Novel 8-arylated purines as inhibitors of glycogen synthase kinase

Article abstract of DOI:10.1016/j.ejmech.2010.04.026

A series of 8-arylated purine derivatives bearing either an aniline or an alkyl amide at position 6 were found to inhibit glycogen synthase kinase-3,with good selectivity over ten kinases. Molecular modeling studies indicated that the most active compounds (8a and 8e),adopt a planar conformation,close to the shape of AMPPNP in the crystal structure of GSK-3. These compounds are stabilized by hydrophobic contacts between the 8-aromatic group and the protein adenine pocket and by electrostatic contacts.

Full text of DOI:10.1016/j.ejmech.2010.04.026

European Journal of Medicinal Chemistry 45 (2010) 3389e3393
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Novel 8-arylated purines as inhibitors of glycogen synthase kinase
,
Nada Ibrahim a, Liliane Mouawad b, Michel Legraverend a
*
a UMR 176 CNRS/Institut Curie, Bât. 110, Institut Curie, Centre Universitaire, 91405 Orsay, France
b U759 INSERM, Bât. 112, Institut Curie, Centre Universitaire, 91405 Orsay, France
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of 8-arylated purine derivatives bearing either an aniline or an alkyl amide at position 6 were
found to inhibit glycogen synthase kinase-3, with good selectivity over ten kinases. Molecular modeling
studies indicated that the most active compounds (8a and 8e), adopt a planar conformation, close to the
shape of AMPPNP in the crystal structure of GSK-3. These compounds are stabilized by hydrophobic
contacts between the 8-aromatic group and the protein adenine pocket and by electrostatic contacts.
Ó 2010 Elsevier Masson SAS. All rights reserved.
Received 18 February 2010
Received in revised form
14 April 2010
Accepted 21 April 2010
Available online 28 April 2010
Keywords:
8-Arylated purines
GSK-3 inhibitors
Molecular modeling
Docking
1. Introduction
pharmacological profile is particularly favorable for drugs aimed at
crossing the blood brain barrier [6]. In addition, most inhibitors of
Glycogen synthase kinase-3 (GSK-3) is a family of serine/thre-
onine protein kinases which are involved in a great number of
physiological events and human diseases, including neurodegen-
erative diseases, diabetes, inflammation and cancers [1].
In Alzheimer’s disease (AD), toxic aggregates are formed in the
brain such as extracellular amyloid deposits (accumulation of b-
protein kinases compete with ATP in the ATP binding site making
contacts with the hinge region of the kinase domain. However, at the
present time, the selectivityof most available GSK-3 inhibitors is poorly
known [7], as well as their potential as future drugs. It is therefore of
interest to continue the search for new inhibitors derived from various
scaffolds to study their ADMET properties and selectivity profile.
Mammalian cells have two glycogen synthase kinase isoforms:
amyloid peptides) and intracellular neurofibrillary tangles that
result from the misfolding of hyperphosphorylated tau protein. Tau
is a microtubule associated protein which promotes the assembly
and stabilization of microtubules in neurons [2]. It has been shown
to be phosphorylated in the normal adult brain, by several protein
kinases, including GSK-3, CDK5 and many others. However,
hyperphosphorylation of tau leads to disassembly of microtubules,
compromising neuronal and synaptic function [3].
In this context, the search for potent and selective inhibitors of GSK-
3 with drug-like properties remains a challenge. Most pharmacological
inhibitors of protein kinases and drugs share common structural
properties [4] such as low molecular weight (<600 Da), a limited
number of H-bond donor atoms (<5), a limited number of H-bond
acceptor atoms (<10), and a rather flat hydrophobic and hetero-
aromatic structure, in which heteroatoms are nitrogens [5]. This
a
and b whose kinase core (308 residues) share 97% sequence
similarity [8], but diverge at the N and C termini [9]. Inhibitors that
target the ATP site of GSK-3 are therefore unlikely to differentiate
the two isoforms [7].
