Scramble validation
12
13b
14b
15
50
ꢀ6.6
The Cat-Scramble validation procedure is based on Fischer’s
randomization test [31]. It is to check whether there is a strong
correlation between chemical structures and bioactivities of
ligands. It is done by randomizing the activity data associated with
the training set of compounds and by generating pharmacophore
hypotheses using the common chemical features and parameters to
develop the original pharmacophore hypothesis. 19 random
spreadsheets are generated for 95% confidence level. If the
randomized data set has the similar or better cost values, RMSD and
correlation in the generation of a pharmacophore model, the
original hypothesis is considered to have been generated by chance.
51b
52
2.5
ꢀ2.6
2.5
ꢀ3.1
ꢀ9.8
ꢀ2.5
2.8
16b
17
10
2
210
49
1.4
52
53
54
18
55
19b
20b
21b
22a,b
23a,b
24
3.5
56a
57
ꢀ4.9
ꢀ1.5
2.3
78
3.2
58
ꢀ7.5
7.4
310
430
600
320
630
270
570
350
88
580
850
62000
5500
560
2300
59b
60a
61
12
1.6
180
1020
52.7
121
2670
953
6450
6350
176
147
3340
23910
32250
146
5080
1.8
223
760
59
13
6.1
1200
1100
60
95
50
1000
290
1100
6400
2400
1200
400
5200
980
380
5.3
1.5
1.0
7.3
8.3
2.8
10
4.1
ꢀ2.4
25b
26a
27a
28a
29
11
62b
63a
64b
65
2.7
ꢀ4.2
ꢀ3.6
ꢀ11
ꢀ18
ꢀ2.0
3.9
2.2.4. Receiver operating characteristic (ROC) studies
360
28
An in-house database was retrieved randomly from Available
Chemicals Directory (ACD) database (1940 molecues) spiked some
known PKB/Akt1 inhibitors (24 molecues, Table 1). The receiver
operating characteristic (ROC) study using the Hypo_1 to screen
this in-house database were further performed [32]. True positive
rate (TPR) and false positive rate (FPR) from a comparison between
in vitro and in silico activities were calculated.
66b
67
30
31
32
33
278
3200
668
300
1100
23270
190
198
68
69
2.0
3.6
ꢀ3.9
70
ꢀ0.6
ꢀ2.8
ꢀ4.5
ꢀ0.71
1.9
34a
35a
36a
37a
2.6
71a
72a
73
ꢀ5.9
3.9
ꢀ2.2
74
TP
TPR ¼
a
The compounds were used as test set.
The compounds were used as active compounds (positive) in ROC studies.
b
TP þ FN

X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
5951
Fig. 1. The 2D structures of 74 PKB/Akt1 inhibitors used to generate the pharmacophore model.
where TP is the number of true positive compounds, FN is the
number of false negative compounds.
where TPR(x) is the percent of the true positives versus the total
positives at rank position x, FPR(x) is the percent of the false posi-
tives versus the total negatives at rank position x.
FP
FPR ¼
TN þ FP
2.3. Molecular docking with FlexiDock program
where TN is the number of true negative compounds, FP is the
number of false positive compounds.
The X-ray crystal structure of PKB/Akt1 (PDB code: 3COB) was
retrieved from the Protein Data Bank (PDB, http://www.pdb.org).
ATP competitive binding pocket was defined to cover all residues
within 4 Ǻ of the ligand in the crystal PKB/Akt1-ligand complex. All
single bonds of residue side chains inside the defined binding
pocket were regarded as flexible. The compounds were allowed to
rotate on all of single bonds and move flexibly within the tentative
The ROC curve is a function of FPR versus the TPR, and the area
under the ROC curve (AUC) value is the important way of measuring
the performance of the test.
N
X
AUC ¼
TPRðxÞ½FPRðxÞ ꢀ FPRðx ꢀ 1Þꢁ
x ¼ 2

5952
X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
binding pocket. The atomic charges were recalculated using the
Kollman all-atom approach for the protein.
randomly, and the resulting data set were used for HypoGen
generation. All parameters were adopted and used in the initial
HypoGen calculation. This procedure was reiterated 19 times. As
shown in Fig. 2C, none of the outcome hypotheses has a lower cost
score than the initial hypothesis (Hypo_1), suggesting that there is
a strong correlation between chemical structures and biological
activities for Hypo_1 and it is not generated by chance.
3. Results and discussion
3.1. HypoGen-generated pharmacophore models
Table 2 displays a set of 10 pharmacophore hypotheses gener-
ated from the training set of 24 PKB/Akt1 inhibitors using Catalyst/
HypoGen. Among these 10 hypotheses, Hypo_1 is characterized by
the highest cost difference of 75.68, the lowest RMSD of 0.987, and
the highest correlation coefficient of 0.948. Noticeably, the cost
range between Hypo_1 and the fixed cost is 17.03, while that
between the null hypothesis and Hypo_1 is 75.68, showing that
Hypo_1 has more than 90% probability of correlating the data.
Fig. 2A shows the topological features of Hypo_1 which consists of
four pharmacophore features, including one H-bond acceptor (HBA,
colored by green), two H-bond donors (HBD, colored by purple),
and one hydrophobic group (HY, colored by cyan).
The predicted results of PKB/Akt inhibitory activities (IC50
values) of the 24 training set molecules are listed in Table 1, and the
plots of predicted versus experimental values are recorded in
Fig. 2B. It is demonstrated that the activities of most compounds are
predicted correctly. For example, as shown in Fig. 2A, the super-
3.2.3. Receiver operating characteristic (ROC) studies
The Hypo_1 was further validated for picking up PKB/akt1
inhibitors from a database, including 1940 inactive compounds and
24 known PKB/akt1 inhibitors (labeled in Table 1). The ROC curves
(Fig. 2D) was obtained with good AUC value of 0.957, which was
equal to the probability that a classifier will rank a randomly chosen
positive instance much higher than a randomly chosen negative
one. Hypo_1 is a useful and reliable tool for identifying novel PKB/
Akt inhibitors in virtual screen.