In addition, numerous inhibitors of GSK-3 are in fact dual inhibi-
tors of CDKs and GSK-3 [7]. This may be explained by the sequence
identity and similarity of the ATP binding site in both kinases [10],
although the structural determinents for selectivity between protein
kinasesremainpoorlyunderstood[2]. However, thedifferenceofbulk
of the isobutyl side chain of Leu132 in GSK-3 and the phenyl ring side
chain of Phe80 in CDK-2 and CDK5, the gatekeeper residues, should
give an easier access to the hydrophobic pocket of GSK-3, adjacent to
the ATP pocket. This difference might be exploited to design selective
inhibitors of GSK-3, as already suggested [11,12].
Recently we have synthesized a library of 8-substituted purines,
that allowed us to identify several inhibitors of GSK-3, derived from
the purine scaffold. We report herein their synthesis, inhibitory
activity in vitro, as well as their mode of interaction with the
protein, resulting from molecular modeling.
Abbreviations: ADMET, adsorption, distribution, metabolism, excretion and
toxicity; CNS, central nervous system; CDK, cyclin-dependent kinase.
* Corresponding author. Tel.: þ33 1 69 86 30 85.
E-mail address: michel.legraverend@curie.u-psud.fr (M. Legraverend).
0223-5234/$ e see front matter Ó 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ejmech.2010.04.026

3390
N. Ibrahim et al. / European Journal of Medicinal Chemistry 45 (2010) 3389e3393
R1
Cl
Cl
Cl
X
N
NH2
Cl
NH2
NH2
N
N
N
N
N
N
i
ii
iv
R
R
N
H
N
H
N
N
N
1
2
3 a-f
24-26
a, c as for 3
3a: R = phenyl
24a: X = O; R1 = cyclopentyl
3b: R = (E)-styryl
3c: R = pyridin-3-yl
24c: X = O; R1 = cyclopentyl
25c: X = O; R1 = cyclohexylmethyl
26c: X = S; R1 = 4-N-phenylacetamide
3d: R = 3-Cl-phenyl
3e: R = 3-CN-phenyl
3f: R = 2-naphtyl
4b,c: NR1R2 = pyrrolidine
R1
N
R2
N
N
5a: R1 = R2 = methyl
6c: NR1R2 = morpholine
N
7b: R1 = H; R2 = ethylmorpholine
8a,b,e: R1 = H; R2 = phenyl
iii
R
3 a-f
N
H
9a: R1 = H; R2 = benzyl
10a,c: R1 = H; R2 = 4-methoxybenzyl
11a-c: R1 = H; R2 = pyridin-3-ylmethyl
12a: R1 = H; R2 = ethanol
4-23
a-f as for 3
13a,c: R1 = H; R2 = 4-N-phenethylmethylsulfonamide
14a-f: R1 = H; R2 = cyclopropanecarboxamide
15a: R1 = H; R2 = isopropanecarboxamide
16a: R1 = H; R2 = isobutanecarboxamide
17a,b: R1 = H; R2 = n-butanecarboxamide
17e,f: R1 = H; R2 = n-butanecarboxamide
18a-f: R1 = H; R2 = N-benzamide
19b-d: R1 = H; R2 = N-4-methoxybenzamide
20b: R1 = H; R2 = N-3-bromobenzamide
21a: R1 = H; R2 = N-pyrrolidine-2-carboxamide
22a: R1 = H; R2 = N-piperidine-2-carboxamide
23a,e,f: R1 = H; R2 = N-acetamide
Scheme 1. Reagents and conditions: (i) NH4OH 2h; (ii) RCOOH (1 eq), NH4Cl (6 eq) in POCl3 (4 ml/100 mg of 2), 100 ꢀC, 24 h; (iii) amine (1e2 eq), Et3N (1e2 eq), EtOH, 70 ꢀC, 2 h; or
aniline (1.4 eq), conc. HCl (50 l), isopropanol (2e3 ml), reflux, 2e3 h; or in an oven-dried Schlenk, under Ar, Pd2dba3 (3 mol%); Xantphos (15 mol%), amide (1.1 eq), K3PO4 (3 eq),
degazed dioxane (2e3 ml), 130 ꢀC, 8e24 h; (iv) alcohol or thiol (excess), Na (1e2 eq), 100 ꢀC, overnight.
m
Cl
N
Cl
N
Cl
N
Cl
N
NO2
Cl
NO2
NH
NH2
NH
N
N
N
N
N
N
i
ii
iii
R
CH3
27
28
30
29
a: R = H
b: R = F
R1
N
R2
N
N
31a: R1 = H, R2 = 4-methoxybenzyl
32a: R1 = H, R2 = 3-pyridylmethyl
33a: NR1R2 = pyrrolidine
N
iv
34b: R1 = H, R2 = benzyl
N
CH3
35b: R1 = H, R2 = 4-methoxybenzyl
36b: NR1R2 = pyrrolidine
R
37b: R1 = H, R2 = N-cyclopropanecarboxamide
31-38
a: R = H
38b: R1 = H, R2 = 4-N-phenethylmethylsulfonamide
b: R = F
Scheme 2. Reagents and conditions: (i) acetic acid/33% aqueous CH3NH2 (1/2 eq), THF, under Ar, 4 ꢀC, 40 min; (ii) aqueous HCl (3.3 eq)/EtOH, Iron powder (3 eq), reflux, 2.5 h; (iii)
ArCHO (2 eq) anhydrous dioxane, freshly prepared 15% FeCl3eSiO2 (2eq), under Ar; 100 ꢀC, 24 h; (iv) amine (1.5 eq), Et3N (1.5 eq), EtOH, 70 ꢀC, 12 h or in an oven dried Schlenk,
under Ar, Pd2dba3 (3 mol%); Xantphos (15 mol%), amide (1.1 eq), Cs2CO3 (3 eq), degazed dioxane (2e3 ml), 130 ꢀC, 8e24 h.
2. Results and discussion
2.1. Chemical synthesis of purine derivatives
Ph
N
Ph
N
The synthesis of compounds 4e26 was performed by the key
step cyclization of 5,6 diaminopyrimidine 2 with various aryl- or
alkenyl carboxylic acids, in POCl3 in the presence of NH4Cl as
previously described (Scheme 1) [13].
S
N
S
N
a: R = 4-Cl
H
N
N
N
b: R = 4-NO2
c: R = 4-OCH3
d: R = 3-F
i
I
R
39
40
The synthesis of 9-methylated-8-substituted purine derivatives
31e38 was achieved by cyclization of 5-amino-4-methyl-
aminopyrimidine derivative 29 with benzaldehyde or 3-fluo-
robenzaldehyde, in the presence of 15% FeCl3eSiO2 (Scheme 2). The
amidation step in both series (14e23 and 37) was catalyzed by
Pd2dba3 in the presence of xantphos [13].
The key step of the synthesis of 7-methylated-8-substituted
purines 41e45, was a one pot copper-catalyzed amidation-cycli-
zation of 5-methylamino-4-iodopyrimidine 39 in the presence of
a trans-diaminocyclohexane ligand as described (Scheme 3) [14].
Physical data and chemical characterization for all these
compounds have already been described in Refs. [13,14].
R1
N
R2
41a: R1 = H, R2 = 4-methoxybenzyl
41d: R1 = H, R2 = 4-methoxybenzyl
42b: R1 = H, R2 = benzyl
N
N
CH3
N
ii
R
43c: R1 = H, R2 = benzyl
N
44a: NR1R2 = pyrrolidine
45b: NR1R2 = pyrrolidine
41-45
Scheme 3. Reagents and conditions: (i) In an oven-dried Schlenk under Ar, amine (1.5
eq), Cs2CO3 (2 eq), CuI (0.1 eq), trans-N,N0-dimethylcyclohexane-1,2-diamine (0.4 eq),
degazed dioxane, 90 ꢀC, 7e8 h (ii) (a) m-chloroperbenzoic acid (3 eq) in dried CH2Cl2,
0
ꢀC, 8 h; (b) amine (2 eq) BuOH, 110 ꢀC, 4e48 h.