3.3. Interaction mode between PKB/Akt1 and ligands proposed by
molecular docking
Although a number of sophisticated techniques are available to
benchmark docking success rate, one particular method has been
the most widely used where one defines the successfully docked
ligand as a top-scored pose with small RMSD value from the
reference coordinates. Herein, we compared with the re-docking
results using FlexiDock [33], FlexX [34], LigandFit [35], CDOCKER
[36] and Libdock [37] base on six PKB/Akt1eligand complex (PDB
code: 3CQU, 3CQW, 3MV5, 3MVH, 3OCB and 3OW4) that was
retrieved from the Protein Data Bank (PDB, http://www.pdb.org).
Each RMSD value of proposed ligands from the reference coordi-
nates was recorded, and the average RMSD value was calculated as
the evaluation criteria for selection of the most preferable docking
program. As shown in Fig. 3a, it is obvious that FlexiDock is the best
program for studying the interaction mode between PKB/Akt1 and
ligands (average RMSD is 0.44).
imposition of Hypo_1 with the most active compound
1
(IC50 ¼ 0.16 nM) indicated that all Hypo_1’s features of HBA, HBD,
and HY are mapped very well onto the compound 1 (prediction
IC50: 0.41 nM; actual IC50: 0.16 nM). In the case of 34, the HBD1 and
HBA features cannot fit well to the corresponding pharmacophore
features while the HBD2 feature is missed (Fig. 2A), leading to
a poor predicted activity but close to its actual activity (prediction
IC50: 62,000 nM; actual IC50: 23,910 nM).
3.2. Prediction capability of the generated pharmacophore model
3.2.1. Test set prediction
Furthermore, the interaction between compound 1 and PKB/
Akt1 (PDB entry code: 3OCB) was studied using FlexiDock
program. The simulated model of compound 1-PKB/Akt1 complex
is depicted in Fig. 3b, showing hydrophobic and hydrogen bond
interaction between the ligand and protein. Two H-bonds are
formed between indazole moiety of compound 1 and Glu228 and
Ala 230 of PKB/Akt1 with a distance of 2.13 Ǻ and 1.43 Ǻ,
respectively. 2-NH2 group in compound 1 can form H-bonds with
Asp292 of PKB/Akt1 with a distance of 2.58 Ǻ. In the fourth H-
bond formed between ligand and protein, compound 1 provides
H-bond acceptor nitrogen atom in pyridinyl ring to form H-bond
interaction with Lys179 of PKB/Akt1 with the H-bond length of
1.63 Ǻ. In addition, the aromatic stack interaction between the
indole ring of compound 1 and Phe161 of PKB/Akt1 serves to
increase the thermal stability of the complex. Such an interactions
mapping is consistent with our previous discussion about Hypo_1,
in which there are four chemical features, including one HBA, two
HBD, and one HY, playing critical roles in the interaction between
PKB/Akt1 and ligand.
In order to verify whether Hypo_1 can also predict the activities
of compounds beyond the training set, we applied a total of 50
compounds representing diverse activity classes and different
structural groups as test set. As shown in Fig. 2B, there is a good line
correlation between the experimental and predicted PKB/akt1
activities of the test set compounds, showing a good correlation
coefficient of 0.803. Thus, Hypo_1 has a good prediction capability
for both training set and test set compounds.
3.2.2. Cat-Scramble validation
With the aid of the Cat-Scramble program, the experimental
activities of compounds in the training set were scrambled
Table 2
Information of statistical significance and predictive power presented in cost values
measured for top 10 hypotheses as a result of HypoGen generation process.
Hypo. no.
Total cost
Dcost
RMSD
Correl.
1
2
3
4
5
6
7
8
9
10
111.47
117.23
118.10
112.77
123.02
123.72
124.86
125.48
125.61
125.99
75.68
69.92
69.05
74.38
64.13
63.43
62.29
61.67
61.54
61.16
0.987
1.275
1.288
1.514
1.507
1.415
1.489
1.598
1.577
1.597
0.948
0.906
0.904
0.861
0.863
0.884
0.869
0.843
0.849
0.845
3.4. Database screening for novel PKB/Akt1 inhibitors
Hypo_l was employed as a 3D search query to screen the in-
house database consisting of the compounds synthesized in our
lab (1024 molecules) and from Maybridge database (w60,000
molecules). Over 80 hits fitted well with features of Hypo_1 were
predicted as potential candidates for further exploration. Consid-
ering that pharmacophore mapping and molecular docking
complement each other and can be synergistically integrated to
Null cost of 10 top-scored hypotheses is 187.15, fixed cost value is 94.43 and
configuration cost is 12.59.

X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
5953
Fig. 2. Hypo_1 was established and validated by cost analysis, prediction capability, Cat-Scramble and receiver operating characteristic (ROC) studies. (A) Hypo_1 and its alignment
to representative compounds. The topological features of Hypo_1: four pharmacophoric features, including one hydrophobic group (Cyan), one H-bond acceptor (Green), and two
H-bond donors (Purple) (a); Hypo_1 aligned to compound 1 (b); Hypo_1 aligned to compound 37 (c); (B) the prediction ability of Hypo_1. The regression of actual versus predicted
activities by the Hypo_1 for training set (a); the regression of actual versus predicted activities by the Hypo_1 for test set inhibitors (b); (C) the difference in costs between the
HypoGen runs (Hypo_1) and the scrambled runs (RM1 w RM19); (D) ROC curves of the database screening Hypo_1. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
improve the drug design and development process, thus, all the hits
obtained by Hypo_1 were further evaluated by docking studies to
check their shape fits in the binding site of protein using FlexiDock.