N. Ibrahim et al. / European Journal of Medicinal Chemistry 45 (2010) 3389e3393
3391
Table 1
Enzyme activity in the presence of the indicated compounds.
Compound
Percent enzyme activity
GSK-3 CDK5
Compound
Percent enzyme activity
GSK-3 CDK5
Compound
Percent enzyme activity
GSK-3 CDK5
4b
4c
5a
6c
7b
8a
8b
8e
87 ꢁ 2
101 ꢁ 1
100 ꢁ 1
85 ꢁ 1
14f
9(b) ꢁ 0
101 ꢁ 0.3
99 ꢁ 0
23f
49 ꢁ 1
98 ꢁ 1
76(b) ꢁ 1
97(b) ꢁ 1
94(b) ꢁ 0.5
99 ꢁ 1
15a
16a
17a
17b
17e
17f
18a
18b
18c
18d
18e
18f
19b
19c
19d
20b
21a
22a
23a
23e
69(b) ꢁ 0
82(b) ꢁ 1.3
63(b) ꢁ 1
55 ꢁ 1
24a
24c
25c
26c
31a
32a
33a
34b
35b
36b
37b
38b
41a
41d
42b
43c
44a
45b
Staur
82(b) ꢁ 0.2
57 ꢁ 2
54 ꢁ 1.8
83 ꢁ 2
102 ꢁ 0.5
103 ꢁ 0.7
103 ꢁ 1
101 ꢁ 1
102 ꢁ 1
102 ꢁ 1
104 ꢁ 1
102 ꢁ 2
102 ꢁ 1
102 ꢁ 2
100 ꢁ 1
104 ꢁ 1
107 ꢁ 1
102 ꢁ 1
101 ꢁ 2
93 ꢁ 1
71 ꢁ 0
74(b) ꢁ 2
101 ꢁ 1
112 ꢁ 1
100 ꢁ 0.7
99 ꢁ 0.3
100 ꢁ 1
96 ꢁ 0.8
96 ꢁ 1.7
100 ꢁ 0.5
99 ꢁ 0.7
97 ꢁ 1.3
105 ꢁ 1
99 ꢁ 0.3
101 ꢁ 0.8
102 ꢁ 0.1
97 ꢁ 2
101 ꢁ 2
63 ꢁ 1
101(b) ꢁ 1
101(b) ꢁ 10
104(b) ꢁ 1.4
100(b) ꢁ 1
101(b) ꢁ 2
101(b) ꢁ 0.9
98(b) ꢁ 0.5
100(b) ꢁ 0.2
102(b) ꢁ 0.4
102(b) ꢁ 0.3
101(b) ꢁ 1.3
102(b) ꢁ 0.6
101(b) ꢁ 2.5
100(b) ꢁ 0.3
103(b) ꢁ 2
99.9(b) ꢁ 0
5 ꢁ 1
34 ꢁ 1
58 ꢁ 1
100 ꢁ 1
63 ꢁ 1
31 ꢁ 0
2 ꢁ 1
67 ꢁ 1
9a
97(b) ꢁ 1.5
101(b) ꢁ 2
99(b) ꢁ 0.4
74(b) ꢁ 0
84 ꢁ 1
84 ꢁ 0.5
93 ꢁ 0.5
95 ꢁ 0.3
84 ꢁ 0.5
96 ꢁ 2
47 ꢁ 1
10a
10c
11a
11b
11c
12a
13a
13c
14a
14b
14d
14e
61 ꢁ 2
60 ꢁ 1
69 ꢁ 1
83 ꢁ 1
72(b) ꢁ 0
50 ꢁ 1
102 ꢁ 3
66 ꢁ 1
96 ꢁ 1
43 ꢁ 1
97(b) ꢁ 0
97(b) ꢁ 1
51(b) ꢁ 0.7
55 ꢁ 1
58 ꢁ 0
71 ꢁ 1
85 ꢁ 0.2
96 ꢁ 0.2
100 ꢁ 1
100 ꢁ 1
103 ꢁ 2
92 ꢁ 1
100(b) ꢁ 0.6
91(b) ꢁ 0.5
68(b) ꢁ 1
79 ꢁ 1
67 ꢁ 2
101 ꢁ 2
98.4 ꢁ 0
17 ꢁ 1
101 ꢁ 1
103 ꢁ 1
30 ꢁ 1
Enzyme activity is expressed in percent of enzyme activity relative to DMSO controls (in the absence of inhibitor). Human GSK-3 was used in the assays, in the presence of 1
33Pe]ATP, and [YRRAAVPPSP SLSRHSSPHQ(pS)EDEEE] (5 (b); CDK5 refer to CDK5/p25. Inhibitor concentration