Finally, a total of nine preferable compounds 75e80, HTS-02143,
BTB-14740 and HTS-08006 (Fig. 4) were identified as preferable
hits in both Hypo_1-based screening and molecular docking
studies, the screening results of which are listed in Table 3. In case
of compounds 79 and 80, both of them showed good Fit-value (79:
10.8 and 80: 10.4) and low docking energy (79: ꢀ3545.0 kal/mol
and 80: ꢀ3766.5 kal/mol) in virtual evaluation, leading to a good
predicted activity for them. As shown in Fig. 5a,b, 7-hydroxyl and
30-hydroxyl of 79 (or 7-hydroxyl and 40-hydroxyl of 80) were
mapped well on HBD_1 and HBD_2 of Hypo_1, while 4-carbonyl
and C-6 prenyl of 79 (or 4-carbonyl and C-8 prenyl of 80) were
mapped well on the HBA and HY features of Hypo_1. As shown in
Fig. 5c,d, three hydrogen bond interactions are formed between
PKB/Akt1 (Lys179, Glu228 and Ala230) and 30,40-dihydroxyl and 4-
carbonyl groups of 79 (or 80). In addition, the hydrophobic inter-
action is observed between C-6 prenyl group of 79 and phenyl
group of Phe161, and between C-8 prenyl group of 80 and phenyl
group of Phe479.
3.5. Chemistry
In order to validate the reliability of a combination of Hypo_1 and
docking studies, all the hits were selected for further studies,

5954
X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
Fig. 3. Interaction mode between PKB/Akt1 and ligands proposed by molecular docking. (a) Redock studies on six PKB/Akt1-ligand complex using Libdock, LigandFit, FlexX,
CDOCKER and Fleixidock programs; (b) Interaction mode between PKB/Akt1 and compound 1 proposed by FlexiDock, hydrogen bond interaction between ligand and enzyme were
highlighted using yellow line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
including preparation and biological evaluation for their PKB/Akt1
inhibitory activities. HTS-02143, BTB-14740 and HTS-08006 were
purchased from Maybridge chemical (Cornwall, United Kingdom).
The synthesis of compounds 75, 76 and 80 were previously reported
[38,39], while the synthesis of compounds 77e79 was reported in
present study as illustrated in Scheme 1. Prenylchalcones 83 were
prepared by Claiseneschmidt condensation of compound 81 with
appropriate benzaldehyde 82 in aqueous alcoholic alkali solution.
Prenylflavanones 84were obtained bycyclization ofcompounds 83 in
a solution of sodium acetate in ethanol. Finally, prenylflavones 77e79
were obtained by dehydrogenation in the presence of I2 and pyridine
and then demethoxymethylation in catalytic amount of 3N HCl in
MeOH/THF(1/1, v/v)usingcorrespondingflavanones 84, successively.
02143 displayed weak activity. Compounds 79 and 80 showed the
best PKB/Akt1 inhibitory activities (IC50 of 79 and 80 were 5.4 and
3.9 mM), which were comparable to that of H-89. Insight into the
observed effects of different substituents of prenylated flavonoids
revealed that the 5-OH on A-ring of prenylflavone (78 is much more
potent than 77), and 30,40-(OH)2 on B-ring of prenylflavone (79, 80
are potent than 78) are essential for their PKB/Akt1 inhibitory
activities, while the position (C-6 or C-8) of prenyl group at A-ring of
prenylflavone has a little effect on the PKB/Akt1 inhibitory activities
as exemplified in compounds 79 and 80. Interestingly, prenyl-
flavones were firstly reported to exhibit significantly PKB/Akt1
inhibitory activities, and would be promising leads for further
structural optimization and pharmacological studies.
3.6. Pharmacology
3.6.2. Cytotoxic activity against cancer cell lines
The cytotoxic activities of compounds 79 and 80 against cancer
cell lines (PC3, OVCAR-8 and HL-60) was further investigated, since
that PKB/Akt1 is an anti-apoptotic protein kinase and the blockade
of its activity leads to cell death. As expected, compounds 79 and 80
showed good anti-proliferative activities against PC3 (79:
3.6.1. PKB/Akt1 inhibitory activities
PKB/Akt1 inhibitory activities of screened hits were evaluated
using HTScanÒ PKB/Akt1 Kinase Assay Kit (Cell Signaling Tech-
nology, Beverly, MA), and compound H-89, a known PKB/Akt1
inhibitor [40] was used as positive control. Table 3 lists the biological
assay results, indicating that five compounds 75, 76 and 78e80
showed moderate or goodPKB/Akt1 inhibitoryactivities, while HTS-
IC50 ¼ 14.6
80: IC50 ¼ 16.6
IC50 ¼ 3.31 M) cell lines (Fig. 6a). Apoptosis was assessed using
m
M; 80: IC50 ¼ 18.6
m
M), OVCAR-8 (79: IC50 ¼ 23.8
mM;
M; 80:
m
M) and HL-60 (79: IC50 ¼ 2.50
m
m
Fig. 4. The 2D structures of preferable screening hits 75e80, HTS-02143, BTB-14740 and HTS-08006.

X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
5955
Table 3
79 and 80 resulted in down-regulation of the phosphorylation of
PKB/Akt (Ser-473). Furthermore, the effects on PKB/Akt down-
The Fit-values, docking energies and PKB/Akt1 inhibitory activities of compounds
75e80, HTS-02143, BTB-14740, HTS-08006 and H-89.
stream substrates were next tested. GSK3
downstream phosphorylation targets of PKB/Akt. Upon treatment
with compound 79 and 80 in OVCAR-8 cells, GSK3 phosphoryla-
tion was inhibited, while no significant changes in overall total
GSK3 and GAPDH protein levels were observed (Fig. 6c).
b is one of the major
Compd.