M) as peptide substrate. GSK-3 refer to GSK-3 or GSK-3
was 1
M. Staur, Staurosporine. Values are means of two experiments. (ꢁSD).
mM
[g
ꢂ
m
a
b
m
Table 2
IC50s and physico-chemical properties of selected inhibitors.
Compd
8a
8e
14d
0.365
14e
0.781
14f
0.169
17e
1.22a
17f
Staur
0.005
IC50
0.016
0.026
>20
cLogP
MW
TPSA
3.89 ꢁ 0.84
3.51 ꢁ 0.91
3.51 ꢁ 0.85
2.33 ꢁ 0.87
3.95 ꢁ 0.84
3.56 ꢁ 0.86
5.17 ꢁ 0.84
4.40 ꢁ 1.42
287
312
279
304
329
320
345
466
66.4
90.2
83.5
107.3
83.5
107.3
83.5
69.4
IC50 values in
mM for compounds 8ae17f and Staurosporine (Staur) as a control were measured against GSK-3
b
in the presence of 10 mM ATP [23]. cLogPs (octanolewater
partition coefficient) were calculated with ACD/ChemSketch Freeware version 12.01 [25]. MW, molecular weight. TPSA (Topological polar surface area) were calculated as the
sum of surface contributions of polar atoms [26].
a
compound precipitation within reaction.
2.2. Enzymatic assays
are weak inhibitors of CDK5 (Table 1). Moravec and coworkers
already noticed that the presence of supplementary small substit-
uents (Cl, OH, SH, NH2, NO2) at position 8 of roscovitine like
compounds, decreased CDK inhibitory activity, whereas bulkier
substituents (propyloxy, bromo or 3-hydroxypropylamino)
cancelled their CDK inhibitory activity [15]. From this point of view,
Biological results are shown in Tables 1 and 2. Structures of GSK-
3 inhibitors are shown in Fig. 1.
The most potent inhibitors (8a, 8e 14d and 14f) in this assay are
shown on Fig. 1. It is interesting to note that these GSK-3 inhibitors,
O
NH
NH
N
N
N
N
N
H
N
H
N
N
R
R
8a: R = H
14d: R = Cl
14e: R = CN
8e: R = CN
O
N
O
N
O
N
NH
N
NH
NH
N
N
N
N
N
H
N
H
N
H
N
CN
14f
17e
17f
Fig. 1. Structures of GSK-3 inhibitors 8a, e, 14d, e, f, 17e, f.

3392
N. Ibrahim et al. / European Journal of Medicinal Chemistry 45 (2010) 3389e3393
Table 3
Fig. 2a and d), the three rings of the compounds are co-planar with
the adenine group of AMPPNP. The aromatic group in position 8 is
embedded in the adenine pocket, the purine core is located in the
AMPPNP sugar area and the substituent in position 6 occupies the
place of phosphates. In the second set (14d, 14e and 17e, Fig. 2b and
e), the purine core and its 8-aromatic group occupy the same
regions as for the first set, but the purine is rotated by 160ꢀ with
respect to the phenyl plane, therefore the compounds are not
planar. Consequently, the 6-substituent (amide) points to the
solvent such as the carbonyl group can establish a hydrogen bond
with Arg141. In the third set (14f and 17f, Fig. 2c and f), the 8-
naphtyl group forms a 60ꢀ angle with the AMPPNP adenine plane. It
occupies both the adenine and sugar pockets, whereas the purine
core occupies the phosphate location. A hydrogen bond is estab-
lished between the carbonyl group and Lys183 and the 6-substit-
uent is exposed to the solvent. Therefore, when the latter is long
and hydrophobic (as in compound 17f), this induces a lower affinity
to the protein.
What is remarkable is that the most active compounds (8a and
8e), which are also the only planar ones, adopt the closest shape to
that of AMPPNP. These compounds are stabilized by hydrophobic
contacts between the 8-aromatic group and the protein adenine
pocket on the one hand, and by two electrostatic contacts
between N3 of the purine and Arg141 and between NH9 and CO of
Ile62, on the other hand. These compounds are well embedded
into the cavity. This is not the case of the second set (14d, 14e and
17e), due to the rotation of the purine, which explains their lower
activity.
Percent activity of 10 protein kinases in the presence of 8a and 14f.
Kinase
% Enzyme activity (relative to
DMSO controls)
Compound IC50 (nM)
14f
91.41 ꢁ 1.20
8a
59.83 ꢁ 0.55
Staurosporine
CDK5/p25
<10
4188
3617
102.40
5089.00
5.88
35.03
3184.00
36.25
48.99
CK1
CK2
a
a
1
91.88 ꢁ 0.80
59.45 ꢁ 1.71
91.57 ꢁ 2.23
99.39 ꢁ 0.81
18.40 ꢁ 1.11
100.33 ꢁ 2.87
99.15 ꢁ 2.36
101.44 ꢁ 0.52
78.18 ꢁ 0.87
100.18 ꢁ 3.12
77.02 ꢁ 1.08
32.67 ꢁ 0.68
109.58 ꢁ 1.32
18.57 ꢁ 0.07
96.24 ꢁ 3.88
93.28 ꢁ 0.79
78.23 ꢁ 0.90
96.72 ꢁ 1.43
EGFR
ERK1
GSK-3
IGF1R
JNK1
b
MAPKAPK2
TYRO3/SKY
Percent enzyme activity were measured against the indicated protein kinases in the
presence of 10 M of compounds 14f and 8a and in the presence of 10 M ATP [23].
Mean of two experiments. IC50 values in nM for Staurosporine as a control. Enzyme
abbreviations: CDK5/p25, Cyclin-dependent kinase 5; CK1 1, Casein kinase I iso-
form alpha; CK2 , Casein kinase II subunit alpha; EGFR, Epidermal growth factor
receptor,; ERK1, Extracellular signal-regulated kinase 1, GSK-3 , Glycogen synthase
kinase-3 ; IGF1R, Insulin-like growth factor 1 receptor; JNK1, c-Jun N-terminal
m
m
a
a
b
b
kinase 1; MAPKAPK2, MAPK-activated protein kinase 2; TYRO3/SKY, Tyrosine-
protein kinase receptor3.
it is interesting to note the weak inhibition of CDK5 by 6c, 13a, 22a
and 24a, which carry a bulky substituent at position 8 (pyridyl or
phenyl), while being not inhibitors of GSK-3.
Compounds 8a and 14f have been tested at 10
against a panel of ten kinases (Table 3) Compound 14f displayed no
inhibitory activity against the enzymes tested but GSK-3 (81.6%
inhibition), withthe exception ofaweakactivityagainst casein kinase
(40.5% inhibition)), and Tyro3 receptor kinase (21.8% inhibition)).
Compound 8a, displayed 81.4% inhibition against GSK-3 , and 40.1%
inhibition against CDK5/p25. A weak inhibition was also observed
against EGFR (32.6%), CK2 (22.9%), and MAPKAPK2 (21.7%).