Fit-valuea
Docking Energy
(Kcal/mol)
PKB/Akt1 inhibitory
activity IC50 (mM)
b
75
76
77
78
79
80
10.2
9.83
7.84
8.06
10.8
10.4
8.65
8.43
9.63
/
ꢀ2806.3
ꢀ2965.4
ꢀ2828.6
ꢀ3319.4
ꢀ3545.0
ꢀ3766.5
ꢀ1608.6
ꢀ1879.5
ꢀ2016.9
/
85.9
143.9
no active
54.48
5.4
b
4. Conclusion
3.9
HTS-02143
BTB-14740
HTS-08006
H-89
>50 (19.6%)
no active
no active
1.8
In this study, we described a rational strategy for identifying
novel PKB/Akt1 inhibitors using a pharmacophore-based virtual
screen protocol. The best pharmacophore model (Hypo_1) was
established using Catalyst/HypoGen program, and showed good
statistical parameters in validation process. The Hypo_1 was
further employed as a 3D search query to screen our in-house
compounds database. Moreover, molecular docking studies were
also performed to improve the reliability and accuracy of virtual
screening. Then, nine preferable hits were prepared and biological
evaluated for their PKB/Akt1 inhibitory activities. Interestingly, five
of the hits exhibited moderate or good PKB/Akt1 inhibitory activi-
ties, indicating that the pharmacophore-based virtual screening is
a rational strategy to discover novel PKB/Akt1 inhibitors. To the best
of our knowledge, the 3D alignment and pharmacophoric features
of PKB/Akt1 inhibitors and its application in virtual screening were
evaluated for the first time. Compounds 79 and 80, the most potent
one, would be good leads to do further structureeactivity research
a
Fit-value indicates how well the features in the pharmacophore overlap the
chemical features in the molecule.
propidium iodide (PI) staining of the sub-G1 cell population, which
gains prominence later in apoptosis. Exposure of OVCAR-8 a cells to
79 and 80 for 72 h led to 40.19% and 49.11% cells undergoing
apoptosis, respectively (Fig. 6b).
Based on the aforementioned experimental results, we
presumed that the inhibition of PKB/Akt pathway contributed to
the cytotoxic activities of compounds 79 and 80. Cell-based assays
were conducted to evaluate the effect of compound 79 and 80 on
PKB/Akt phosphorylation. Antibodies specific for PKB/Akt phos-
phorylated at residue Ser-473 were used in immunoblotting
experiments. Treatment of OVCAR-8 cells with both of compound
Fig. 5. The virtual evaluated results of preferable hits 79 and 80 by Hypo_1 and molecular docking studies. (a) The compound 79 was aligned to Hypo_1; (b) the compound 80 was
aligned to Hypo_1; (c) interaction mode between PKB/Akt1 qand compound 79, hydrogen bond interaction between ligand and enzyme were highlighted using yellow line; (d)
interaction mode between PKB/Akt1 and compound 80. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

5956
X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
Scheme 1. The synthetic route for compounds 77e79. Reagents and conditions: (a) 10% KOH (H2O/EtOH ¼ 1/4, v/v); (b): NaOAc, EtOH, reflux; (c) I2, pyridine, 90 ꢂC; (d): 3N HCl,
MeOH/THF, reflux.
to find high potent inhibitors, since that the scaffold was the first
time reported to exhibit PKB/Akt1 inhibitory activities.
atmosphere at room temperature for 36 h. Then, the reaction
mixture was poured into iceewater, acidified to pH w5 with 2 N
HCl, and extracted with ethyl acetate. The organic phase was
washed with brine, dried over anhydrous Na2SO4 and concen-
trated in vacuo. The residue was purified using silica gel column
chromatography (petroleum ether:ethyl acetate ¼ 8:1, v/v) to give
desired compounds 83.
5. Experimental
5.1. Chemistry
Melting points were obtained on a B-540 Büchi melting point
apparatus and are uncorrected. 1H NMR spectra and 13C NMR were
recorded on a 400 MHz 1H and 500 MHz 1H (125 MHz 13C) spec-
5.1.1.1. 2-Hydroxy-20,4,40-trimethoxymethoxy-5-(3,3-dimethylallyl)-
chalcone (83a). The product was obtained as a yellow oil (62%). 1H
NMR (CDCl3,
d
): 1.74 (s, 3H), 1.76 (s, 3H), 3.29 (d, 2H, J ¼ 6.8 Hz), 3.48
tromter, respectively. Chemical shifts are expressed as
d values
(s, 3H), 3.50 (s, 3H), 3.52 (s, 3H), 5.21 (s, 2H), 5.25 (s, 2H), 5.28 (s,
2H), 5.29 (m, 1H), 6.64 (s, 1H), 6.76 (dd, 1H, J ¼ 2.0, 8.0 Hz), 6.87 (d,
1H, J ¼ 2.0 Hz), 7.58 (d, 1H, J ¼ 16.0 Hz), 7.61 (d, 1H, J ¼ 8.0 Hz), 7.62
(s, 1H), 8.17 (d, 1H, J ¼ 16.0 Hz), 13.41 (s, 1H, OH). ESI-MS: m/z
[M þ H]þ 473.
relative to tetramethylsilane (TMS). Mass spectral data were ob-
tained on an Esquire-LC-00075 spectrometer. Elemental analyses
were performed on a Flash EA 1112 elemental analyzer. Compounds
75, 76 and 80 were previously synthesized [32].
5.1.1. General procedure for the synthesis of compounds (83aec)
To a cold solution of the acetophenone 81 (1.55 mmol) and
appropriate benzaldehyde 82 (1.62 mmol) in H2OeEtOH (1/4, v/v,
3 mL), KOH (600 mg) in H2OeEtOH (1/4, v/v, 3 mL) was added
with stirring. The resulting mixture was stirred under N2
5.1.1.2. 2-Hydroxy-20,4,40,6-tetramethoxymethoxy-5-(3,3-
dimethylallyl)-chalcone (83b). The product was obtained as
a yellow oil (52%). 1H NMR (CDCl3,
d): 1.69 (s, 3H,), 1.81 (s, 3H), 3.34
(d, 2H, J ¼ 6.8 Hz), 3.48 (s, 3H), 3.50 (s, 3H), 3.53 (s, 3H), 3.54 (s, 3H),
Fig. 6. The cytotoxic activities of 79 and 80 and the preliminary action mechanisms. (a) Cytotoxic activities of 79 and 80 against HL-60, OVCAR-8 and PC-3 cell lines; (b) cell
apoptotic index analyses of OVCAR-8 cells exposed to 79 and 80; (c) treatment with 79 and 80-caused decrease of PKB/Akt and GSK3 phosphorylation in OVCAR-8 cells.
b

X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
5957
5.20 (m, 1H), 5.20 (s, 2H), 5.23 (s, 2H), 5.27 (s, 4H), 6.43 (s, 1H,), 6.76
(dd, 1H, J ¼ 2.0, 8.0 Hz), 6.83 (d, 1H, J ¼ 2.0 Hz), 7.63 (d, 1H,
J ¼ 8.0 Hz), 7.82 (d, 1H, J ¼ 15.6 Hz), 8.17 (d, 1H, J ¼ 15.6 Hz), 13.88 (s,
1H). ESI-MS: m/z [M þ H]þ 533.
the solvent, the residue was purified using silica gel column chro-
matography (petroleum ether: ethyl acetate ¼ 2:1, v/v) to give
desired compounds 77e79.