mM concentration
b
Among the most active compounds, 8a, 8e and 14f, the differ-
ence of activity may be explained by some difference between the
cavities of the two proteins and more specifically around the G-loop
2
a
b
which is located in the
, and strands 1 and 2 in CDK5. This loop is more structured in
GSK-3 than in CDK5 (Fig. 3a and b). Indeed, in the former, the 3rd
b-sheet A and relates strands 2 and 3 in GSK-
3b
a
b
strand of sheet A starts at Phe66 and its backbone establishes an H-
bond with the facing residue, Leu88, of the end of strand 4, whereas
in CDK5, the corresponding residues Tyr15 and Arg36, respectively,
are far from each other. In CDK5, Arg36, interacts with a neigh-
2.3. Molecular modeling studies
Molecular modeling studies [24] allowed us to cluster the active
compounds presented in Fig.1 in three different sets corresponding
to their pose into the protein. In the first set (compounds 8a and 8e,
boring negative cluster (Glu38 to Asp42) absent in GSK-3
b. In
addition, an H-bond relates Tyr15 and Glu51 (in helix 1).
Fig. 2. Structure of one representative compound from each set of poses. Superimposition on the AMPPNP in GSK-3
yellow and the AMPPNP is colored according to the atom type: cyan for carbon, blue for nitrogen, red for oxygen and tan for phosphorus. (def) The compounds are shown into the
cavity of GSK-3 , 8a (d), 14d (e) and 14f (f). They are colored according to the atom type. The surface of the cavity and the residues in stick are colored according to the residue type, i.
b of compounds 8a (a), 14d (b) and 14f (c). The inhibitors are in
b
e., white, lime, blue and red for hydrophobic, polar, positively and negatively charged residues, respectively (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article).

N. Ibrahim et al. / European Journal of Medicinal Chemistry 45 (2010) 3389e3393
3393
Fig. 3. Comparison between the cavities of GSK-3
(3 and 4 in GSK-3 and 2 and 3 in CDK5) of sheet A are in orange, the negatively charged residues are in blue, the loop BeC (that relates sheets B and C) is in green and the ligands (14f
docked in GSK-3 and Roscovitine in the crystal structure of CDK5) are in red. In (a) and (b) the key residues are drawn as sticks to show the structure of the G-loop. In (c) and (d), the
b (a,c) and CDK5 (b,d). The pdb codes of the proteins are 1J1B and 1UNL, respectively. The G-loop is inyellow, the two following strands
b
b
residues of this loop and loop BeC, as well as the ligands are in hard spheres to show the difference of closure of this part of the cavity. The figures were made with VMD [21], and the
cartoons were drawn using STRIDE [22] (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
Consequently, the G-loop of CDK5, which is not maintained in the
sheet, is bent over the cavity, and comes closer by about 3 Å to the
facing loop BeC. Therefore, there is not enough room in this cavity
to accommodate the 6-substituent (cyclopropyl group) of
compound 14f (Fig. 3c and d) whereas this is not the case for the
shorter compounds 8a and 8e which are good inhibitors of GSK-3
(weak inhibitors of CDK5, Table 1).
References
[1] P. Cohen, M. Goeder, Nat. Rev. Drug Discov. 3 (2004) 479e487.
[2] M.P. Mazanetz, P.M. Fischer, Nature 6 (2007) 464e479.
[3] K. Blennow, M.J. de Leon, H. Zetterberg, Alzheimer’s disease. Lancet 368
(2006) 387e403.
[4] C.A. Lipinski, F. Lombardo, B.W. Dominy, P.J. Feeney, Adv. Drug Deliv. Rev. 23
(1997) 3e25.
[5] P. Ertl, S. Jelfs, J. Mühlbacher, A. Schuffenhauer, P. Selzer, J. Med. Chem. 49
(2006) 4568e4573.
[6] L.K. Chico, L.J. Van Eldik, D.M. Watterson, Nat. Rev. DrugDiscov. 8 (2009) 892e909.
[7] L. Meijer, M. Flajolet, P. Greengard, Trends Pharmacol. Sci. 25 (2004) 471e480.
[8] E. ter Haar, J.T. Coll, D.A. Austen, H.-M. Hsiao, L. Swenson, J. Jain, Nat. Struct.
Mol. Biol. 8 (2001) 593e596.
3. Conclusion
[9] J.A. Bertrand, S. Thieffine, A. Vulpetti, C. Cristiani, B. Valsasina, S. Knapp, H.
M. Kalisz, M. Flocco, J. Mol. Biol. 333 (2003) 393e407.
[10] A. Vulpetti, P. Crivori, A. Cameron, J. Bertrand, M.G. Brasca, R. D’Alessio, P.
J. Pevarello, Chem. Inf. Model. 45 (2005) 1282e1290.
[11] P. Polychronopoulos, P. Magiatis, A.-L. Skaltsounis, V. Myrianthopoulos,
E. Mikros, A. Tarricone, A. Musacchio, S.M. Roe, L. Pearl, M. Leost, P. Greengard,
L. Meijer, J. Med. Chem. 47 (2004) 935e946.
The presence of a bulky hydrophobic substituent at position 8
together with an amide group at position 6 seems favorable to
increase the anti-GSK-3 activity vs CDK5. However, the presence of
a long hydrophobic chain at position 6 appeared unfavorable, being
oriented outside the protein.
[12] J.J.-L. Liao, J. Med. Chem. 50 (2007) 409e424.
Physico-chemical properties of CNS-penetrant small molecules
require a more restricted profile than the classical Lipinski “Rule of Five”
[6]. Thus most of the brain-penetrant small molecules have a molecular
weight <400 Da, cLogP < 4 and PSA (polar surface area) 80 Å2 [6,16], as
the inhibitors identified in this work (Fig.1, Table 2), and particularly for
compound 14f. In addition, these GSK-3 purine inhibitors do not inhibit
CDK5, contrary to known purine inhibitors of CDK such as Roscovitine
and analogs which do not inhibit GSK-3 [17e20].
[13] N. Ibrahim, M. Legraverend, J. Comb. Chem. 11 (2009) 658e666.
[14] N. Ibrahim, M. Legraverend, J. Org. Chem. 74 (2009) 463e465.
[15] J. Moravec, V. Krystof, J. Hanus, L. Havlicek, D. Moravcova, K. Fuksova, M. Kuzma,
R. Lenobel, M. Otyepka, M. Strnad, Bioorg. Med. Chem. Lett. 13 (2003) 2993e2996.
[16] P. Ertl, B. Rohde, P. Selzer, J. Med. Chem. 43 (2000) 3714e3717.
[17] L. Vandromme, M. Legraverend, S. Kreimerman, O. Lozach, L. Meijer, D.
S. Grierson, Bioorg. Med. Chem. 15 (2007) 130e141.
[18] K. Bettayeb, H. Sallam, Y. Ferandin, F. Popowycz, G. Fournet, M. Hassan, A. Echalier,
P. Bernard, J. Endicott, B. Joseph, L. Meijer, Mol. Cancer Ther. 7 (2008) 2713e2724.
[19] K. Bettayeb, N. Oumata, A. Echalier, Y. Ferandin, J. Endicott, H. Galons,
L. Meijer, Oncogene 27 (2008) 5797e5807.
These compounds may find interest as inhibitors of GSK-3, one of
the several protein kinases that hold tremendous promise as ther-
apeutic targets for CNS disorders such as an Alzheimer’s disease [6].
[20] N. Oumata, K. Bettayeb, Y. Ferandin, L. Demange, A. Lopez-Giral, M.-
L. Goddard, V. Myrianthopoulos, E. Mikros, M. Flajolet, P. Greengard, L. Meijer,
H. Galons, J. Med. Chem. 51 (2008) 5229e5242.
[21] W. Humphrey, A. Dalke, K. Shulten, J. Mol. Graph. 14 (1996) 33e38.
[22] D. Frishman, P. Argos, Protein Struct. Funct. Genet. 23 (1995) 566e579.
[23] The enzymatic assays were performed in Reaction Biology Corporation,
Malvern, PA.
[24] Docking experiments were carried out with Gold, v. 3.2.
[25] For LogP calculations, see. http://www.acdlabs.com/download/chemsketch/.
[26] For TPSA calculation, see. http://www.molinspiration.com/cgi-bin/properties
It is based on the methodology published by Ertl [16].
Acknowledgments
We gratefully acknowledge Institut Curie for financial support
(NI).