5.1.3.1. 20,40,7-Trihydroxy-6-(3,3-dimethylallyl)-flavone
yellow solid (68%), mp: 192e194 ꢂC; 1H NMR (Acetone-d6,
(77). Pale
): 1.74
5.1.1.3. 2-Hydroxy-30,4,40,6-tetramethoxymethoxy-5-(3,3-dimethyla-
d
llyl)-chalcone (83c). The product was obtained as a yellow oil
(s, 3H),1.76 (s, 3H), 3.41 (d, 2H, J ¼ 7.5 Hz), 5.40 (m,1H), 6.55 (dd,1H,
J ¼ 2.0, 8.5 Hz), 6.61 (d, 1H, J ¼ 2.5 Hz), 7.02 (s, 1H), 7.08 (s, 1H), 7.78
(d, 1H, J ¼ 8.5 Hz), 7.83 (s, 1H), 8.86 (s, 1H, OH), 9.40 (s,1H, OH), 9.58
(65%). 1H NMR (CDCl3,
d): 1.68 (s, 3H,), 1.76 (s, 3H), 3.32 (d, 2H,
J ¼ 6.8 Hz), 3.47 (s, 3H), 3.52 (s, 6H), 3.53 (s, 3H), 5.21 (m, 1H), 5.25
(s, 2H), 5.27 (s, 2H), 5.28 (s, 4H), 6.45 (s, 1H),7.17 (d, 1H, J ¼ 8.8 Hz),
7.21 (dd, 1H, J ¼ 8.8, 1.6 Hz), 7.50 (d, 1H, J ¼ 1.6 Hz), 7.73 (d, 1H,
J ¼ 15.6 Hz), 7.84 (d, 1H, J ¼ 15.6 Hz), 13.68 (s, 1H, OH). ESI-MS: m/z
[M þ H]þ 533.
(s, 1H, OH). 13C NMR (Acetone-d6,
d): 181.2, 166.5,162.3, 160.1, 157.8,
136.2, 131.5, 130.1, 126.5, 124.6, 123.9, 118.2, 112.1, 109.7, 107.6, 104.3,
102.1, 27.6, 22.2, 18.1. ESI-MS: m/z [M þ H]þ 339. Anal. calcd. for
C20H18O5: C, 70.99; H, 5.36; Found C, 70.87; H, 5.49.
5.1.2. General procedure for the synthesis of compounds (84aec)
A solution of compound 83 (0.8 mmol) and sodium acetate
(500 mg) in ethanol (5 mL) containing 3 drops of water was
refluxed for 24 h. The mixture was poured into cold water and
extracted with ethyl acetate. The organic phase was washed with
brine and dried over anhydrous Na2SO4. After removing the solvent,
the residue was purified using silica gel column chromatography
(petroleum ether/ethyl acetate ¼ 4:1) to afford the desired
compounds 84.
5.1.3.2. 20,40,5,7-Tetrahydroxy-6-(3,3-dimethylallyl)-flavone
(78). Pale yellow solid (41%), mp: >240 ꢂC (des.); 1H NMR
(Acetone-d6,
d
): 1.60 (s, 3H), 1.73 (s, 3H), 3.30 (d, 2H, J ¼ 7.2 Hz),
5.23 (m, 1H), 6.49 (dd, 1H, J ¼ 2.0, 8.0 Hz), 6.51 (s, 1H), 6.55 (d, 2H,
J ¼ 2.0 Hz), 7.00 (s, 1H), 7.75 (d, 1H, J ¼ 8.0), 8.96 (s, 1H, OH), 9.45
(s,1H, OH), 9.50 (s, 1H, OH), 13.35 (s, 1H, OH). 13C NMR (Acetone-d6,
d): 182.1, 163.6, 162.3, 161.7, 159.2, 158.9, 156.1, 130.8, 129.6, 123.3,
110.6, 109.7, 108.7, 106.8, 105.9, 103.2, 94.6, 25.5, 21.2, 17.8. ESI-MS:
m/z [M þ H]þ 355. Anal. calcd. for C20H18O6: C, 67.79; H, 5.12; Found
C, 67.66; H, 5.30.
5.1.2.1. 20,40,7-Trimethoxymethoxy-6-(3,3-dimethylallyl)-flavanone
(84a). The product was obtained as a pale yellow solid (51%),
5.1.3.3. 30,40,5,7-Tetrahydroxy-6-(3,3-dimethylallyl)-flavone
mp.:57e59 ꢂC. 1H NMR (CDCl3,
d): 1.71 (s, 3H), 173 (s, 3H), 2.76 (dd,
(79). Pale yellow solid (35%), mp: >210 ꢂC (des.); 1H NMR
1H, J ¼ 3.0, 17.0 Hz), 2.91 (dd, 1H, J ¼ 13.0, 17.0 Hz), 3.28 (d, 2H,
J ¼ 7.0 Hz), 3.45 (s, 3H), 3.47 (s, 3H), 3.49 (s, 3H), 5.17 (s, 2H), 5.18 (s,
2H), 5.23 (s, 2H), 5.26 (m, 1H), 5.76 (dd, J ¼ 3.0, 13.0 Hz), 6.69 (s, 1H),
6.79 (dd, 1H, J ¼ 2.0, 8.5 Hz), 6.85 (d, 1H, J ¼ 2.0 Hz), 7.50 (d, 1H,
J ¼ 8.5 Hz), 7.71 (s, 1H). ESI-MS: m/z [M þ H]þ 473.
(Acetone-d6,
d
): 1.65 (s, 3H), 1.78 (s, 3H), 3.35 (d, 2H, J ¼ 7.2 Hz),
5.28 (m, 1H), 6.56 (s, 1H), 6.60 (s, 1H), 6.99 (d, 1H, J ¼ 8.5 Hz), 7.44
(dd, 1H, J ¼ 2.0, 8.5 Hz), 7.48 (d, 1H, J ¼ 2.0 Hz), 8.98 (s, 1H, OH), 9.47
(s,1H, OH), 9.52 (s, 1H, OH), 13.28 (s, 1H, OH). 13C NMR (Acetone-d6,
d): 183.8, 165.2, 162.8, 161.3, 157.5 150.1, 147.4, 131.8, 124.6, 123.7,
120.0, 116.5, 114.3, 112.2, 105.4, 104.5, 94.3, 25.7, 22.1, 18.0. ESI-MS:
m/z [M þ H]þ 355. Anal. calcd. for C20H18O6: C, 67.79; H, 5.12; Found
C, 67.62; H, 4.97.
5.1.2.2. 20,40,5,7-Tetramethoxymethoxy-6-(3,3-dimethylallyl)-flava-
none (84b). The product was obtained as a pale yellow solid (55%).
mp.: 83e84 ꢂC. 1H-NMR (CDCl3,
d): 1.68 (s, 3H), 1.79 (s, 3H), 2.76
(dd, 1H, J ¼ 3.0, 16.5 Hz), 2.91 (dd, 1H, J ¼ 12.5, 16.5 Hz), 3.41 (d,
2H, J ¼ 7.0 Hz), 3.46 (s, 3H), 3.48 (s, 3H), 3.50 (s, 3H), 3.63 (s, 3H),
5.19 (s, 2H), 5.20 (s, 2H), 5.21 (s, 2H), 5.22 (m, 1H), 5.23 (s, 2H),
5.72 (dd, J ¼ 3.0, 12.5 Hz), 6.55 (s, 1H), 6.79 (dd, 1H, J ¼ 2.0, 8.5 Hz),
6.86 (d, 1H, J ¼ 2.0 Hz), 7.48 (d, 1H, J ¼ 8.5 Hz). ESI-MS: m/z
[M þ H]þ 533.
5.2. Pharmacological assay
5.2.1. PKB/Akt1 assay
In vitro kinase assays were carried out using HTScanÒ PKB/Akt1
Kinase Assay Kit (Cell Signaling Technology, Beverly, MA). Active
recombinant Akt1 kinase (GST-fusion protein, 4 ng) in 8
2.5 ꢃ kinase buffer [62.5 mM TriseHCl (pH 7.5), 25 mM MgCl2,
12.5 mM -glycerophosphate, 0.25 mM Na3VO4, 5 mM dithio-
threitol (DTT)], was mixed with 2 L of dimethyl sulfoxide (DMSO)
vehicle or each of the compound (indicated concentrations), incu-
bated at room temperature for 5 min and 10 L of ATP/substrate
cocktail (20 mM ATP, 3 mM eNOS served as substrate) was added.
After incubation at room temperature for 30 min, add 20 L of
mL of
5.1.2.3. 30,40,5,7-Tetramethoxymethoxy-6-(3,3-dimethylallyl)-flava-
none (84c). The product was obtained as a pale yellow syrup (45%).
b
m
1H NMR (CDCl3,
d
): 1.66 (s, 3H), 1.77 (s, 3H), 2.81 (dd, 1H, J ¼ 2.8,
17.2 Hz), 2.30 (dd, 1H, J ¼ 13.2, 17.2 Hz), 3.32 (d, 2H, J ¼ 6.8 Hz), 3.47
(s, 3H), 3.52 (s, 3H), 3.53 (s, 6H), 5.21 (m, 1H), 5.23 (s, 2H), 5.25 (s,
4H), 5.26 (s, 2H), 5.33 (dd, 1H, J ¼ 2.8 Hz, 13.2 Hz), 6.56 (s, 1H), 7.04
(dd, 1H, J ¼ 2.0, 8.0 Hz), 7.19 (d, 1H, J ¼ 8.0 Hz), 7.31 (d, 1H,
J ¼ 2.0 Hz). ESI-MS: m/z [M þ H]þ 533.
m
m
50 mM ethylenediaminetetraacetic acid (EDTA) (pH 8.0) and
terminate the reaction. Then, PKB/Akt1 kinase activity was
analyzed according to the manufacturer’s instructions.
5.1.3. General procedure for the synthesis of compounds 77e79
A stirred solution of corresponding 84 (0.3 mmol) and iodine
(0.3 mmol) in dry pyridine (4 mL) was heated to 90 ꢂC for 6 h. The
mixture was then poured into cold water, acidified to pH w 6 with
2 N HCl, then extracted with ethyl acetate. The organic phase was
washed with saturated sodium thiosulfate and water, successively.
Then the organic layer was washed with brine and dried over
anhydrous Na2SO4 and concentrated. The residue was dissolved in
methanol/THF (1/1, v/v, 5 mL) without purification, and 3N HCl
(0.5 mL) was added. The mixture was stirred at 50 ꢂC for 6 h. After
cooling to room temperature, the reaction mixture was poured into
cold water and extracted with ethyl acetate. The organic phase was
washed with brine, dried over anhydrous Na2SO4. After removal of
5.2.2. Cytotoxic activity assay
The cytotoxic activity of the tested compounds in PC3, OVCAR-8
and HL-60 cells was measured using the MTT (3-(4,5-Dimethylthia-
zol-2-yl)-2,5-diphenyltetrazolium bromide, a tetrazole) method.
Cells were seeded in 96-well microtiter plates (at a density of
4000 cells per well) for overnight attachment and exposed to each of
the compound (1.0 w 100.0
mL in RPIM 1640 medium; Sigma-Aldrich) was added (20.0
and plates were incubated for a further 4 h at 37 ꢂC. The purple for-
mazan crystals were dissolved in 100.0 L of DMSO. After 5 min, the
mM) for 72 h. The MTT solution (5.0 mg/
ml/well),
m
plates were read on an automated microplate spectrophotometer
(Bio-Tek Instruments, Winooski, VT) at 570 nm. Assays were

5958
X. Dong et al. / European Journal of Medicinal Chemistry 46 (2011) 5949e5958
[11] D. Mahadevan, G. Powis, E.A. Mash, B. George, V.M. Gokhale, S. Zhang,
K. Shakalya, L. Du-Cuny, M. Berggren, M.A. Ali, U. Jana, N. Ihle, S. Moses,
C. Franklin, S. Narayan, N. Shirahatti, E.J. Meuillet, Mol. Cancer Ther. 7 (2008)
2621e2632.
performed in triplicate on three independent experiments. The
concentration of drug inhibiting 50% of cells (IC50) was calculated
using the software of doseeeffect analysis with microcomputers.
[12] C.W. Lindsley, S.F. Barnett, M. Yaroschak, M.T. Bilodeau, M.E. Layton, Curr. Top.
Med. Chem. 7 (2007) 1349e1363.
[13] C.W. Lindsley, Z. Zhao, W.H. Leister, R.G. Robinson, S.F. Barnett, D. Defeo-Jones,
R.E. Jones, G.D. Hartman, J.R. Huff, H.E. Huber, M.E. Duggan, Bioorg. Med.
Chem. Lett. 15 (2005) 761e764.
[14] Q. Li, T. Li, G. Zhu, J. Gong, A. Claibone, C. Dalton, Y. Luo, E.F. Johnson, Y. Shi,
X. Liu, V. Klinghofer, J.L. Bauch, K.C. Marsh, J.J. Bousk, S. Arries, R. de Jong,
T. Oltersdorf, V.S. Stoll, C.G. Jakob, S.H. Rosenberg, V.L. Giranda, Bioorg. Med.
Chem. Lett. 16 (2006) 1679e1685.
[15] G. Zhu, J. Gong, A. Claiborne, K.W. Woods, V.B. Gandhi, S. Thomas, Y. Luo,
X. Liu, Y. Shi, R. Guan, S.R. Magnone, V. Klinghofer, E.F. Johnson, J. Bouska,
A. Shoemaker, A. Oleksijew, V.S. Stoll, R. de Jong, T. Oltersdorf, Q. Li,
S.H. Rosenberga, V.L. Giranda, Bioorg. Med. Chem. Lett. 16 (2006) 3150e3155.
[16] K.W. Woods, J.P. Fischer, A. Claiborne, T. Li, S.A. Thomas, G. Zhu, R.B. Diebold,
X. Liu, Y. Shi, V. Klinghofer, E.K. Han, R. Guan, S.R. Magnone, E.F. Johnson,
J.J. Bouska, A.M. Olson, R. de Jong, T. Oltersdorf, Y. Luo, S.H. Rosenberg,
V.L. Giranda, Q. Li, Bioorg. Med. Chem. Lett. 14 (2006) 6832e6848.
[17] G. Zhu, J. Gong, V.B. Gandhi, K. Woods, Y. Luo, X. Liu, R. Guan, V. Klinghofer,
E.F. Johnson, V.S. Stoll, M. Mamo, Q. Li, S.H. Rosenberg, V.L. Giranda, Bioorg.
Med. Chem. Lett. 15 (2007) 2441e2452.
[18] G. Zhu, V.B. Gandhi, R. de Jong, Y. Luo, X. Liu, Y. Shi, R. Guan, S.R. Magnone,
V. Klinghofer, E.F. Johnson, J. Bouska, A. Shoemaker, A. Oleksijew, K. Jarvis,
C. Park, R. de Jong, T. Oltersdorf, Q. Li, S.H. Rosenberg, V.L. Giranda., Bioorg.
Med. Chem. Lett. 16 (2006) 3424e3429.
[19] O. Dror, A. Shulman-Peleg, R. Nussinov, H.L. Wolfson, Curr. Med. Chem. 11
(2004) 71e90.
5.2.3. Apoptosis assay
To analysis apoptosis, OVCAR-8 cells (5 ꢃ 104 cells/ml, 5 mL)
were cultured in 25 cm2 flasks and treated with vehicle to test
compounds (20 mM). After 48 h of treatments, cells were harvested
and quickly washed twice with ice-cold PBS. Detection of apoptosis
by FACSCalibur flow cytometer (Becton Dickinson, Lincoln Park, NJ)
was performed using the Annexin V-FITC/Propidium iodide (PI)
apoptosis detection kit (BioVision, Mountain View, CA).
5.2.4. Western blotting assay
OVCAR-8 cells were harvested, washed with ice-cold
1 ꢃ phosphate-buffered saline (PBS), and lysed in immunoprecipi-
tation assay buffer [150 mM NaCl, 50 mMTris, 2 mM ethyleneglycol-
bis(b-aminoethylether), 2 mM EDTA, 25 mM NaF, 25 mM b-glycer-
ophosphate, 0.2% TritonX-100, 0.3% Nonidet P-40, and 0.1 mM
phenylmethylsulfonyl fluoride]. Cellular debris were pelleted by
centrifugation at 13,000 rpm for 30 min at 4 ꢂC. The concentrations
of the total lysate protein were measured using a standard Bradford
assay (Bio-Rad, San Diego, CA). For Western blot analysis, 40 w 80 mg
[20] S. Vadivelan, B.N. Sinha, G. Rambabu, K. Boppana, S.A. Jagarlapudi, J. Mol.
Graph. Model. 26 (2008) 935e946.
[21] A. Rinaldi, EMBO Rep. 9 (2008) 1073e1077.
of protein from the total cell lysate was electrophoresed by
SDSePAGE. The proteins were then transferred to nitrocellulose
membrane (Pierce Chemical) and probed with primary antibodies
followed by horseradish peroxidase-labeled secondary antibodies.
Proteins were visualized using enhanced chemiluminescence (Pierce
Chemical).
[22] L. Di, E.H. Kerns, Curr. Opin. Drug Discov. Devel. 8 (2005) 495e504.
[23] S.L. Dixon, Chem. Biol. Drug Des. 67 (2006) 370e372.
[24] T.A. Esbenshade, G.B. Fox, K.M. Krueger, T.R. Miller, C.H. Kang, L.I. Denny,
D.G. Witte, B.B. Yao, L. Pan, J. Wetter, K. Marsh, Y.L. Bennani, M.D. Cowart,
J.P. Sullivan, A.A. Hancock, J. Pharmacol. Exp. Ther. 313 (2005) 165e175.
[25] H.J. Boehm, M. Boehringer, D. Bur, H. Gmuender, W. Huber, W. Klaus,
D. Kostrewa, H. Kuehne, T. Luebbers, N. Meunier-Keller, F. Mueller, J. Med.
Chem. 43 (2000) 2664e2674.
Acknowledgments
[26] H. Yu, Z. Wang, L. Zhang, J. Zhang, Q. Huang, Bioorg. Med. Chem. Lett. 17
(2007) 2126e2133.
This study was financially supported by the China Postdoctoral
Science Foundation (NO. 20090460103), National Natural Science
Foundation of China (No. 201003734) and National Key Tech Project
for Major Creation of New drugs (NO. 2009ZX09501-003).
[27] G. Zhu, V.B. Gandhi, J. Gong, S. Thomas, K.W. Woods, X. Song, T. Li,
R.B. Diebold, Y. Luo, X. Liu, R. Guan, V. Klinghofer, E.F. Johnson, J. Bouska,
A. Olson, K.C. Marsh, V.S. Stoll, M. Mamo, J. Polakowski, T.J. Campbell,
R.L. Martin, G.A. Gintant, T.D. Penning, Q. Li, S.H. Rosenberg, V.L. Giranda,
J. Med. Chem. 50 (2007) 2990e3003.
[28] B.R. Brooks, R.E. Brucolleri, B.D. Olafson, D.J. States, S.S., M. Karplus, J. Comput.
Chem. 4 (1983) 187e214.
[29] A. Smellie, S.D. Kahn, S.L. Teig, J. Chem. Inf. Comp. Sci. 35 (1995) 285e294.
[30] A.K. Debnath, J. Med. Chem. 45 (2002) 41e53.
[31] W.M. Kazmierski, W. Andrews, E. Furfine, A. Spaltenstein, L. Wright, Bioorg.
Med. Chem. Lett. 14 (2004) 5689e5692.
[32] T. Fawcett, Pattern. Recogn. Lett. 27 (2006) 861e874.
[33] Sybyl6.9 (2002), Molecular Modelling System, Tripos Associates: St. Louis, MO,
USA.
[34] M. Rarey, B. Kramer, T. Lengauer, G. Klebe, J. Mol. Biol. 261 (1996) 470e489.
[35] C.M. Venkatachalam, X. Jiang, T. Oldfield, M. Waldman, J. Mol. Graphics Model.
21 (2003) 289e307.
[36] G. Wu, D.H. Robertson, C.L. Brooks 3rd, M. Vieth, J. Comput. Chem. 24 (2003)
1549e1562.
[37] S.N. Rao, M.S. Head, A. Kulkarni, J.M. LaLonde, J. Chem. Inf. Model. 47 (2007)
2159e2171.
[38] X. Dong, J. Chen, C. Jiang, T. Liu, Y. Hu, Arch. Pharm. 342 (2009) 428e432.
[39] X. Dong, L. Qi, C. Jiang, J. Chen, E. Wei, Y. Hu, Bioorg. Med. Chem. Lett. 19
(2009) 3196e3198.
References
[1] J. Brognard, A.S. Clark, Y. Ni, P.A. Dennis, Cancer Res. 61 (2001) 3986e3997.
[2] J.Q. Cheng, B. Ruggeri, W.M. Klein, G. Sonoda, D.A. Altomare, D.K. Watson,
J.R. Testa, Proc. Natl. Acad. Sci. U.S.A 93 (1996) 3636e3641.
[3] D. Haas-Kogan, N. Shalev, M. Wong, G. Mills, G. Yount, D. Stokoe, Curr. Biol. 8
(1998) 1195e1198.
[4] S.P. Staal, Proc. Natl. Acad. Sci. U.S.A 84 (1987) 5034e5037.
[5] E. Fayard, L. Tintignac, A. Baudry, B. Hemmings, J. Cell. Sci. 118 (2005)
5675e5678.
[6] S. Masure, B. Haefner, J.J. Wesselink, E. Hoefnagel, E. Mortier, P. Verhasselt,
A. Tuytelaars, R. Gordon, A. Richardson, Eur. J. Biochem. 265 (1999) 353e360.
[7] R.T. Joseph, B. Alfonso, Proc. Natl. Acad. Sci. U.S.A 98 (2001) 10983e10985.
[8] E. Oki, H. Baba, E. Tokunaga, T. Nakamura, N. Ueda, M. Futatsugi, K. Mashino,
M. Yamamoto, M. Ikebe, Y. Kakeji, Y. Maehara, Int. J. Cancer 117 (2005) 376e380.
[9] C. Sourbier, V. Lindner, H. Lang, A. Agouni, E. Schordan, S. Danilin, S. Rothhut,
D. Jacqmin, J. Helwig, T. Massfelder, Cancer Res. 66 (2006) 5130e5142.
[10] E. Tokunaga, A. Kataoka, Y. Kimura, E. Oki, K. Mashino, K. Nishida, T. Koga,
M. Morita, Y. Kakeji, H. Baba, S. Ohno, Y. Maehar, Eur. J. Cancer 42 (2006)
629e635.
[40] C.K. Chandra, M. Vincent, AKT crystal structure and AKT-specific inhibitors,
Oncogene 24 (2005) 7493e7501.